AI-Driven Predictive Maintenance for Industry 4.0: A Systematic Review of Models, Methods, and Challenges

AI-Driven Predictive Maintenance for Industry 4.0: A Systematic Review of Models, Methods, and Challenges | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article AI-Driven Predictive Maintenance for Industry 4.0: A Systematic Review of Models, Methods, and Challenges Abdelhafid Hamdi Alaoui, Anwar Meddaoui, Mustapha Hain This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7618987/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 18 Feb, 2026 Read the published version in The International Journal of Advanced Manufacturing Technology → Version 1 posted 4 You are reading this latest preprint version Abstract In the context of the transition to Industry 4.0, Predictive Maintenance (PdM) emerges as a key strategy to anticipate failures, reduce operational costs, and optimize the availability of industrial assets. This study presents a systematic review of recent works focused on approaches, methods, and challenges related to PdM, with particular emphasis on the integration of Artificial Intelligence (AI), the Internet of Things (IoT), and Big Data. A distinctive contribution of this research lies in the development of a taxonomy of maintenance strategies, tracing the evolution from corrective and preventive approaches to predictive and prescriptive paradigms, thereby providing a structured framework for positioning PdM within Industry 4.0. In addition, the review is guided by a set of research questions formulated to better capture the stakes and challenges associated with PdM implementation at both the technical and organizational levels. The analysis classifies scientific contributions based on prediction models (physics-based, knowledge-based, data-driven, and hybrid), evaluates machine learning algorithms (Random Forest, SVM, deep neural networks, transformers, etc.), and identifies the main technical and industrial limitations. The findings reveal that despite technological advances, significant obstacles persist in real-time deployment, model robustness, heterogeneous data management, and cybersecurity. The article also outlines promising perspectives for future research, with particular attention to prescriptive maintenance, digital twins, and explainable artificial intelligence (XAI). Predictive maintenance Artificial Intelligence XAI Machine Learning Remaining Useful Life Internet of Things Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 1. Introduction The rise of Industry 4.0 has significantly transformed production systems, operational management, and industrial maintenance strategies. This fourth industrial revolution, driven by the integration of advanced digital technologies such as the Internet of Things (IoT), Big Data, Artificial Intelligence (AI), and real-time connectivity requires companies to rethink their practices to meet increasing demands for performance, flexibility, and competitiveness (Heng et al. 2009) Among the critical functions in modern industry, maintenance holds a strategic position. Far from being limited to repairs or periodic interventions, it has evolved into a proactive and intelligent process. Predictive Maintenance (PdM), in particular, relies on sensor data, failure history, and operational conditions to anticipate breakdowns, estimate the Remaining Useful Life (RUL) of components, and optimize maintenance interventions (Susto et al. 2015) In recent years, there has been a surge of research focused on PdM, especially with the emergence of machine learning (ML) and deep learning (DL) techniques, which can model complex and nonlinear behaviors. Algorithms such as convolutional neural networks (CNN), recurrent neural networks (LSTMs), and more recently transformers, are increasingly applied to detect degradation patterns and predict failures from vibration, thermal, or acoustic signals (Zonta et al. 2020). Nevertheless, several technical, economic, and methodological challenges remain: managing massive and heterogeneous data, validating models in real-world industrial settings, the lack of standardization, and organizational resistance to change(Physics-informed machine learning | Nature Reviews Physics). In this context, this article aims to: Conduct a systematic review of major scientific contributions from 2018 to 2023; Classify PdM approaches according to their theoretical foundations (physical, knowledge-based, data-driven); Identify major technical bottlenecks and research perspectives related to intelligent predictive maintenance This critical and structured analysis aims to provide researchers, engineers, and decision-makers with a clear overview of current trends and key enablers for establishing PdM as a cornerstone of the factory of the future. 2. Theoretical Foundations and Terminology a. Types of Maintenance Industrial maintenance can be grouped by timing and approach. Preventive maintenance, which includes systematic maintenance (SM), condition-based maintenance (CBM), and predictive maintenance (PdM), aims to avoid failures, while corrective maintenance addresses issues after breakdowns occur. Corrective maintenance is triggered after a failure has occurred. It is divided into curative (complete failure) and troubleshooting (partial failure) (Smith and Hinchcliffe 2003). Systematic maintenance is based on fixed time or usage intervals, carried out before any observable degradation signs appear. Condition-Based Maintenance (CBM) monitors real-time performance indicators (e.g., temperature, vibration) without predictive modeling(Mobley 2002). Predictive Maintenance (PdM), on the other hand, anticipates failure by analyzing historical and real-time data using advanced techniques such as machine learning and physical modeling (Heng et al. 2009; Lei et al. 2018). a. Core and Advanced Concepts: Predictive Maintenance (PdM) is part of a broader paradigm known as Prognostics and Health Management (PHM), which aims to assess the health status of a system and predict its future failures (Heng et al. 2009). A central concept in PHM is the Remaining Useful Life (RUL), defined as the expected time remaining before a system or component reaches a failure threshold (Remaining useful life estimation -A review on the statistical data driven approaches 2011). PdM is a strategic discipline within industrial asset management that aims to anticipate equipment failures by analyzing historical and real-time data. As described in the figure 1, key distinction from Condition-Based Maintenance (CBM) lies in PdM’s ability to predict future degradation trends rather than solely reacting to current conditions. The modeling techniques used in PdM can be broadly categorized into four families: physics-based, knowledge-based, data-driven, and hybrid models. Physics-Based Models Physics-based models are grounded in the first principles of physics and engineering, such as mechanics, thermodynamics, and material science. These models simulate degradation mechanisms including fatigue, crack growth, corrosion, and wear by employing mathematical formulations that reflect real-world physical behavior. This category is central to the Physics-of-Failure (PoF) methodology, which is extensively applied in structural health monitoring and component reliability assessment (Prognostics and Health Management of Electronics). Such models require minimal historical data and offer high interpretability. However, they are limited by their dependence on domain-specific expertise and the complexity of accurately modeling real-world systems(Intelligent Fault Diagnosis and Prognosis for Engineering Systems | Wiley Online Books) Knowledge-Based Models Knowledge-based models leverage human expertise and logical reasoning to evaluate equipment condition. These models often rely on rule-based systems or expert systems consisting of a knowledge base and inference engine. Rules are defined explicitly, allowing the system to mimic expert decision-making (Chen and Chen 2011). These models are valuable when data are limited but expert knowledge is abundant. However, they may struggle with scalability and adaptability in high-dimensional or rapidly evolving environments(Jardine et al. 2006). Data-Driven Models Data-driven approaches utilize statistical learning and artificial intelligence particularly machine learning (ML) and deep learning (DL) to uncover degradation patterns and predict Remaining Useful Life (RUL). These models rely on large datasets from sensors, logs, and monitoring systems, and they are well-suited to environments aligned with Industry 4.0 paradigms (Susto et al. 2015). Data-driven models are capable of learning complex, non-linear relationships between multivariate inputs and output targets. Nonetheless, their performance hinges on the quality, quantity, and labeling of available data. Moreover, many of these models function as black boxes, making their predictions difficult to interpret in safety-critical systems (Zhang et al. 2019). Hybrid and Physics-Informed Models Hybrid models aim to combine the strengths of physics-based and data-driven approaches. A prominent class of these models is Physics-Informed Machine Learning (PIML), which integrates physical laws directly into machine learning architectures, such as neural networks or Gaussian processes. These methods enable learning under physical constraints, thereby enhancing model robustness, generalizability, and interpretability(Willard et al. 2020). Hybrid models are particularly effective when datasets are sparse or noisy and when model transparency is essential. However, they are technically complex to design and deploy, requiring expertise in both physical modeling and AI (Zonta et al. 2020; Physics-informed machine learning | Nature Reviews Physics). b. Maintenance Evolution in the Industry 4.0 Paradigm With the rise of Industry 4.0, maintenance has embarked on a digital transformation journey. This evolution can be described using the maintenance 1.0 to maintenance 4.0 framework (see table 1). Transitioning to Maintenance 4.0 involves integrating smart sensors, cloud-based platforms, self-learning algorithms, and interactive visualization interfaces (Pech et al., 2021). These technologies transform data into actionable decisions, enabling connected, dynamic, and cost-effective maintenance strategies. Table 1. Maintenance evolution Generation Main focus Enabling Technology Maintenance 1.0 Corrective None Maintenance 2.0 Preventive Timers, visual inspection Maintenance 3.0 Condition-Based Sensors, SCADA, PLCs Maintenance 4.0 Predictive & Prescriptive AI, IoT, Cloud, Digital Twins 3. Related work In recent years, predictive maintenance (PdM) has emerged as a critical enabler of intelligent asset management within Industry 4.0 environments. The proliferation of sensor networks, the Internet of Things (IoT), and artificial intelligence (AI) techniques has significantly advanced PdM capabilities beyond traditional condition-based maintenance approaches. A comprehensive and widely cited systematic literature review by (Zonta et al. 2020 ) provides a structured mapping of the scientific landscape surrounding PdM in the context of Industry 4.0. Their study, which analyzed over 100 articles, identifies the main technological pillars of PdM as cyber-physical systems (CPS), big data analytics, machine learning, and cloud computing. The authors emphasize that AI particularly machine learning (ML) plays a pivotal role in enabling automated fault diagnosis and Remaining Useful Life (RUL) estimation through the analysis of multivariate time series data. Furthermore, Zonta et al. categorize PdM solutions into three main architectural layers: data acquisition, data processing, and decision-making. In their review, it is noted that a majority of PdM systems remain focused on the operational layer, with limited integration between data-driven insights and prescriptive maintenance actions. The authors highlight several key research challenges, including data quality issues, model generalizability across equipment types, lack of standardization in architectures, and the need for explainable AI to enhance trust and transparency in industrial settings. This review also stresses the importance of combining PdM with cybersecurity, particularly in distributed IoT environments, and calls for further research in integrating edge computing and digital twins to support real-time, scalable, and interpretable maintenance strategies. Their findings form a foundational reference point for subsequent research efforts aimed at advancing PdM solutions through hybrid modeling approaches and AI integration. 4. Matériel & Method This section outlines the systematic review methodology adopted to identify, select, and synthesize peer-reviewed literature on predictive maintenance (PdM) in the context of Industry 4.0. The approach was structured according to the PRISMA framework (Moher et al. 2009), focusing on targeted and reproducible search, selection, and filtering processes within four major academic databases. a. Research Questions The review was guided by structured research questions designed to explore technological, methodological, and industrial aspects of PdM. These questions, summarized in Table 2, served as a thematic framework for selecting and analyzing the literature. Table 2. Research Questions Research Questions Research Question RQ 1 What are the main channels and trends in the dissemination of research related to Predictive Maintenance (PdM) in Industry 4.0? RQ2 What are the dominant approaches used in predictive maintenance in Industry 4.0? RQ3 What types of AI algorithms are most frequently used in PdM? RQ4 How do different models and architectures in PdM target specific industrial applications, and which variables are most critical in these applications? b. Research Strategy The search strategy adopted in this review follows a two-step process inspired by Zonta et al. (2020), combining an initial exploratory search and a refined database-specific search. We initially selected Google Scholar as a starting point because it allows free-text queries and returns a high number of potentially relevant publications by searching across full texts, titles, and abstracts. This exploratory phase enabled us to identify relevant terminology, refine our search string, and detect early patterns of duplication across databases. However, since many of the documents retrieved via Google Scholar are already indexed in specialized scientific databases such as IEEE Xplore, Elsevier, and Springer-Link , this research was limited to the systematic selection of articles to these three scientific publishers, which provide peer-reviewed content with traceable metadata and reliable indexing. The initial research phase involved defining the scope and objectives of the review (figure 2), focusing on identifying and analyzing scientific contributions related to : Predictive maintenance methods in the context of Industry 4.0 Applications of artificial intelligence, machine learning, and deep learning in PdM Real-time implementation challenges and RUL estimation models The literature search targeted peer-reviewed publications from the beginning of 2019 to the end of 2024, ensuring the inclusion of the most recent and relevant studies. The following databases were selected due to their relevance in engineering and computer science: IEEE Xplore Science-Direct (Elsevier) Springer-Link Google Scholar (for complementary coverage) Search Terms and Query Design The search queries were constructed using combinations of keywords and Boolean operators. The final query string was adapted to each database’s syntax and included terms such as: This query aimed to ensure the inclusion of articles addressing both methodological aspects and industrial applications of PdM, particularly those involving condition monitoring systems, prognostics, and AI-based modeling. Inclusion and Exclusion Criteria To ensure quality and consistency in the selected studies, the following inclusion criteria were applied (see table 3) : Articles published in English Peer-reviewed journal and conference papers Publications between 2019 and 2024 Explicit focus on predictive maintenance and AI-based diagnostics Inclusion of practical use cases, algorithms, or architectures Table 3. Quality Filtering Criteria for Industry 4.0 and Predictive Maintenance Publications Criteria Type Included Excluded if... 1 Publication Period Published between 2019 and 2024 Published before 2019 2 Language Article is written in English Article is written in another language 3 Publication Type Peer-reviewed journal or conference paper Patent, thesis, dissertation, book chapter, or non-peer-reviewed material 4 Content Quality Includes technical contributions, such as algorithms, architectures, or case studies Lacks experimental methods, clear results, or methodological rigor 5 Content Quality Presenting a clear abstract, full text, and a detailed methodology. Publications without abstract, full text, or clear methodology. 6 Focus Area Dedicated to predictive maintenance and AI-based diagnostics in Industry 4.0. Focused only on preventive or corrective maintenance The exclusion criteria eliminated: Patents, theses, and dissertations Articles focused solely on preventive or corrective maintenance Studies lacking technical or experimental contributions Publications without abstract, full text, or clear methodology Selection and Filtering Process As described in the figure 3, the selection process consisted of four stages: Initial retrieval of publications based on keyword queries (approx. 3,907 results); Duplicate removal using citation management software; Abstract and title screening to assess relevance; Full-text review to apply inclusion/exclusion criteria. 5. Results and discussions In this section, we present the results and discussion based on the research questions previously defined, with the objective of answering the main questions guiding this systematic review RQ1 – What are the main channels and trends in the dissemination of research related to Predictive Maintenance (PdM) in Industry 4.0? The dissemination of research on Predictive Maintenance (PdM) within Industry 4.0 is primarily channeled through peer-reviewed journals, with a significantly smaller proportion presented at conferences. As indicated in Figure 5, journal articles account for 86 % of the publications, while conference proceedings constitute only 14%. This distribution suggests that PdM has evolved into a mature research area, where detailed methodological studies and rigorous validation are preferred over preliminary conference communications. Regarding publishers, figure 4 and table 4 highlight Elsevier’s dominant role, with 42 publications, far surpassing Springer (9 publications), IEEE (10 publications), and other publishers (3 publication). This concentration reflects Elsevier’s strong presence in fields such as industrial engineering, computer science, and reliability engineering, which are closely aligned with PdM research. The temporal analysis shown in Figure 6 reveals a clear upward trend in publications from 2019 to 2024, with a particularly notable increase from 5 publications in 2019 to 18 publications in 2024. This growth reflects both the technological advances in AI and Industry 4.0, and the heightened industrial interest in deploying PdM solutions for cost reduction and operational efficiency. Overall, the data suggest that scientific journals, particularly those published by Elsevier, serve as the principal dissemination channels for PdM research, supporting the development and standardization of methodologies. Conferences, while less prevalent, still play a crucial role in introducing novel concepts and fostering collaboration within the PdM community. RQ2–What are the dominant approaches used in predictive maintenance in Industry 4.0? Predictive maintenance in Industry 4.0 has experienced significant methodological evolution in recent years. As reported by Zhang et al. (2021), data-driven approaches, particularly those leveraging deep learning, have become increasingly prominent due to their capacity to identify complex patterns in large-scale sensor data. This is especially true in the context of remaining useful life (RUL) prediction, where architectures like BiGRU and CNN-LSTM have shown strong performance. As summarized in Table 5, the majority of reviewed studies adopt data-driven strategies, reflecting their widespread applicability across industrial domains. Table 4. Selected Articles sorted by year No. Article Type Year Publisher Name 1 Franciosi et al. Journal 2021 ELSEVIER 17th IFAC Symposium on Information Control Problems in Manufacturing INCOM 2021 2 Kans et al. Journal 2020 ELSEVIER 4th IFAC Workshop on Advanced Maintenance Engineering, Services and Technologies - AMEST 2020 3 Crespo Márquez et al. Journal 2023 ELSEVIER Computers in Industry 4 Ma et al. Journal 2023 ELSEVIER Mechanical Systems and Signal Processing 5 Chowdhury et al. Journal 2022 ELSEVIER Computers & Industrial Engineering 6 Nikolakis et al. Journal 2020 ELSEVIER 30th International Conference on Flexible Automation and Intelligent Manufacturing (FAIM2021) 7 You et al. Journal 2022 ELSEVIER 3rd International Conference on Industry 4.0 and Smart Manufacturing 8 Ragab et al. Journal 2021 ELSEVIER Neurocomputing 9 Calabrese et al. Journal 2019 ELSEVIER 25th International Conference on Production Research Manufacturing Innovation: Cyber Physical Manufacturing August 9-14, 2019 | Chicago, Illinois (USA) 10 Traini et al. Journal 2019 ELSEVIER 9th IFAC Conference on Manufacturing Modelling, Management and Control MIM 2019 11 Lv et al. Journal 2023 ELSEVIER Advanced Engineering Informatics 12 Joseph et al. Journal 2022 ELSEVIER 5th IFAC Workshop on Advanced Maintenance Engineering, Services and Technologies AMEST 2022 13 Taşcı et al. Journal 2023 ELSEVIER Computers & Industrial Engineering 14 Vargas et al. Journal 2023 ELSEVIER Engineering Applications of Artificial Intelligence 15 Gupta et al. Journal 2023 ELSEVIER Computers & Industrial Engineering 16 Tessoni et al. Journal 2022 ELSEVIER 3rd International Conference on Industry 4.0 and Smart Manufacturing 17 Daniyan et al. Journal 2020 ELSEVIER Learning Factories across the value chain – from innovation to service – The 10th Conference on Learning Factories 2020 18 Uhlmann et al. Journal 2021 ELSEVIER 54th CIRP CMS 2021 - Towards Digitalized Manufacturing 4.0 19 Rihi et al. Journal 2022 ELSEVIER Knowledge-Based and Intelligent Information & Engineering Systems: Proceedings of the 26th International Conference KES2022 20 Nentwich et al. Journal 2021 ELSEVIER 54th CIRP CMS 2021 - Towards Digitalized Manufacturing 4.0 21 Bilal Yıldız et al. Journal 2023 ELSEVIER Advanced Engineering Informatics 22 Lee et al. Journal 2019 ELSEVIER 26th CIRP Conference on Life Cycle Engineering (LCE) Purdue University, West Lafayette, IN, USA May 7-9, 2019 23 Killeen et al. Journal 2019 ELSEVIER The 10th International Conference on Ambient Systems, Networks and Technologies (ANT 2019) 24 Luo et al. Journal 2020 ELSEVIER Robotics and Computer-Integrated Manufacturing 25 Feng et al. Journal 2023 ELSEVIER Journal of Manufacturing Systems 26 Cao et al. Journal 2022 ELSEVIER Robotics and Computer-Integrated Manufacturing 27 Coelho et al. Journal 2022 ELSEVIER 3rd International Conference on Industry 4.0 and Smart Manufacturing 28 Chinta et al. Journal 2023 ELSEVIER Advanced Engineering Informatics 29 Jiang et al. Journal 2022 ELSEVIER Computers & Industrial Engineering 30 Farahani et al. Journal 2022 ELSEVIER Journal of Manufacturing Processes 31 Drakaki et al. Journal 2021 ELSEVIER Proceedings of the 2nd International Conference on Industry 4.0 and Smart Manufacturing (ISM 2020) 32 Jain et al. Journal 2020 ELSEVIER 4th IFAC Workshop on Advanced Maintenance Engineering, Services and Technologies - AMEST 2020 33 Huynh et al. Journal 2022 ELSEVIER Reliability Engineering & System Safety 34 Zonta et al. Journal 2022 ELSEVIER Journal of Manufacturing Systems 35 Wen et al. Journal 2022 ELSEVIER Measurement 36 Einabadi et al. Journal 2019 ELSEVIER 9th IFAC Conference on Manufacturing Modelling, Management and Control MIM 2019 37 Shaheen et al. Journal 2023 ELSEVIER Engineering Applications of Artificial Intelligence 38 Hillenbrand et al. Journal 2021 IOP Publishing IOP Conference Series: Materials Science and Engineering 39 Augustyn et al. Journal 2023 ELSEVIER Acta Mechanica et Automatica 40 Lv et al. Journal 2023 ELSEVIER IET Conference Proceedings 41 Rajasekar et al. Conference 2023 IEEE 2023 15th International Conference on Developments in eSystems Engineering (DeSE) 42 Singha et al. Conference 2020 IEEE 2020 IEEE 17th India Council International Conference (INDICON) 43 Kotsiopoulos et al. Journal 2021 Springer The International Journal of Advanced Manufacturing Technology 44 Fordal et al. Journal 2023 Springer Advances in Manufacturing 45 Vicêncio et al. Conference 2021 Springer Industrial IoT Technologies and Applications 46 Wang et al. Journal 2024 Elsevier Reliability Engineering & System Safety 47 Herzog et al. Conference 2024 IEEE 28th International Conference on Methods and Models in Automation and Robotics (MMAR) 48 Kolvig-Raun et al. Journal 2024 IEEE IEEE Robotics and Automation Letters 49 Mardianto et al. Conference 2024 IEEE 8th Int. Conf. on Information Technology, Information Systems and Electrical Engineering (ICITISEE) 50 Patel et al. Conference 2024 IEEE/ACM 17th International Conference on Utility and Cloud Computing (UCC) 51 Gougam et al. Journal 2024 Springer Journal of the Brazilian Society of Mechanical Sciences and Engineering 52 Kondo et al. Journal 2024 Elsevier Computers & Industrial Engineering 53 Meddaoui et al. Journal 2023 Springer The International Journal of Advanced Manufacturing Technology 54 Meddaoui et al. Journal 2024 Springer The International Journal of Advanced Manufacturing Technology 55 Yuhua Yin, et al. Journal 2024 Springer Chinese Journal of Mechanical Engineering 56 Javad Isavand, et al. Journal 2024 Elsevier Measurement 57 Anouar et al. Conférence 2024 IEEE 21st Int. Multi-Conference on Systems, Signals & Devices (SSD) 58 Ugur Ileri, et al. Journal 2024 MDPI Applied Sciences 59 Rengaraj R, et al. Conférence 2024 IEEE 2nd Int. Conference on Networking and Communications (ICNWC) 60 Haobin Wen, et al. Journal 2024 MDPI Applied Sciences 61 Zekai Si, et al. Journal 2024 Springer Arabian Journal for Science and Engineering 62 Patrick Seebold, et al. Conférence 2024 IEEE 3rd Int. Conference on Computing and Machine Intelligence (ICMI) 63 Huan Wang et al. Journal 2024 IEEE IEEE Internet of Things Journal 64 Pooja KamatN et al. Journal 2024 Springer Journal of the Brazilian Society of Mechanical Sciences and Engineering However, Wang et al. (2021) noted that purely data-driven methods often face challenges related to interpretability and robustness in practical settings. To address these limitations, hybrid approaches have emerged, combining physical modeling with machine learning techniques. Notable examples include the hybrid digital twin frameworks proposed by Yu et al. (2022), which integrate physical simulations with data-driven analysis to enhance both accuracy and model transparency. As shown in Table 5, several studies, such as those by Calabrese et al. (2019) and Zonta et al. (2022), reflect this growing trend towards hybrid solutions. In specific contexts, physical model-based methods remain relevant, particularly in industries requiring high levels of interpretability and compliance with engineering standards. For instance, Ouafi et al. (2020), listed in Table 5, utilized gamma process models for multi-component systems, emphasizing the value of mathematical rigor and domain knowledge. Overall, the current literature compiled indicates that while data-driven techniques dominate research in predictive maintenance for Industry 4.0, hybrid methods are increasingly seen as essential for bridging the gap between predictive performance and industrial feasibility. Meanwhile, physical and statistical models continue to hold value in domains where transparency and explainability are crucial. Table 5. Articles by approach used Prediction Classification Articles Data-Driven (Traini et al. 2019; Einabadi et al. 2019; Daniyan et al. 2020; Nikolakis et al. 2020; Singha et al. 2020; She and Jia 2021a; Kotsiopoulos et al. 2021; Hillenbrand et al. 2021; Ragab et al. 2021; Joseph et al. 2022a; Wen et al. 2022; Rihi et al. 2022; Tessoni and Amoretti 2022; Chowdhury et al. 2022; Jiang et al. 2022; Jiang et al. 2022; Gupta et al. 2023a; Shaheen et al. 2023; Lv et al. 2023; Fordal et al. 2023; Chinta et al. 2023; Augustyn and Fidali 2023; Meddaoui et al. 2023; Ileri et al. 2024; Isavand et al. 2024; Gougam et al. 2024; BOUROKBA et al. 2024; Kolvig-Raun et al. 2024; Kolvig-Raun et al. 2024; Seebold et al. 2024; Kamat et al. 2024; Meddaoui et al. 2024; L. Wang et al. 2024; Herzog and Bartecki 2024; Mardianto and Mauritsius 2024; Patel et al. 2024) Physical model -based Huynh et al. (Huynh et al. 2022) Hybrid (Calabrese et al. 2019; Einabadi et al. 2019; Killeen et al. 2019; Luo et al. 2020; Jain et al. 2020; Kans et al. 2020; Sfar et al. 2020; Drakaki et al. 2021; Franciosi et al. 2021; Nentwich and Reinhart 2021; Uhlmann et al. 2021; Theissler et al. 2021; Coelho et al. 2022; You et al. 2022; Zonta et al. 2022; Cao et al. 2022; Farahani et al. 2022; Vargas et al. 2023; Polenghi et al. 2023; Bilal Yıldız and Soylu 2023; Crespo Márquez et al. 2023; Feng et al. 2023; Ma et al. 2023; Wen et al. 2024; Yin et al. 2024; R et al. 2024; H. Wang et al. 2024; H. Wang et al. 2024; Kondo et al. 2024; Si et al. 2024) RQ3 – What types of AI algorithms are most frequently used in PdM? An analysis of the selected research articles shows that a wide spectrum of artificial intelligence (AI) algorithms is employed in predictive maintenance (PdM) for Industry 4.0 applications. As summarized in Table 6 , the dominant approaches can be broadly grouped into deep learning (DL), classical machine learning (ML), artificial neural networks (ANN), hybrid combinations, and architecture-level solutions integrating IoT, digital twins (DT), and cloud computing. Deep learning techniques have emerged as the most frequently adopted methods in PdM studies. Authors such as (She and Jia 2021b), (Kotsiopoulos et al. 2021), and (Zonta et al. 2022) have demonstrated the effectiveness of DL architectures including BiGRU, CNN, and LSTM in capturing complex time dependencies and multi-dimensional patterns in sensor data. These models are particularly prominent in remaining useful life (RUL) prediction tasks, reflecting their strength in time-series modeling. Alongside DL, classical machine learning algorithms remain extensively used, especially for classification tasks and anomaly detection. Techniques such as Random Forest (RF), support vector machines (SVM), and ensemble methods are frequently cited, as seen in studies by Traini et al. (2019), Lee et al. (2019), and Gupta et al. (2023). These methods are valued for their lower computational demands and greater interpretability compared to deep learning models. Table 6. AI and algorithms used for PdM Algorithmes Identifiers DEEP LEARNING (She and Jia 2021b; Kotsiopoulos et al. 2021; Ragab et al. 2021; Joseph et al. 2022b; Zonta et al. 2022; Chowdhury et al. 2022; Jiang et al. 2022; Chinta et al. 2023; Ileri et al. 2024; Ileri et al. 2024; BOUROKBA et al. 2024; Seebold et al. 2024; Kamat et al. 2024; H. Wang et al. 2024; Mardianto and Mauritsius 2024; Si et al. 2024) ANN (Einabadi et al. 2019; Daniyan et al. 2020; Shaheen et al. 2023; Lv et al. 2023; Meddaoui et al. 2023) ML (RF…) (Traini et al. 2019; Lee et al. 2019; Jain et al. 2020; Nikolakis et al. 2020; Singha et al. 2020; Hillenbrand et al. 2021; Theissler et al. 2021; Gupta et al. 2023b; Augustyn and Fidali 2023; Taşcı et al. 2023; Isavand et al. 2024; R et al. 2024; Herzog and Bartecki 2024) Statistical +ML +DL (Sfar et al. 2020; Wen et al. 2022; Coelho et al. 2022; Tessoni and Amoretti 2022; Huynh et al. 2022; Wen et al. 2024; Wen et al. 2024; Yin et al. 2024; Gougam et al. 2024; Kolvig-Raun et al. 2024; Meddaoui et al. 2024) DL +ML Rihi et al (L. Wang et al. 2024; Patel et al. 2024) IoT Architecte + DT + Cloud computing (Killeen et al. 2019; Luo et al. 2020; Farahani et al. 2022; Feng et al. 2023; Fordal et al. 2023; Kondo et al. 2024) Artificial neural networks (ANN), although sometimes grouped under ML, are reported separately in several studies due to their distinct architecture. Einabadi et al. (2019), Daniyan et al. (2020), and Shaheen et al. (2023) utilized ANN-based models for fault diagnosis and health state classification, highlighting their simplicity and efficiency for smaller datasets. A notable trend in the literature is the development of hybrid models that combine statistical techniques, ML, and DL. Works by Sfar et al. (2020), Wen et al. (2022), and Huynh et al. (2022) exemplify this approach, aiming to leverage the strengths of different methods to improve prediction accuracy and model robustness. Hybrid methods are particularly relevant for bridging the gap between purely data-driven approaches and the need for physical interpretability. Beyond pure algorithmic strategies, some studies propose architecture-level solutions integrating IoT, digital twins (DT), and cloud computing for predictive maintenance. As shown in Table 6 , Killeen et al. (2019), Luo et al. (2020), and Farahani et al. (2022) explored frameworks that combine advanced sensing networks, digital replicas of assets, and scalable cloud platforms. Such architectures are designed to enable real-time analytics, remote monitoring, and scalable deployment across industrial systems. Overall, the data consolidated in Table 6 confirms that while deep learning dominates current PdM research, machine learning, ANN, and hybrid solutions continue to play essential roles. The choice of algorithm often depends on the specific industrial context, data availability, and the trade-off between predictive accuracy and model interpretability. RQ4_How do different models and architectures in PdM target specific industrial applications, and which variables are most critical in these applications? An analysis of the selected research articles shows that PdM models and architectures are strongly shaped by specific industrial applications and the variables critical to each context. As summarized in Table 7, manufacturing studies often use deep learning methods like BiGRU or CNN for tool wear and machine health prediction, relying on variables such as vibration, cutting forces, and spindle current (e.g., Zhang et al., 2021; Fordal et al., 2023). In the automotive sector, machine learning models analyze fleet data, focusing on variables like temperature, vibration, and diagnostic codes (Theissler et al., 2021). Mining applications require robust models like CNNs due to harsh conditions, using variables such as vibration and motor currents (Rojas et al., 2022). Hybrid approaches, as seen in Yu et al. (2022), combine physical models and data-driven analytics to enhance prediction accuracy and explainability, leveraging both physical parameters and statistical features. Architectural solutions integrating IoT, digital twins (DT), and cloud computing enable broader data collection and real-time analysis, adapting to varied industrial needs (Killeen et al., 2019; Luo et al., 2020). Overall, variable selection in PdM reflects both the physical characteristics of machinery and the operational environment, confirming that effective PdM requires models tailored to specific industrial contexts. Table 7. models, application and architecture presentation Authors Title Year Case Variable Franciosi et al. A maintenance scheduling optimization model for a multi-component machine in a digitalized manufacturing context 2021 Critical Multi-component Welding Machine in manufacturing Downtime periods, Component health states, Failure predictions, Maintenance costs Kans et al. A remote laboratory for Maintenance 4.0 training and education 2020 Machinery Fault Simulator (Rotating Machinery) for training and education Vibration signals, simulated fault data (unbalance, misalignment, resonance, bearing faults, cracked shaft, gearbox faults) Crespo Márquez et al. Simulating dynamic RUL based CBM scheduling. A case study in the railway sector 2023 "Railway Sector - Fleet of Trains (Rolling Stock) "Detected anomalies, Estimated RUL, Risk levels, Workshop CBM capacity, Stock and flow variables: Ma et al. A hybrid-driven probabilistic state space model for tool wear monitoring 2023 CNC Milling of stainless steel using ball nose cutter Cutting forces (Fx, Fy, Fz), Acoustic Emission (AE), Vibration (Vx, Vy, Vz) Chowdhury et al. Internet of Things resource monitoring through proactive fault prediction 2022 IoT Resources (Multiple domains: AWS IoT, Network Faults, Machine Element Faults) Sensor readings: vibration, force, network logs, usage metrics (depending on dataset) Nikolakis et al. A microservice architecture for predictive analytics in manufacturing 2020 Robotic manipulator (robot box) for belt tension monitoring Electric current, Position signals (robot joints), Statistical features (mean, RMS, min, max, skewness, kurtosis) You et al. Advances of Digital Twins for Predictive Maintenance 2022 Various industrial equipment (gearbox, bearings, milling machine, turbines, satellites, etc.) Varies by case: vibration, acoustic emission, stress, force, temperature, geometry, multi-physics simulation outputs Ragab et al. Attention-based sequence to sequence model for machine remaining useful life prediction 2021 Turbofan Engines (C-MAPSS dataset) Temperature, Pressure, Speed, Time-series sensor data Calabrese et al. Prognostic Health Management of Production Systems. New Proposed Approach and Experimental Evidences 2019 Rotating Component in Packaging Machine Current signals (mA) from actuator, Time-domain features (RMS, Kurtosis, Mean, Skewness, Variance, Crest Factor) Traini et al. Machine Learning Framework for Predictive Maintenance in Milling 2019 Milling operations on Matsuura MC-510V CNC machine (NASA milling dataset) Vibration signals, Acoustic Emission (AE), Spindle motor current, Cutting parameters (speed, depth of cut, feed), Flank wear measurements Lv et al. Predictive maintenance decision-making for variable faults with non-equivalent costs of fault severities 2023 Railcar Wheel Bearings Bearing temperature data (historical), Threshold-based states (Healthy, Degradation, Failure), Statistical features Joseph et al. A Predictive Maintenance Application for A Robot Cell using LSTM Model 2022 Robot Cell (Gluing Station) in Automotive Industry Temperature, Doser volume, Torque (Nm), Pressure (bar), Alarm logs Taşcı et al. Remaining useful lifetime prediction for predictive maintenance in manufacturing 2023 Consumer Goods Manufacturing Assembly Line Sensor readings: Weight, Speed, Temperature, Electric current, Vacuum, Air pressure, Statistical features Vargas et al. A hybrid feature learning approach based on convolutional kernels for ATM fault prediction using event-log data 2023 ATM machines (68 machines over 2 years) Event-log commands and responses, Count of events per 10-minute interval (e.g. PrepareWithdrawal, Initialize, CloseShutter, etc.) Gupta et al. Predictive maintenance of baggage handling conveyors using IoT 2023 Airport Baggage Handling Conveyors (S-Lifts) Vibration signals, Acoustic signals, RMS, Kurtosis, Skewness, Frequency-domain features Tessoni et al. Advanced statistical and machine learning methods for multi-step multivariate time series forecasting in predictive maintenance 2022 Multiple public datasets: FRED, Air Quality, Appliances Energy, Beijing PM2.5, Gas Turbine Emission Time-series data (varies by dataset): e.g. sensor measurements, energy usage, pollution levels, gas emissions Daniyan et al. Artificial intelligence for predictive maintenance in the railcar learning factories 2020 Railcar Wheel Bearings Bearing temperature data (historical), Threshold-based states (Healthy, Degradation, Failure), Statistical features Uhlmann et al. Holistic Concept Towards a Reference Architecture Model for Predictive Maintenance 2021 Axis test rig (ball screw spindle) Vibration signals (x, y, z axes), Variance, Time-series features Rihi et al. Predictive maintenance in mining industry: grinding mill case study 2022 Grinding Mill in Mining Industry Vibration signals, Temperature, Electric signals, Ultrasonic measurements, Time-domain, Frequency-domain, Time-frequency features Nentwich et al. Towards Data Acquisition for Predictive Maintenance of Industrial Robots 2021 Six-axis Articulated Industrial Robots (e.g. KUKA KR210) Vibration signals, Motor currents, RMS of acceleration, Trajectory-specific measurements Bilal Yıldız et al. Integrating preventive and predictive maintenance policies with system dynamics: A decision table approach 2023 AI4I 2020 Predictive Maintenance Dataset (simulation study) Air temperature, Process temperature, Rotational speed, Torque, Tool wear, Failure modes (HDF, OF, PF, TWF), Maintenance budgets, Time between failures Lee et al. Predictive Maintenance of Machine Tool Systems Using Artificial Intelligence Techniques Applied to Machine Condition Data 2019 Milling Process (Cutting Tool) & Bearing Test Rig Milling signals: DC spindle current, AC spindle current, Table vibration, Spindle vibration, Acoustic Emission (AE); Bearing signals: Vibration signals at 20 kHz Killeen et al. IoT-based predictive maintenance for fleet management 2019 Public Transport Buses (Société de Transport de l’Outaouais, Canada) J1939 sensor data: wheel speed, axle angle, engine speed, engine torque, oil temperature, coolant temperature, driver pedal positions Luo et al. A hybrid predictive maintenance approach for CNC machine tool driven by Digital Twin 2020 CNC Milling Machine Tool – Cutting Tool Wear Vibration signals, Cutting force, Acoustic Emission (AE), Temperature, Relative slip speed, Stress, Statistical features (RMS, Kurtosis, Skewness) Feng et al. Multi-level predictive maintenance of smart manufacturing systems driven by digital twin: A matheuristics approach 2023 Offshore Oil and Gas Production System – Subsea Christmas Trees (XTs) Remaining Useful Life (RUL) of components, Oil production rates, Maintenance costs, Downtime losses, Spare parts inventory, ROV availability Cao et al. KSPMI: A Knowledge-based System for Predictive Maintenance in Industry 4.0 2022 Semiconductor Manufacturing Process (UCI SECOM Dataset) 590 process variables (temperatures, pressures, times), event timestamps, chronicle patterns (sequences of events leading to failures) Coelho et al. Predictive maintenance on sensorized stamping presses by time series segmentation, anomaly detection, and classification algorithms 2022 Stamping Press in Bosch Thermo Technology Oil Temperature, Oil Level, Engine Actual Speed, Shaft Vibration, Pressure Left, Pressure Right, Tool Height, plus aggregated features (mean, max, std, etc.) Chinta et al. Optimal feature selection on Serial Cascaded deep learning for predictive maintenance system in automotive industry with fused optimization algorithm 2023 Automotive Industry (Simulated & real datasets) Sensor telemetry (e.g. voltage, rotation, pressure, vibration), Degradation indices, Hidden neuron counts (optimized), Data from 5 public datasets (e.g. turbofan engines, Genesis linear drive, etc.) Jiang et al. A2-LSTM for predictive maintenance of industrial equipment based on machine learning 2022 Horizontal Machining Center – Aircraft Manufacturing Electrical signals (currents, voltages, active/reactive power, power factor, frequency), 28 attributes reduced to 15 selected attributes Farahani et al. A data-driven predictive maintenance framework for injection molding process 2022 Injection Molding Process – Cooling System Monitoring Mold temperature, Cavity pressure, Strain gauge data, Barrel pressure, Clamp force, Screw position, Derived features (RMS, Variance, PCA components) Drakaki et al. Recent Developments Towards Industry 4.0 Oriented Predictive Maintenance in Induction Motors 2021 Induction Motors (IM) in Industrial Settings Motor current signals, Vibration signals, Time-domain, Frequency-domain, Time-frequency domain features, Raw data from sensors Jain et al. Distributed Diagnostics, Prognostics and Maintenance Planning: Realizing Industry 4.0 2020 Milling Cutter System on manufacturing shop floor Force signals, Vibration signals, Acoustic Emission signals Huynh et al. A predictive maintenance model for k-out-of-n:F continuously deteriorating systems subject to stochastic and economic dependencies 2022 Generic – applicable to various industrial systems Degradation levels, parameters of Gamma processes (shape, scale), Inspection cost, Replacement costs, Downtime costs, System configuration Zonta et al. A predictive maintenance model for optimizing production schedule using deep neural networks 2022 Simulated industrial machines – CNCs and generic production equipment using Microsoft Azure PdM dataset Telemetry Data: Voltage, Rotation, Pressure, Vibration; Degradation Indices; Production tasks and job scheduling constraints; Maintenance costs Wen et al. Recent advances and trends of predictive maintenance from data-driven machine prognostics perspective 2022 Review article (no specific single case study) Generic variables: vibration, temperature, current, pressure, time-series signals, health indicators Einabadi et al. Dynamic Predictive Maintenance in industry 4.0 based on real time information: Case study in automotive industries 2019 CNC Machine in Fiat Power Train Technologies (FPT) automotive plant Sensor data: Temperature, Vibration (mm/s), Energy consumption (kW), Remaining Useful Life (RUL), Maintenance costs, Set-up costs Shaheen et al. Data-driven failure prediction and RUL estimation of mechanical components using accumulative artificial neural networks 2023 Simulated mechanical component with predefined failure threshold (0.3) Degradation level (simulated data), single variable (e.g. temperature, pressure, speed depending on component) Hillenbrand et al. Investigation of defects in roll contacts of machine elements with Acoustic Emission and Unsupervised Machine Learning 2021 Axial Ball Bearings on test bench Acoustic Emission signals (AE), RMS, Mean Frequency, Formfactor, Kurtosi Augustyn et al. Method of Machining Centre Sliding System Fault Detection using Torque Signals and Autoencoder 2023 Sliding System of Machining Centres (X, Y, Z, A axes) Torque signals from servomotors (measured current converted to torque) Rajasekar et al. Prediction of Component Level Degradation in a Hydraulic Rig using Machine Learning Methods 2023 Multi-component hydraulic rig. The system is monitored to identify and classify component degradations and failures. Multivariate sensor signals (asynchronous) Singha et al. Predictive Failure Analysis of Spindle Motor & Cutting Oil Condition Monitoring of Grinding Machine using Artificial Intelligence Models 2020 Grinding Machine (focus on spindle motor & cutting oil condition) Vibration signals, Temperature, pH levels of cutting oil, Sulphur concentration Kotsiopoulos et al. Deep multi-sensorial data analysis for production monitoring in hard metal industry 2021 Hard Metal Industry (CNC lathe machining) Vibration (X, Z axis), Laser microprofilometry, Ultrasound scans, FFT features Fordal et al. Application of sensor data based predictive maintenance and artificial neural networks to enable Industry 4.0 2023 Splitting Saw in Talgø MøreTre AS (Norway) Blade temperature (IR sensor), Current (Power usage), Vibration RMS, Temperature from vibration sensor, Aggregated features (min, max, mean, std), Alarm thresholds Vicêncio et al. An Intelligent Predictive Maintenance Approach Based on End-of-Line Test Logfiles in the Automotive Industry 2021 Automotive Industry – End-of-Line (EoL) Testing Systems. Focus on predicting malfunctions in final quality testing equipment, Log message contents (event type, timestamp, operational context). Wang et al. Dynamic predictive maintenance strategy for RUL prediction via DL ensemble 2024 NASA Turbofan engine (C-MAPSS dataset) RUL, mission cycles, reliability, stock & maintenance decisions Herzog et al. Predictive maintenance for electrical motors: AI algorithms 2024 Electrical motors (fault detection & isolation) Vibration signals, classification features (SVM, perceptron, KNN) Kolvig-Raun et al. Joint Stress Estimation and RUL Prediction for Collaborative Robots 2024 Collaborative robots (UR e-series) Joint stress, load case, torque, temperature, RUL Mardianto et al. Augmenting Machinery Health Monitoring System with LSTM 2024 Turbomachinery equipment RUL, vibration data, MAPE, online monitoring Patel et al. Blockchain-based Predictive Maintenance Scheme for Smart Agriculture 2024 Smart Agriculture machinery Sensor data (IoT), crop machinery conditions, faults Gougam et al. Bearing faults classification using signal processing + ML 2024 Mechanical test rig (rotating machinery) Vibration signals, MODWPT features, health indicators Kondo et al. Industrial edge computing architecture for Local Digital Twin 2024 Assembly line (Brazilian subsidiary) Edge computing, DT parameters, yield prediction Meddaoui et al. Benefits of predictive maintenance in manufacturing excellence 2023 Industrial production process Physical parameters, downtime, costs, ML algorithms Meddaoui et al. Advanced ML for PdM: RUL prediction & reliability 2024 Aerospace maintenance dataset Feature engineering, ML models, RUL, failures Yuhua Yin, et al. A Dual-Task Learning Approach for Bearing Anomaly Detection and State Evaluation of Safe Region 2024 Bearings anomaly detection Safe region evaluation, anomaly features, SVDD parameters Javad Isavand, et al. A reduced-order machine-learning-based method for fault recognition in tool condition monitoring 2024 Tool condition monitoring (machining, 200 min operation, 5 tools) Frequency-domain & time-frequency features (FFT, STFT, EMD, VMD) Anouar Bourokba, et al. A Shapley based XAI approach for a turbofan RUL estimation 2024 Turbofan engines (aircraft) RUL, sensor parameters with Shapley contributions Ugur Ileri, et al. An Efficient Approach for Automatic Fault Classification Based on Data Balance and One-Dimensional Deep Learning 2024 Predictive maintenance datasets – fault classification Raw PdM datasets, normalized features, balanced groups Rengaraj R, G.R. et al. Design and Implementation of Real Time Monitoring System of Diesel Generator using IoT 2024 Diesel Generators (DG) with IoT sensors Vibration, operational data, IoT sensor streams Haobin Wen, et al. Early Prediction of Remaining Useful Life for Rolling Bearings Based on Envelope Spectral Indicator and Bayesian Filter 2024 Bearings RUL (rolling-element) Envelope Spectral Indicator (ESI), vibration signals, fault frequencies Zekai Si, et al. Efficient Tool Wear Prediction in Manufacturing: BiLPReS Hybrid Model with Performer Encoder 2024 Tool wear prediction (PHM2010 dataset) Force, acceleration, acoustic emission (multichannel sensors) Patrick Seebold, et al. Explainable AI-based Shapley Additive Explanations for Remaining Useful Life Prediction using NASA Turbofan Engine Dataset 2024 NASA Turbofan engine dataset 21 sensors + operating settings, SHAP features Huan Wang, et al. Fourier Feature Refiner Network With Soft Thresholding for Machinery Fault Diagnosis Under Highly Noisy Conditions 2024 Bearings – noisy industrial vibration Vibration signals (time + frequency domain, Fourier features) Pooja Kamat, et al. Vibration-based anomaly pattern mining for remaining useful life (RUL) prediction in bearings 2024 Bearings RUL prediction Sensor data (vibration), features (time + frequency domain), BiLSTM, CNN-LSTM, Conv-LSTM To address our research questions on what themes, dominate predictive maintenance (PdM) in the Industry 4.0 context, how they interconnect, and how they evolve over time, we analyzed the co-occurrence structure of author-supplied keywords (minimum occurrence ≥ 2) and produced complementary VOSviewer visualizations: a clustered network, item-focused network views, and a density map. Together they reveal a coherent, maturing research landscape in which PdM and machine learning form the conceptual and methodological cores, connected to deep-learning prognostics and Industry 4.0 implementation strands The co-occurrence map (figure7) clearly shows that machine learning and predictive maintenance occupy central positions, acting as hubs connecting the entire network. Their high link strength indicates that they co-occur with a large number of other keywords, confirming their foundational role in this research domain. The clustered co-occurrence network (Figure 7) contains 21 items, 85 links, and a total link strength of 113, forming four stable thematic clusters: Cluster 1: Algorithms & Predictive Modeling. Contains machine learning, prediction algorithms, accuracy, and monitoring. This is the algorithmic backbone of PdM: supervised learning methods, performance criteria, and operational monitoring. Cluster 2: Deep Learning & Prognostics. Includes deep learning, bidirectional LSTM (BiLSTM), feature extraction, fault diagnosis, remaining useful life (RUL), and terms from prognostics and health management. This cluster captures the data-driven prognostics pipeline (signal → features → sequential DL → diagnosis/RUL). Cluster 3: PdM, Monitoring & Conceptual Core. Organized around predictive maintenance and predictive model, closely tied to condition monitoring. It functions as the semantic anchor linking methods to operational use. Cluster 4 : Applications & Industry 4.0 : Contains industry 4.0, maintenance engineering, and robot sensing systems. This cluster confirms that PdM research is now deeply embedded in the Industry 4.0 paradigm, with a strong orientation toward sensor integration, robotics, and practical deployment. Two central hubs knit the map together: predictive maintenance (C3) and machine learning (C1). Their high connectivity and proximity to multiple clusters show that PdM is implemented primarily via ML-centred analytics, while the C2 and C4 clusters bring, respectively, advanced prognostics and industrial realization into the fold. The density map established in the figure 8 complements the structural views by showing where the highest concentration of items and links lies: The brightest hotspots sit on machine learning, predictive maintenance, and deep learning the epicenter of current research. Secondary but distinct hotspots appear over remaining useful life, condition monitoring, confirming that prognostics-oriented DL is a major focal area. Industry 4.0, maintenance engineering, and robot sensing systems register as application-side hotspots, indicating tangible interest in instrumentation and deployment. This intensity pattern triangulates the earlier findings: method (ML/DL), task (RUL/diagnosis) and context (Industry 4.0) jointly defines the field’s present center of gravity. The brightest hotspots sit on machine learning, predictive maintenance, and deep learning the epicenter of current research. The co-occurrence network (figure 9) highlights three key thematic dynamics shaping predictive maintenance (PdM) research. First, predictive maintenance and machine learning occupy central positions, confirming their dual role as the conceptual and methodological cores of the field. PdM integrates monitoring and diagnostic activities, while machine learning provides the computational framework that drives data-based decision-making and connects theory to industrial applications. The literature shows an increasing standardization of analytical workflows. The co-occurrence of deep learning, BiLSTM, and remaining useful life (RUL) prediction illustrates a consolidated pipeline: feature extraction, sequential neural architectures, and RUL estimation. This convergence reflects the transition from experimental studies to a recognized methodological benchmark for prognostics and health management. The explicit presence of industry 4.0 and its association with maintenance engineering and robot sensing systems indicate a clear shift toward practical implementation. Research is now focusing on embedding PdM within smart factories and cyber-physical production systems, emphasizing automation, connectivity, and real-time decision-making. This trend signals the move toward fully integrated, autonomous, and data-driven maintenance ecosystems, bridging the gap between theoretical development and industrial deployment. 6. Conclusion This review has explored the rapidly evolving landscape of Predictive Maintenance (PdM) within the context of Smart Industry and Industry 4.0. Drawing on a diverse body of recent literature, it is evident that data-driven methodologies, particularly deep learning techniques, have become the dominant force in contemporary PdM research. These approaches offer remarkable capabilities in handling complex, high-dimensional sensor data, enabling accurate predictions of equipment health and remaining useful life (RUL). However, challenges such as data scarcity, model interpretability, and deployment robustness remain significant barriers to widespread industrial adoption. Hybrid approaches that integrate physical knowledge with machine learning have emerged as a promising pathway, aiming to combine the predictive power of AI with the transparency and domain understanding required in industrial environments. Furthermore, the increasing incorporation of digital twin technologies signals a shift towards more holistic, context-aware maintenance solutions. Across sectors from manufacturing and automotive to mining and railways, the adoption of AI-driven PdM demonstrates clear potential for reducing downtime, optimizing maintenance schedules, and enhancing operational efficiency. Yet, the variability of industrial contexts demands tailored solutions, and no universal approach has yet been established. Looking ahead, future research should prioritize developing explainable AI techniques, leveraging transfer learning to overcome data limitations, and aligning predictive models with economic and operational impacts. Only through such integrated efforts can Predictive Maintenance fully realize its transformative potential within Smart Industry, contributing to more resilient, efficient, and intelligent industrial systems. Ultimately, PdM stands at the intersection of technological innovation and practical industrial needs. As Industry 4.0 continues to reshape production landscapes, the development of robust, interpretable, and cost-effective PdM solutions will be pivotal in defining the future of smart manufacturing and maintenance. Declarations Competing Interest: Authors of this paper have no financial interests and have no relevant financial or non-financial interests to disclose.” Funding : The authors declare that no funds, no data set used, grants, or other support were received during the preparation of this manuscript. Author Contributions: All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Hamdi Alaoui. A, Meddaoui. A, and Hain, M. The first draft of the manuscript was written by Mr. Hamdi and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. References Augustyn D, Fidali M. 2023. Method of Machining Centre Sliding System Fault Detection using Torque Signals and Autoencoder. Acta Mechanica et Automatica. 17(3):445–451. https://doi.org/10.2478/ama-2023-0051 Bilal Yıldız G, Soylu B. 2023. Integrating preventive and predictive maintenance policies with system dynamics: A decision table approach. Advanced Engineering Informatics. 56:101952. https://doi.org/10.1016/j.aei.2023.101952 Bourokba A, El Hamdi R, Njah M. 2024. A Shapley based XAI approach for a turbofan RUL estimation. In: 2024 21st International Multi-Conference on Systems, Signals & Devices (SSD) [Internet]. [place unknown]; [accessed 2025 Sep 1]; p. 832–837. https://doi.org/10.1109/SSD61670.2024.10548499 Calabrese F, Regattieri A, Botti L, Galizia FG. 2019. Prognostic Health Management of Production Systems. New Proposed Approach and Experimental Evidences. Procedia Manufacturing. 39:260–269. https://doi.org/10.1016/j.promfg.2020.01.333 Cao Q, Zanni-Merk C, Samet A, Reich C, Beuvron F de B de, Beckmann A, Giannetti C. 2022. KSPMI: A Knowledge-based System for Predictive Maintenance in Industry 4.0. Robotics and Computer-Integrated Manufacturing. 74:102281. https://doi.org/10.1016/j.rcim.2021.102281 Chen C-S, Chen J-S. 2011. Rotor fault diagnosis system based on sGA-based individual neural networks. Expert Systems with Applications. 38(9):10822–10830. https://doi.org/10.1016/j.eswa.2011.02.074 Chinta VS, Kethi Reddi S, Yarramsetty N. 2023. Optimal feature selection on Serial Cascaded deep learning for predictive maintenance system in automotive industry with fused optimization algorithm. Advanced Engineering Informatics. 57:102105. https://doi.org/10.1016/j.aei.2023.102105 Chowdhury A, Raut S, Pal A. 2022. Internet of Things resource monitoring through proactive fault prediction. Computers & Industrial Engineering. 169:108265. https://doi.org/10.1016/j.cie.2022.108265 Coelho D, Costa D, Rocha EM, Almeida D, Santos JP. 2022. Predictive maintenance on sensorized stamping presses by time series segmentation, anomaly detection, and classification algorithms. Procedia Computer Science. 200:1184–1193. https://doi.org/10.1016/j.procs.2022.01.318 Crespo Márquez A, Marcos Alberca JA, Crespo del Castillo A. 2023. Simulating dynamic RUL based CBM scheduling. A case study in the railway sector. Computers in Industry. 148:103914. https://doi.org/10.1016/j.compind.2023.103914 Daniyan I, Mpofu K, Oyesola M, Ramatsetse B, Adeodu A. 2020. Artificial intelligence for predictive maintenance in the railcar learning factories. Procedia Manufacturing. 45:13–18. https://doi.org/10.1016/j.promfg.2020.04.032 Drakaki M, Karnavas YL, Tzionas P, Chasiotis ID. 2021. Recent Developments Towards Industry 4.0 Oriented Predictive Maintenance in Induction Motors. Procedia Computer Science. 180:943–949. https://doi.org/10.1016/j.procs.2021.01.345 Einabadi B, Baboli A, Ebrahimi M. 2019. Dynamic Predictive Maintenance in industry 4.0 based on real time information: Case study in automotive industries. IFAC-PapersOnLine. 52(13):1069–1074. https://doi.org/10.1016/j.ifacol.2019.11.337 Farahani S, Khade V, Basu S, Pilla S. 2022. A data-driven predictive maintenance framework for injection molding process. Journal of Manufacturing Processes. 80:887–897. https://doi.org/10.1016/j.jmapro.2022.06.013 Feng Q, Zhang Y, Sun B, Guo X, Fan D, Ren Y, Song Y, Wang Z. 2023. Multi-level predictive maintenance of smart manufacturing systems driven by digital twin: A matheuristics approach. Journal of Manufacturing Systems. 68:443–454. https://doi.org/10.1016/j.jmsy.2023.05.004 Fordal JM, Schjølberg P, Helgetun H, Skjermo TØ, Wang Y, Wang C. 2023. Application of sensor data based predictive maintenance and artificial neural networks to enable Industry 4.0. Adv Manuf. 11(2):248–263. https://doi.org/10.1007/s40436-022-00433-x Franciosi C, Miranda S, Veneroso CR, Riemma S. 2021. A maintenance scheduling optimization model for a multi-component machine in a digitalized manufacturing context. IFAC-PapersOnLine. 54(1):1254–1259. https://doi.org/10.1016/j.ifacol.2021.08.150 Gougam F, Afia A, Soualhi A, Touzout W, Rahmoune C, Benazzouz D. 2024. Bearing faults classification using a new approach of signal processing combined with machine learning algorithms. J Braz Soc Mech Sci Eng. 46(2):65. https://doi.org/10.1007/s40430-023-04645-5 Gupta V, Mitra R, Koenig F, Kumar M, Tiwari MK. 2023a. Predictive maintenance of baggage handling conveyors using IoT. Computers & Industrial Engineering. 177:109033. https://doi.org/10.1016/j.cie.2023.109033 Gupta V, Mitra R, Koenig F, Kumar M, Tiwari MK. 2023b. Predictive maintenance of baggage handling conveyors using IoT. Computers & Industrial Engineering. Heng A, Zhang S, Tan ACC, Mathew J. 2009. Rotating machinery prognostics: State of the art, challenges and opportunities. Mechanical Systems and Signal Processing. 23(3):724–739. https://doi.org/10.1016/j.ymssp.2008.06.009 Herzog T, Bartecki K. 2024. Predictive maintenance for electrical motors: Current approach and usage of artificial intelligence algorithms. In: 2024 28th International Conference on Methods and Models in Automation and Robotics (MMAR) [Internet]. [place unknown]; [accessed 2025 Aug 30]; p. 494–498. https://doi.org/10.1109/MMAR62187.2024.10680808 Hillenbrand J, Detroy J, Fleischer J. 2021. Investigation of defects in roll contacts of machine elements with Acoustic Emission and Unsupervised Machine Learning. IOP Conf Ser: Mater Sci Eng. 1193(1):012085. https://doi.org/10.1088/1757-899X/1193/1/012085 Huynh KT, Vu HC, Nguyen TD, Ho AC. 2022. A predictive maintenance model for k-out-of-n:F continuously deteriorating systems subject to stochastic and economic dependencies. Reliability Engineering & System Safety. 226:108671. https://doi.org/10.1016/j.ress.2022.108671 Ileri U, Altun Y, Narin A. 2024. An Efficient Approach for Automatic Fault Classification Based on Data Balance and One-Dimensional Deep Learning. Applied Sciences. 14(11):4899. https://doi.org/10.3390/app14114899 Intelligent Fault Diagnosis and Prognosis for Engineering Systems | Wiley Online Books. [accessed 2025 Jun 26]. https://onlinelibrary.wiley.com/doi/book/10.1002/9780470117842 Isavand J, Kasaei A, Peplow A, Wang X, Yan J. 2024. A reduced-order machine-learning-based method for fault recognition in tool condition monitoring. Measurement. 224:113906. https://doi.org/10.1016/j.measurement.2023.113906 Jain AK, Chouksey P, Parlikad AK, Lad BK. 2020. Distributed Diagnostics, Prognostics and Maintenance Planning: Realizing Industry 4.0. IFAC-PapersOnLine. 53(3):354–359. https://doi.org/10.1016/j.ifacol.2020.11.057 Jardine AKS, Lin D, Banjevic D. 2006. A review on machinery diagnostics and prognostics implementing condition-based maintenance. Mechanical Systems and Signal Processing. 20(7):1483–1510. https://doi.org/10.1016/j.ymssp.2005.09.012 Jiang Y, Dai P, Fang P, Zhong RY, Zhao X, Cao X. 2022. A2-LSTM for predictive maintenance of industrial equipment based on machine learning. Computers & Industrial Engineering. 172:108560. https://doi.org/10.1016/j.cie.2022.108560 Joseph D, Gallege T, Bekar ET, Dudas C, Skoogh A. 2022a. A Predictive Maintenance Application for A Robot Cell using LSTM Model. 5th IFAC Workshop on Advanced Maintenance Engineering, Services and Technologies AMEST 2022. Joseph D, Gallege T, Bekar ET, Dudas C, Skoogh A. 2022b. A Predictive Maintenance Application for A Robot Cell using LSTM Model. IFAC-PapersOnLine. 55(19):115–120. https://doi.org/10.1016/j.ifacol.2022.09.193 Kamat P, Kumar S, Sugandhi R. 2024. Vibration-based anomaly pattern mining for remaining useful life (RUL) prediction in bearings. J Braz Soc Mech Sci Eng. 46(5):290. https://doi.org/10.1007/s40430-024-04872-4 Kans M, Campos J, Håkansson L. 2020. A remote laboratory for Maintenance 4.0 training and education. IFAC-PapersOnLine. 53(3):101–106. https://doi.org/10.1016/j.ifacol.2020.11.016 Killeen P, Ding B, Kiringa I, Yeap T. 2019. IoT-based predictive maintenance for fleet management. Procedia Computer Science. 151:607–613. https://doi.org/10.1016/j.procs.2019.04.184 Kolvig-Raun ES, Kjærgaard MB, Brorsen R. 2024. Joint Stress Estimation and Remaining Useful Life Prediction for Collaborative Robots to Support Predictive Maintenance. IEEE Robotics and Automation Letters. 9(4):3554–3561. https://doi.org/10.1109/LRA.2024.3368296 Kondo RE, Andrade WJ, de Mello Henequim C, Lazzaretti AE, de Souza Britto A, de Freitas Rocha Loures E, Santos EAP, Reynoso-Meza G. 2024. An industrial edge computing architecture for Local Digital Twin. Computers & Industrial Engineering. 193:110257. https://doi.org/10.1016/j.cie.2024.110257 Kotsiopoulos T, Leontaris L, Dimitriou N, Ioannidis D, Oliveira F, Sacramento J, Amanatiadis S, Karagiannis G, Votis K, Tzovaras D, Sarigiannidis P. 2021. Deep multi-sensorial data analysis for production monitoring in hard metal industry. Int J Adv Manuf Technol. 115(3):823–836. https://doi.org/10.1007/s00170-020-06173-1 Lee WJ, Wu H, Yun H, Kim H, Jun MBG, Sutherland JW. 2019. Predictive Maintenance of Machine Tool Systems Using Artificial Intelligence Techniques Applied to Machine Condition Data. Procedia CIRP. 80:506–511. https://doi.org/10.1016/j.procir.2018.12.019 Lei Y, Li Naipeng, Guo L, Li Ningbo, Yan T, Lin J. 2018. Machinery health prognostics: A systematic review from data acquisition to RUL prediction. Mechanical Systems and Signal Processing. 104:799–834. https://doi.org/10.1016/j.ymssp.2017.11.016 Luo W, Hu T, Ye Y, Zhang C, Wei Y. 2020. A hybrid predictive maintenance approach for CNC machine tool driven by Digital Twin. Robotics and Computer-Integrated Manufacturing. 65:101974. Lv Y, Guo X, Zhou Q, Qian L, Liu J. 2023. Predictive maintenance decision-making for variable faults with non-equivalent costs of fault severities. Advanced Engineering Informatics. 56:102011. https://doi.org/10.1016/j.aei.2023.102011 Ma Z, Zhao M, Dai X, Chen Y. 2023. A hybrid-driven probabilistic state space model for tool wear monitoring. Mechanical Systems and Signal Processing. 200:110599. https://doi.org/10.1016/j.ymssp.2023.110599 Mardianto FD, Mauritsius T. 2024. Augmenting the Internet-based Machinery Health Monitoring System (MHMS) Through Integration of Long Short-Term Memory (LSTM) Model for Enhanced Prediction of Remaining Useful Lifetime (RUL). In: 2024 8th International Conference on Information Technology, Information Systems and Electrical Engineering (ICITISEE) [Internet]. [place unknown]; [accessed 2025 Aug 30]; p. 447–452. https://doi.org/10.1109/ICITISEE63424.2024.10730046 Meddaoui A, Hachmoud A, Hain M. 2024. Advanced ML for predictive maintenance: a case study on remaining useful life prediction and reliability enhancement. Int J Adv Manuf Technol. 132(1):323–335. https://doi.org/10.1007/s00170-024-13351-y Meddaoui A, Hain M, Hachmoud A. 2023. The benefits of predictive maintenance in manufacturing excellence: a case study to establish reliable methods for predicting failures. Int J Adv Manuf Technol. 128(7):3685–3690. https://doi.org/10.1007/s00170-023-12086-6 Mobley RK. 2002. 2 - Financial Implications and Cost Justification. In: Mobley RK, editor. An Introduction to Predictive Maintenance (Second Edition) [Internet]. Burlington: Butterworth-Heinemann; [accessed 2023 Oct 15]; p. 23–42. https://doi.org/10.1016/B978-075067531-4/50002-6 Moher D, Liberati A, Tetzlaff J, Altman DG. 2009. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 339:b2535. https://doi.org/10.1136/bmj.b2535 Nentwich C, Reinhart G. 2021. Towards Data Acquisition for Predictive Maintenance of Industrial Robots. Procedia CIRP. 104:62–67. https://doi.org/10.1016/j.procir.2021.11.011 Nikolakis N, Marguglio A, Veneziano G, Greco P, Panicucci S, Cerquitelli T, Macii E, Andolina S, Alexopoulos K. 2020. A microservice architecture for predictive analytics in manufacturing. Procedia Manufacturing. 51:1091–1097. https://doi.org/10.1016/j.promfg.2020.10.153 Patel M, Tanwar S, Jindal A, Alshammari N, Pervaiz H, Ahmed H. 2024. A Blockchain-based Predictive Maintenance Scheme for Smart Agriculture. In: 2024 IEEE/ACM 17th International Conference on Utility and Cloud Computing (UCC) [Internet]. [place unknown]; [accessed 2025 Aug 30]; p. 405–410. https://doi.org/10.1109/UCC63386.2024.00062 (PDF) Evolution of Technical Systems Maintenance Approaches – Review and a Case Study. ResearchGate [Internet]. [accessed 2025 Jun 26]. https://doi.org/10.1007/978-3-319-97490-3_16 Pech M, Vrchota J, Bednář J. 2021. Predictive Maintenance and Intelligent Sensors in Smart Factory: Review. Sensors. 21(4):1470. https://doi.org/10.3390/s21041470 Physics-informed machine learning | Nature Reviews Physics. [accessed 2025 Jun 25]. https://www.nature.com/articles/s42254-021-00314-5?utm_source=chatgpt.com Polenghi A, Roda I, Pesenti V, Pasanisi D, Macchi M, Cortinovis D, Chebat F. 2023. Norm-based data labelling in supervised learning for fault detection and diagnostics of rotating elements towards maintenance servitisation. IFAC-PapersOnLine. 56(2):1282–1287. https://doi.org/10.1016/j.ifacol.2023.10.1761 Prognostics and Health Management of Electronics. Wiley.com [Internet]. [accessed 2025 Jun 26]. https://www.wiley.com/en-us/Prognostics+and+Health+Management+of+Electronics%3A+Fundamentals%2C+Machine+Learning%2C+and+the+Internet+of+Things-p-9781119515357 R Rengaraj, Venkatakrishnan GR, Rajalakshmi M, R Ranganath, R SG, V S. 2024. Design and Implementation of Real Time Monitoring System of Diesel Generator using IoT. In: 2024 2nd International Conference on Networking and Communications (ICNWC) [Internet]. [place unknown]; [accessed 2025 Sep 1]; p. 1–8. https://doi.org/10.1109/ICNWC60771.2024.10537268 Ragab M, Chen Z, Wu M, Kwoh C-K, Yan R, Li X. 2021. Attention-based sequence to sequence model for machine remaining useful life prediction. Neurocomputing. 466:58–68. https://doi.org/10.1016/j.neucom.2021.09.022 Remaining useful life estimation – A review on the statistical data driven approaches. 2011. European Journal of Operational Research. 213(1):1–14. https://doi.org/10.1016/j.ejor.2010.11.018 Rihi A, Baïna S, Mhada F, Elbachari E, Tagemouati H, Guerboub M, Benzakour I. 2022. Predictive maintenance in mining industry: grinding mill case study. Procedia Computer Science. 207:2483–2492. https://doi.org/10.1016/j.procs.2022.09.306 Seebold P, Kaye MK, Kim S, Nam CS. 2024. Explainable AI-based Shapley Additive Explanations for Remaining Useful Life Prediction using NASA Turbofan Engine Dataset. In: 2024 IEEE 3rd International Conference on Computing and Machine Intelligence (ICMI) [Internet]. [place unknown]; [accessed 2025 Sep 1]; p. 1–5. https://doi.org/10.1109/ICMI60790.2024.10586061 Sfar M, Yaacoubi S, Dellagi S, El Mountassir M, Turki S. 2020. Towards a predictive maintenance based on ultrasonic guided waves non-destructive testing technique: a case study. In: 13ème Conférence Internationale de MOdélisation, Optimisation et SIMulation : “ Nouvelles avancées et défis pour des industries durables et avisées ” : MOSIM 20, 12-14 Novembre 2020 : actes de conférence = 13 th International Conference on MOdeling, Optimization and SIMulation : “New Advances and Challenges for Sustainable and Smart Industries” : MOSIM 20, 12-14 November 2020 : Conference proceedings [Internet]. Agadir (en ligne), Morocco: Ali Siadat (président du comité d’organisation) and Lahcen Mifdal (président du comité d’organisation) and Ilias Majdouline (président du comité d’organisation); [accessed 2025 Jun 22]; p. 54–62. https://hal.science/hal-03192678 Shaheen B, Kocsis Á, Németh I. 2023. Data-driven failure prediction and RUL estimation of mechanical components using accumulative artificial neural networks. Engineering Applications of Artificial Intelligence. 119:105749. https://doi.org/10.1016/j.engappai.2022.105749 She D, Jia M. 2021a. A BiGRU method for remaining useful life prediction of machinery. Measurement. She D, Jia M. 2021b. A BiGRU method for remaining useful life prediction of machinery. Measurement. 167:108277. https://doi.org/10.1016/j.measurement.2020.108277 Si Z, Si S, Mu D. 2024. Efficient Tool Wear Prediction in Manufacturing: BiLPReS Hybrid Model with Performer Encoder. Arab J Sci Eng. 49(11):15193–15204. https://doi.org/10.1007/s13369-024-08943-5 Singha A, Ali J, Khera V. 2020. Predictive Failure Analysis of Spindle Motor & Cutting Oil Condition Monitoring of Grinding Machine using Artificial Intelligence Models. In: 2020 IEEE 17th India Council International Conference (INDICON) [Internet]. New Delhi, India: IEEE; [accessed 2025 Jun 20]; p. 1–7. https://doi.org/10.1109/INDICON49873.2020.9342533 Smith AM, Hinchcliffe GR. 2003. RCM--Gateway to World Class Maintenance. [place unknown]: Elsevier. Susto GA, Schirru A, Pampuri S, McLoone S, Beghi A. 2015. Machine Learning for Predictive Maintenance: A Multiple Classifier Approach. IEEE Trans Ind Inf. 11(3):812–820. https://doi.org/10.1109/TII.2014.2349359 Taşcı B, Omar A, Ayvaz S. 2023. Remaining useful lifetime prediction for predictive maintenance in manufacturing. Computers & Industrial Engineering. 184:109566. https://doi.org/10.1016/j.cie.2023.109566 Tessoni V, Amoretti M. 2022. Advanced statistical and machine learning methods for multi-step multivariate time series forecasting in predictive maintenance. Procedia Computer Science. 200:748–757. https://doi.org/10.1016/j.procs.2022.01.273 Theissler A, Pérez-Velázquez J, Kettelgerdes M, Elger G. 2021. Predictive maintenance enabled by machine learning: Use cases and challenges in the automotive industry. Reliability Engineering & System Safety. 215:107864. https://doi.org/10.1016/j.ress.2021.107864 Traini E, Bruno G, D’antonio G, Lombardi F. 2019. Machine learning framework for predictive maintenance in milling. IFAC-PapersOnLine. 52(13):177–182. Uhlmann E, Polte J, Koutrakis N-S. 2021. Holistic Concept Towards a Reference Architecture Model for Predictive Maintenance. Procedia CIRP. 104:1430–1433. https://doi.org/10.1016/j.procir.2021.11.241 Vargas VM, Rosati R, Hervás-Martínez C, Mancini A, Romeo L, Gutiérrez PA. 2023. A hybrid feature learning approach based on convolutional kernels for ATM fault prediction using event-log data. Engineering Applications of Artificial Intelligence. Wang H, Luo W, Liu Z, Zhang J, Zuo M. 2024. Fourier Feature Refiner Network With Soft Thresholding for Machinery Fault Diagnosis Under Highly Noisy Conditions. IEEE Internet of Things Journal. 11(13):22880–22891. https://doi.org/10.1109/JIOT.2024.3363216 Wang L, Zhu Z, Zhao X. 2024. Dynamic predictive maintenance strategy for system remaining useful life prediction via deep learning ensemble method. Reliability Engineering & System Safety. 245:110012. https://doi.org/10.1016/j.ress.2024.110012 Wen H, Zhang L, Sinha JK. 2024. Early Prediction of Remaining Useful Life for Rolling Bearings Based on Envelope Spectral Indicator and Bayesian Filter. Applied Sciences. 14(1):436. https://doi.org/10.3390/app14010436 Wen Y, Rahman MF, Xu H, Tseng T-LB. 2022. Recent advances and trends of predictive maintenance from data-driven machine prognostics perspective. Measurement. 187:110276. Willard J, Jia X, Xu S, Steinbach M, Kumar V. 2020. Integrating physics-based modeling with machine learning: A survey. arXiv preprint arXiv:200304919. 1(1):1–34. Yin Y, Liu Z, Guo B, Zuo M. 2024. A Dual-Task Learning Approach for Bearing Anomaly Detection and State Evaluation of Safe Region. Chin J Mech Eng. 37(1):4. https://doi.org/10.1186/s10033-023-00978-3 You Y, Chen C, Hu F, Liu Y, Ji Z. 2022. Advances of Digital Twins for Predictive Maintenance. Procedia Computer Science. 200:1471–1480. https://doi.org/10.1016/j.procs.2022.01.348 Zhang W, Yang D, Wang H. 2019. Data-Driven Methods for Predictive Maintenance of Industrial Equipment: A Survey. IEEE Systems Journal. 13(3):2213–2227. https://doi.org/10.1109/JSYST.2019.2905565 Zonta T, da Costa CA, Zeiser FA, de Oliveira Ramos G, Kunst R, da Rosa Righi R. 2022. A predictive maintenance model for optimizing production schedule using deep neural networks. Journal of Manufacturing Systems. 62:450–462. https://doi.org/10.1016/j.jmsy.2021.12.013 Zonta T, Da Costa CA, Da Rosa Righi R, De Lima MJ, Da Trindade ES, Li GP. 2020. Predictive maintenance in the Industry 4.0: A systematic literature review. Computers & Industrial Engineering. 150:106889. https://doi.org/10.1016/j.cie.2020.106889 Cite Share Download PDF Status: Published Journal Publication published 18 Feb, 2026 Read the published version in The International Journal of Advanced Manufacturing Technology → Version 1 posted Reviewers agreed at journal 10 Nov, 2025 Reviewers invited by journal 18 Sep, 2025 Editor assigned by journal 18 Sep, 2025 First submitted to journal 15 Sep, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7618987","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":517102384,"identity":"b05fc7d6-dfbd-41f2-bd3c-dbe404136ff6","order_by":0,"name":"Abdelhafid Hamdi Alaoui","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Abdelhafid","middleName":"Hamdi","lastName":"Alaoui","suffix":""},{"id":517102386,"identity":"c2c2e3cf-7d21-48c1-bc8a-7ad8a929b676","order_by":1,"name":"Anwar Meddaoui","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABFklEQVRIie2QsUrDQBiA/9LRpllbhOQVLgSiBWleJSGQLgkKghQs5B5A94o+xJVA5oOAtwSz/nIOiujUQRCCgoLX/Urp5nDf8HMc/8fHHYDB8K/hkRrkZH8l3VuBeveifXvlfc6hcoYieXkdnLXOkV2/MVhMQ5sexB8aZfTU+OMGpBrvvj8g0p8s0wDhPomX3CpHugzmbExBxgyj4FAp6gCB/KX9CLjFdIqLefmtlILhrFPKQ8Fa0WGPFqGrKl8ahWBebSoRwWxT4RHhWaCUuse4VekqHp52E0qkd9Osz707kngMswv1FhGvaqs61igOpskjnUt3KGar5/XP1CWtqBAWl6Ejrkvc+tVEe9vfum8wGAyGHfwBPFZtwBRXCx4AAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0001-9308-1679","institution":"Hassan II University of Casablanca National School of Arts and Crafts: Universite Hassan II Casablanca Ecole Nationale Superieure d'Arts et Metiers","correspondingAuthor":true,"prefix":"","firstName":"Anwar","middleName":"","lastName":"Meddaoui","suffix":""},{"id":517102387,"identity":"f60be664-c038-424b-8272-066465f8125f","order_by":2,"name":"Mustapha Hain","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Mustapha","middleName":"","lastName":"Hain","suffix":""}],"badges":[],"createdAt":"2025-09-15 09:47:02","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7618987/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7618987/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00170-026-17531-w","type":"published","date":"2026-02-18T15:57:23+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":92516599,"identity":"6ff71e8d-a418-413b-97e9-d019c557ccef","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"xml","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7163,"visible":true,"origin":"","legend":"","description":"","filename":"jamtJAMTD2505315.xml","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/8973690f682cdf789adc023c.xml"},{"id":92516612,"identity":"7898ba44-8b56-4cec-89a0-b2783cb9f96e","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":967,"visible":true,"origin":"","legend":"","description":"","filename":"JAMTD2505315156239.go.xml","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/888f7608c3a40014fb8d3fa0.xml"},{"id":92518332,"identity":"83b29787-c030-4211-b4ea-3d6560a086f5","added_by":"auto","created_at":"2025-09-30 14:35:41","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":843,"visible":true,"origin":"","legend":"","description":"","filename":"JAMTD2505315Import.xml","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/d6515e1a776a5395bc9610e7.xml"},{"id":92517951,"identity":"604a1f63-908b-4f65-a4b6-6005b3be3ec4","added_by":"auto","created_at":"2025-09-30 14:27:41","extension":"xml","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":258254,"visible":true,"origin":"","legend":"","description":"","filename":"JAMTD25053150enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/bfa8817c49c5dcc5c55bb6b2.xml"},{"id":92516617,"identity":"14fd2ed4-c6fb-4290-a75a-37ed3caf5b5a","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":375179,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/d70a7fbf4ab816b5dda763a1.jpeg"},{"id":92516613,"identity":"d23a6b20-3e7d-4ffb-a61a-8251b5962096","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":121146,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/2673fd808b687e869953dc4b.jpeg"},{"id":92517949,"identity":"8e89295a-a9ee-41a2-a80a-5d6a27616982","added_by":"auto","created_at":"2025-09-30 14:27:41","extension":"jpeg","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":651208,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/e7fad39cbfcb1edaf92f57fa.jpeg"},{"id":92519542,"identity":"07f9f3b3-1a7f-4701-8bb2-7eb65809c753","added_by":"auto","created_at":"2025-09-30 14:43:41","extension":"jpeg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":440594,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/f3a87a9b621d80294e1e6500.jpeg"},{"id":92516606,"identity":"a45d7f6c-0ad4-4dc5-8198-ad8622b06729","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":582954,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/fa9ce9ee3e9bc0788cb0dddd.png"},{"id":92516611,"identity":"1f1da22e-4fcf-4824-896f-3b77cdb4613e","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"jpeg","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":484501,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/bfc805a8abaee4941e9a9c9a.jpeg"},{"id":92519541,"identity":"26b68f64-0fde-40da-abdc-6a0ba095fd97","added_by":"auto","created_at":"2025-09-30 14:43:41","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55109,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/5a3bd9a05467758414d9d8a0.png"},{"id":92517952,"identity":"da8594c6-406f-4259-a377-adf16d507744","added_by":"auto","created_at":"2025-09-30 14:27:41","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":98127,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/2975292177c5d9eed53095cb.png"},{"id":92516621,"identity":"444b4f40-f359-4ee8-8931-dbb14029d269","added_by":"auto","created_at":"2025-09-30 14:19:42","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":110455,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/0ac0dd71f35049e8add2c390.png"},{"id":92517955,"identity":"c55dd7d5-2e00-4b5a-9455-aea98196d5f7","added_by":"auto","created_at":"2025-09-30 14:27:41","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":107931,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/9b9d7be76d6ab2d25d35275c.png"},{"id":92516619,"identity":"671543f2-af15-46d5-bddf-2fe7c8749ecf","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":51953,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/bd5d4672a7d32d792255d8b0.png"},{"id":92518334,"identity":"a4c8cb79-4dc0-43de-86c9-9136d90d32a3","added_by":"auto","created_at":"2025-09-30 14:35:41","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":92727,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/5759b7a605fc95163590f71e.png"},{"id":92516622,"identity":"3b8d0703-7c00-46f9-91cb-47cf741ba7d8","added_by":"auto","created_at":"2025-09-30 14:19:42","extension":"xml","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":256875,"visible":true,"origin":"","legend":"","description":"","filename":"JAMTD25053150structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/04327d6531ab6fd800f548fd.xml"},{"id":92516620,"identity":"b2734d19-a143-4f9a-aa1e-a896bf4e52e2","added_by":"auto","created_at":"2025-09-30 14:19:42","extension":"html","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":263991,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/9e2807b97cfed98fe1d42796.html"},{"id":92516597,"identity":"7fb2135c-e064-4069-b9ed-c1c63b5b0949","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":164578,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eWorkflow of predictive, preventive, and corrective maintenance operations\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/f4d49d6b6cc59aad1c121e04.png"},{"id":92516596,"identity":"36b1709b-0320-4348-a189-c2b7b0c79a05","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":107319,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSearch string\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/759c6028764e4985a796f183.png"},{"id":92517944,"identity":"2303c808-4a96-4e73-855e-f4f7e22cc4cb","added_by":"auto","created_at":"2025-09-30 14:27:41","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":188386,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eScreening and selection across databases\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/59599a779248d34da897c5cc.png"},{"id":92517943,"identity":"2a5260d4-f3e2-43e2-a1a4-ff330b386f77","added_by":"auto","created_at":"2025-09-30 14:27:41","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":17830,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDistribution of publication by publisher\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/edf6a122eaed52e0916c0fcc.png"},{"id":92516604,"identity":"802ea55d-9fcf-4cc4-9dbf-022fc462d078","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":54265,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDistribution of publication by type (Conference paper or Journal article)\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/3d881aeca0ab1788b081e0af.png"},{"id":92517946,"identity":"77ab4292-4d12-43be-a364-ee5aaa7ca403","added_by":"auto","created_at":"2025-09-30 14:27:41","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":43143,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDistribution of publications by year\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/0a2c0c8da57551d20b29b04e.png"},{"id":92518331,"identity":"9d3fbd75-fbe8-4e5a-b896-95757e2645c5","added_by":"auto","created_at":"2025-09-30 14:35:41","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":465753,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eKeyword co-occurrence network\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/cb7a782d6f55bfe7b72f457d.png"},{"id":92516602,"identity":"cc5e3b23-666f-4588-8dbe-d0829665581a","added_by":"auto","created_at":"2025-09-30 14:19:41","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":497023,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eKeyword density map\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/3088aa01f09e71134c58ca5e.png"},{"id":92517947,"identity":"3a77395b-8a1d-4f93-93b6-dd53645f6f45","added_by":"auto","created_at":"2025-09-30 14:27:41","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":413939,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFocused network views of PdM, ML, and DL\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/722f070a3bfd1d6e9d62976c.png"},{"id":103251126,"identity":"59909643-e440-4a15-a212-5a977f8601c5","added_by":"auto","created_at":"2026-02-23 16:04:28","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3485070,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7618987/v1/5993605d-a916-40ab-beab-159d6f8dd886.pdf"}],"financialInterests":"","formattedTitle":"AI-Driven Predictive Maintenance for Industry 4.0: A Systematic Review of Models, Methods, and Challenges","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eThe rise of Industry 4.0 has significantly transformed production systems, operational management, and industrial maintenance strategies. This fourth industrial revolution, driven by the integration of advanced digital technologies such as the Internet of Things (IoT), Big Data, Artificial Intelligence (AI), and real-time connectivity requires companies to rethink their practices to meet increasing demands for performance, flexibility, and competitiveness (Heng et al. 2009)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAmong the critical functions in modern industry, maintenance holds a strategic position. Far from being limited to repairs or periodic interventions, it has evolved into a proactive and intelligent process. Predictive Maintenance (PdM), in particular, relies on sensor data, failure history, and operational conditions to anticipate breakdowns, estimate the Remaining Useful Life (RUL) of components, and optimize maintenance interventions (Susto et al. 2015)\u003c/p\u003e\n\u003cp\u003eIn recent years, there has been a surge of research focused on PdM, especially with the emergence of machine learning (ML) and deep learning (DL) techniques, which can model complex and nonlinear behaviors. Algorithms such as convolutional neural networks (CNN), recurrent neural networks (LSTMs), and more recently transformers, are increasingly applied to detect degradation patterns and predict failures from vibration, thermal, or acoustic signals (Zonta et al. 2020).\u003c/p\u003e\n\u003cp\u003eNevertheless, several technical, economic, and methodological challenges remain: managing massive and heterogeneous data, validating models in real-world industrial settings, the lack of standardization, and organizational resistance to change(Physics-informed machine learning | Nature Reviews Physics).\u003c/p\u003e\n\u003cp\u003eIn this context, this article aims to:\u003c/p\u003e\n\u003cul style=\"list-style-type: circle;\"\u003e\n \u003cli\u003eConduct a systematic review of major scientific contributions from 2018 to 2023;\u003c/li\u003e\n \u003cli\u003eClassify PdM approaches according to their theoretical foundations (physical, knowledge-based, data-driven);\u003c/li\u003e\n \u003cli\u003eIdentify major technical bottlenecks and research perspectives related to intelligent predictive maintenance\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThis critical and structured analysis aims to provide researchers, engineers, and decision-makers with a clear overview of current trends and key enablers for establishing PdM as a cornerstone of the factory of the future.\u003c/p\u003e"},{"header":"2.\tTheoretical Foundations and Terminology","content":"\u003cp\u003e\u003cstrong\u003ea. Types of Maintenance\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIndustrial maintenance can be grouped by timing and approach. Preventive maintenance, which includes systematic maintenance (SM), condition-based maintenance (CBM), and predictive maintenance (PdM), aims to avoid failures, while corrective maintenance addresses issues after breakdowns occur. Corrective maintenance is triggered after a failure has occurred. It is divided into curative (complete failure) and troubleshooting (partial failure) (Smith and Hinchcliffe 2003).\u003c/p\u003e\n\u003cp\u003eSystematic maintenance is based on fixed time or usage intervals, carried out before any observable degradation signs appear.\u003c/p\u003e\n\u003cp\u003eCondition-Based Maintenance (CBM) monitors real-time performance indicators (e.g., temperature, vibration) without predictive modeling(Mobley 2002).\u003c/p\u003e\n\u003cp\u003ePredictive Maintenance (PdM), on the other hand, anticipates failure by analyzing historical and real-time data using advanced techniques such as machine learning and physical modeling (Heng et al. 2009; Lei et al. 2018).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea. Core and Advanced Concepts:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePredictive Maintenance (PdM) is part of a broader paradigm known as Prognostics and Health Management (PHM), which aims to assess the health status of a system and predict its future failures (Heng et al. 2009). A central concept in PHM is the Remaining Useful Life (RUL), defined as the expected time remaining before a system or component reaches a failure threshold (Remaining useful life estimation -A review on the statistical data driven approaches 2011).\u003c/p\u003e\n\u003cp\u003ePdM is a strategic discipline within industrial asset management that aims to anticipate equipment failures by analyzing historical and real-time data. As described in the figure 1, key distinction from Condition-Based Maintenance (CBM) lies in PdM\u0026rsquo;s ability to predict future degradation trends rather than solely reacting to current conditions. The modeling techniques used in PdM can be broadly categorized into four families: physics-based, knowledge-based, data-driven, and hybrid models.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003ePhysics-Based Models\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePhysics-based models are grounded in the first principles of physics and engineering, such as mechanics, thermodynamics, and material science. These models simulate degradation mechanisms including fatigue, crack growth, corrosion, and wear by employing mathematical formulations that reflect real-world physical behavior. This category is central to the Physics-of-Failure (PoF) methodology, which is extensively applied in structural health monitoring and component reliability assessment (Prognostics and Health Management of Electronics).\u003c/p\u003e\n\u003cp\u003eSuch models require minimal historical data and offer high interpretability. However, they are limited by their dependence on domain-specific expertise and the complexity of accurately modeling real-world systems(Intelligent Fault Diagnosis and Prognosis for Engineering Systems | Wiley Online Books)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eKnowledge-Based Models\u003c/em\u003e\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eKnowledge-based models leverage human expertise and logical reasoning to evaluate equipment condition. These models often rely on rule-based systems or expert systems consisting of a knowledge base and inference engine. Rules are defined explicitly, allowing the system to mimic expert decision-making (Chen and Chen 2011).\u003c/p\u003e\n\u003cp\u003eThese models are valuable when data are limited but expert knowledge is abundant. However, they may struggle with scalability and adaptability in high-dimensional or rapidly evolving environments(Jardine et al. 2006).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;Data-Driven Models\u003c/em\u003e\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eData-driven approaches utilize statistical learning and artificial intelligence particularly machine learning (ML) and deep learning (DL) to uncover degradation patterns and predict Remaining Useful Life (RUL). These models rely on large datasets from sensors, logs, and monitoring systems, and they are well-suited to environments aligned with Industry 4.0 paradigms (Susto et al. 2015).\u003c/p\u003e\n\u003cp\u003eData-driven models are capable of learning complex, non-linear relationships between multivariate inputs and output targets. Nonetheless, their performance hinges on the quality, quantity, and labeling of available data. Moreover, many of these models function as black boxes, making their predictions difficult to interpret in safety-critical systems (Zhang et al. 2019).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;Hybrid and Physics-Informed Models\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHybrid models aim to combine the strengths of physics-based and data-driven approaches. A prominent class of these models is Physics-Informed Machine Learning (PIML), which integrates physical laws directly into machine learning architectures, such as neural networks or Gaussian processes. These methods enable learning under physical constraints, thereby enhancing model robustness, generalizability, and interpretability(Willard et al. 2020).\u003c/p\u003e\n\u003cp\u003eHybrid models are particularly effective when datasets are sparse or noisy and when model transparency is essential. However, they are technically complex to design and deploy, requiring expertise in both physical modeling and AI (Zonta et al. 2020; Physics-informed machine learning | Nature Reviews Physics).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb. Maintenance Evolution in the Industry 4.0 Paradigm\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWith the rise of Industry 4.0, maintenance has embarked on a digital transformation journey. This evolution can be described using the maintenance 1.0 to maintenance 4.0 framework (see table 1). Transitioning to Maintenance 4.0 involves integrating smart sensors, cloud-based platforms, self-learning algorithms, and interactive visualization interfaces (Pech et al., 2021). These technologies transform data into actionable decisions, enabling connected, dynamic, and cost-effective maintenance strategies.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eTable 1. Maintenance evolution\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGeneration\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMain focus\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEnabling Technology\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eMaintenance 1.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eCorrective\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eNone\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eMaintenance 2.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003ePreventive\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eTimers, visual inspection\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eMaintenance 3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eCondition-Based\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eSensors, SCADA, PLCs\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eMaintenance 4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003ePredictive \u0026amp; Prescriptive\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 201px;\"\u003e\n \u003cp\u003eAI, IoT, Cloud, Digital Twins\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"3. Related work","content":"\u003cp\u003eIn recent years, predictive maintenance (PdM) has emerged as a critical enabler of intelligent asset management within Industry 4.0 environments. The proliferation of sensor networks, the Internet of Things (IoT), and artificial intelligence (AI) techniques has significantly advanced PdM capabilities beyond traditional condition-based maintenance approaches.\u003c/p\u003e\u003cp\u003eA comprehensive and widely cited systematic literature review by (Zonta et al. \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) provides a structured mapping of the scientific landscape surrounding PdM in the context of Industry 4.0. Their study, which analyzed over 100 articles, identifies the main technological pillars of PdM as cyber-physical systems (CPS), big data analytics, machine learning, and cloud computing. The authors emphasize that AI particularly machine learning (ML) plays a pivotal role in enabling automated fault diagnosis and Remaining Useful Life (RUL) estimation through the analysis of multivariate time series data.\u003c/p\u003e\u003cp\u003eFurthermore, Zonta et al. categorize PdM solutions into three main architectural layers: data acquisition, data processing, and decision-making. In their review, it is noted that a majority of PdM systems remain focused on the operational layer, with limited integration between data-driven insights and prescriptive maintenance actions. The authors highlight several key research challenges, including data quality issues, model generalizability across equipment types, lack of standardization in architectures, and the need for explainable AI to enhance trust and transparency in industrial settings.\u003c/p\u003e\u003cp\u003eThis review also stresses the importance of combining PdM with cybersecurity, particularly in distributed IoT environments, and calls for further research in integrating edge computing and digital twins to support real-time, scalable, and interpretable maintenance strategies. Their findings form a foundational reference point for subsequent research efforts aimed at advancing PdM solutions through hybrid modeling approaches and AI integration.\u003c/p\u003e"},{"header":"4. Matériel \u0026 Method","content":"\u003cp\u003eThis section outlines the systematic review methodology adopted to identify, select, and synthesize peer-reviewed literature on predictive maintenance (PdM) in the context of Industry 4.0. The approach was structured according to the PRISMA framework (Moher et al. 2009), focusing on targeted and reproducible search, selection, and filtering processes within four major academic databases.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;a. \u003cstrong\u003eResearch Questions\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe review was guided by structured research questions designed to explore technological, methodological, and industrial aspects of PdM. These questions, summarized in Table 2, served as a thematic framework for selecting and analyzing the literature.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eTable 2. Research Questions\u003c/strong\u003e\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"576\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eResearch Questions\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 472px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eResearch Question\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eRQ 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 472px;\"\u003e\n \u003cp\u003eWhat are the main channels and trends in the dissemination of research related to Predictive Maintenance (PdM) in Industry 4.0?\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eRQ2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 472px;\"\u003e\n \u003cp\u003eWhat are the dominant approaches used in predictive maintenance in Industry 4.0?\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eRQ3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 472px;\"\u003e\n \u003cp\u003eWhat types of AI algorithms are most frequently used in PdM?\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eRQ4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 472px;\"\u003e\n \u003cp\u003eHow do different models and architectures in PdM target specific industrial applications, and which variables are most critical in these applications?\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eb. Research Strategy\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe search strategy adopted in this review follows a two-step process inspired by Zonta et al. (2020), combining an initial exploratory search and a refined database-specific search.\u003c/p\u003e\n\u003cp\u003eWe initially selected Google Scholar as a starting point because it allows free-text queries and returns a high number of potentially relevant publications by searching across full texts, titles, and abstracts. This exploratory phase enabled us to identify relevant terminology, refine our search string, and detect early patterns of duplication across databases.\u003c/p\u003e\n\u003cp\u003eHowever, since many of the documents retrieved via Google Scholar are already indexed in specialized scientific databases such as \u003cem\u003eIEEE Xplore, Elsevier, and Springer-Link\u003c/em\u003e, this research was limited to the systematic selection of articles to these three scientific publishers, which provide peer-reviewed content with traceable metadata and reliable indexing. The initial research phase involved defining the scope and objectives of the review (figure 2), focusing on identifying and analyzing scientific contributions related to :\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003ePredictive maintenance methods in the context of Industry 4.0\u003c/li\u003e\n \u003cli\u003eApplications of artificial intelligence, machine learning, and deep learning in PdM\u003c/li\u003e\n \u003cli\u003eReal-time implementation challenges and RUL estimation models\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThe literature search targeted peer-reviewed publications from the beginning of 2019 to the end of 2024, ensuring the inclusion of the most recent and relevant studies. The following databases were selected due to their relevance in engineering and computer science:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\u003cem\u003eIEEE Xplore\u003c/em\u003e\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eScience-Direct (Elsevier)\u003c/em\u003e\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eSpringer-Link\u003c/em\u003e\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eGoogle Scholar (for complementary coverage)\u003c/em\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSearch Terms and Query Design\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe search queries were constructed using combinations of keywords and Boolean operators. The final query string was adapted to each database\u0026rsquo;s syntax and included terms such as:\u003c/p\u003e\n\u003cp\u003eThis query aimed to ensure the inclusion of articles addressing both methodological aspects and industrial applications of PdM, particularly those involving condition monitoring systems, prognostics, and AI-based modeling.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eInclusion and Exclusion Criteria\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo ensure quality and consistency in the selected studies, the following inclusion criteria were applied (see table 3) :\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eArticles published in English\u003c/li\u003e\n \u003cli\u003ePeer-reviewed journal and conference papers\u003c/li\u003e\n \u003cli\u003ePublications between 2019 and 2024\u003c/li\u003e\n \u003cli\u003eExplicit focus on predictive maintenance and AI-based diagnostics\u003c/li\u003e\n \u003cli\u003eInclusion of practical use cases, algorithms, or architectures\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eTable 3. Quality Filtering Criteria for Industry 4.0 and Predictive Maintenance Publications\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"614\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCriteria\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eType\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 236px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIncluded\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 217px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eExcluded if...\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003ePublication Period\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 236px;\"\u003e\n \u003cp\u003ePublished between 2019 and 2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 217px;\"\u003e\n \u003cp\u003ePublished before 2019\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e2\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003eLanguage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 236px;\"\u003e\n \u003cp\u003eArticle is written in English\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 217px;\"\u003e\n \u003cp\u003eArticle is written in another language\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e3\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003ePublication Type\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 236px;\"\u003e\n \u003cp\u003ePeer-reviewed journal or conference paper\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 217px;\"\u003e\n \u003cp\u003ePatent, thesis, dissertation, book chapter, or non-peer-reviewed material\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003eContent Quality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 236px;\"\u003e\n \u003cp\u003eIncludes technical contributions, such as algorithms, architectures, or case studies\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 217px;\"\u003e\n \u003cp\u003eLacks experimental methods, clear results, or methodological rigor\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003eContent Quality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 236px;\"\u003e\n \u003cp\u003ePresenting a clear abstract, full text, and a detailed methodology.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 217px;\"\u003e\n \u003cp\u003ePublications without abstract, full text, or clear methodology.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 91px;\"\u003e\n \u003cp\u003eFocus Area\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 236px;\"\u003e\n \u003cp\u003eDedicated to predictive maintenance and AI-based diagnostics in Industry 4.0.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 217px;\"\u003e\n \u003cp\u003eFocused only on preventive or corrective maintenance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eThe exclusion criteria eliminated:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003ePatents, theses, and dissertations\u003c/li\u003e\n \u003cli\u003eArticles focused solely on preventive or corrective maintenance\u003c/li\u003e\n \u003cli\u003eStudies lacking technical or experimental contributions\u003c/li\u003e\n \u003cli\u003ePublications without abstract, full text, or clear methodology\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSelection and Filtering Process\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs described in the figure 3, the selection process consisted of four stages:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eInitial retrieval of publications based on keyword queries (approx. 3,907 results);\u003c/li\u003e\n \u003cli\u003eDuplicate removal using citation management software;\u003c/li\u003e\n \u003cli\u003eAbstract and title screening to assess relevance;\u003c/li\u003e\n \u003cli\u003eFull-text review to apply inclusion/exclusion criteria.\u003c/li\u003e\n\u003c/ul\u003e"},{"header":"5. Results and discussions","content":"\u003cp\u003eIn this section, we present the results and discussion based on the research questions previously defined, with the objective of answering the main questions guiding this systematic review\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRQ1 \u0026ndash; What are the main channels and trends in the dissemination of research related to Predictive Maintenance (PdM) in Industry 4.0?\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe dissemination of research on Predictive Maintenance (PdM) within Industry 4.0 is primarily channeled through peer-reviewed journals, with a significantly smaller proportion presented at conferences. As indicated in Figure 5, journal articles account for 86 % of the publications, while conference proceedings constitute only 14%. This distribution suggests that PdM has evolved into a mature research area, where detailed methodological studies and rigorous validation are preferred over preliminary conference communications.\u003c/p\u003e\n\u003cp\u003eRegarding publishers, figure 4 and table 4 highlight Elsevier\u0026rsquo;s dominant role, with 42 publications, far surpassing Springer (9 publications), IEEE (10 publications), and other publishers (3 publication). This concentration reflects Elsevier\u0026rsquo;s strong presence in fields such as industrial engineering, computer science, and reliability engineering, which are closely aligned with PdM research.\u003c/p\u003e\n\u003cp\u003eThe temporal analysis shown in Figure 6 reveals a clear upward trend in publications from 2019 to 2024, with a particularly notable increase from 5 publications in 2019 to 18 publications in 2024. This growth reflects both the technological advances in AI and Industry 4.0, and the heightened industrial interest in deploying PdM solutions for cost reduction and operational efficiency.\u003c/p\u003e\n\u003cp\u003eOverall, the data suggest that scientific journals, particularly those published by Elsevier, serve as the principal dissemination channels for PdM research, supporting the development and standardization of methodologies. Conferences, while less prevalent, still play a crucial role in introducing novel concepts and fostering collaboration within the PdM community.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRQ2\u0026ndash;What are the dominant approaches used in predictive maintenance in Industry 4.0?\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePredictive maintenance in Industry 4.0 has experienced significant methodological evolution in recent years. As reported by Zhang et al. (2021), data-driven approaches, particularly those leveraging deep learning, have become increasingly prominent due to their capacity to identify complex patterns in large-scale sensor data. This is especially true in the context of remaining useful life (RUL) prediction, where architectures like BiGRU and CNN-LSTM have shown strong performance. As summarized in Table 5, the majority of reviewed studies adopt data-driven strategies, reflecting their widespread applicability across industrial domains.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cstrong\u003eTable 4. Selected Articles sorted by year\u003c/strong\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"634\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003eNo.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eArticle\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eType\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eYear\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePublisher\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eName\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eFranciosi et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e17th IFAC Symposium on Information Control Problems in Manufacturing INCOM 2021\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eKans et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e4th IFAC Workshop on Advanced Maintenance Engineering, Services and Technologies - AMEST 2020\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eCrespo M\u0026aacute;rquez et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eComputers in Industry\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eMa et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eMechanical Systems and Signal Processing\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eChowdhury et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eComputers \u0026amp; Industrial Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eNikolakis et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e30th International Conference on Flexible Automation and Intelligent Manufacturing (FAIM2021)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eYou et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e3rd International Conference on Industry 4.0 and Smart Manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eRagab et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eNeurocomputing\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eCalabrese et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e25th International Conference on Production Research Manufacturing Innovation: Cyber Physical Manufacturing August 9-14, 2019 | Chicago, Illinois (USA)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eTraini et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e9th IFAC Conference on Manufacturing Modelling, Management and Control MIM 2019\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eLv et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eAdvanced Engineering Informatics\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eJoseph et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e5th IFAC Workshop on Advanced Maintenance Engineering, Services and Technologies AMEST 2022\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eTaşcı et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eComputers \u0026amp; Industrial Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eVargas et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eEngineering Applications of Artificial Intelligence\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eGupta et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eComputers \u0026amp; Industrial Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eTessoni et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e3rd International Conference on Industry 4.0 and Smart Manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eDaniyan et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eLearning Factories across the value chain \u0026ndash; from innovation to service \u0026ndash; The 10th Conference on Learning Factories 2020\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eUhlmann et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e54th CIRP CMS 2021 - Towards Digitalized Manufacturing 4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eRihi et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eKnowledge-Based and Intelligent Information \u0026amp; Engineering Systems: Proceedings of the 26th International Conference KES2022\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eNentwich et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e54th CIRP CMS 2021 - Towards Digitalized Manufacturing 4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eBilal Yıldız et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eAdvanced Engineering Informatics\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eLee et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e26th CIRP Conference on Life Cycle Engineering (LCE) Purdue University, West Lafayette, IN, USA May 7-9, 2019\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eKilleen et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eThe 10th International Conference on Ambient Systems, Networks and Technologies (ANT 2019)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eLuo et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eRobotics and Computer-Integrated Manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eFeng et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eJournal of Manufacturing Systems\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eCao et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eRobotics and Computer-Integrated Manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eCoelho et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e3rd International Conference on Industry 4.0 and Smart Manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eChinta et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eAdvanced Engineering Informatics\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eJiang et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eComputers \u0026amp; Industrial Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eFarahani et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eJournal of Manufacturing Processes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eDrakaki et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eProceedings of the 2nd International Conference on Industry 4.0 and Smart Manufacturing (ISM 2020)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e32\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eJain et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e4th IFAC Workshop on Advanced Maintenance Engineering, Services and Technologies - AMEST 2020\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eHuynh et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eReliability Engineering \u0026amp; System Safety\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eZonta et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eJournal of Manufacturing Systems\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eWen et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eMeasurement\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eEinabadi et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e9th IFAC Conference on Manufacturing Modelling, Management and Control MIM 2019\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eShaheen et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eEngineering Applications of Artificial Intelligence\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eHillenbrand et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIOP Publishing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eIOP Conference Series: Materials Science and Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e39\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eAugustyn et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eActa Mechanica et Automatica\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eLv et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eELSEVIER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eIET Conference Proceedings\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eRajasekar et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e2023 15th International Conference on Developments in eSystems Engineering (DeSE)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eSingha et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e2020 IEEE 17th India Council International Conference (INDICON)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eKotsiopoulos et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eThe International Journal of Advanced Manufacturing Technology\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eFordal et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eAdvances in Manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eVic\u0026ecirc;ncio et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eIndustrial IoT Technologies and Applications\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eWang et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eElsevier\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eReliability Engineering \u0026amp; System Safety\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eHerzog et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e28th International Conference on Methods and Models in Automation and Robotics (MMAR)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eKolvig-Raun et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eIEEE Robotics and Automation Letters\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eMardianto et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e8th Int. Conf. on Information Technology, Information Systems and Electrical Engineering (ICITISEE)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003ePatel et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE/ACM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e17th International Conference on Utility and Cloud Computing (UCC)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eGougam et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eJournal of the Brazilian Society of Mechanical Sciences and Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eKondo et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eElsevier\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eComputers \u0026amp; Industrial Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eMeddaoui et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eThe International Journal of Advanced Manufacturing Technology\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eMeddaoui et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eThe International Journal of Advanced Manufacturing Technology\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e55\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eYuhua Yin, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eChinese Journal of Mechanical Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eJavad Isavand, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eElsevier\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eMeasurement\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eAnouar et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConf\u0026eacute;rence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e21st Int. Multi-Conference on Systems, Signals \u0026amp; Devices (SSD)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eUgur Ileri, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eMDPI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eApplied Sciences\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eRengaraj R, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConf\u0026eacute;rence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e2nd Int. Conference on Networking and Communications (ICNWC)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eHaobin Wen, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eMDPI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eApplied Sciences\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eZekai Si, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eArabian Journal for Science and Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003ePatrick Seebold, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eConf\u0026eacute;rence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003e3rd Int. Conference on Computing and Machine Intelligence (ICMI)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eHuan Wang et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eIEEE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eIEEE Internet of Things Journal\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 44px;\"\u003e\n \u003cp\u003e64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003ePooja KamatN et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eJournal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 60px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 90px;\"\u003e\n \u003cp\u003eSpringer\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 266px;\"\u003e\n \u003cp\u003eJournal of the Brazilian Society of Mechanical Sciences and Engineering\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eHowever, Wang et al. (2021) noted that purely data-driven methods often face challenges related to interpretability and robustness in practical settings. To address these limitations, hybrid approaches have emerged, combining physical modeling with machine learning techniques. Notable examples include the hybrid digital twin frameworks proposed by Yu et al. (2022), which integrate physical simulations with data-driven analysis to enhance both accuracy and model transparency. As shown in Table 5, several studies, such as those by Calabrese et al. (2019) and Zonta et al. (2022), reflect this growing trend towards hybrid solutions.\u003c/p\u003e\n\u003cp\u003eIn specific contexts, physical model-based methods remain relevant, particularly in industries requiring high levels of interpretability and compliance with engineering standards. For instance, Ouafi et al. (2020), listed in Table 5, utilized gamma process models for multi-component systems, emphasizing the value of mathematical rigor and domain knowledge. Overall, the current literature compiled indicates that while data-driven techniques dominate research in predictive maintenance for Industry 4.0, hybrid methods are increasingly seen as essential for bridging the gap between predictive performance and industrial feasibility. Meanwhile, physical and statistical models continue to hold value in domains where transparency and explainability are crucial.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eTable 5. Articles by approach used\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"614\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 161px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrediction Classification\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 454px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eArticles\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 161px;\"\u003e\n \u003cp\u003eData-Driven\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 454px;\"\u003e\n \u003cp\u003e\u0026nbsp; (Traini et al. 2019; Einabadi et al. 2019; Daniyan et al. 2020; Nikolakis et al. 2020; Singha et al. 2020; She and Jia 2021a; Kotsiopoulos et al. 2021; Hillenbrand et al. 2021; Ragab et al. 2021; Joseph et al. 2022a; Wen et al. 2022; Rihi et al. 2022; Tessoni and Amoretti 2022; Chowdhury et al. 2022; Jiang et al. 2022; Jiang et al. 2022; Gupta et al. 2023a; Shaheen et al. 2023; Lv et al. 2023; Fordal et al. 2023; Chinta et al. 2023; Augustyn and Fidali 2023; Meddaoui et al. 2023; Ileri et al. 2024; Isavand et al. 2024; Gougam et al. 2024; BOUROKBA et al. 2024; Kolvig-Raun et al. 2024; Kolvig-Raun et al. 2024; Seebold et al. 2024; Kamat et al. 2024; Meddaoui et al. 2024; L. Wang et al. 2024; Herzog and Bartecki 2024; Mardianto and Mauritsius 2024; Patel et al. 2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 161px;\"\u003e\n \u003cp\u003ePhysical model -based\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 454px;\"\u003e\n \u003cp\u003eHuynh et al. (Huynh et al. 2022)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 161px;\"\u003e\n \u003cp\u003eHybrid\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 454px;\"\u003e\n \u003cp\u003e(Calabrese et al. 2019; Einabadi et al. 2019; Killeen et al. 2019; Luo et al. 2020; Jain et al. 2020; Kans et al. 2020; Sfar et al. 2020; Drakaki et al. 2021; Franciosi et al. 2021; Nentwich and Reinhart 2021; Uhlmann et al. 2021; Theissler et al. 2021; Coelho et al. 2022; You et al. 2022; Zonta et al. 2022; Cao et al. 2022; Farahani et al. 2022; Vargas et al. 2023; Polenghi et al. 2023; Bilal Yıldız and Soylu 2023; Crespo M\u0026aacute;rquez et al. 2023; Feng et al. 2023; Ma et al. 2023; Wen et al. 2024; Yin et al. 2024; R et al. 2024; H. Wang et al. 2024; H. Wang et al. 2024; Kondo et al. 2024; Si et al. 2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eRQ3 \u0026ndash; What types of AI algorithms are most frequently used in PdM?\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAn analysis of the selected research articles shows that a wide spectrum of artificial intelligence (AI) algorithms is employed in predictive maintenance (PdM) for Industry 4.0 applications. As summarized in \u003cstrong\u003eTable 6\u003c/strong\u003e, the dominant approaches can be broadly grouped into deep learning (DL), classical machine learning (ML), artificial neural networks (ANN), hybrid combinations, and architecture-level solutions integrating IoT, digital twins (DT), and cloud computing.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDeep learning techniques have emerged as the most frequently adopted methods in PdM studies. Authors such as (She and Jia 2021b), (Kotsiopoulos et al. 2021), and (Zonta et al. 2022) have demonstrated the effectiveness of DL architectures including BiGRU, CNN, and LSTM in capturing complex time dependencies and multi-dimensional patterns in sensor data. These models are particularly prominent in remaining useful life (RUL) prediction tasks, reflecting their strength in time-series modeling. Alongside DL, classical machine learning algorithms remain extensively used, especially for classification tasks and anomaly detection. Techniques such as Random Forest (RF), support vector machines (SVM), and ensemble methods are frequently cited, as seen in studies by Traini et al. (2019), Lee et al. (2019), and Gupta et al. (2023). These methods are valued for their lower computational demands and greater interpretability compared to deep learning models.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eTable 6. AI and algorithms used for PdM\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAlgorithmes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 425px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIdentifiers\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003eDEEP LEARNING\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 425px;\"\u003e\n \u003cp\u003e(She and Jia 2021b; Kotsiopoulos et al. 2021; Ragab et al. 2021; Joseph et al. 2022b; Zonta et al. 2022; Chowdhury et al. 2022; Jiang et al. 2022; Chinta et al. 2023; Ileri et al. 2024; Ileri et al. 2024; BOUROKBA et al. 2024; Seebold et al. 2024; Kamat et al. 2024; H. Wang et al. 2024; Mardianto and Mauritsius 2024; Si et al. 2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003eANN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 425px;\"\u003e\n \u003cp\u003e(Einabadi et al. 2019; Daniyan et al. 2020; Shaheen et al. 2023; Lv et al. 2023; Meddaoui et al. 2023)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003eML (RF\u0026hellip;)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 425px;\"\u003e\n \u003cp\u003e(Traini et al. 2019; Lee et al. 2019; Jain et al. 2020; Nikolakis et al. 2020; Singha et al. 2020; Hillenbrand et al. 2021; Theissler et al. 2021; Gupta et al. 2023b; Augustyn and Fidali 2023; Taşcı et al. 2023; Isavand et al. 2024; R et al. 2024; Herzog and Bartecki 2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003eStatistical +ML +DL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 425px;\"\u003e\n \u003cp\u003e(Sfar et al. 2020; Wen et al. 2022; Coelho et al. 2022; Tessoni and Amoretti 2022; Huynh et al. 2022; Wen et al. 2024; Wen et al. 2024; Yin et al. 2024; Gougam et al. 2024; Kolvig-Raun et al. 2024; Meddaoui et al. 2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003eDL +ML\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 425px;\"\u003e\n \u003cp\u003eRihi et al (L. Wang et al. 2024; Patel et al. 2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 170px;\"\u003e\n \u003cp\u003eIoT Architecte + DT + Cloud computing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 425px;\"\u003e\n \u003cp\u003e(Killeen et al. 2019; Luo et al. 2020; Farahani et al. 2022; Feng et al. 2023; Fordal et al. 2023; Kondo et al. 2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eArtificial neural networks (ANN), although sometimes grouped under ML, are reported separately in several studies due to their distinct architecture. Einabadi et al. (2019), Daniyan et al. (2020), and Shaheen et al. (2023) utilized ANN-based models for fault diagnosis and health state classification, highlighting their simplicity and efficiency for smaller datasets.\u003c/p\u003e\n\u003cp\u003eA notable trend in the literature is the development of hybrid models that combine statistical techniques, ML, and DL. Works by Sfar et al. (2020), Wen et al. (2022), and Huynh et al. (2022) exemplify this approach, aiming to leverage the strengths of different methods to improve prediction accuracy and model robustness. Hybrid methods are particularly relevant for bridging the gap between purely data-driven approaches and the need for physical interpretability.\u003c/p\u003e\n\u003cp\u003eBeyond pure algorithmic strategies, some studies propose architecture-level solutions integrating IoT, digital twins (DT), and cloud computing for predictive maintenance. As shown in \u003cstrong\u003eTable 6\u003c/strong\u003e, Killeen et al. (2019), Luo et al. (2020), and Farahani et al. (2022) explored frameworks that combine advanced sensing networks, digital replicas of assets, and scalable cloud platforms. Such architectures are designed to enable real-time analytics, remote monitoring, and scalable deployment across industrial systems.\u003c/p\u003e\n\u003cp\u003eOverall, the data consolidated in \u003cstrong\u003eTable 6\u003c/strong\u003e confirms that while deep learning dominates current PdM research, machine learning, ANN, and hybrid solutions continue to play essential roles. The choice of algorithm often depends on the specific industrial context, data availability, and the trade-off between predictive accuracy and model interpretability.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRQ4_How do different models and architectures in PdM target specific industrial applications, and which variables are most critical in these applications?\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAn analysis of the selected research articles shows that PdM models and architectures are strongly shaped by specific industrial applications and the variables critical to each context. As summarized in Table 7, manufacturing studies often use deep learning methods like BiGRU or CNN for tool wear and machine health prediction, relying on variables such as vibration, cutting forces, and spindle current (e.g., Zhang et al., 2021; Fordal et al., 2023).\u003c/p\u003e\n\u003cp\u003eIn the automotive sector, machine learning models analyze fleet data, focusing on variables like temperature, vibration, and diagnostic codes (Theissler et al., 2021). Mining applications require robust models like CNNs due to harsh conditions, using variables such as vibration and motor currents (Rojas et al., 2022).\u003c/p\u003e\n\u003cp\u003eHybrid approaches, as seen in Yu et al. (2022), combine physical models and data-driven analytics to enhance prediction accuracy and explainability, leveraging both physical parameters and statistical features. Architectural solutions integrating IoT, digital twins (DT), and cloud computing enable broader data collection and real-time analysis, adapting to varied industrial needs (Killeen et al., 2019; Luo et al., 2020).\u003c/p\u003e\n\u003cp\u003eOverall, variable selection in PdM reflects both the physical characteristics of machinery and the operational environment, confirming that effective PdM requires models tailored to specific industrial contexts.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eTable 7. models, application and architecture presentation\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"593\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAuthors\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTitle\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eYear\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCase\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eFranciosi et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA maintenance scheduling optimization model for a multi-component machine in a digitalized manufacturing context\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eCritical Multi-component Welding Machine in manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eDowntime periods, Component health states, Failure predictions, Maintenance costs\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eKans et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA remote laboratory for Maintenance 4.0 training and education\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eMachinery Fault Simulator (Rotating Machinery) for training and education\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals, simulated fault data (unbalance, misalignment, resonance, bearing faults, cracked shaft, gearbox faults)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eCrespo M\u0026aacute;rquez et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eSimulating dynamic RUL based CBM scheduling. A case study in the railway sector\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003e\u0026quot;Railway Sector - Fleet of Trains (Rolling Stock)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003e\u0026quot;Detected anomalies, Estimated RUL, Risk levels, Workshop CBM capacity, Stock and flow variables:\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eMa et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA hybrid-driven probabilistic state space model for tool wear monitoring\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eCNC Milling of stainless steel using ball nose cutter\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eCutting forces (Fx, Fy, Fz), Acoustic Emission (AE), Vibration (Vx, Vy, Vz)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eChowdhury et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eInternet of Things resource monitoring through proactive fault prediction\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eIoT Resources (Multiple domains: AWS IoT, Network Faults, Machine Element Faults)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eSensor readings: vibration, force, network logs, usage metrics (depending on dataset)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eNikolakis et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA microservice architecture for predictive analytics in manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eRobotic manipulator (robot box) for belt tension monitoring\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eElectric current, Position signals (robot joints), Statistical features (mean, RMS, min, max, skewness, kurtosis)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eYou et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eAdvances of Digital Twins for Predictive Maintenance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eVarious industrial equipment (gearbox, bearings, milling machine, turbines, satellites, etc.)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVaries by case: vibration, acoustic emission, stress, force, temperature, geometry, multi-physics simulation outputs\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eRagab et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eAttention-based sequence to sequence model for machine remaining useful life prediction\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eTurbofan Engines (C-MAPSS dataset)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eTemperature, Pressure, Speed, Time-series sensor data\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eCalabrese et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003ePrognostic Health Management of Production Systems. New Proposed Approach and Experimental Evidences\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eRotating Component in Packaging Machine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eCurrent signals (mA) from actuator, Time-domain features (RMS, Kurtosis, Mean, Skewness, Variance, Crest Factor)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eTraini et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eMachine Learning Framework for Predictive Maintenance in Milling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eMilling operations on Matsuura MC-510V CNC machine (NASA milling dataset)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals, Acoustic Emission (AE), Spindle motor current, Cutting parameters (speed, depth of cut, feed), Flank wear measurements\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eLv et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003ePredictive maintenance decision-making for variable faults with non-equivalent costs of fault severities\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eRailcar Wheel Bearings\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eBearing temperature data (historical), Threshold-based states (Healthy, Degradation, Failure), Statistical features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eJoseph et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA Predictive Maintenance Application for A Robot Cell using LSTM Model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eRobot Cell (Gluing Station) in Automotive Industry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eTemperature, Doser volume, Torque (Nm), Pressure (bar), Alarm logs\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eTaşcı et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eRemaining useful lifetime prediction for predictive maintenance in manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eConsumer Goods Manufacturing Assembly Line\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eSensor readings: Weight, Speed, Temperature, Electric current, Vacuum, Air pressure, Statistical features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eVargas et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA hybrid feature learning approach based on convolutional kernels for ATM fault prediction using event-log data\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eATM machines (68 machines over 2 years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eEvent-log commands and responses, Count of events per 10-minute interval (e.g. PrepareWithdrawal, Initialize, CloseShutter, etc.)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eGupta et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003ePredictive maintenance of baggage handling conveyors using IoT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eAirport Baggage Handling Conveyors (S-Lifts)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals, Acoustic signals, RMS, Kurtosis, Skewness, Frequency-domain features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eTessoni et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eAdvanced statistical and machine learning methods for multi-step multivariate time series forecasting in predictive maintenance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eMultiple public datasets: FRED, Air Quality, Appliances Energy, Beijing PM2.5, Gas Turbine Emission\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eTime-series data (varies by dataset): e.g. sensor measurements, energy usage, pollution levels, gas emissions\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eDaniyan et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eArtificial intelligence for predictive maintenance in the railcar learning factories\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eRailcar Wheel Bearings\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eBearing temperature data (historical), Threshold-based states (Healthy, Degradation, Failure), Statistical features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eUhlmann et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eHolistic Concept Towards a Reference Architecture Model for Predictive Maintenance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eAxis test rig (ball screw spindle)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals (x, y, z axes), Variance, Time-series features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eRihi et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003ePredictive maintenance in mining industry: grinding mill case study\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eGrinding Mill in Mining Industry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals, Temperature, Electric signals, Ultrasonic measurements, Time-domain, Frequency-domain, Time-frequency features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eNentwich et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eTowards Data Acquisition for Predictive Maintenance of Industrial Robots\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eSix-axis Articulated Industrial Robots (e.g. KUKA KR210)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals, Motor currents, RMS of acceleration, Trajectory-specific measurements\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eBilal Yıldız et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eIntegrating preventive and predictive maintenance policies with system dynamics: A decision table approach\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eAI4I 2020 Predictive Maintenance Dataset (simulation study)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eAir temperature, Process temperature, Rotational speed, Torque, Tool wear, Failure modes (HDF, OF, PF, TWF), Maintenance budgets, Time between failures\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eLee et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003ePredictive Maintenance of Machine Tool Systems Using Artificial Intelligence Techniques Applied to Machine Condition Data\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eMilling Process (Cutting Tool) \u0026amp; Bearing Test Rig\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eMilling signals: DC spindle current, AC spindle current, Table vibration, Spindle vibration, Acoustic Emission (AE); Bearing signals: Vibration signals at 20 kHz\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eKilleen et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eIoT-based predictive maintenance for fleet management\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003ePublic Transport Buses (Soci\u0026eacute;t\u0026eacute; de Transport de l\u0026rsquo;Outaouais, Canada)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eJ1939 sensor data: wheel speed, axle angle, engine speed, engine torque, oil temperature, coolant temperature, driver pedal positions\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eLuo et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA hybrid predictive maintenance approach for CNC machine tool driven by Digital Twin\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eCNC Milling Machine Tool \u0026ndash; Cutting Tool Wear\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals, Cutting force, Acoustic Emission (AE), Temperature, Relative slip speed, Stress, Statistical features (RMS, Kurtosis, Skewness)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eFeng et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eMulti-level predictive maintenance of smart manufacturing systems driven by digital twin: A matheuristics approach\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eOffshore Oil and Gas Production System \u0026ndash; Subsea Christmas Trees (XTs)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eRemaining Useful Life (RUL) of components, Oil production rates, Maintenance costs, Downtime losses, Spare parts inventory, ROV availability\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eCao et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eKSPMI: A Knowledge-based System for Predictive Maintenance in Industry 4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eSemiconductor Manufacturing Process (UCI SECOM Dataset)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003e590 process variables (temperatures, pressures, times), event timestamps, chronicle patterns (sequences of events leading to failures)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eCoelho et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003ePredictive maintenance on sensorized stamping presses by time series segmentation, anomaly detection, and classification algorithms\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eStamping Press in Bosch Thermo Technology\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eOil Temperature, Oil Level, Engine Actual Speed, Shaft Vibration, Pressure Left, Pressure Right, Tool Height, plus aggregated features (mean, max, std, etc.)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eChinta et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eOptimal feature selection on Serial Cascaded deep learning for predictive maintenance system in automotive industry with fused optimization algorithm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eAutomotive Industry (Simulated \u0026amp; real datasets)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eSensor telemetry (e.g. voltage, rotation, pressure, vibration), Degradation indices, Hidden neuron counts (optimized), Data from 5 public datasets (e.g. turbofan engines, Genesis linear drive, etc.)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eJiang et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA2-LSTM for predictive maintenance of industrial equipment based on machine learning\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eHorizontal Machining Center \u0026ndash; Aircraft Manufacturing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eElectrical signals (currents, voltages, active/reactive power, power factor, frequency), 28 attributes reduced to 15 selected attributes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eFarahani et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA data-driven predictive maintenance framework for injection molding process\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eInjection Molding Process \u0026ndash; Cooling System Monitoring\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eMold temperature, Cavity pressure, Strain gauge data, Barrel pressure, Clamp force, Screw position, Derived features (RMS, Variance, PCA components)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eDrakaki et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eRecent Developments Towards Industry 4.0 Oriented Predictive Maintenance in Induction Motors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eInduction Motors (IM) in Industrial Settings\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eMotor current signals, Vibration signals, Time-domain, Frequency-domain, Time-frequency domain features, Raw data from sensors\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eJain et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eDistributed Diagnostics, Prognostics and Maintenance Planning: Realizing Industry 4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eMilling Cutter System on manufacturing shop floor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eForce signals, Vibration signals, Acoustic Emission signals\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eHuynh et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA predictive maintenance model for k-out-of-n:F continuously deteriorating systems subject to stochastic and economic dependencies\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eGeneric \u0026ndash; applicable to various industrial systems\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eDegradation levels, parameters of Gamma processes (shape, scale), Inspection cost, Replacement costs, Downtime costs, System configuration\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eZonta et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA predictive maintenance model for optimizing production schedule using deep neural networks\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eSimulated industrial machines \u0026ndash; CNCs and generic production equipment using Microsoft Azure PdM dataset\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eTelemetry Data: Voltage, Rotation, Pressure, Vibration; Degradation Indices; Production tasks and job scheduling constraints; Maintenance costs\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eWen et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eRecent advances and trends of predictive maintenance from data-driven machine prognostics perspective\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eReview article (no specific single case study)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eGeneric variables: vibration, temperature, current, pressure, time-series signals, health indicators\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eEinabadi et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eDynamic Predictive Maintenance in industry 4.0 based on real time information: Case study in automotive industries\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eCNC Machine in Fiat Power Train Technologies (FPT) automotive plant\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eSensor data: Temperature, Vibration (mm/s), Energy consumption (kW), Remaining Useful Life (RUL), Maintenance costs, Set-up costs\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eShaheen et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eData-driven failure prediction and RUL estimation of mechanical components using accumulative artificial neural networks\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eSimulated mechanical component with predefined failure threshold (0.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eDegradation level (simulated data), single variable (e.g. temperature, pressure, speed depending on component)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eHillenbrand et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eInvestigation of defects in roll contacts of machine elements with Acoustic Emission and Unsupervised Machine Learning\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eAxial Ball Bearings on test bench\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eAcoustic Emission signals (AE), RMS, Mean Frequency, Formfactor, Kurtosi\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eAugustyn et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eMethod of Machining Centre Sliding System Fault Detection using Torque Signals and Autoencoder\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eSliding System of Machining Centres (X, Y, Z, A axes)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eTorque signals from servomotors (measured current converted to torque)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eRajasekar et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003ePrediction of Component Level Degradation in a Hydraulic Rig using Machine Learning Methods\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eMulti-component hydraulic rig. The system is monitored to identify and classify component degradations and failures.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eMultivariate sensor signals (asynchronous)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eSingha et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003ePredictive Failure Analysis of Spindle Motor \u0026amp; Cutting Oil Condition Monitoring of Grinding Machine using Artificial Intelligence Models\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eGrinding Machine (focus on spindle motor \u0026amp; cutting oil condition)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals, Temperature, pH levels of cutting oil, Sulphur concentration\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eKotsiopoulos et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eDeep multi-sensorial data analysis for production monitoring in hard metal industry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eHard Metal Industry (CNC lathe machining)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration (X, Z axis), Laser microprofilometry, Ultrasound scans, FFT features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eFordal et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eApplication of sensor data based predictive maintenance and artificial neural networks to enable Industry 4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eSplitting Saw in Talg\u0026oslash; M\u0026oslash;reTre AS (Norway)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eBlade temperature (IR sensor), Current (Power usage), Vibration RMS, Temperature from vibration sensor, Aggregated features (min, max, mean, std), Alarm thresholds\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eVic\u0026ecirc;ncio et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eAn Intelligent Predictive Maintenance Approach Based on End-of-Line Test Logfiles in the Automotive Industry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eAutomotive Industry \u0026ndash; End-of-Line (EoL) Testing Systems. Focus on predicting malfunctions in final quality testing equipment,\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eLog message contents (event type, timestamp, operational context).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eWang et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eDynamic predictive maintenance strategy for RUL prediction via DL ensemble\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eNASA Turbofan engine (C-MAPSS dataset)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eRUL, mission cycles, reliability, stock \u0026amp; maintenance decisions\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eHerzog et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003ePredictive maintenance for electrical motors: AI algorithms\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eElectrical motors (fault detection \u0026amp; isolation)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals, classification features (SVM, perceptron, KNN)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eKolvig-Raun et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eJoint Stress Estimation and RUL Prediction for Collaborative Robots\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eCollaborative robots (UR e-series)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eJoint stress, load case, torque, temperature, RUL\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eMardianto et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eAugmenting Machinery Health Monitoring System with LSTM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eTurbomachinery equipment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eRUL, vibration data, MAPE, online monitoring\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003ePatel et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eBlockchain-based Predictive Maintenance Scheme for Smart Agriculture\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eSmart Agriculture machinery\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eSensor data (IoT), crop machinery conditions, faults\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eGougam et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eBearing faults classification using signal processing + ML\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eMechanical test rig (rotating machinery)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals, MODWPT features, health indicators\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eKondo et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eIndustrial edge computing architecture for Local Digital Twin\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eAssembly line (Brazilian subsidiary)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eEdge computing, DT parameters, yield prediction\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eMeddaoui et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eBenefits of predictive maintenance in manufacturing excellence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eIndustrial production process\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003ePhysical parameters, downtime, costs, ML algorithms\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eMeddaoui et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eAdvanced ML for PdM: RUL prediction \u0026amp; reliability\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eAerospace maintenance dataset\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eFeature engineering, ML models, RUL, failures\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eYuhua Yin, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA Dual-Task Learning Approach for Bearing Anomaly Detection and State Evaluation of Safe Region\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eBearings anomaly detection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eSafe region evaluation, anomaly features, SVDD parameters\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eJavad Isavand, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA reduced-order machine-learning-based method for fault recognition in tool condition monitoring\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eTool condition monitoring (machining, 200 min operation, 5 tools)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eFrequency-domain \u0026amp; time-frequency features (FFT, STFT, EMD, VMD)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eAnouar Bourokba, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eA Shapley based XAI approach for a turbofan RUL estimation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eTurbofan engines (aircraft)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eRUL, sensor parameters with Shapley contributions\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eUgur Ileri, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eAn Efficient Approach for Automatic Fault Classification Based on Data Balance and One-Dimensional Deep Learning\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003ePredictive maintenance datasets \u0026ndash; fault classification\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eRaw PdM datasets, normalized features, balanced groups\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eRengaraj R, G.R. et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eDesign and Implementation of Real Time Monitoring System of Diesel Generator using IoT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eDiesel Generators (DG) with IoT sensors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration, operational data, IoT sensor streams\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eHaobin Wen, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eEarly Prediction of Remaining Useful Life for Rolling Bearings Based on Envelope Spectral Indicator and Bayesian Filter\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eBearings RUL (rolling-element)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eEnvelope Spectral Indicator (ESI), vibration signals, fault frequencies\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eZekai Si, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eEfficient Tool Wear Prediction in Manufacturing: BiLPReS Hybrid Model with Performer Encoder\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eTool wear prediction (PHM2010 dataset)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eForce, acceleration, acoustic emission (multichannel sensors)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003ePatrick Seebold, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eExplainable AI-based Shapley Additive Explanations for Remaining Useful Life Prediction using NASA Turbofan Engine Dataset\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eNASA Turbofan engine dataset\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003e21 sensors + operating settings, SHAP features\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eHuan Wang, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eFourier Feature Refiner Network With Soft Thresholding for Machinery Fault Diagnosis Under Highly Noisy Conditions\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eBearings \u0026ndash; noisy industrial vibration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eVibration signals (time + frequency domain, Fourier features)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003ePooja Kamat, et al.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 160px;\"\u003e\n \u003cp\u003eVibration-based anomaly pattern mining for remaining useful life (RUL) prediction in bearings\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 43px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eBearings RUL prediction\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eSensor data (vibration), features (time + frequency domain), BiLSTM, CNN-LSTM, Conv-LSTM\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eTo address our research questions on what themes, dominate predictive maintenance (PdM) in the Industry 4.0 context, how they interconnect, and how they evolve over time, we analyzed the co-occurrence structure of author-supplied keywords (minimum occurrence \u0026ge; 2) and produced complementary VOSviewer visualizations: a clustered network, item-focused network views, and a density map. Together they reveal a coherent, maturing research landscape in which PdM and machine learning form the conceptual and methodological cores, connected to deep-learning prognostics and Industry 4.0 implementation strands\u003c/p\u003e\n\u003cp\u003eThe co-occurrence map (figure7) clearly shows that machine learning and predictive maintenance occupy central positions, acting as hubs connecting the entire network. Their high link strength indicates that they co-occur with a large number of other keywords, confirming their foundational role in this research domain.\u003c/p\u003e\n\u003cp\u003eThe clustered co-occurrence network (Figure 7) contains 21 items, 85 links, and a total link strength of 113, forming four stable thematic clusters:\u003c/p\u003e\n\u003cp\u003eCluster 1: Algorithms \u0026amp; Predictive Modeling.\u003c/p\u003e\n\u003cp\u003eContains machine learning, prediction algorithms, accuracy, and monitoring. This is the algorithmic backbone of PdM: supervised learning methods, performance criteria, and operational monitoring.\u003c/p\u003e\n\u003cp\u003eCluster 2: \u0026nbsp; Deep Learning \u0026amp; Prognostics.\u003c/p\u003e\n\u003cp\u003eIncludes deep learning, bidirectional LSTM (BiLSTM), feature extraction, fault diagnosis, remaining useful life (RUL), and terms from prognostics and health management. This cluster captures the data-driven prognostics pipeline (signal \u0026rarr; features \u0026rarr; sequential DL \u0026rarr; diagnosis/RUL).\u003c/p\u003e\n\u003cp\u003eCluster 3: \u0026nbsp; PdM, Monitoring \u0026amp; Conceptual Core.\u003c/p\u003e\n\u003cp\u003eOrganized around predictive maintenance and predictive model, closely tied to condition monitoring. It functions as the semantic anchor linking methods to operational use.\u003c/p\u003e\n\u003cp\u003eCluster 4 : Applications \u0026amp; Industry 4.0 :\u003c/p\u003e\n\u003cp\u003eContains industry 4.0, maintenance engineering, and robot sensing systems. This cluster confirms that PdM research is now deeply embedded in the Industry 4.0 paradigm, with a strong orientation toward sensor integration, robotics, and practical deployment.\u003c/p\u003e\n\u003cp\u003eTwo central hubs knit the map together: predictive maintenance (C3) and machine learning (C1). Their high connectivity and proximity to multiple clusters show that PdM is implemented primarily via ML-centred analytics, while the C2 and C4 clusters bring, respectively, advanced prognostics and industrial realization into the fold.\u003c/p\u003e\n\u003cp\u003eThe density map established in the figure 8 complements the structural views by showing where the highest concentration of items and links lies:\u003c/p\u003e\n\u003cp\u003eThe brightest hotspots sit on machine learning, predictive maintenance, and deep learning the epicenter of current research.\u003c/p\u003e\n\u003cp\u003eSecondary but distinct hotspots appear over remaining useful life, condition monitoring, confirming that prognostics-oriented DL is a major focal area.\u003c/p\u003e\n\u003cp\u003eIndustry 4.0, maintenance engineering, and robot sensing systems register as application-side hotspots, indicating tangible interest in instrumentation and deployment.\u003c/p\u003e\n\u003cp\u003eThis intensity pattern triangulates the earlier findings: method (ML/DL), task (RUL/diagnosis) and context (Industry 4.0) jointly defines the field\u0026rsquo;s present center of gravity.\u003c/p\u003e\n\u003cp\u003eThe brightest hotspots sit on machine learning, predictive maintenance, and deep learning the epicenter of current research. The co-occurrence network (figure 9) highlights three key thematic dynamics shaping predictive maintenance (PdM) research. First, predictive maintenance and machine learning occupy central positions, confirming their dual role as the conceptual and methodological cores of the field. PdM integrates monitoring and diagnostic activities, while machine learning provides the computational framework that drives data-based decision-making and connects theory to industrial applications.\u003c/p\u003e\n\u003cp\u003eThe literature shows an increasing standardization of analytical workflows. The co-occurrence of deep learning, BiLSTM, and remaining useful life (RUL) prediction illustrates a consolidated pipeline: feature extraction, sequential neural architectures, and RUL estimation. This convergence reflects the transition from experimental studies to a recognized methodological benchmark for prognostics and health management.\u003c/p\u003e\n\u003cp\u003eThe explicit presence of industry 4.0 and its association with maintenance engineering and robot sensing systems indicate a clear shift toward practical implementation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eResearch is now focusing on embedding PdM within smart factories and cyber-physical production systems, emphasizing automation, connectivity, and real-time decision-making. This trend signals the move toward fully integrated, autonomous, and data-driven maintenance ecosystems, bridging the gap between theoretical development and industrial deployment.\u003c/p\u003e"},{"header":"6. Conclusion","content":"\u003cp\u003eThis review has explored the rapidly evolving landscape of Predictive Maintenance (PdM) within the context of Smart Industry and Industry 4.0. Drawing on a diverse body of recent literature, it is evident that data-driven methodologies, particularly deep learning techniques, have become the dominant force in contemporary PdM research. These approaches offer remarkable capabilities in handling complex, high-dimensional sensor data, enabling accurate predictions of equipment health and remaining useful life (RUL). However, challenges such as data scarcity, model interpretability, and deployment robustness remain significant barriers to widespread industrial adoption.\u003c/p\u003e\u003cp\u003eHybrid approaches that integrate physical knowledge with machine learning have emerged as a promising pathway, aiming to combine the predictive power of AI with the transparency and domain understanding required in industrial environments. Furthermore, the increasing incorporation of digital twin technologies signals a shift towards more holistic, context-aware maintenance solutions.\u003c/p\u003e\u003cp\u003eAcross sectors from manufacturing and automotive to mining and railways, the adoption of AI-driven PdM demonstrates clear potential for reducing downtime, optimizing maintenance schedules, and enhancing operational efficiency. Yet, the variability of industrial contexts demands tailored solutions, and no universal approach has yet been established.\u003c/p\u003e\u003cp\u003eLooking ahead, future research should prioritize developing explainable AI techniques, leveraging transfer learning to overcome data limitations, and aligning predictive models with economic and operational impacts. Only through such integrated efforts can Predictive Maintenance fully realize its transformative potential within Smart Industry, contributing to more resilient, efficient, and intelligent industrial systems.\u003c/p\u003e\u003cp\u003eUltimately, PdM stands at the intersection of technological innovation and practical industrial needs. As Industry 4.0 continues to reshape production landscapes, the development of robust, interpretable, and cost-effective PdM solutions will be pivotal in defining the future of smart manufacturing and maintenance.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eCompeting Interest:\u003c/h2\u003e\u003cp\u003eAuthors of this paper have no financial interests and have no relevant financial or non-financial interests to disclose.\u0026rdquo;\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eFunding :\u003c/h2\u003e\u003cp\u003eThe authors declare that no funds, no data set used, grants, or other support were received during the preparation of this manuscript.\u003c/p\u003e\u003ch2\u003eAuthor Contributions:\u003c/h2\u003e\u003cp\u003eAll authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Hamdi Alaoui. A, Meddaoui. A, and Hain, M. The first draft of the manuscript was written by Mr. Hamdi and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAugustyn D, Fidali M. 2023. Method of Machining Centre Sliding System Fault Detection using Torque Signals and Autoencoder. Acta Mechanica et Automatica. 17(3):445\u0026ndash;451. https://doi.org/10.2478/ama-2023-0051\u003c/li\u003e\n\u003cli\u003eBilal Yıldız G, Soylu B. 2023. Integrating preventive and predictive maintenance policies with system dynamics: A decision table approach. Advanced Engineering Informatics. 56:101952. https://doi.org/10.1016/j.aei.2023.101952\u003c/li\u003e\n\u003cli\u003eBourokba A, El Hamdi R, Njah M. 2024. A Shapley based XAI approach for a turbofan RUL estimation. In: 2024 21st International Multi-Conference on Systems, Signals \u0026amp; Devices (SSD) [Internet]. [place unknown]; [accessed 2025 Sep 1]; p. 832\u0026ndash;837. https://doi.org/10.1109/SSD61670.2024.10548499\u003c/li\u003e\n\u003cli\u003eCalabrese F, Regattieri A, Botti L, Galizia FG. 2019. Prognostic Health Management of Production Systems. New Proposed Approach and Experimental Evidences. Procedia Manufacturing. 39:260\u0026ndash;269. https://doi.org/10.1016/j.promfg.2020.01.333\u003c/li\u003e\n\u003cli\u003eCao Q, Zanni-Merk C, Samet A, Reich C, Beuvron F de B de, Beckmann A, Giannetti C. 2022. KSPMI: A Knowledge-based System for Predictive Maintenance in Industry 4.0. Robotics and Computer-Integrated Manufacturing. 74:102281. https://doi.org/10.1016/j.rcim.2021.102281\u003c/li\u003e\n\u003cli\u003eChen C-S, Chen J-S. 2011. Rotor fault diagnosis system based on sGA-based individual neural networks. Expert Systems with Applications. 38(9):10822\u0026ndash;10830. https://doi.org/10.1016/j.eswa.2011.02.074\u003c/li\u003e\n\u003cli\u003eChinta VS, Kethi Reddi S, Yarramsetty N. 2023. Optimal feature selection on Serial Cascaded deep learning for predictive maintenance system in automotive industry with fused optimization algorithm. Advanced Engineering Informatics. 57:102105. https://doi.org/10.1016/j.aei.2023.102105\u003c/li\u003e\n\u003cli\u003eChowdhury A, Raut S, Pal A. 2022. Internet of Things resource monitoring through proactive fault prediction. Computers \u0026amp; Industrial Engineering. 169:108265. https://doi.org/10.1016/j.cie.2022.108265\u003c/li\u003e\n\u003cli\u003eCoelho D, Costa D, Rocha EM, Almeida D, Santos JP. 2022. Predictive maintenance on sensorized stamping presses by time series segmentation, anomaly detection, and classification algorithms. Procedia Computer Science. 200:1184\u0026ndash;1193. https://doi.org/10.1016/j.procs.2022.01.318\u003c/li\u003e\n\u003cli\u003eCrespo M\u0026aacute;rquez A, Marcos Alberca JA, Crespo del Castillo A. 2023. Simulating dynamic RUL based CBM scheduling. A case study in the railway sector. Computers in Industry. 148:103914. https://doi.org/10.1016/j.compind.2023.103914\u003c/li\u003e\n\u003cli\u003eDaniyan I, Mpofu K, Oyesola M, Ramatsetse B, Adeodu A. 2020. Artificial intelligence for predictive maintenance in the railcar learning factories. Procedia Manufacturing. 45:13\u0026ndash;18. https://doi.org/10.1016/j.promfg.2020.04.032\u003c/li\u003e\n\u003cli\u003eDrakaki M, Karnavas YL, Tzionas P, Chasiotis ID. 2021. Recent Developments Towards Industry 4.0 Oriented Predictive Maintenance in Induction Motors. Procedia Computer Science. 180:943\u0026ndash;949. https://doi.org/10.1016/j.procs.2021.01.345\u003c/li\u003e\n\u003cli\u003eEinabadi B, Baboli A, Ebrahimi M. 2019. Dynamic Predictive Maintenance in industry 4.0 based on real time information: Case study in automotive industries. IFAC-PapersOnLine. 52(13):1069\u0026ndash;1074. https://doi.org/10.1016/j.ifacol.2019.11.337\u003c/li\u003e\n\u003cli\u003eFarahani S, Khade V, Basu S, Pilla S. 2022. A data-driven predictive maintenance framework for injection molding process. Journal of Manufacturing Processes. 80:887\u0026ndash;897. https://doi.org/10.1016/j.jmapro.2022.06.013\u003c/li\u003e\n\u003cli\u003eFeng Q, Zhang Y, Sun B, Guo X, Fan D, Ren Y, Song Y, Wang Z. 2023. Multi-level predictive maintenance of smart manufacturing systems driven by digital twin: A matheuristics approach. Journal of Manufacturing Systems. 68:443\u0026ndash;454. https://doi.org/10.1016/j.jmsy.2023.05.004\u003c/li\u003e\n\u003cli\u003eFordal JM, Schj\u0026oslash;lberg P, Helgetun H, Skjermo T\u0026Oslash;, Wang Y, Wang C. 2023. Application of sensor data based predictive maintenance and artificial neural networks to enable Industry 4.0. Adv Manuf. 11(2):248\u0026ndash;263. https://doi.org/10.1007/s40436-022-00433-x\u003c/li\u003e\n\u003cli\u003eFranciosi C, Miranda S, Veneroso CR, Riemma S. 2021. A maintenance scheduling optimization model for a multi-component machine in a digitalized manufacturing context. IFAC-PapersOnLine. 54(1):1254\u0026ndash;1259. https://doi.org/10.1016/j.ifacol.2021.08.150\u003c/li\u003e\n\u003cli\u003eGougam F, Afia A, Soualhi A, Touzout W, Rahmoune C, Benazzouz D. 2024. Bearing faults classification using a new approach of signal processing combined with machine learning algorithms. J Braz Soc Mech Sci Eng. 46(2):65. https://doi.org/10.1007/s40430-023-04645-5\u003c/li\u003e\n\u003cli\u003eGupta V, Mitra R, Koenig F, Kumar M, Tiwari MK. 2023a. Predictive maintenance of baggage handling conveyors using IoT. Computers \u0026amp; Industrial Engineering. 177:109033. https://doi.org/10.1016/j.cie.2023.109033\u003c/li\u003e\n\u003cli\u003eGupta V, Mitra R, Koenig F, Kumar M, Tiwari MK. 2023b. Predictive maintenance of baggage handling conveyors using IoT. Computers \u0026amp; Industrial Engineering.\u003c/li\u003e\n\u003cli\u003eHeng A, Zhang S, Tan ACC, Mathew J. 2009. Rotating machinery prognostics: State of the art, challenges and opportunities. Mechanical Systems and Signal Processing. 23(3):724\u0026ndash;739. https://doi.org/10.1016/j.ymssp.2008.06.009\u003c/li\u003e\n\u003cli\u003eHerzog T, Bartecki K. 2024. Predictive maintenance for electrical motors: Current approach and usage of artificial intelligence algorithms. In: 2024 28th International Conference on Methods and Models in Automation and Robotics (MMAR) [Internet]. [place unknown]; [accessed 2025 Aug 30]; p. 494\u0026ndash;498. https://doi.org/10.1109/MMAR62187.2024.10680808\u003c/li\u003e\n\u003cli\u003eHillenbrand J, Detroy J, Fleischer J. 2021. Investigation of defects in roll contacts of machine elements with Acoustic Emission and Unsupervised Machine Learning. IOP Conf Ser: Mater Sci Eng. 1193(1):012085. https://doi.org/10.1088/1757-899X/1193/1/012085\u003c/li\u003e\n\u003cli\u003eHuynh KT, Vu HC, Nguyen TD, Ho AC. 2022. A predictive maintenance model for k-out-of-n:F continuously deteriorating systems subject to stochastic and economic dependencies. Reliability Engineering \u0026amp; System Safety. 226:108671. https://doi.org/10.1016/j.ress.2022.108671\u003c/li\u003e\n\u003cli\u003eIleri U, Altun Y, Narin A. 2024. An Efficient Approach for Automatic Fault Classification Based on Data Balance and One-Dimensional Deep Learning. Applied Sciences. 14(11):4899. https://doi.org/10.3390/app14114899\u003c/li\u003e\n\u003cli\u003eIntelligent Fault Diagnosis and Prognosis for Engineering Systems | Wiley Online Books. [accessed 2025 Jun 26]. https://onlinelibrary.wiley.com/doi/book/10.1002/9780470117842\u003c/li\u003e\n\u003cli\u003eIsavand J, Kasaei A, Peplow A, Wang X, Yan J. 2024. A reduced-order machine-learning-based method for fault recognition in tool condition monitoring. Measurement. 224:113906. https://doi.org/10.1016/j.measurement.2023.113906\u003c/li\u003e\n\u003cli\u003eJain AK, Chouksey P, Parlikad AK, Lad BK. 2020. Distributed Diagnostics, Prognostics and Maintenance Planning: Realizing Industry 4.0. IFAC-PapersOnLine. 53(3):354\u0026ndash;359. https://doi.org/10.1016/j.ifacol.2020.11.057\u003c/li\u003e\n\u003cli\u003eJardine AKS, Lin D, Banjevic D. 2006. A review on machinery diagnostics and prognostics implementing condition-based maintenance. Mechanical Systems and Signal Processing. 20(7):1483\u0026ndash;1510. https://doi.org/10.1016/j.ymssp.2005.09.012\u003c/li\u003e\n\u003cli\u003eJiang Y, Dai P, Fang P, Zhong RY, Zhao X, Cao X. 2022. A2-LSTM for predictive maintenance of industrial equipment based on machine learning. Computers \u0026amp; Industrial Engineering. 172:108560. https://doi.org/10.1016/j.cie.2022.108560\u003c/li\u003e\n\u003cli\u003eJoseph D, Gallege T, Bekar ET, Dudas C, Skoogh A. 2022a. A Predictive Maintenance Application for A Robot Cell using LSTM Model. 5th IFAC Workshop on Advanced Maintenance Engineering, Services and Technologies AMEST 2022.\u003c/li\u003e\n\u003cli\u003eJoseph D, Gallege T, Bekar ET, Dudas C, Skoogh A. 2022b. A Predictive Maintenance Application for A Robot Cell using LSTM Model. IFAC-PapersOnLine. 55(19):115\u0026ndash;120. https://doi.org/10.1016/j.ifacol.2022.09.193\u003c/li\u003e\n\u003cli\u003eKamat P, Kumar S, Sugandhi R. 2024. Vibration-based anomaly pattern mining for remaining useful life (RUL) prediction in bearings. J Braz Soc Mech Sci Eng. 46(5):290. https://doi.org/10.1007/s40430-024-04872-4\u003c/li\u003e\n\u003cli\u003eKans M, Campos J, H\u0026aring;kansson L. 2020. A remote laboratory for Maintenance 4.0 training and education. IFAC-PapersOnLine. 53(3):101\u0026ndash;106. https://doi.org/10.1016/j.ifacol.2020.11.016\u003c/li\u003e\n\u003cli\u003eKilleen P, Ding B, Kiringa I, Yeap T. 2019. IoT-based predictive maintenance for fleet management. Procedia Computer Science. 151:607\u0026ndash;613. https://doi.org/10.1016/j.procs.2019.04.184\u003c/li\u003e\n\u003cli\u003eKolvig-Raun ES, Kj\u0026aelig;rgaard MB, Brorsen R. 2024. Joint Stress Estimation and Remaining Useful Life Prediction for Collaborative Robots to Support Predictive Maintenance. IEEE Robotics and Automation Letters. 9(4):3554\u0026ndash;3561. https://doi.org/10.1109/LRA.2024.3368296\u003c/li\u003e\n\u003cli\u003eKondo RE, Andrade WJ, de Mello Henequim C, Lazzaretti AE, de Souza Britto A, de Freitas Rocha Loures E, Santos EAP, Reynoso-Meza G. 2024. An industrial edge computing architecture for Local Digital Twin. Computers \u0026amp; Industrial Engineering. 193:110257. https://doi.org/10.1016/j.cie.2024.110257\u003c/li\u003e\n\u003cli\u003eKotsiopoulos T, Leontaris L, Dimitriou N, Ioannidis D, Oliveira F, Sacramento J, Amanatiadis S, Karagiannis G, Votis K, Tzovaras D, Sarigiannidis P. 2021. Deep multi-sensorial data analysis for production monitoring in hard metal industry. Int J Adv Manuf Technol. 115(3):823\u0026ndash;836. https://doi.org/10.1007/s00170-020-06173-1\u003c/li\u003e\n\u003cli\u003eLee WJ, Wu H, Yun H, Kim H, Jun MBG, Sutherland JW. 2019. Predictive Maintenance of Machine Tool Systems Using Artificial Intelligence Techniques Applied to Machine Condition Data. Procedia CIRP. 80:506\u0026ndash;511. https://doi.org/10.1016/j.procir.2018.12.019\u003c/li\u003e\n\u003cli\u003eLei Y, Li Naipeng, Guo L, Li Ningbo, Yan T, Lin J. 2018. Machinery health prognostics: A systematic review from data acquisition to RUL prediction. Mechanical Systems and Signal Processing. 104:799\u0026ndash;834. https://doi.org/10.1016/j.ymssp.2017.11.016\u003c/li\u003e\n\u003cli\u003eLuo W, Hu T, Ye Y, Zhang C, Wei Y. 2020. A hybrid predictive maintenance approach for CNC machine tool driven by Digital Twin. Robotics and Computer-Integrated Manufacturing. 65:101974.\u003c/li\u003e\n\u003cli\u003eLv Y, Guo X, Zhou Q, Qian L, Liu J. 2023. Predictive maintenance decision-making for variable faults with non-equivalent costs of fault severities. Advanced Engineering Informatics. 56:102011. https://doi.org/10.1016/j.aei.2023.102011\u003c/li\u003e\n\u003cli\u003eMa Z, Zhao M, Dai X, Chen Y. 2023. A hybrid-driven probabilistic state space model for tool wear monitoring. Mechanical Systems and Signal Processing. 200:110599. https://doi.org/10.1016/j.ymssp.2023.110599\u003c/li\u003e\n\u003cli\u003eMardianto FD, Mauritsius T. 2024. Augmenting the Internet-based Machinery Health Monitoring System (MHMS) Through Integration of Long Short-Term Memory (LSTM) Model for Enhanced Prediction of Remaining Useful Lifetime (RUL). In: 2024 8th International Conference on Information Technology, Information Systems and Electrical Engineering (ICITISEE) [Internet]. [place unknown]; [accessed 2025 Aug 30]; p. 447\u0026ndash;452. https://doi.org/10.1109/ICITISEE63424.2024.10730046\u003c/li\u003e\n\u003cli\u003eMeddaoui A, Hachmoud A, Hain M. 2024. Advanced ML for predictive maintenance: a case study on remaining useful life prediction and reliability enhancement. Int J Adv Manuf Technol. 132(1):323\u0026ndash;335. https://doi.org/10.1007/s00170-024-13351-y\u003c/li\u003e\n\u003cli\u003eMeddaoui A, Hain M, Hachmoud A. 2023. The benefits of predictive maintenance in manufacturing excellence: a case study to establish reliable methods for predicting failures. Int J Adv Manuf Technol. 128(7):3685\u0026ndash;3690. https://doi.org/10.1007/s00170-023-12086-6\u003c/li\u003e\n\u003cli\u003eMobley RK. 2002. 2 - Financial Implications and Cost Justification. In: Mobley RK, editor. An Introduction to Predictive Maintenance (Second Edition) [Internet]. Burlington: Butterworth-Heinemann; [accessed 2023 Oct 15]; p. 23\u0026ndash;42. https://doi.org/10.1016/B978-075067531-4/50002-6\u003c/li\u003e\n\u003cli\u003eMoher D, Liberati A, Tetzlaff J, Altman DG. 2009. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 339:b2535. https://doi.org/10.1136/bmj.b2535\u003c/li\u003e\n\u003cli\u003eNentwich C, Reinhart G. 2021. Towards Data Acquisition for Predictive Maintenance of Industrial Robots. Procedia CIRP. 104:62\u0026ndash;67. https://doi.org/10.1016/j.procir.2021.11.011\u003c/li\u003e\n\u003cli\u003eNikolakis N, Marguglio A, Veneziano G, Greco P, Panicucci S, Cerquitelli T, Macii E, Andolina S, Alexopoulos K. 2020. A microservice architecture for predictive analytics in manufacturing. Procedia Manufacturing. 51:1091\u0026ndash;1097. https://doi.org/10.1016/j.promfg.2020.10.153\u003c/li\u003e\n\u003cli\u003ePatel M, Tanwar S, Jindal A, Alshammari N, Pervaiz H, Ahmed H. 2024. A Blockchain-based Predictive Maintenance Scheme for Smart Agriculture. In: 2024 IEEE/ACM 17th International Conference on Utility and Cloud Computing (UCC) [Internet]. [place unknown]; [accessed 2025 Aug 30]; p. 405\u0026ndash;410. https://doi.org/10.1109/UCC63386.2024.00062\u003c/li\u003e\n\u003cli\u003e(PDF) Evolution of Technical Systems Maintenance Approaches \u0026ndash; Review and a Case Study. ResearchGate [Internet]. [accessed 2025 Jun 26]. https://doi.org/10.1007/978-3-319-97490-3_16\u003c/li\u003e\n\u003cli\u003ePech M, Vrchota J, Bedn\u0026aacute;ř J. 2021. Predictive Maintenance and Intelligent Sensors in Smart Factory: Review. Sensors. 21(4):1470. https://doi.org/10.3390/s21041470\u003c/li\u003e\n\u003cli\u003ePhysics-informed machine learning | Nature Reviews Physics. [accessed 2025 Jun 25]. https://www.nature.com/articles/s42254-021-00314-5?utm_source=chatgpt.com\u003c/li\u003e\n\u003cli\u003ePolenghi A, Roda I, Pesenti V, Pasanisi D, Macchi M, Cortinovis D, Chebat F. 2023. Norm-based data labelling in supervised learning for fault detection and diagnostics of rotating elements towards maintenance servitisation. IFAC-PapersOnLine. 56(2):1282\u0026ndash;1287. https://doi.org/10.1016/j.ifacol.2023.10.1761\u003c/li\u003e\n\u003cli\u003ePrognostics and Health Management of Electronics. Wiley.com [Internet]. [accessed 2025 Jun 26]. https://www.wiley.com/en-us/Prognostics+and+Health+Management+of+Electronics%3A+Fundamentals%2C+Machine+Learning%2C+and+the+Internet+of+Things-p-9781119515357\u003c/li\u003e\n\u003cli\u003eR Rengaraj, Venkatakrishnan GR, Rajalakshmi M, R Ranganath, R SG, V S. 2024. Design and Implementation of Real Time Monitoring System of Diesel Generator using IoT. In: 2024 2nd International Conference on Networking and Communications (ICNWC) [Internet]. [place unknown]; [accessed 2025 Sep 1]; p. 1\u0026ndash;8. https://doi.org/10.1109/ICNWC60771.2024.10537268\u003c/li\u003e\n\u003cli\u003eRagab M, Chen Z, Wu M, Kwoh C-K, Yan R, Li X. 2021. Attention-based sequence to sequence model for machine remaining useful life prediction. Neurocomputing. 466:58\u0026ndash;68. https://doi.org/10.1016/j.neucom.2021.09.022\u003c/li\u003e\n\u003cli\u003eRemaining useful life estimation \u0026ndash; A review on the statistical data driven approaches. 2011. European Journal of Operational Research. 213(1):1\u0026ndash;14. https://doi.org/10.1016/j.ejor.2010.11.018\u003c/li\u003e\n\u003cli\u003eRihi A, Ba\u0026iuml;na S, Mhada F, Elbachari E, Tagemouati H, Guerboub M, Benzakour I. 2022. Predictive maintenance in mining industry: grinding mill case study. Procedia Computer Science. 207:2483\u0026ndash;2492. https://doi.org/10.1016/j.procs.2022.09.306\u003c/li\u003e\n\u003cli\u003eSeebold P, Kaye MK, Kim S, Nam CS. 2024. Explainable AI-based Shapley Additive Explanations for Remaining Useful Life Prediction using NASA Turbofan Engine Dataset. In: 2024 IEEE 3rd International Conference on Computing and Machine Intelligence (ICMI) [Internet]. [place unknown]; [accessed 2025 Sep 1]; p. 1\u0026ndash;5. https://doi.org/10.1109/ICMI60790.2024.10586061\u003c/li\u003e\n\u003cli\u003eSfar M, Yaacoubi S, Dellagi S, El Mountassir M, Turki S. 2020. Towards a predictive maintenance based on ultrasonic guided waves non-destructive testing technique: a case study. In: 13\u0026egrave;me Conf\u0026eacute;rence Internationale de MOd\u0026eacute;lisation, Optimisation et SIMulation : \u0026ldquo; Nouvelles avanc\u0026eacute;es et d\u0026eacute;fis pour des industries durables et avis\u0026eacute;es \u0026rdquo; : MOSIM 20, 12-14 Novembre 2020 : actes de conf\u0026eacute;rence = 13 th International Conference on MOdeling, Optimization and SIMulation : \u0026ldquo;New Advances and Challenges for Sustainable and Smart Industries\u0026rdquo; : MOSIM 20, 12-14 November 2020 : Conference proceedings [Internet]. Agadir (en ligne), Morocco: Ali Siadat (pr\u0026eacute;sident du comit\u0026eacute; d\u0026rsquo;organisation) and Lahcen Mifdal (pr\u0026eacute;sident du comit\u0026eacute; d\u0026rsquo;organisation) and Ilias Majdouline (pr\u0026eacute;sident du comit\u0026eacute; d\u0026rsquo;organisation); [accessed 2025 Jun 22]; p. 54\u0026ndash;62. https://hal.science/hal-03192678\u003c/li\u003e\n\u003cli\u003eShaheen B, Kocsis \u0026Aacute;, N\u0026eacute;meth I. 2023. Data-driven failure prediction and RUL estimation of mechanical components using accumulative artificial neural networks. Engineering Applications of Artificial Intelligence. 119:105749. https://doi.org/10.1016/j.engappai.2022.105749\u003c/li\u003e\n\u003cli\u003eShe D, Jia M. 2021a. A BiGRU method for remaining useful life prediction of machinery. Measurement.\u003c/li\u003e\n\u003cli\u003eShe D, Jia M. 2021b. A BiGRU method for remaining useful life prediction of machinery. Measurement. 167:108277. https://doi.org/10.1016/j.measurement.2020.108277\u003c/li\u003e\n\u003cli\u003eSi Z, Si S, Mu D. 2024. Efficient Tool Wear Prediction in Manufacturing: BiLPReS Hybrid Model with Performer Encoder. Arab J Sci Eng. 49(11):15193\u0026ndash;15204. https://doi.org/10.1007/s13369-024-08943-5\u003c/li\u003e\n\u003cli\u003eSingha A, Ali J, Khera V. 2020. Predictive Failure Analysis of Spindle Motor \u0026amp; Cutting Oil Condition Monitoring of Grinding Machine using Artificial Intelligence Models. In: 2020 IEEE 17th India Council International Conference (INDICON) [Internet]. New Delhi, India: IEEE; [accessed 2025 Jun 20]; p. 1\u0026ndash;7. https://doi.org/10.1109/INDICON49873.2020.9342533\u003c/li\u003e\n\u003cli\u003eSmith AM, Hinchcliffe GR. 2003. RCM--Gateway to World Class Maintenance. [place unknown]: Elsevier.\u003c/li\u003e\n\u003cli\u003eSusto GA, Schirru A, Pampuri S, McLoone S, Beghi A. 2015. Machine Learning for Predictive Maintenance: A Multiple Classifier Approach. IEEE Trans Ind Inf. 11(3):812\u0026ndash;820. https://doi.org/10.1109/TII.2014.2349359\u003c/li\u003e\n\u003cli\u003eTaşcı B, Omar A, Ayvaz S. 2023. Remaining useful lifetime prediction for predictive maintenance in manufacturing. Computers \u0026amp; Industrial Engineering. 184:109566. https://doi.org/10.1016/j.cie.2023.109566\u003c/li\u003e\n\u003cli\u003eTessoni V, Amoretti M. 2022. Advanced statistical and machine learning methods for multi-step multivariate time series forecasting in predictive maintenance. Procedia Computer Science. 200:748\u0026ndash;757. https://doi.org/10.1016/j.procs.2022.01.273\u003c/li\u003e\n\u003cli\u003eTheissler A, P\u0026eacute;rez-Vel\u0026aacute;zquez J, Kettelgerdes M, Elger G. 2021. Predictive maintenance enabled by machine learning: Use cases and challenges in the automotive industry. Reliability Engineering \u0026amp; System Safety. 215:107864. https://doi.org/10.1016/j.ress.2021.107864\u003c/li\u003e\n\u003cli\u003eTraini E, Bruno G, D\u0026rsquo;antonio G, Lombardi F. 2019. Machine learning framework for predictive maintenance in milling. IFAC-PapersOnLine. 52(13):177\u0026ndash;182.\u003c/li\u003e\n\u003cli\u003eUhlmann E, Polte J, Koutrakis N-S. 2021. Holistic Concept Towards a Reference Architecture Model for Predictive Maintenance. Procedia CIRP. 104:1430\u0026ndash;1433. https://doi.org/10.1016/j.procir.2021.11.241\u003c/li\u003e\n\u003cli\u003eVargas VM, Rosati R, Herv\u0026aacute;s-Mart\u0026iacute;nez C, Mancini A, Romeo L, Guti\u0026eacute;rrez PA. 2023. A hybrid feature learning approach based on convolutional kernels for ATM fault prediction using event-log data. Engineering Applications of Artificial Intelligence.\u003c/li\u003e\n\u003cli\u003eWang H, Luo W, Liu Z, Zhang J, Zuo M. 2024. Fourier Feature Refiner Network With Soft Thresholding for Machinery Fault Diagnosis Under Highly Noisy Conditions. IEEE Internet of Things Journal. 11(13):22880\u0026ndash;22891. https://doi.org/10.1109/JIOT.2024.3363216\u003c/li\u003e\n\u003cli\u003eWang L, Zhu Z, Zhao X. 2024. Dynamic predictive maintenance strategy for system remaining useful life prediction via deep learning ensemble method. Reliability Engineering \u0026amp; System Safety. 245:110012. https://doi.org/10.1016/j.ress.2024.110012\u003c/li\u003e\n\u003cli\u003eWen H, Zhang L, Sinha JK. 2024. Early Prediction of Remaining Useful Life for Rolling Bearings Based on Envelope Spectral Indicator and Bayesian Filter. Applied Sciences. 14(1):436. https://doi.org/10.3390/app14010436\u003c/li\u003e\n\u003cli\u003eWen Y, Rahman MF, Xu H, Tseng T-LB. 2022. Recent advances and trends of predictive maintenance from data-driven machine prognostics perspective. Measurement. 187:110276.\u003c/li\u003e\n\u003cli\u003eWillard J, Jia X, Xu S, Steinbach M, Kumar V. 2020. Integrating physics-based modeling with machine learning: A survey. arXiv preprint arXiv:200304919. 1(1):1\u0026ndash;34.\u003c/li\u003e\n\u003cli\u003eYin Y, Liu Z, Guo B, Zuo M. 2024. A Dual-Task Learning Approach for Bearing Anomaly Detection and State Evaluation of Safe Region. Chin J Mech Eng. 37(1):4. https://doi.org/10.1186/s10033-023-00978-3\u003c/li\u003e\n\u003cli\u003eYou Y, Chen C, Hu F, Liu Y, Ji Z. 2022. Advances of Digital Twins for Predictive Maintenance. Procedia Computer Science. 200:1471\u0026ndash;1480. https://doi.org/10.1016/j.procs.2022.01.348\u003c/li\u003e\n\u003cli\u003eZhang W, Yang D, Wang H. 2019. Data-Driven Methods for Predictive Maintenance of Industrial Equipment: A Survey. IEEE Systems Journal. 13(3):2213\u0026ndash;2227. https://doi.org/10.1109/JSYST.2019.2905565\u003c/li\u003e\n\u003cli\u003eZonta T, da Costa CA, Zeiser FA, de Oliveira Ramos G, Kunst R, da Rosa Righi R. 2022. A predictive maintenance model for optimizing production schedule using deep neural networks. Journal of Manufacturing Systems. 62:450\u0026ndash;462. https://doi.org/10.1016/j.jmsy.2021.12.013\u003c/li\u003e\n\u003cli\u003eZonta T, Da Costa CA, Da Rosa Righi R, De Lima MJ, Da Trindade ES, Li GP. 2020. Predictive maintenance in the Industry 4.0: A systematic literature review. Computers \u0026amp; Industrial Engineering. 150:106889. https://doi.org/10.1016/j.cie.2020.106889\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"the-international-journal-of-advanced-manufacturing-technology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jamt","sideBox":"Learn more about [The International Journal of Advanced Manufacturing Technology](https://www.springer.com/journal/170)","snPcode":"170","submissionUrl":"https://submission.nature.com/new-submission/170/3","title":"The International Journal of Advanced Manufacturing Technology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Predictive maintenance, Artificial Intelligence, XAI, Machine Learning, Remaining Useful Life, Internet of Things","lastPublishedDoi":"10.21203/rs.3.rs-7618987/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7618987/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn the context of the transition to Industry 4.0, Predictive Maintenance (PdM) emerges as a key strategy to anticipate failures, reduce operational costs, and optimize the availability of industrial assets. This study presents a systematic review of recent works focused on approaches, methods, and challenges related to PdM, with particular emphasis on the integration of Artificial Intelligence (AI), the Internet of Things (IoT), and Big Data.\u003c/p\u003e\u003cp\u003eA distinctive contribution of this research lies in the development of a taxonomy of maintenance strategies, tracing the evolution from corrective and preventive approaches to predictive and prescriptive paradigms, thereby providing a structured framework for positioning PdM within Industry 4.0. In addition, the review is guided by a set of research questions formulated to better capture the stakes and challenges associated with PdM implementation at both the technical and organizational levels. The analysis classifies scientific contributions based on prediction models (physics-based, knowledge-based, data-driven, and hybrid), evaluates machine learning algorithms (Random Forest, SVM, deep neural networks, transformers, etc.), and identifies the main technical and industrial limitations. The findings reveal that despite technological advances, significant obstacles persist in real-time deployment, model robustness, heterogeneous data management, and cybersecurity. The article also outlines promising perspectives for future research, with particular attention to prescriptive maintenance, digital twins, and explainable artificial intelligence (XAI).\u003c/p\u003e","manuscriptTitle":"AI-Driven Predictive Maintenance for Industry 4.0: A Systematic Review of Models, Methods, and Challenges","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-30 14:19:36","doi":"10.21203/rs.3.rs-7618987/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2025-11-10T15:56:01+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-18T12:52:50+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-18T04:03:15+00:00","index":"","fulltext":""},{"type":"submitted","content":"The International Journal of Advanced Manufacturing Technology","date":"2025-09-15T05:46:21+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"the-international-journal-of-advanced-manufacturing-technology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jamt","sideBox":"Learn more about [The International Journal of Advanced Manufacturing Technology](https://www.springer.com/journal/170)","snPcode":"170","submissionUrl":"https://submission.nature.com/new-submission/170/3","title":"The International Journal of Advanced Manufacturing Technology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"7d299d89-c890-496e-9200-f3878975084b","owner":[],"postedDate":"September 30th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-02-23T16:01:30+00:00","versionOfRecord":{"articleIdentity":"rs-7618987","link":"https://doi.org/10.1007/s00170-026-17531-w","journal":{"identity":"the-international-journal-of-advanced-manufacturing-technology","isVorOnly":false,"title":"The International Journal of Advanced Manufacturing Technology"},"publishedOn":"2026-02-18 15:57:23","publishedOnDateReadable":"February 18th, 2026"},"versionCreatedAt":"2025-09-30 14:19:36","video":"","vorDoi":"10.1007/s00170-026-17531-w","vorDoiUrl":"https://doi.org/10.1007/s00170-026-17531-w","workflowStages":[]},"version":"v1","identity":"rs-7618987","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7618987","identity":"rs-7618987","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}