Advanced ML for Predictive Maintenance: Case Study on Remaining Useful Life Prediction and Reliability Enhancement

Advanced ML for Predictive Maintenance: Case Study on Remaining Useful Life Prediction and Reliability Enhancement | 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 Advanced ML for Predictive Maintenance: Case Study on Remaining Useful Life Prediction and Reliability Enhancement Anwar Meddaoui, Adil Hachmoud, Mustapha Hain This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3875020/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 11 Mar, 2024 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 order to achieve an optimal system performance, decision makers are continually faced with the responsibility of making choices that will enhance availability and reduce failures cost. To realize this goal, it is crucial to ensure the timely maintenance of equipment, which often poses a significant challenge. However, the adoption of Predictive Maintenance (PdM) technology can offer a solution by enabling real-time maintenance, resulting in various benefits such as reduced downtime, cost savings, and enhanced production quality. Machine learning (ML) techniques are increasingly being used in the field of predictive maintenance to predict failures and calculate estimated remaining useful life (RUL) of equipment. A case study is proposed in this research paper based on a maintenance dataset from the aerospace industry. It experiments and compare multiple combination of feature engineering techniques and advanced ML models with the aim to propose the most efficient techniques for prediction. Moreover, future research papers can focus on the challenge of validating this proposed model in different industrial environments. Predictive maintenance remaining useful life prediction machine learning Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction In today's industrial landscape, companies face numerous challenges that can significantly impact their operations. One of the key challenges is the maintenance of machines and equipment, which is crucial for ensuring optimal performance and minimizing downtime. Traditional maintenance approaches, such as preventive maintenance, often result in unnecessary maintenance actions and high costs. However, with the advent of predictive maintenance, companies now have the opportunity to address these challenges proactively. Predictive maintenance utilizes advanced technologies, such as data analytics and machine learning, to monitor equipment in real-time and predict potential failures before they occur. By implementing predictive maintenance strategies, companies can reduce maintenance costs, increase equipment reliability, and improve overall operational efficiency. This approach enables companies to shift from reactive maintenance practices to a proactive and preventive maintenance approach, thereby overcoming the challenges associated with unplanned downtime, excessive maintenance, and unpredictable failures. The traditional approach of preventive maintenance is commonly employed in maintenance management to prevent failures by following a predetermined schedule. However, this approach has its limitations in terms of cost-effectiveness, as it does not consider the dynamic condition of production equipment. Often, it results in either excessive or insufficient maintenance, leading to unnecessary replacements or disruptions in the production process [ 1 ]. Recent studies have demonstrated the significance of predictive maintenance in optimizing the industrial performance of the company, especially with the use of artificial intelligence [ 2 ]. The objective of predictive maintenance is to reduce costs associated with maintenance activities. Traditional maintenance approaches, such as preventive maintenance, often result in unnecessary maintenance actions, leading to increased costs for equipment replacements and unproductive downtime [ 3 ]. The Remaining Useful Life (RUL) refers to the estimated amount of time or usage an asset has before it is expected to fail or become unreliable. It serves as a crucial metric for maintenance planning and decision-making processes. Accurately predicting the RUL of assets enables organizations to optimize their maintenance schedules, plan for timely replacements or repairs, and minimize disruptions in operations. The concept of RUL is utilized to predict life-span of components (of a service system) with the purpose of minimizing catastrophic failure events in both manufacturing and service sectors. Prognostic is defined as the estimation of RUL (or time-to failure) of a component or system which can be filtered by existing or future failure modes [ 4 ]. The interaction between Machine Learning (ML) and Predictive Maintenance (PdM) is a key aspect in driving the effectiveness of PdM strategies. ML algorithms and techniques play a crucial role in analyzing large volumes of data collected from various sensors and sources to detect patterns, anomalies, and potential failure indicators. ML algorithms can be trained on historical data to learn the relationship between different variables and failure events. This allows for the development of predictive models that can accurately forecast the likelihood of equipment failure or the remaining useful life (RUL) of assets. By integrating ML with PdM, organizations can benefit from more accurate and timely predictions of equipment failures. This enables them to schedule maintenance activities proactively, optimize resource allocation, and reduce downtime. ML algorithms can also assist in identifying early warning signs of potential failures, allowing maintenance teams to intervene before a breakdown occurs. 2. Literature Review 2.1. PdM and predicting RUL: Predictive maintenance (PdM) is able to assess the condition of equipment to detect signs of failure and anticipate them, PdM could also bring several potential benefits in terms of dependability and cost performance and other benefits. Different approaches are proposed in the literature. Initially grounded in data, physics models, or existing knowledge, these approaches face persistent challenges and limitations. Specifically, there is a need to overcome the reliance on a specific context, incorporate data and business knowledge while accounting for the unique challenges of applying established solutions to new situations, address difficulties related to data analysis, and effectively manage uncertainties [ 5 ]. RUL is a crucial concept in the field of predictive maintenance, as it refers to the estimated time remaining before an asset or component is expected to fail or become unreliable. Accurately predicting the RUL of assets is essential for effective maintenance planning and decision-making. To predict the RUL, various techniques and algorithms can be employed. These methods typically utilize historical data, sensor measurements, and other relevant information to analyze the degradation patterns and behavior of the asset. Zonta et al. [ 6 ] described in their research that a smart industry is using the ML techniques in predictive maintenance to model RUL and predict failures. For Ferreira, C. and Gonçalves, G. [ 7 ] Over the past decade, there has been a significant increase in the research and application of data-driven models, particularly machine learning (ML) methods, for predicting the Remaining Useful Life (RUL) of equipment. This field of study has gained significant momentum since 2019, as evidenced by the publication of high-quality research papers in reputable journals. However, there is a noticeable gap in the literature regarding papers that directly address the RUL prediction process and provide a comprehensive evaluation of the advantages and drawbacks of the available methods. This lack of research hinders the understanding of the RUL prediction process and limits the potential for improvement in this area. Addressing these fundamental aspects, such as the RUL prediction process and the strengths and limitations of different methods, is crucial for advancing the field of RUL prediction. By systematically evaluating and comparing the available methods, researchers and practitioners can gain valuable insights and develop more effective and reliable RUL prediction models. 2.2. Machine Learning models: ML has emerged as a valuable approach in the field of Predictive Maintenance (PdM) and has been applied in various contexts. One example is the work of Cakir, Guvenc, and Mistikoglu [ 8 ], who utilized ML models to develop a method for classifying bearing damage. They collected data from sources such as vibration, sound, rotational speed, current, and temperature to monitor the condition of the bearings. ML algorithms, including Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbors (K-NN), Artificial Neural Network (ANN or NN) and Decision Tree (DT), were employed for analysis. The results demonstrated accuracies exceeding 99% using all of the algorithms. Other industries and products have also integrated ML into their PdM systems. These include heavy equipment engines [ 9 ], roads [ 10 ], and high voltage circuit breakers [ 11 ], among others. 2.3. Describing used ML models: 2.3.1. NN and K-NN NN is inspired by the human brain's structure and functioning. It consists of interconnected nodes or "neurons" that process and transmit information. It uses a nonlinear activation function to model complex relationships between input and output variables (Fig. 1 ). K-NN: is generally used for classification and regression tasks. It works by comparing a new data point with its nearest neighbors based on a distance metric. is a non-parametric algorithm that classifies a new data point based on the majority vote of its k nearest neighbors. It measures the proximity between data points using distance metrics such as Euclidean, Manhattan, or Minkowski distance. According to Lei et al. [ 12 ], NNs have demonstrated promising performance in predicting RUL of complex systems. It is worth noting that different types of NNs, including recurrent neural networks (RNNs), fuzzy neural networks (FNNs), and deep convolutional neural networks, have been successfully applied in machinery RUL estimation and health prognostics [ 13 ], [ 14 ], [ 15 ], [ 16 ], [ 17 ]. 2.3.2. RF and SVM RF is combining multiple decision trees to make predictions. Each decision tree is built using a random subset of the training data and a random subset of features (Fig. 2 ). SVM: is used for classification and regression tasks. It aims to find an optimal hyperplane that separates different classes or predicts a continuous output based on features. SVM models are effective in handling high-dimensional data and can handle both linear and nonlinear relationships (Fig. 3 ). These models can also be applied for classification purposes. In one study, Lingitz et al. conducted a comparative analysis of different regression techniques to estimate lead time in semiconductor manufacturing [ 18 ]. Wescoat et al. used RF to predict an anomalous condition on a cobot end-effector from purposeful failure data [ 19 ]. Additionally, Yang et al. employed SVM regression for fault prediction and detection in a wind turbine system [ 20 ]. 2.4. Using feature engineering in predictive maintenance Feature engineering (FE) plays a vital role in the realm of machine learning and data analysis. It entails the conversion of raw data into novel, meaningful features that amplify the effectiveness of predictive models. This encompasses the identification, generation, or alteration of pertinent variables (features) extracted from the original dataset. The ultimate objective is to enhance the model's capability to generate precise predictions or classifications. For The selection of optimal features is accomplished by either extracting them from the normalized maintenance data. This meticulous process ensures that the most pertinent features are identified. Subsequently, the finest features are meticulously selected to bolster the overall effectiveness and performance of the predictive maintenance system that is being provided [ 21 ]. 3. Case Study: A Case study using a maintenance dataset from the aerospace industry 3.1. Engine data used in the case study: The Element used in this case study is turbofan. The engine dataset comprises multiple multivariate time series, with each series representing a different engine. This wear and variation are considered normal and not indicative of a fault condition. The dataset includes three operational settings that significantly impact engine performance and containing other 26 features which corresponding to the different sensor variable (temperature, vibration, and others). 3.2. Experimental set design The experimental set design is showed in Fig. 4 , it is composed of four modules. The implemented scenarios enable the testing of 160 combinations of feature engineering and prediction models. This configuration takes the dataset of data in text file format as input. The dataset contains 21 sensors values (temperature, vibration, pressure...) and 3 attributes settings, each line represents one cycle of the tested engines. 3.2.1. Module 1: Extract Transform Load Module ETL (Extract Transform Load) as the module (M1) allows the extraction of data from text files, formatting them according to the corresponding data type, eliminating redundancies, removing rows containing missing values and loading data into database. 3.2.2. Module 2: The Feature Engineering module The Feature Engineering module (M2) functions differently based on whether it is integrated into the Regression or Classification pipeline. In the case of regression, the objective is to forecast the Remaining Useful Life (RUL) for an engine using sensor values at a specific moment (cycle). To achieve this, the module utilizes the max cycle (failure cycle) of each engine and cycle for each row to calculate the corresponding RUL for each cycle using the following formula: or the classification algorithm, the goal is to anticipate whether the engine is susceptible to failure within a 30-cycle period. The module employs the previously calculated RUL for regression to generate an additional binary attribute using the following formula: Classification is used when the target variable is categorical (binary), and the goal is to predict the class membership of an instance, the aim of our model is to help to predict failures occurrence. Regression, on the other hand, is used when the target variable is numerical, and the aim is to predict the Remaining Useful Life (RUL) of the studied equipment. The aim of this section is to show to predictive maintenance decision-makers that the choice of the best prediction model depends on the target being sought (failure, time, ...). Before normalizing the dataset's data, they were subjected to the K-means clustering algorithm, which allowed us to identify six clusters corresponding to different operating regimes of the engines (see Table 1 ). These clusters are characterized by combinations of values of the attributes "setting 1", "setting 2" and "setting 3" (Fig. 5 ). This allowed us to proceed with data normalization by operating regime (cluster) using the standard deviation and mean of each cluster. Table 1 The six clusters corresponding to different operating regime of the studied engines. Attribute cluster_0 cluster_1 cluster_2 cluster_3 cluster_4 cluster_5 Setting1 0,00150 42,00298 20,00300 25,00304 10,00297 35,00305 Setting2 0,00049 0,84049 0,70052 0,62050 0,25050 0,84050 Setting3 100 100 100 60 100 100 3.2.3. Module3: Feature selection This module allows for the selection of different percentages (20%, 40%, 60%, 80%, 100%) of the most influential attributes in prediction, using two methods: Principal Component Analysis (PCA) and AVG. PCA is applied to both Regression and Classification models. Weight by correlation is applied to Regression. AVG) average of weight by correlation, weight by Gini index, weight by information gain, and weight by information gain ratio is applied to classification algorithm. 3.2.4. Module 4: Application of Models and Hyperparameter Tuning Module: This module involves applying different prediction models to the various combinations generated by the previous modules: The classification models are: SVM, NN, RF, K-NN, the regression models are: LR, NN, RF, K-NN. Hyperparameter tuning involves selecting the best values among the different configurations of model parameters that yielded the best results. For each model, we chose the most influential parameter; for the SVM model, parameter C (regularization) was tested for values (0.001, 0.01, 0.1, 1, 2, 4, 8). For the NN model, the training cycles parameter was varied between 100 and 200. For the RF model, different values of the 'Number of Trees' parameter were tested within the range of integers between 10 and 20. The LR model is only influenced by the chosen feature selection method. The Cross Validation (CV) method with 20 folds is used for the model validation. The evaluation of regression models considers the following criteria: Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mean Squared Error (MSE). The evaluation of classification models considers the following criteria: accuracy, specificity, sensitivity, recall and precision. 3.3. Results and accuracy The different combinations of feature engineering and prediction models tested reveal that the best combination depends on the prediction scenario. It will be presented in this paper the results following the two scenarios, classification (target: failure and no failure) and regression (target: RUL). Classification: In terms of accuracy, the K-NN model applied to normalized data of 80% of attributes using the feature selection average method (by means of Weight by Correlation, Weight by Gini Index, Weight by Information Gain and Weight by Information Gain Ratio), is the best model with 95,92% of accuracy (see table 2). Otherwise, the Receiver Operating Characteristic curve (ROC), showed in the Fig. 6 , as a graphical representation that illustrates the performance, reveals the difference between the four predictive models evaluated in the same conditions as K-NN. Table 2. Performance indicators of K-NN as the best model applied on engine failure dataset. If the objective is to minimize false negatives (engines not reported as likely to break down) then the model evaluation criterion to be taken into consideration is "recall" (see table 3), in this case the best combination would be SVM model applied to normalized data of 40% of attributes using the feature selection average method (by means of Weight by Correlation, Weight by Gini Index, Weight by Information Gain and Weight by Information Gain Ratio) with 86,44% recall. Table 3. Performance indicators of SVM Regression: As mentioned in the table 4, the best combination in the regression axis in terms of Root Mean Squared Error (RMSE) is the RF model, applied to normalized data and a set of 20% of attributes using the PCA feature selection method. Table 4. Performance indicators of K-NN as the best model applied on engine failure dataset. 4. Conclusion Predictive maintenance is an advanced maintenance strategy that utilizes data analysis techniques to forecast equipment failures. By continuously monitoring equipment and analyzing historical data, predictive maintenance can identify patterns and predict potential issues before they occur. In this case study, the focus is on turbofan engine Dataset, where the performance of various powerful machine learning (ML) models is compared in accurately predicting the Remaining Useful Life (RUL) of the engines. The main objective is to determine the most effective model for maintenance prediction. The analysis of different feature engineering and prediction models reveals that the best combination depends on the specific prediction scenario, whether classification or regression. For classification, the K-Nearest Neighbors (K-NN) model applied to 80% of normalized attributes with feature selection achieves the highest accuracy of 95.92%. In regression, the Random Forest (RF) model with Principal Component Analysis (PCA) feature selection, applied to 20% of attributes, demonstrates the lowest Root Mean Squared Error (RMSE) of 34.39. Implementing predictive maintenance presents several challenges, such as ensuring data reliability, real-time data processing capabilities, and overcoming technical obstacles. Additionally, there is a need to develop simplified predictive maintenance models that are easily adoptable by industrial companies. Future research efforts could explore the potential of evolutionary algorithms in feature selection combined with other health indicators and deep learning models to address these challenges in predictive maintenance. Declarations Funding : The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. Data set used in the case study are based is confidential. Competing Interest: Authors of this paper have no financial interests and have no relevant financial or non-financial interests to disclose. Author Contributions: All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Professor Hachmoud Adil and Professor Meddaoui Anwar. References Ding, S.H. et al, Maintenance strategy optimization—literature review and directions, Int. J. Adv. Manuf. Technol., 2014. Toumi H, Meddaoui A, Hain M, The influence of predictive maintenance in industry 4.0: a systematic literature review. 2nd International Conference on Innovative Research in Applied Science, Engineering and Technology (IRASET), 2022. Meddaoui, A., Hachmoud, A., Hain, M., The benefits of predictive maintenance in manufacturing excellence: a case study to establish reliable methods for predicting failures, The International Journal of Advanced Manufacturing Technology, Volume 128, pages 3685–3690, 2023. Medjaher, K., Tobon-Mejia, D. A., Zerhouni, N., Remaining Useful Life Estimation of Critical Components with Application to Bearings. IEEE Journal of Transactions on Reliability, 61(2):292-302, 2012. Hafsi, M., Hamour, N., Ouchani, S., Predictive Maintenance for Smart Industrial Systems: A Roadmap, Procedia Computer Science 220, 645–650, 2023 Zonta, T., Da Costa, C. A., Da Rosa Righi, R., de Lima, M. J., Da Trindade E. S., Li, G. P., Predictive maintenance in the industry 4.0: A systematic literature review, Computers & Industrial Engineering 150, 2020. Ferreira, C., Gonçalves, G., Remaining Useful Life prediction and challenges: A literature review on the use of Machine Learning Methods, Journal of Manufacturing Systems, Volume 63, 550-562, 2022. Cakir, M. Guvenc, M. A., Mistikoglu, S., “The experimental application of popular machine learning algorithms on predictive maintenance and the design of IIoT based condition monitoring system,” Computers & Industrial Engineering, vol. 151, p. 106948, 2021. Joelian, A., “Engine replacement scheduling optimization using Data Mining,” Journal of Physics Conference Series, 2020. Silva, N., Soares, J., Shah, V., Santos M. Y, Rodrigues, H., “Anomaly detection in roads with a data mining approach,” Procedia computer science, pp. 415-422, 2017. Dehghanian, P., Guan, Y., Kezunovic, M., “Real-time life-cycle assessment of high-voltage circuit breakers for maintenance using online condition monitoring data,” IEEE Transactions on Industry Applications, pp. 1135-1146, 2018. Lei, Y., Li, N., Guo, L., Li, N., Yan, T., Lin, J., Machinery health prognostics: A systematic review from data acquisition to RUL prediction. Mech. Syst. Signal Process. 104, 799–834., 2018. Zemouri, R., Racoceanu, D., Zerhouni, N., 2003. Recurrent radial basis function network for time-series prediction. Eng. Appl. Artif. Intell. 16 (5–6), 453–463, 2003. Zemouri, R., Gouriveau, R., Towards accurate and reproducible predictions for prognostic: An approach combining a RRBF network and an AutoRegressive model. IFAC Proc. Vol. (IFAC-PapersOnline) 1 (PART 1), 140–145, 2010. Pan, Y., Er, M.J., Li, X., Yu, H., Gouriveau, R., Machine health condition prediction via online dynamic fuzzy neural networks. Eng. Appl. Artif. Intell. 35, 105–113, 2014. Lu, Y., Sun, L., Zhang, X., Feng, F., Kang, J., Fu, G., Condition based maintenance optimization for offshore wind turbine considering opportunities based on neural network approach. Appl. Ocean Res. 74, 69–79, 2018. Xia, M., Li, T., Shu, T., Wan, J., De Silva, C.W., Wang, Z., A two-stage approach for the remaining useful life prediction of bearings using deep neural networks. IEEE Trans. Ind. Inform. 15 (6), 3703–3711, 2019. Lingitz, L., Gallina, V., Ansari, F., Gyulai, D., Pfeiffer, A., and Sihn, W., “Lead Time Prediction Using Machine Learning Algorithms: A Case Study by a Semiconductor Manufacturer,” Procedia CIRP, 72, pp. 1051–1056, 2018. Wescoat, E., Krugh, M., Mears, L., Random Forest regression for predicting an anomalous condition on a UR10 cobot end-effector from purposeful failure data Procedia Manufacturing, 53, 644–655, 2021. Yang, C., Liu, J., Zeng, Y., and Xie, G., “Real-Time Condition Monitoring and Fault Detection of Components Based on MachineLearning Reconstruction Model,” Renew. Energy, 133, pp. 433–441, 2019. Chinta, V. S., Reddi, S. K., Yarramsetty, N., Optimal feature selection on Serial Cascaded deep learning for predictive maintenance system in automotive industry with fused optimization algorithm, Advanced Engineering Informatics, 24, PP. 102–105, 2023. Cite Share Download PDF Status: Published Journal Publication published 11 Mar, 2024 Read the published version in The International Journal of Advanced Manufacturing Technology → Version 1 posted Reviewers agreed at journal 22 Jan, 2024 Reviewers invited by journal 20 Jan, 2024 Editor assigned by journal 19 Jan, 2024 First submitted to journal 18 Jan, 2024 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-3875020","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":268354705,"identity":"cf704d47-c147-4196-b7ed-75b3efb40171","order_by":0,"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":268354706,"identity":"7a5882fd-751b-4728-bb4c-4200961e3380","order_by":1,"name":"Adil Hachmoud","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Adil","middleName":"","lastName":"Hachmoud","suffix":""},{"id":268354707,"identity":"bf9cb02b-1994-4302-a3ac-c65ee0e7241d","order_by":2,"name":"Mustapha Hain","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Mustapha","middleName":"","lastName":"Hain","suffix":""}],"badges":[],"createdAt":"2024-01-18 07:25:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3875020/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3875020/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00170-024-13351-y","type":"published","date":"2024-03-11T15:01:22+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":50049897,"identity":"2cec2b1d-8ce2-4931-a528-68366d383379","added_by":"auto","created_at":"2024-01-23 16:35:00","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":44696,"visible":true,"origin":"","legend":"\u003cp\u003eThe structure of an NN model\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3875020/v1/0d0c22fc03c4a88e52626303.jpg"},{"id":50049895,"identity":"a5d5d937-4e3e-4941-ba17-f6c5ff9dfc87","added_by":"auto","created_at":"2024-01-23 16:35:00","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":50366,"visible":true,"origin":"","legend":"\u003cp\u003eThe structure of an RF model\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3875020/v1/a4722d3756b386014d49e8af.jpg"},{"id":50049896,"identity":"dd3b87bd-2bc4-4756-a756-78e4c300b1e7","added_by":"auto","created_at":"2024-01-23 16:35:00","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":86750,"visible":true,"origin":"","legend":"\u003cp\u003eThe structure of an SVM model\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3875020/v1/d5633e032747277a7cba6231.jpg"},{"id":50049900,"identity":"5b4a71b4-cab7-4427-8595-e70443198bf2","added_by":"auto","created_at":"2024-01-23 16:35:00","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":125402,"visible":true,"origin":"","legend":"\u003cp\u003eThe proposed experimental set based on the four main modules.\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3875020/v1/e6cff94e8260a63ce179228a.jpg"},{"id":50050811,"identity":"0850a414-d712-4231-bb58-387072bde566","added_by":"auto","created_at":"2024-01-23 16:43:00","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":75606,"visible":true,"origin":"","legend":"\u003cp\u003eThe six clusters corresponding to different operating regime of the studied engines.\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3875020/v1/080e09fac6cdd3419d1fae38.jpg"},{"id":50049898,"identity":"b03a78f7-f5d5-4b40-b6ba-74ae6b42d271","added_by":"auto","created_at":"2024-01-23 16:35:00","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":113297,"visible":true,"origin":"","legend":"\u003cp\u003eThe ROC curve showing the most relevant predictive model in classification.\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3875020/v1/3b4358cec33cc03d7bf12980.jpg"},{"id":52907441,"identity":"512d3607-b2a9-4cfe-acac-228ada90646d","added_by":"auto","created_at":"2024-03-18 15:12:15","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":685723,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3875020/v1/5f8e0c5b-d481-4c68-acf5-781bcb7140a2.pdf"}],"financialInterests":"","formattedTitle":"Advanced ML for Predictive Maintenance: Case Study on Remaining Useful Life Prediction and Reliability Enhancement","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eIn today's industrial landscape, companies face numerous challenges that can significantly impact their operations. One of the key challenges is the maintenance of machines and equipment, which is crucial for ensuring optimal performance and minimizing downtime. Traditional maintenance approaches, such as preventive maintenance, often result in unnecessary maintenance actions and high costs. However, with the advent of predictive maintenance, companies now have the opportunity to address these challenges proactively. Predictive maintenance utilizes advanced technologies, such as data analytics and machine learning, to monitor equipment in real-time and predict potential failures before they occur. By implementing predictive maintenance strategies, companies can reduce maintenance costs, increase equipment reliability, and improve overall operational efficiency. This approach enables companies to shift from reactive maintenance practices to a proactive and preventive maintenance approach, thereby overcoming the challenges associated with unplanned downtime, excessive maintenance, and unpredictable failures. The traditional approach of preventive maintenance is commonly employed in maintenance management to prevent failures by following a predetermined schedule. However, this approach has its limitations in terms of cost-effectiveness, as it does not consider the dynamic condition of production equipment. Often, it results in either excessive or insufficient maintenance, leading to unnecessary replacements or disruptions in the production process [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eRecent studies have demonstrated the significance of predictive maintenance in optimizing the industrial performance of the company, especially with the use of artificial intelligence [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. The objective of predictive maintenance is to reduce costs associated with maintenance activities. Traditional maintenance approaches, such as preventive maintenance, often result in unnecessary maintenance actions, leading to increased costs for equipment replacements and unproductive downtime [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. The Remaining Useful Life (RUL) refers to the estimated amount of time or usage an asset has before it is expected to fail or become unreliable. It serves as a crucial metric for maintenance planning and decision-making processes. Accurately predicting the RUL of assets enables organizations to optimize their maintenance schedules, plan for timely replacements or repairs, and minimize disruptions in operations. The concept of RUL is utilized to predict life-span of components (of a service system) with the purpose of minimizing catastrophic failure events in both manufacturing and service sectors. Prognostic is defined as the estimation of RUL (or time-to failure) of a component or system which can be filtered by existing or future failure modes [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe interaction between Machine Learning (ML) and Predictive Maintenance (PdM) is a key aspect in driving the effectiveness of PdM strategies. ML algorithms and techniques play a crucial role in analyzing large volumes of data collected from various sensors and sources to detect patterns, anomalies, and potential failure indicators. ML algorithms can be trained on historical data to learn the relationship between different variables and failure events. This allows for the development of predictive models that can accurately forecast the likelihood of equipment failure or the remaining useful life (RUL) of assets. By integrating ML with PdM, organizations can benefit from more accurate and timely predictions of equipment failures. This enables them to schedule maintenance activities proactively, optimize resource allocation, and reduce downtime. ML algorithms can also assist in identifying early warning signs of potential failures, allowing maintenance teams to intervene before a breakdown occurs.\u003c/p\u003e"},{"header":"2. Literature Review","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1. PdM and predicting RUL:\u003c/h2\u003e \u003cp\u003ePredictive maintenance (PdM) is able to assess the condition of equipment to detect signs of failure and anticipate them, PdM could also bring several potential benefits in terms of dependability and cost performance and other benefits. Different approaches are proposed in the literature. Initially grounded in data, physics models, or existing knowledge, these approaches face persistent challenges and limitations. Specifically, there is a need to overcome the reliance on a specific context, incorporate data and business knowledge while accounting for the unique challenges of applying established solutions to new situations, address difficulties related to data analysis, and effectively manage uncertainties [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eRUL is a crucial concept in the field of predictive maintenance, as it refers to the estimated time remaining before an asset or component is expected to fail or become unreliable. Accurately predicting the RUL of assets is essential for effective maintenance planning and decision-making. To predict the RUL, various techniques and algorithms can be employed. These methods typically utilize historical data, sensor measurements, and other relevant information to analyze the degradation patterns and behavior of the asset. Zonta et al. [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] described in their research that a smart industry is using the ML techniques in predictive maintenance to model RUL and predict failures.\u003c/p\u003e \u003cp\u003eFor Ferreira, C. and Gon\u0026ccedil;alves, G. [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e] Over the past decade, there has been a significant increase in the research and application of data-driven models, particularly machine learning (ML) methods, for predicting the Remaining Useful Life (RUL) of equipment. This field of study has gained significant momentum since 2019, as evidenced by the publication of high-quality research papers in reputable journals. However, there is a noticeable gap in the literature regarding papers that directly address the RUL prediction process and provide a comprehensive evaluation of the advantages and drawbacks of the available methods. This lack of research hinders the understanding of the RUL prediction process and limits the potential for improvement in this area. Addressing these fundamental aspects, such as the RUL prediction process and the strengths and limitations of different methods, is crucial for advancing the field of RUL prediction. By systematically evaluating and comparing the available methods, researchers and practitioners can gain valuable insights and develop more effective and reliable RUL prediction models.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Machine Learning models:\u003c/h2\u003e \u003cp\u003eML has emerged as a valuable approach in the field of Predictive Maintenance (PdM) and has been applied in various contexts. One example is the work of Cakir, Guvenc, and Mistikoglu [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e], who utilized ML models to develop a method for classifying bearing damage. They collected data from sources such as vibration, sound, rotational speed, current, and temperature to monitor the condition of the bearings. ML algorithms, including Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbors (K-NN), Artificial Neural Network (ANN or NN) and Decision Tree (DT), were employed for analysis. The results demonstrated accuracies exceeding 99% using all of the algorithms. Other industries and products have also integrated ML into their PdM systems. These include heavy equipment engines [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e], roads [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], and high voltage circuit breakers [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], among others.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3. Describing used ML models:\u003c/h2\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e2.3.1. NN and K-NN\u003c/h2\u003e \u003cp\u003eNN is inspired by the human brain's structure and functioning. It consists of interconnected nodes or \"neurons\" that process and transmit information. It uses a nonlinear activation function to model complex relationships between input and output variables (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). K-NN: is generally used for classification and regression tasks. It works by comparing a new data point with its nearest neighbors based on a distance metric. is a non-parametric algorithm that classifies a new data point based on the majority vote of its k nearest neighbors. It measures the proximity between data points using distance metrics such as Euclidean, Manhattan, or Minkowski distance. According to Lei et al. [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e], NNs have demonstrated promising performance in predicting RUL of complex systems. It is worth noting that different types of NNs, including recurrent neural networks (RNNs), fuzzy neural networks (FNNs), and deep convolutional neural networks, have been successfully applied in machinery RUL estimation and health prognostics [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e], [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e], [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e2.3.2. RF and SVM\u003c/h2\u003e \u003cp\u003eRF is combining multiple decision trees to make predictions. Each decision tree is built using a random subset of the training data and a random subset of features (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). SVM: is used for classification and regression tasks. It aims to find an optimal hyperplane that separates different classes or predicts a continuous output based on features. SVM models are effective in handling high-dimensional data and can handle both linear and nonlinear relationships (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). These models can also be applied for classification purposes. In one study, Lingitz et al. conducted a comparative analysis of different regression techniques to estimate lead time in semiconductor manufacturing [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Wescoat et al. used RF to predict an anomalous condition on a cobot end-effector from purposeful failure data [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Additionally, Yang et al. employed SVM regression for fault prediction and detection in a wind turbine system [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.4. Using feature engineering in predictive maintenance\u003c/h2\u003e \u003cp\u003eFeature engineering (FE) plays a vital role in the realm of machine learning and data analysis. It entails the conversion of raw data into novel, meaningful features that amplify the effectiveness of predictive models. This encompasses the identification, generation, or alteration of pertinent variables (features) extracted from the original dataset. The ultimate objective is to enhance the model's capability to generate precise predictions or classifications. For The selection of optimal features is accomplished by either extracting them from the normalized maintenance data. This meticulous process ensures that the most pertinent features are identified. Subsequently, the finest features are meticulously selected to bolster the overall effectiveness and performance of the predictive maintenance system that is being provided [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Case Study: A Case study using a maintenance dataset from the aerospace industry","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1. Engine data used in the case study:\u003c/h2\u003e\n \u003cp\u003eThe Element used in this case study is turbofan. The engine dataset comprises multiple multivariate time series, with each series representing a different engine. This wear and variation are considered normal and not indicative of a fault condition. The dataset includes three operational settings that significantly impact engine performance and containing other 26 features which corresponding to the different sensor variable (temperature, vibration, and others).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2. Experimental set design\u003c/h2\u003e\n \u003cp\u003eThe experimental set design is showed in Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e, it is composed of four modules. The implemented scenarios enable the testing of 160 combinations of feature engineering and prediction models. This configuration takes the dataset of data in text file format as input. The dataset contains 21 sensors values (temperature, vibration, pressure...) and 3 attributes settings, each line represents one cycle of the tested engines.\u003c/p\u003e\n \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e\n \u003ch2\u003e3.2.1. Module 1: Extract Transform Load Module\u003c/h2\u003e\n \u003cp\u003eETL (Extract Transform Load) as the module (M1) allows the extraction of data from text files, formatting them according to the corresponding data type, eliminating redundancies, removing rows containing missing values and loading data into database.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec13\" class=\"Section3\"\u003e\n \u003ch2\u003e3.2.2. Module 2: The Feature Engineering module\u003c/h2\u003e\n \u003cp\u003eThe Feature Engineering module (M2) functions differently based on whether it is integrated into the Regression or Classification pipeline. In the case of regression, the objective is to forecast the Remaining Useful Life (RUL) for an engine using sensor values at a specific moment (cycle). To achieve this, the module utilizes the max cycle (failure cycle) of each engine and cycle for each row to calculate the corresponding RUL for each cycle using the following formula:\u003c/p\u003e\n \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" style=\"width: 737px; height: 38.5411px;\" width=\"737\" height=\"38.5411\"\u003e\u003cbr\u003e\u003c/p\u003e\n \u003cp\u003eor the classification algorithm, the goal is to anticipate whether the engine is susceptible to failure within a 30-cycle period. The module employs the previously calculated RUL for regression to generate an additional binary attribute using the following formula:\u003c/p\u003e\n \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" style=\"width: 686px; height: 55.0857px;\" width=\"686\" height=\"55.0857\"\u003e\u003cbr\u003e\u003c/p\u003e\n \u003cp\u003eClassification is used when the target variable is categorical (binary), and the goal is to predict the class membership of an instance, the aim of our model is to help to predict failures occurrence. Regression, on the other hand, is used when the target variable is numerical, and the aim is to predict the Remaining Useful Life (RUL) of the studied equipment. The aim of this section is to show to predictive maintenance decision-makers that the choice of the best prediction model depends on the target being sought (failure, time, ...).\u003c/p\u003e\n \u003cp\u003eBefore normalizing the dataset\u0026apos;s data, they were subjected to the K-means clustering algorithm, which allowed us to identify six clusters corresponding to different operating regimes of the engines (see Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e). These clusters are characterized by combinations of values of the attributes \u0026quot;setting 1\u0026quot;, \u0026quot;setting 2\u0026quot; and \u0026quot;setting 3\u0026quot; (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e). This allowed us to proceed with data normalization by operating regime (cluster) using the standard deviation and mean of each cluster.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003cdiv align=\"char\" class=\"colspec\"\u003e\u003cbr\u003e\u003c/div\u003e\u0026nbsp;\u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eThe six clusters corresponding to different operating regime of the studied engines.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"7\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAttribute\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ecluster_0\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ecluster_1\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ecluster_2\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ecluster_3\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ecluster_4\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ecluster_5\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSetting1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0,00150\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e42,00298\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20,00300\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e25,00304\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10,00297\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e35,00305\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSetting2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0,00049\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0,84049\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0,70052\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0,62050\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0,25050\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0,84050\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSetting3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e100\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e100\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e100\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e100\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e100\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e\n \u003ch2\u003e3.2.3. Module3: Feature selection\u003c/h2\u003e\n \u003cp\u003eThis module allows for the selection of different percentages (20%, 40%, 60%, 80%, 100%) of the most influential attributes in prediction, using two methods:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003ePrincipal Component Analysis (PCA) and AVG. PCA is applied to both Regression and Classification models.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eWeight by correlation is applied to Regression.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eAVG) average of weight by correlation, weight by Gini index, weight by information gain, and weight by information gain ratio is applied to classification algorithm.\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e\n \u003ch2\u003e3.2.4. Module 4: Application of Models and Hyperparameter Tuning Module:\u003c/h2\u003e\n \u003cp\u003eThis module involves applying different prediction models to the various combinations generated by the previous modules: The classification models are: SVM, NN, RF, K-NN, the regression models are: LR, NN, RF, K-NN.\u003c/p\u003e\n \u003cp\u003eHyperparameter tuning involves selecting the best values among the different configurations of model parameters that yielded the best results. For each model, we chose the most influential parameter; for the SVM model, parameter C (regularization) was tested for values (0.001, 0.01, 0.1, 1, 2, 4, 8). For the NN model, the training cycles parameter was varied between 100 and 200. For the RF model, different values of the \u0026apos;Number of Trees\u0026apos; parameter were tested within the range of integers between 10 and 20. The LR model is only influenced by the chosen feature selection method. The Cross Validation (CV) method with 20 folds is used for the model validation.\u003c/p\u003e\n \u003cp\u003eThe evaluation of regression models considers the following criteria: Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mean Squared Error (MSE). The evaluation of classification models considers the following criteria: accuracy, specificity, sensitivity, recall and precision.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\n \u003ch2\u003e3.3. Results and accuracy\u003c/h2\u003e\n \u003cp\u003eThe different combinations of feature engineering and prediction models tested reveal that the best combination depends on the prediction scenario. It will be presented in this paper the results following the two scenarios, classification (target: failure and no failure) and regression (target: RUL).\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eClassification:\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003cp\u003eIn terms of accuracy, the K-NN model applied to normalized data of 80% of attributes using the feature selection average method (by means of Weight by Correlation, Weight by Gini Index, Weight by Information Gain and Weight by Information Gain Ratio), is the best model with 95,92% of accuracy (see table 2). Otherwise, the Receiver Operating Characteristic curve (ROC), showed in the Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e, as a graphical representation that illustrates the performance, reveals the difference between the four predictive models evaluated in the same conditions as K-NN.\u003c/p\u003e\n \u003cp\u003eTable 2. Performance indicators of K-NN as the best model applied on engine failure dataset.\u003c/p\u003e\n \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" style=\"width: 799px; height: 133.167px;\" width=\"799\" height=\"133.167\"\u003e\u003cbr\u003e\u003c/p\u003e\n \u003cp\u003eIf the objective is to minimize false negatives (engines not reported as likely to break down) then the model evaluation criterion to be taken into consideration is \u0026quot;recall\u0026quot; (see table 3), in this case the best combination would be SVM model applied to normalized data of 40% of attributes using the feature selection average method (by means of Weight by Correlation, Weight by Gini Index, Weight by Information Gain and Weight by Information Gain Ratio) with 86,44% recall.\u003c/p\u003e\n \u003cp\u003eTable 3. Performance indicators of SVM\u003c/p\u003e\n \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" style=\"width: 761px; height: 138.364px;\" width=\"761\" height=\"138.364\"\u003e\u003cbr\u003e\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eRegression:\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003cp\u003eAs mentioned in the table 4, the best combination in the regression axis in terms of Root Mean Squared Error (RMSE) is the RF model, applied to normalized data and a set of 20% of attributes using the PCA feature selection method.\u003c/p\u003e\n \u003cp\u003eTable 4. Performance indicators of K-NN as the best model applied on engine failure dataset.\u003c/p\u003e\n \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" style=\"width: 790px; height: 100.249px;\" width=\"790\" height=\"100.249\"\u003e\u003cbr\u003e\u003c/p\u003e\n\u003c/div\u003e"},{"header":"4. Conclusion","content":"\u003cp\u003ePredictive maintenance is an advanced maintenance strategy that utilizes data analysis techniques to forecast equipment failures. By continuously monitoring equipment and analyzing historical data, predictive maintenance can identify patterns and predict potential issues before they occur.\u003c/p\u003e \u003cp\u003eIn this case study, the focus is on turbofan engine Dataset, where the performance of various powerful machine learning (ML) models is compared in accurately predicting the Remaining Useful Life (RUL) of the engines. The main objective is to determine the most effective model for maintenance prediction.\u003c/p\u003e \u003cp\u003eThe analysis of different feature engineering and prediction models reveals that the best combination depends on the specific prediction scenario, whether classification or regression. For classification, the K-Nearest Neighbors (K-NN) model applied to 80% of normalized attributes with feature selection achieves the highest accuracy of 95.92%. In regression, the Random Forest (RF) model with Principal Component Analysis (PCA) feature selection, applied to 20% of attributes, demonstrates the lowest Root Mean Squared Error (RMSE) of 34.39.\u003c/p\u003e \u003cp\u003eImplementing predictive maintenance presents several challenges, such as ensuring data reliability, real-time data processing capabilities, and overcoming technical obstacles. Additionally, there is a need to develop simplified predictive maintenance models that are easily adoptable by industrial companies. Future research efforts could explore the potential of evolutionary algorithms in feature selection combined with other health indicators and deep learning models to address these challenges in predictive maintenance.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003e\u003cu\u003eFunding :\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that no funds, grants, or other support were received during the preparation of this manuscript.\u003c/p\u003e\n\u003cp\u003eData set used in the case study are based is confidential.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cu\u003eCompeting Interest:\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAuthors of this paper have no financial interests and have no relevant financial or non-financial interests to disclose.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cu\u003eAuthor Contributions:\u003c/u\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Professor Hachmoud Adil and Professor Meddaoui Anwar.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eDing, S.H. et al, Maintenance strategy optimization\u0026mdash;literature review and directions, Int. J. Adv. Manuf. Technol., 2014.\u003c/li\u003e\n\u003cli\u003eToumi H, Meddaoui A, Hain M, The influence of predictive maintenance in industry 4.0: a systematic literature review. 2nd International Conference on Innovative Research in Applied Science, Engineering and Technology (IRASET), 2022.\u003c/li\u003e\n\u003cli\u003eMeddaoui, A., Hachmoud, A., Hain, M., The benefits of predictive maintenance in manufacturing excellence: a case study to establish reliable methods for predicting failures, The International Journal of Advanced Manufacturing Technology, Volume 128, pages 3685\u0026ndash;3690, 2023.\u003c/li\u003e\n\u003cli\u003eMedjaher, K., Tobon-Mejia, D. A., Zerhouni, N., Remaining Useful Life Estimation of Critical Components with Application to Bearings. IEEE Journal of Transactions on Reliability, 61(2):292-302, 2012.\u003c/li\u003e\n\u003cli\u003eHafsi, M., Hamour, N., Ouchani, S., Predictive Maintenance for Smart Industrial Systems: A Roadmap, Procedia Computer Science 220, 645\u0026ndash;650, 2023\u003c/li\u003e\n\u003cli\u003eZonta, T., Da Costa, C. A., Da Rosa Righi, R., de Lima, M. J., Da Trindade E. S., Li, G. P., Predictive maintenance in the industry 4.0: A systematic literature review, Computers \u0026amp; Industrial Engineering 150, 2020.\u003c/li\u003e\n\u003cli\u003eFerreira, C., Gon\u0026ccedil;alves, G., Remaining Useful Life prediction and challenges: A literature review on the use of Machine Learning Methods, Journal of Manufacturing Systems, Volume 63, 550-562, 2022.\u003c/li\u003e\n\u003cli\u003eCakir, M. Guvenc, M. A., Mistikoglu, S., \u0026ldquo;The experimental application of popular machine learning algorithms on predictive maintenance and the design of IIoT based condition monitoring system,\u0026rdquo; Computers \u0026amp; Industrial Engineering, vol. 151, p. 106948, 2021. \u003c/li\u003e\n\u003cli\u003eJoelian, A., \u0026ldquo;Engine replacement scheduling optimization using Data Mining,\u0026rdquo; Journal of Physics Conference Series, 2020. \u003c/li\u003e\n\u003cli\u003eSilva, N., Soares, J., Shah, V., Santos M. Y, Rodrigues, H., \u0026ldquo;Anomaly detection in roads with a data mining approach,\u0026rdquo; Procedia computer science, pp. 415-422, 2017. \u003c/li\u003e\n\u003cli\u003eDehghanian, P., Guan, Y., Kezunovic, M., \u0026ldquo;Real-time life-cycle assessment of high-voltage circuit breakers for maintenance using online condition monitoring data,\u0026rdquo; IEEE Transactions on Industry Applications, pp. 1135-1146, 2018.\u003c/li\u003e\n\u003cli\u003eLei, Y., Li, N., Guo, L., Li, N., Yan, T., Lin, J., Machinery health prognostics: A systematic review from data acquisition to RUL prediction. Mech. Syst. Signal Process. 104, 799\u0026ndash;834., 2018.\u003c/li\u003e\n\u003cli\u003eZemouri, R., Racoceanu, D., Zerhouni, N., 2003. Recurrent radial basis function network for time-series prediction. Eng. Appl. Artif. Intell. 16 (5\u0026ndash;6), 453\u0026ndash;463, 2003.\u003c/li\u003e\n\u003cli\u003eZemouri, R., Gouriveau, R., Towards accurate and reproducible predictions for prognostic: An approach combining a RRBF network and an AutoRegressive model. IFAC Proc. Vol. (IFAC-PapersOnline) 1 (PART 1), 140\u0026ndash;145, 2010.\u003c/li\u003e\n\u003cli\u003ePan, Y., Er, M.J., Li, X., Yu, H., Gouriveau, R., Machine health condition prediction via online dynamic fuzzy neural networks. Eng. Appl. Artif. Intell. 35, 105\u0026ndash;113, 2014.\u003c/li\u003e\n\u003cli\u003eLu, Y., Sun, L., Zhang, X., Feng, F., Kang, J., Fu, G., Condition based maintenance optimization for offshore wind turbine considering opportunities based on neural network approach. Appl. Ocean Res. 74, 69\u0026ndash;79, 2018.\u003c/li\u003e\n\u003cli\u003eXia, M., Li, T., Shu, T., Wan, J., De Silva, C.W., Wang, Z., A two-stage approach for the remaining useful life prediction of bearings using deep neural networks. IEEE Trans. Ind. Inform. 15 (6), 3703\u0026ndash;3711, 2019.\u003c/li\u003e\n\u003cli\u003eLingitz, L., Gallina, V., Ansari, F., Gyulai, D., Pfeiffer, A., and Sihn, W., \u0026ldquo;Lead Time Prediction Using Machine Learning Algorithms: A Case Study by a Semiconductor Manufacturer,\u0026rdquo; Procedia CIRP, 72, pp. 1051\u0026ndash;1056, 2018.\u003c/li\u003e\n\u003cli\u003eWescoat, E., Krugh, M., Mears, L., Random Forest regression for predicting an anomalous condition on a UR10 cobot end-effector from purposeful failure data Procedia Manufacturing, 53, 644\u0026ndash;655, 2021.\u003c/li\u003e\n\u003cli\u003eYang, C., Liu, J., Zeng, Y., and Xie, G., \u0026ldquo;Real-Time Condition Monitoring and Fault Detection of Components Based on MachineLearning Reconstruction Model,\u0026rdquo; Renew. Energy, 133, pp. 433\u0026ndash;441, 2019.\u003c/li\u003e\n\u003cli\u003eChinta, V. S., Reddi, S. K., Yarramsetty, N., Optimal feature selection on Serial Cascaded deep learning for predictive maintenance system in automotive industry with fused optimization algorithm, Advanced Engineering Informatics, 24, PP. 102\u0026ndash;105, 2023.\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, remaining useful life, prediction, machine learning","lastPublishedDoi":"10.21203/rs.3.rs-3875020/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3875020/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn order to achieve an optimal system performance, decision makers are continually faced with the responsibility of making choices that will enhance availability and reduce failures cost. To realize this goal, it is crucial to ensure the timely maintenance of equipment, which often poses a significant challenge. However, the adoption of Predictive Maintenance (PdM) technology can offer a solution by enabling real-time maintenance, resulting in various benefits such as reduced downtime, cost savings, and enhanced production quality. Machine learning (ML) techniques are increasingly being used in the field of predictive maintenance to predict failures and calculate estimated remaining useful life (RUL) of equipment. A case study is proposed in this research paper based on a maintenance dataset from the aerospace industry. It experiments and compare multiple combination of feature engineering techniques and advanced ML models with the aim to propose the most efficient techniques for prediction. Moreover, future research papers can focus on the challenge of validating this proposed model in different industrial environments.\u003c/p\u003e","manuscriptTitle":"Advanced ML for Predictive Maintenance: Case Study on Remaining Useful Life Prediction and Reliability Enhancement","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-01-23 16:34:55","doi":"10.21203/rs.3.rs-3875020/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2024-01-23T04:31:43+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-01-20T16:59:58+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-01-19T08:44:36+00:00","index":"","fulltext":""},{"type":"submitted","content":"The International Journal of Advanced Manufacturing Technology","date":"2024-01-18T12:44:06+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":"January 23rd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2024-03-18T15:06:59+00:00","versionOfRecord":{"articleIdentity":"rs-3875020","link":"https://doi.org/10.1007/s00170-024-13351-y","journal":{"identity":"the-international-journal-of-advanced-manufacturing-technology","isVorOnly":false,"title":"The International Journal of Advanced Manufacturing Technology"},"publishedOn":"2024-03-11 15:01:22","publishedOnDateReadable":"March 11th, 2024"},"versionCreatedAt":"2024-01-23 16:34:55","video":"","vorDoi":"10.1007/s00170-024-13351-y","vorDoiUrl":"https://doi.org/10.1007/s00170-024-13351-y","workflowStages":[]},"version":"v1","identity":"rs-3875020","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3875020","identity":"rs-3875020","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}