Harmonic Mitigation of the Microgrid Using Multi Hidden Layer Recurrent Neural Network Controlled SAPF

Harmonic Mitigation of the Microgrid Using Multi Hidden Layer Recurrent Neural Network Controlled SAPF | 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 Harmonic Mitigation of the Microgrid Using Multi Hidden Layer Recurrent Neural Network Controlled SAPF Divya Banu. P, G Sundar This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3971795/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Recently, renewable energy systems are more prevalent than traditional energy systems. Especially, Photovoltaic (PV) systems and wind power systems are playing an important part in meeting the world's energy needs. The present needs of more nonlinear loads drastically create power quality issues in the grid-connected system. As a result, it causes an undesired power quality defect that takes the manner of an alteration in the amplitude and patterns of voltage and current in the power system. The shunt active power filter (SAPF) delivers the appropriate current in shunt to grids while suppressing harmonics created by grid-tied nonlinear loads. This paper proposes a multi hidden layer recurrent neural network (RNN) controller for the developing a reference signal in synchronous reference frame theory. RNN is proposed for its high efficacy compared to neural networks, due to self-loops and memory. The effectiveness of the proposed multi-hidden layer RNN-based SAPF system is compared with the performance of neural networks and traditional PI controller-based SAPF systems in terms of harmonic mitigation. Effective harmonic mitigation by proposed multi-hidden layer RNN results current THD of proposed microgrid to 1.74%. Power quality Photovoltaic wind power systems SAPF neural networks multi hidden layer RNN Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 1. Introduction Renewable energy integration has garnered considerable interest due to its zero-fuel price, cleanliness, accessibility, and simplicity of installation. Among the many renewable energy sources, photovoltaic (PV) and wind turbines (WT) are among the most enticing due to their abundant accessibility in nature, technological progress, and monetary advantages. The hybrid combination of both scattered energy sources reduces mutual interruptions caused by their unfavourable characteristics, thereby enhancing system dependability. Nonlinear loads such as switched mode power supply, uninterrupted power supply, adjustable speed drive and converter-based electronics applications are causing redundant power quality (PQ) disturbances in grid-tied systems [ 1 – 3 ]. PQ issues have been the focus of a lot of research, with potential remedies including power filters and customized power devices. Recently, active filtering systems for PQ enrichment have established to be more effective than passive approaches due to their faster response time, lower size, and greater performance. Moreover, active filtering adapts to changes in network properties automatically, limiting the potential of resonance among the filter and the network impedance [ 4 ]. Active power filters outperform other harmonic compensation systems in terms of real tracking performance, compensation capabilities, and harmonic suppression. APF is one of the most essential tools for addressing the power grid's harmonic problem [ 5 ]. Efficiency of harmonic suppression is decided by the compensation current produced by SAPF, where DC link voltage controller plays crucial role. Accuracy of DC link voltage controller decides the accuracy of compensation. Hence in this article various soft computing methods are analysed as DC link voltage controller to enrich mitigation of harmonics in microgrid system. The universal approximation of neural networks (NN), which allows them to estimate any indefinite smooth nonlinear function, makes them a popular choice for nonlinear systems. In the active power filter [ 6 ], NN was employed to approximate the nonlinear portion. Many researchers have adjusted the topology of neural networks in recent study to increase approximation precision, like boosting feedback items or increasing the quantity of hidden layers. For a class of dynamic systems, a double-hidden-layer neural network was created [ 7 ]. Self-loops in RNNs occur when some processed data is fed back into the same or previous layers. The RNN uses its internal state to store this data, providing the model with modifiable memory that can be carried over across time. Consequently, a recursive network has greater efficacy than a feed-forward network. In this article multi-layer RNN is proposed for compensation. In order to approximate the unidentified nonlinear component of APF, a novel multi-hidden-layer RNN is being developed with numerous computational layers that can approximate any incessant function with arbitrary precision. When compared to the single-hidden-layer NN, the double-hidden-layer RNN is a deep neural network that is able to approximate with high accuracy while using fewer nodes. Hence in this research multi-hidden-layer RNN is proposed for SAPF control in grid connected PV-Wind hybrid power system with nonlinear load. In this article proposed microgrid is presented section 2 . SAPF and it’s control algorithm are discussed in section 3 . Section 4 discusses various controllers as DC link voltage controller in SAPF. Simulation results are analysed in section 5 and conclusion is presented in section 6 . 2. Proposed Microgrid with SAPF It is planned to connect PV and wind power systems to the grid. The low voltage output of the photovoltaic panels can be further increased by linking the array of panels to a DC-DC boost converter. Several approaches for tracking maximum power points (MPPT) have been addressed in the literature, the most common of which are incremental conductance and perturbation and observation [ 8 ]. In this analysis incremental conductance algorithm is employed to maximize the output power of the solar panel. A wind power system consists of a permanent magnet synchronous generator coupled to an adjustable speed wind turbine [ 9 ]. The rectifier converts the output of a three-phase voltage into a direct current voltage. A DC-DC buck-boost converter is linked to the rectifier's output to keep the DC voltage stable regardless of changes in the rectified DC voltage. When the rectifier's output is routed to a DC-DC buck-boost converter, the output voltage is stabilised regardless of the DC voltage being rectified. Using the tip speed ratio method, the wind power system maximum power was determined. Through the utilisation of a shunt connected transformer, SAPF is capable to inject the compensating current in parallel with the source. Figure 1 illustrates the PV and wind energy are the green energy system connected with the grid. The quantity of energy present in the DC-link has a significant impact on the reimbursement of the SAPF, an inverter which comprises IGBT and DC-link capacitor. 3. Shunt active power filter Active filters can be divided into two primary categories: shunt type and series type. SAPFs have been shown to reduce current harmonics well. They have a good harmonic reduction performance and are simple to implement. Shunt Active Power Filters with the proper control can solve all source current-related power quality problems [ 8 ]. The SAPF is controlled so that it will fully satisfy the nonlinear loads' need for reactive power, maintaining unity power factor at the source side. The reference current waveforms for the SAPF can be created using a variety of control techniques. One of the well-liked control strategies that is simple to put into practice and results in a sufficient decrease in THD is control based on Synchronous reference frame (SRF) algorithm. Accuracy of compensation or quality of harmonic mitigation of SAPF is decided by the DC link voltage controller. In this research two soft computing methods such as ANN and multi-layer RNN are proposed as DC link voltage controller. Performance of proposed multi-layer RNN and ANN based system are compared with conventional PI controller-based system. 3.1. Control algorithm for shunt active filter Figure 2 depicts the present SAPF control system based on the synchronous reference detection mechanism. In the beginning, using the park transform [ 10 ], the three-phase line currents ia, ib, and ic undergo a conversion from the three-phase (a-b-c) reference frame to the two-phase (d-q) stationary reference frame currents I d and I q. The park transformation for abc-dq computation is shown in Eq. ( 1 ). $$\left[{i}_{d} {i}_{q} \right]=\frac{2}{3}\left[sinsin \omega t coscos \omega t sinsin \omega t-\frac{2\pi }{3} coscos \omega t-\frac{2\pi }{3} sinsin \omega t+\frac{2\pi }{3} coscos \omega t+\frac{2\pi }{3} \right]\left[{i}_{a} {i}_{b} {i}_{c} \right]$$ 1 The fundamental current element is converted to an incessant component by the park transformation, whereas the harmonic current elements experience a change in the frequency spectrum. The incessant element can be removed by using a Butterworth low pass filter (50Hz). Higher order harmonics will be helped to be removed by the low pass filter. By applying the inverse transform on parks synchronised with network frequency, harmonic current references can be generated. The current in ( d-q ) reference frame is given by [ 11 ]: $$\left[{{i}_{a}}^{*} {{i}_{b}}^{*} {{i}_{c}}^{*} \right]=\left[sinsin \omega t coscos \omega t sinsin \omega t-\frac{2\pi }{3} sinsin \omega t+\frac{2\pi }{3} coscos \omega t-\frac{2\pi }{3} coscos \omega t+\frac{2\pi }{3} \right]\left[{i}_{d} {i}_{q} \right]$$ 2 To improve accuracy of compensation by tuning I d and regulate the DC link voltage V dc , various controllers proposed in this analysis. 4. DC link voltage controller The DC side capacitor's primary function is to supply the real power difference between the load and source during the transient period [ 12 ], but it also serves as a steady-state DC voltage regulator. Whenever the load varies, the voltage across the dc link capacitor also shifts. A controller is employed to achieve the sought-after voltage regulation. In conventional system, PI controller is employed. In this analysis artificial intelligent controllers such as ANN and recurrent neural network (RNN) are proposed as DC link voltage controller. 4.1 PI controller in SAPF PI controller is a basic controller greatly applied in different uses of the power framework. This basic regulator sets aside least effort to react. Steady-state error is dense effectively by using the PI regulator. In SAPF, DC link voltage error is considered as a contribution to the PI regulator [ 13 ]. Here is an expression for a PI controller, $$U\left(s\right)={K}_{p}E\left(s\right)+\frac{{K}_{i}}{s}E\left(s\right)$$ 3 The gains of the controller, K p & K i , are denoted, whereas E(s) and U(s) represent the user input and outcome of the controller, respectively. The gain esteems tuned utilizing Ziegler-Nichols' technique. As stated in previous section PI controller produces tune I d * reference current. 4.2 ANN in SAPF In this analysis, an ANN controller is proposed as DC link voltage controller as in the active power filter component [ 14 ]. Synchronous reference frame theory is employed in active power filters for diminishing harmonic, and DC voltage error is treated by a PI controller. In this analysis, an ANN controller is developed to control I d * to enrich quality of power. The back-propagation algorithm of the Mean Square Error (MSE) amid the yield and the target data is used to train the ANN. In this investigation, the Levenberg-Marquardt approach is used to train the ANN, since it takes the least amount of time. The training set for ANN was created off-line using data obtained from a PI controller-based system. To obtain an accurate output from this analysis to train network, 7230 datas are assumed. Figure 6 depicts a Structure of three layer ANN. To minimise the duration of processing as much as possible, the hidden layer is built with only sixteen neurons. ANN applied in SAPF system ends training with 34 epochs with MSE of 33 X 10 − 1 in 1s. It is proposed in SAPF to enhance quality of power. 4.3 Multiple layer RNN in SAPF RNNs are a sort of neural network that uses previous data from a time series to make predictions about the future [ 15 ]. Recurrent Neural Networks (RNNs) function based on the principle of retaining the output of a particular layer and reintroducing it as input in order to predict the output of that layer. Delaying inputs and outputs is a typical method for storing temporal information using static neural networks. This representation, however, is constrained subsequently it can only store a limited number of the previously measured outputs and enforced inputs. In addition, it frequently necessitates unrealistically high memory requirements, which prevents it from being used for all but comparatively lesser order dynamical systems. The usage of dynamic or recurrent neural networks has been investigated by the international research community as a very effective and promising alternative. In contrast to static neural networks, RNN feature at least one feedback loop. This type of neural network's topologies, learning techniques, and applications are covered in one of the earliest assessments in [ 6 ]. There, it is indicated that feedback-containing neural networks have existed since the very beginning of ANN. Indeed, McCulloch and Pitts created systems for time-dependent and time-delayed feedforward networks in [ 16 ], however these networks were executed using threshold logic neurons. After that, they expanded their system to include dynamic memory networks which comprises feedback. Later, commonly regarded as the first study on these types of mechanisms, these networks were modelled as determinate mechanisms with a regular language. Among various methods of training RNN, the scaled conjugate gradient technique employs a rapid strategy and makes more deliberate decisions regarding the search direction and step size. Hence in this analysis scaled conjugate gradient method is proposed for its fast performance, which determines the generation of compensation signal in SAPF. Multi hidden layer RNN proposed in SAPF is exposed in Fig. 5 . The RNN in the SAPF system training converge in 1m with an MSE of 6.03 X 10 − 1 and 5000 epochs. In SAPF, this trained RNN is recommended to increase power quality. 5. Simulation Results and Analysis Entire power system with SAPF is analysed using MATLAB/Simulink as shown in Fig. 6 . Renewable energy system with grid supplies nonlinear load of 10kw. Specification of analysed system is three phase, 410V, 50 Hz, 10kW load. Performance of power system with nonlinear load is shown in Fig. 7. From the Fig. 7, it is observed that nonlinear load results high distortions in source current. Influence of nonlinear load changes shape of source current and result high harmonics of 17.92%. Performance of SAPF is analysed using PI, ANN and proposed RNN based DC link controllers under variable nonlinear load. Performance of SAPF using PI controller is presented in Fig. 8. From the Fig. 8, it is clear that source current shape is improved with the help of SAPF. As depicted in Fig. 8 source voltage THD (THDV) which is 2.18% and THD of source current (THDI) is 3.34% which are pretty much lesser than 5% of IEEE standard. Performance of SAPF using ANN controller is presented in Fig. 9 . Figure 9 shows that, ANN based SAPF effectively improves source current wave shape and reduces THD. THDV of ANN based SAPF is 2.01% while it was 2.18% by PI based SAPF. Meantime effectiveness of SAPF in source current harmonic extenuation is proven by ANN based SAPF as depicted in figure 9 (d). By employing ANN in SPAF, THDI is abridged to 2.27%, which is lesser than 3.34% THDI by PI based SPAF. Performance of SAPF using proposed RNN controller is portrayed in figure 10. Figure 10 shows the efficacy of proposed RNN controller in SAPF, accompanied by the improvement in current shape, THD is effectually mitigated in both voltage and current. RNN in SAPF results THDV and THDI are 1.7% and 1.74% respectively, which are very less compared to PI and ANN based SAPF. Comparative performance of all analysed systems is offered in Table 1 and Fig. 11 . Table 1 Comparative performance of all analysed systems Controller THDV(%) THDI(%) PI 2.18 3.34 LM-ANN 2.01 2.27 RNN 1.70 1.74 From the Table 1 and Fig. 11 , it is clear THDV is 2.18%. 2.01% and 1.70% by PI, ANN and RNN based SAPF respectively. THDI is 3.34%. 2.27% and 1.74% by PI, ANN and RNN based SAPF respectively. From this Table it is noted harmonic reduction source voltage and current by RNN based SAPF is high in contrast to other two systems. Compared to uncompensated system THDI of 17.94%, all controller based SAPF mitigates THD within IEEE standard 519. 6. Conclusion In this analysis enhancement of power quality in grid connected PV and wind system is examined with the aid of SAPF. In recent days soft computing methods plays major role in enriching performance of SAPF. In this research multilayer RNN is proposed as DC link voltage controller to mitigate harmonics effectively. Performance of anticipated system is authenticated with the PI and ANN controller. Power system with nonlinear load and SAPF for harmonic extenuation is analysed using MATLAB/ Simulink. Performance of PI, ANN and RNN based SAPF are analysed in terms of source current, THDI and THDV. The results of the simulation analysis show that THDV by PI and ANN are in the range of 2%, while proposed multi-layer RNN mitigate it to 1.7%. Uncompensated system THDI is 17.94%. Compared to uncompensated system 81% and 87% of THDI is abridged by PI and ANN based SAPF. Extreme harmonic extenuation of 90% compared to uncompensated system is achieved by while proposed multi-layer RNN based SAPF. According to the findings of the investigation, that all controllers based SAPF limits THDI and THDV within IEEE standard 519. Proposed multi-layer RNN based SAPF offers better harmonic reduction compared to PI and ANN based SAPF systems. Declarations Declaration of interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article. Author Contribution All authors have contributed equally to the work References Y.Yan, K.Chen, H.Geng, W. Fan, and X Zhou, “A Review on Intelligent Detection and Classification of Power Quality Disturbances: Trends, Methodologies, and Prospects”. CMES-Computer Modeling in Engineering & Sciences, vol.137, no.2,p1345-1379,2023. M. Mozaffari, K. Doshi, and Y. Yilmaz, “Real-Time Detection and Classification of Power Quality Disturbances”. Sensors, vol.22, no.20, p.7958,2022. DOI: https://doi.org/10.3390/s22207958 G.S.Chawda et al ., “Comprehensive review on detection and classification of power quality disturbances in utility grid with renewable energy penetration”. IEEE Access, vol.8, pp.146807–146830,2020.DOI: 10.1109/ACCESS.2020.3014732 A. Baitha, and N. Gupta, “A comparative analysis of passive filters for power quality improvement”. In 2015 International Conference on Technological Advancements in Power and Energy (TAP Energy) , IEEE, Kollam, India, June,2015, (pp. 327–332).DOI: 10.1109/TAPENERGY.2015.7229640 D.Li, T.Wang, W.Pan, X.Ding, and J.Gong, “A comprehensive review of improving power quality using active power filters”. Electric Power Systems Research, vol.199, p.107389,2021. DOI: https://doi.org/10.1016/j.epsr.2021.107389 S.Agrawal, D.K. Palwalia, and M.Kumar, “Performance analysis of ANN based three-phase four-wire shunt active power filter for harmonic mitigation under distorted supply voltage conditions”. IETE Journal of Research, vol.68, no.1, pp.566–574, 2022.DOI:https://doi.org/10.1080/03772063.2019.1617198 Q.Liu, and S.Kim, “Temperature-field-angle dependent critical current estimation of commercial second generation high temperature superconducting conductor using double hidden layer Bayesian regularized neural network”. Superconductor Science and Technology, vol.35, no.3, p.035001,2022.DOI: 10.1088/1361-6668/ac45a2 K.Ishaque, Z.Salam, and G.Lauss, “The performance of perturb and observe and incremental conductance maximum power point tracking method under dynamic weather conditions”. Applied Energy, vol.119, pp.228–236,2014. DOI: https://doi.org/10.1016/j.apenergy.2013.12.054 H.W.Kim, S.S. Kim and H.S.Ko,, “Modeling and control of PMSG-based variable-speed wind turbine”. Electric Power Systems Research, vol.80, no.1, pp.46–52,2010. DOI: https://doi.org/10.1016/j.epsr.2009.08.003 C.J.O’Rourke, M.M.Qasim, M.R. Overlin, and J.L.Kirtley, “A geometric interpretation of reference frames and transformations: dq0, Clarke, and park”. IEEE Transactions on Energy Conversion, vol.34, no.4, pp.2070–2083,2019.DOI: 10.1109/TEC.2019.2941175 A.S.N.Mokhtar, B.B.I.Reaz, M.Maruffuzaman, and M.A.M.Ali,, “Inverse Park transformation using cordic and phase-locked loop”. Rev Roumaine des Sciences Techniques-Series Electrotechnique et Energetique, vol.57, no.4, pp.422–431,2012. R.L.de Araujo Ribeiro, T.D.O.A.Rocha, R.M.de Sousa, E.C.dos Santos, and A.M.N Lima, “A robust DC-link voltage control strategy to enhance the performance of shunt active power filters without harmonic detection schemes”. IEEE Transactions on Industrial Electronics, vol. 62 , no.2, pp.803–813,2014.DOI: 10.1109/TIE.2014.2345329 P. Karuppanan, and K.K.Mahapatra, “PI and fuzzy logic controllers for shunt active power filter—A report”. ISA transactions, vol.51, no.1, pp.163–169, 2012.DOI:https://doi.org/10.1016/j.isatra.2011.09.004 V.Deshpande, P.Modi and A.V.Sant, “Analysis of Levenberg Marquardt-ANN based reference current generation for control of shunt active power filter”. Materials Today: Proceedings , vol. 62 , pp.7104–7108, 2022.DOI:https://doi.org/10.1016/j.matpr.2022.02.030 P.N.Srinivasu, et al. , “Using recurrent neural networks for predicting type-2 diabetes from genomic and tabular data”. Diagnostics, vol.12, no.12, p. 3067,2022.DOI:https://doi.org/10.3390/diagnostics12123067 S.Chakraverty et al .,, “McCulloch–Pitts neural network model”. Concepts of soft computing: fuzzy and ANN with programming , pp.167–173,2019.DOI: https://doi.org/10.1007/978-981-13-7430-2_11 Additional Declarations No competing interests reported. Supplementary Files AuthorsInfo.docx Cite Share Download PDF Status: Posted Version 1 posted 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-3971795","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":273857502,"identity":"36693ecb-9a8e-4fb2-94ed-97160a6dfe81","order_by":0,"name":"Divya Banu. P","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABA0lEQVRIiWNgGAWjYDCCw0DM2ABiMR988KECRDM3EKuFLdlwxhmQFkYCWg7AtfCoSfO2McC5OAHfcR4ziZ877PIMbvewSfPOq43mbwdq+VGxDacWycM8ZpK9Z5KLDe6cPWw5d9vx3BmHGRsYe87cxqnF4DBbmgRvG3Pihht5iTfebjuW2wDUwszYhl+L5N+2eqCWHAMJ3jnHcucT1sJ8DOjrwyAtRpK8DTW5GwhpkTzMfNha9szxxJl3jgED+diB3I1ALQfx+YXv/MHGm293VCf23W4GRmVNXe6884cPPvhRgVsLELBIgCkIeRhMHsCnHgiYPyBpqSOgeBSMglEwCkYiAADOM2ZyLoX0NQAAAABJRU5ErkJggg==","orcid":"","institution":"SNS College of Engineering","correspondingAuthor":true,"prefix":"","firstName":"Divya","middleName":"Banu.","lastName":"P","suffix":""},{"id":273857503,"identity":"0dd6f040-2d66-4869-801b-0cca74e75fe6","order_by":1,"name":"G Sundar","email":"","orcid":"","institution":"Sri Shakthi Institute of Engineering and Technology","correspondingAuthor":false,"prefix":"","firstName":"G","middleName":"","lastName":"Sundar","suffix":""}],"badges":[],"createdAt":"2024-02-20 05:45:53","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3971795/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3971795/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":51476103,"identity":"0fa35599-ba03-44d5-b521-1bc959eb932c","added_by":"auto","created_at":"2024-02-22 09:27:23","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":93643,"visible":true,"origin":"","legend":"\u003cp\u003eProposed microgrid with SAPF\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/793b250214d1251d685f4506.png"},{"id":51476104,"identity":"21f9a999-ac58-48ce-96a5-0e587fbf4637","added_by":"auto","created_at":"2024-02-22 09:27:23","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":61194,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic of SAPF\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/4642bc2bb95b1a6b909d2337.png"},{"id":51476106,"identity":"021fcb91-e2a9-4604-858d-09c41c13c87c","added_by":"auto","created_at":"2024-02-22 09:27:23","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":171573,"visible":true,"origin":"","legend":"\u003cp\u003eSynchronous reference detection method for SAPF System\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/52b5447ed51b38e76cb51de4.jpeg"},{"id":51476638,"identity":"c4b86299-a28f-47c7-b0a8-3be568167d57","added_by":"auto","created_at":"2024-02-22 09:35:23","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":12321,"visible":true,"origin":"","legend":"\u003cp\u003eStructure of neural network\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/1459527a0b64384a48ce0b42.png"},{"id":51476109,"identity":"155dcb0a-d4fd-40a1-8fce-6a38911323b6","added_by":"auto","created_at":"2024-02-22 09:27:23","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":39862,"visible":true,"origin":"","legend":"\u003cp\u003eStructure of proposed Multi hidden layer RNN\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/5779196bf887f67ee32eeae7.png"},{"id":51476115,"identity":"8a3b38ca-fd84-4df3-a2d3-aaf2b40116b7","added_by":"auto","created_at":"2024-02-22 09:27:24","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":62567,"visible":true,"origin":"","legend":"\u003cp\u003eSimulation model of SAPF in microgrid\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/9cef9169b125f52f04eba9e0.png"},{"id":51476111,"identity":"d03fddf3-c879-442e-85c3-371e7c744f8e","added_by":"auto","created_at":"2024-02-22 09:27:24","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":710247,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance of power system with nonlinear load without filter (a ) Uncompensated source current (b) THD in Uncompensated source current\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/83c43a65d1a8ab965526bb84.jpeg"},{"id":51476114,"identity":"87c07103-8124-4ade-952b-87aaafdd1428","added_by":"auto","created_at":"2024-02-22 09:27:24","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":493062,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance of SAPF using PI controller (a ) Source Voltage (b) Source current (c) THDV(d) THDI\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/601ec250d275d3904a062355.png"},{"id":51476108,"identity":"9de24dd5-988e-42d3-83b8-1da3023cff63","added_by":"auto","created_at":"2024-02-22 09:27:23","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":528977,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance of SAPF using ANN controller (a ) Source Voltage (b) Source current (c) THDV (d) THDI\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/a3b4e4240eaf38b26b09059f.png"},{"id":51476105,"identity":"dd72a984-d2d4-4912-8a66-34e8643f0d1f","added_by":"auto","created_at":"2024-02-22 09:27:23","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":542645,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance of SAPF using proposed RNN controller (a ) Source Voltage (b) Source current (c) THDV (d) THDI\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/3915dfb8bd0054ee7d5078e0.png"},{"id":51476639,"identity":"4b1729f4-76cf-4dd1-a7fa-9dceebe74cc3","added_by":"auto","created_at":"2024-02-22 09:35:23","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":51567,"visible":true,"origin":"","legend":"\u003cp\u003eComparative performance of all analysed controllers in SAPF (a) THDV (b) THDI\u003c/p\u003e","description":"","filename":"11.png","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/d9550f1983227eaf35359c95.png"},{"id":55689481,"identity":"97c7ea4a-8213-4b33-bcac-e48093ad4fbf","added_by":"auto","created_at":"2024-05-01 22:03:30","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2316398,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/ad1f1bb3-4a8e-4184-8c33-77a243cf8ecc.pdf"},{"id":51476102,"identity":"9868bda9-0e8b-41c5-83d2-eb1fc8d1aa37","added_by":"auto","created_at":"2024-02-22 09:27:23","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":197780,"visible":true,"origin":"","legend":"","description":"","filename":"AuthorsInfo.docx","url":"https://assets-eu.researchsquare.com/files/rs-3971795/v1/cde3b31abf0da8a588247cd0.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eHarmonic Mitigation of the Microgrid Using Multi Hidden Layer Recurrent Neural Network Controlled SAPF\u003c/p\u003e","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eRenewable energy integration has garnered considerable interest due to its zero-fuel price, cleanliness, accessibility, and simplicity of installation. Among the many renewable energy sources, photovoltaic (PV) and wind turbines (WT) are among the most enticing due to their abundant accessibility in nature, technological progress, and monetary advantages. The hybrid combination of both scattered energy sources reduces mutual interruptions caused by their unfavourable characteristics, thereby enhancing system dependability.\u003c/p\u003e \u003cp\u003eNonlinear loads such as switched mode power supply, uninterrupted power supply, adjustable speed drive and converter-based electronics applications are causing redundant power quality (PQ) disturbances in grid-tied systems [\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. PQ issues have been the focus of a lot of research, with potential remedies including power filters and customized power devices. Recently, active filtering systems for PQ enrichment have established to be more effective than passive approaches due to their faster response time, lower size, and greater performance. Moreover, active filtering adapts to changes in network properties automatically, limiting the potential of resonance among the filter and the network impedance [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Active power filters outperform other harmonic compensation systems in terms of real tracking performance, compensation capabilities, and harmonic suppression. APF is one of the most essential tools for addressing the power grid's harmonic problem [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Efficiency of harmonic suppression is decided by the compensation current produced by SAPF, where DC link voltage controller plays crucial role. Accuracy of DC link voltage controller decides the accuracy of compensation. Hence in this article various soft computing methods are analysed as DC link voltage controller to enrich mitigation of harmonics in microgrid system.\u003c/p\u003e \u003cp\u003eThe universal approximation of neural networks (NN), which allows them to estimate any indefinite smooth nonlinear function, makes them a popular choice for nonlinear systems. In the active power filter [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e], NN was employed to approximate the nonlinear portion. Many researchers have adjusted the topology of neural networks in recent study to increase approximation precision, like boosting feedback items or increasing the quantity of hidden layers. For a class of dynamic systems, a double-hidden-layer neural network was created [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eSelf-loops in RNNs occur when some processed data is fed back into the same or previous layers. The RNN uses its internal state to store this data, providing the model with modifiable memory that can be carried over across time. Consequently, a recursive network has greater efficacy than a feed-forward network.\u003c/p\u003e \u003cp\u003eIn this article multi-layer RNN is proposed for compensation. In order to approximate the unidentified nonlinear component of APF, a novel multi-hidden-layer RNN is being developed with numerous computational layers that can approximate any incessant function with arbitrary precision. When compared to the single-hidden-layer NN, the double-hidden-layer RNN is a deep neural network that is able to approximate with high accuracy while using fewer nodes. Hence in this research multi-hidden-layer RNN is proposed for SAPF control in grid connected PV-Wind hybrid power system with nonlinear load.\u003c/p\u003e \u003cp\u003eIn this article proposed microgrid is presented section \u003cspan refid=\"Sec2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. SAPF and it\u0026rsquo;s control algorithm are discussed in section \u003cspan refid=\"Sec3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Section \u003cspan refid=\"Sec5\" class=\"InternalRef\"\u003e4\u003c/span\u003e discusses various controllers as DC link voltage controller in SAPF. Simulation results are analysed in section \u003cspan refid=\"Sec9\" class=\"InternalRef\"\u003e5\u003c/span\u003e and conclusion is presented in section \u003cspan refid=\"Sec10\" class=\"InternalRef\"\u003e6\u003c/span\u003e.\u003c/p\u003e"},{"header":"2. Proposed Microgrid with SAPF","content":"\u003cp\u003eIt is planned to connect PV and wind power systems to the grid. The low voltage output of the photovoltaic panels can be further increased by linking the array of panels to a DC-DC boost converter. Several approaches for tracking maximum power points (MPPT) have been addressed in the literature, the most common of which are incremental conductance and perturbation and observation [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. In this analysis incremental conductance algorithm is employed to maximize the output power of the solar panel.\u003c/p\u003e \u003cp\u003eA wind power system consists of a permanent magnet synchronous generator coupled to an adjustable speed wind turbine [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. The rectifier converts the output of a three-phase voltage into a direct current voltage. A DC-DC buck-boost converter is linked to the rectifier's output to keep the DC voltage stable regardless of changes in the rectified DC voltage. When the rectifier's output is routed to a DC-DC buck-boost converter, the output voltage is stabilised regardless of the DC voltage being rectified. Using the tip speed ratio method, the wind power system maximum power was determined.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThrough the utilisation of a shunt connected transformer, SAPF is capable to inject the compensating current in parallel with the source. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates the PV and wind energy are the green energy system connected with the grid. The quantity of energy present in the DC-link has a significant impact on the reimbursement of the SAPF, an inverter which comprises IGBT and DC-link capacitor.\u003c/p\u003e"},{"header":"3. Shunt active power filter","content":"\u003cp\u003eActive filters can be divided into two primary categories: shunt type and series type. SAPFs have been shown to reduce current harmonics well. They have a good harmonic reduction performance and are simple to implement. Shunt Active Power Filters with the proper control can solve all source current-related power quality problems [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe SAPF is controlled so that it will fully satisfy the nonlinear loads' need for reactive power, maintaining unity power factor at the source side. The reference current waveforms for the SAPF can be created using a variety of control techniques. One of the well-liked control strategies that is simple to put into practice and results in a sufficient decrease in THD is control based on Synchronous reference frame (SRF) algorithm. Accuracy of compensation or quality of harmonic mitigation of SAPF is decided by the DC link voltage controller. In this research two soft computing methods such as ANN and multi-layer RNN are proposed as DC link voltage controller. Performance of proposed multi-layer RNN and ANN based system are compared with conventional PI controller-based system.\u003c/p\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e3.1. Control algorithm for shunt active filter\u003c/h2\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e depicts the present SAPF control system based on the synchronous reference detection mechanism. In the beginning, using the park transform [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], the three-phase line currents ia, ib, and ic undergo a conversion from the three-phase (a-b-c) reference frame to the two-phase (d-q) stationary reference frame currents \u003cem\u003eI\u003c/em\u003e\u003csub\u003e\u003cem\u003ed\u003c/em\u003e\u003c/sub\u003eand \u003cem\u003eI\u003c/em\u003e\u003csub\u003e\u003cem\u003eq.\u003c/em\u003e\u003c/sub\u003eThe park transformation for abc-dq computation is shown in Eq.\u0026nbsp;(\u003cspan refid=\"Equ1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003cdiv id=\"Equ1\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e\n$$\\left[{i}_{d} {i}_{q} \\right]=\\frac{2}{3}\\left[sinsin \\omega t coscos \\omega t sinsin \\omega t-\\frac{2\\pi }{3} coscos \\omega t-\\frac{2\\pi }{3} sinsin \\omega t+\\frac{2\\pi }{3} coscos \\omega t+\\frac{2\\pi }{3} \\right]\\left[{i}_{a} {i}_{b} {i}_{c} \\right]$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThe fundamental current element is converted to an incessant component by the park transformation, whereas the harmonic current elements experience a change in the frequency spectrum. The incessant element can be removed by using a Butterworth low pass filter (50Hz). Higher order harmonics will be helped to be removed by the low pass filter. By applying the inverse transform on parks synchronised with network frequency, harmonic current references can be generated.\u003c/p\u003e \u003cp\u003eThe current in (\u003cem\u003ed-q\u003c/em\u003e) reference frame is given by [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]:\u003cdiv id=\"Equ2\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ2\" name=\"EquationSource\"\u003e\n$$\\left[{{i}_{a}}^{*} {{i}_{b}}^{*} {{i}_{c}}^{*} \\right]=\\left[sinsin \\omega t coscos \\omega t sinsin \\omega t-\\frac{2\\pi }{3} sinsin \\omega t+\\frac{2\\pi }{3} coscos \\omega t-\\frac{2\\pi }{3} coscos \\omega t+\\frac{2\\pi }{3} \\right]\\left[{i}_{d} {i}_{q} \\right]$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e2\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo improve accuracy of compensation by tuning I\u003csub\u003ed\u003c/sub\u003e and regulate the DC link voltage V\u003csub\u003edc\u003c/sub\u003e, various controllers proposed in this analysis.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. DC link voltage controller","content":"\u003cp\u003eThe DC side capacitor's primary function is to supply the real power difference between the load and source during the transient period [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e], but it also serves as a steady-state DC voltage regulator. Whenever the load varies, the voltage across the dc link capacitor also shifts. A controller is employed to achieve the sought-after voltage regulation. In conventional system, PI controller is employed. In this analysis artificial intelligent controllers such as ANN and recurrent neural network (RNN) are proposed as DC link voltage controller.\u003c/p\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e4.1 PI controller in SAPF\u003c/h2\u003e \u003cp\u003ePI controller is a basic controller greatly applied in different uses of the power framework. This basic regulator sets aside least effort to react. Steady-state error is dense effectively by using the PI regulator. In SAPF, DC link voltage error is considered as a contribution to the PI regulator [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Here is an expression for a PI controller,\u003cdiv id=\"Equ3\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ3\" name=\"EquationSource\"\u003e\n$$U\\left(s\\right)={K}_{p}E\\left(s\\right)+\\frac{{K}_{i}}{s}E\\left(s\\right)$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThe gains of the controller, K\u003csub\u003ep\u003c/sub\u003e \u0026amp; K\u003csub\u003ei\u003c/sub\u003e, are denoted, whereas E(s) and U(s) represent the user input and outcome of the controller, respectively. The gain esteems tuned utilizing Ziegler-Nichols' technique. As stated in previous section PI controller produces tune I\u003csub\u003ed\u003c/sub\u003e\u003csup\u003e*\u003c/sup\u003ereference current.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e4.2 ANN in SAPF\u003c/h2\u003e \u003cp\u003eIn this analysis, an ANN controller is proposed as DC link voltage controller as in the active power filter component [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Synchronous reference frame theory is employed in active power filters for diminishing harmonic, and DC voltage error is treated by a PI controller. In this analysis, an ANN controller is developed to control I\u003csub\u003ed\u003c/sub\u003e\u003csup\u003e*\u003c/sup\u003eto enrich quality of power.\u003c/p\u003e \u003cp\u003eThe back-propagation algorithm of the Mean Square Error (MSE) amid the yield and the target data is used to train the ANN. In this investigation, the Levenberg-Marquardt approach is used to train the ANN, since it takes the least amount of time. The training set for ANN was created off-line using data obtained from a PI controller-based system. To obtain an accurate output from this analysis to train network, 7230 datas are assumed. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e depicts a Structure of three layer ANN.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo minimise the duration of processing as much as possible, the hidden layer is built with only sixteen neurons. ANN applied in SAPF system ends training with 34 epochs with MSE of 33 X 10\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e in 1s. It is proposed in SAPF to enhance quality of power.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Multiple layer RNN in SAPF\u003c/h2\u003e \u003cp\u003eRNNs are a sort of neural network that uses previous data from a time series to make predictions about the future [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Recurrent Neural Networks (RNNs) function based on the principle of retaining the output of a particular layer and reintroducing it as input in order to predict the output of that layer.\u003c/p\u003e \u003cp\u003eDelaying inputs and outputs is a typical method for storing temporal information using static neural networks. This representation, however, is constrained subsequently it can only store a limited number of the previously measured outputs and enforced inputs. In addition, it frequently necessitates unrealistically high memory requirements, which prevents it from being used for all but comparatively lesser order dynamical systems. The usage of dynamic or recurrent neural networks has been investigated by the international research community as a very effective and promising alternative. In contrast to static neural networks, RNN feature at least one feedback loop. This type of neural network's topologies, learning techniques, and applications are covered in one of the earliest assessments in [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. There, it is indicated that feedback-containing neural networks have existed since the very beginning of ANN. Indeed, McCulloch and Pitts created systems for time-dependent and time-delayed feedforward networks in [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], however these networks were executed using threshold logic neurons. After that, they expanded their system to include dynamic memory networks which comprises feedback. Later, commonly regarded as the first study on these types of mechanisms, these networks were modelled as determinate mechanisms with a regular language. Among various methods of training RNN, the scaled conjugate gradient technique employs a rapid strategy and makes more deliberate decisions regarding the search direction and step size. Hence in this analysis scaled conjugate gradient method is proposed for its fast performance, which determines the generation of compensation signal in SAPF. Multi hidden layer RNN proposed in SAPF is exposed in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe RNN in the SAPF system training converge in 1m with an MSE of 6.03 X 10\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e and 5000 epochs. In SAPF, this trained RNN is recommended to increase power quality.\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Simulation Results and Analysis","content":"\u003cp\u003eEntire power system with SAPF is analysed using MATLAB/Simulink as shown in Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e. Renewable energy system with grid supplies nonlinear load of 10kw. Specification of analysed system is three phase, 410V, 50 Hz, 10kW load.\u003c/p\u003e\n\u003cp\u003ePerformance of power system with nonlinear load is shown in Fig. 7.\u003c/p\u003e\n\u003cp\u003eFrom the Fig.\u0026nbsp;7, it is observed that nonlinear load results high distortions in source current. Influence of nonlinear load changes shape of source current and result high harmonics of 17.92%.\u003c/p\u003e\n\u003cp\u003ePerformance of SAPF is analysed using PI, ANN and proposed RNN based DC link controllers under variable nonlinear load. Performance of SAPF using PI controller is presented in Fig. 8.\u003c/p\u003e\n\u003cp\u003eFrom the Fig. 8, it is clear that source current shape is improved with the help of SAPF. As depicted in Fig. 8 source voltage THD (THDV) which is 2.18% and THD of source current (THDI) is 3.34% which are pretty much lesser than 5% of IEEE standard. Performance of SAPF using ANN controller is presented in Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eFigure 9 shows that, ANN based SAPF effectively improves source current wave shape and reduces THD. THDV of ANN based SAPF is 2.01% while it was 2.18% by PI based SAPF. Meantime effectiveness of SAPF in source current harmonic extenuation is proven by ANN based SAPF as depicted in figure 9 (d). By employing ANN in SPAF, THDI is abridged to 2.27%, which is lesser than 3.34% THDI by PI based SPAF. \u0026nbsp;Performance of SAPF using proposed RNN controller is portrayed in figure 10.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFigure 10 shows the efficacy of proposed RNN controller in SAPF, accompanied by the improvement in current shape, THD is effectually mitigated in both voltage and current. RNN in SAPF results THDV and THDI are 1.7% and 1.74% respectively, which are very less compared to PI and ANN based SAPF. Comparative performance of all analysed systems is offered in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e and Fig. \u003cspan class=\"InternalRef\"\u003e11\u003c/span\u003e.\u0026nbsp;\u003c/p\u003e\n\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\u003eComparative performance of all analysed systems\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eController\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTHDV(%)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTHDI(%)\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\u003ePI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.34\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLM-ANN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.27\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRNN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.74\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003eFrom the Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e and Fig. \u003cspan class=\"InternalRef\"\u003e11\u003c/span\u003e, it is clear THDV is 2.18%. 2.01% and 1.70% by PI, ANN and RNN based SAPF respectively. THDI is 3.34%. 2.27% and 1.74% by PI, ANN and RNN based SAPF respectively. From this Table it is noted harmonic reduction source voltage and current by RNN based SAPF is high in contrast to other two systems. Compared to uncompensated system THDI of 17.94%, all controller based SAPF mitigates THD within IEEE standard 519.\u003c/p\u003e"},{"header":"6. Conclusion","content":"\u003cp\u003eIn this analysis enhancement of power quality in grid connected PV and wind system is examined with the aid of SAPF. In recent days soft computing methods plays major role in enriching performance of SAPF. In this research multilayer RNN is proposed as DC link voltage controller to mitigate harmonics effectively. Performance of anticipated system is authenticated with the PI and ANN controller. Power system with nonlinear load and SAPF for harmonic extenuation is analysed using MATLAB/ Simulink. Performance of PI, ANN and RNN based SAPF are analysed in terms of source current, THDI and THDV. The results of the simulation analysis show that THDV by PI and ANN are in the range of 2%, while proposed multi-layer RNN mitigate it to 1.7%. Uncompensated system THDI is 17.94%. Compared to uncompensated system 81% and 87% of THDI is abridged by PI and ANN based SAPF. Extreme harmonic extenuation of 90% compared to uncompensated system is achieved by while proposed multi-layer RNN based SAPF. According to the findings of the investigation, that all controllers based SAPF limits THDI and THDV within IEEE standard 519. Proposed multi-layer RNN based SAPF offers better harmonic reduction compared to PI and ANN based SAPF systems.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eDeclaration of interests\u003c/h2\u003e \u003cp\u003eThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.\u003c/p\u003e \u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAll authors have contributed equally to the work\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eY.Yan, K.Chen, H.Geng, W. Fan, and X Zhou, \u0026ldquo;A Review on Intelligent Detection and Classification of Power Quality Disturbances: Trends, Methodologies, and Prospects\u0026rdquo;. CMES-Computer Modeling in Engineering \u0026amp; Sciences, vol.137, no.2,p1345-1379,2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Mozaffari, K. Doshi, and Y. Yilmaz, \u0026ldquo;Real-Time Detection and Classification of Power Quality Disturbances\u0026rdquo;. Sensors, vol.22, no.20, p.7958,2022. DOI: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/s22207958\u003c/span\u003e\u003cspan address=\"10.3390/s22207958\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG.S.Chawda \u003cem\u003eet al\u003c/em\u003e., \u0026ldquo;Comprehensive review on detection and classification of power quality disturbances in utility grid with renewable energy penetration\u0026rdquo;. IEEE Access, vol.8, pp.146807\u0026ndash;146830,2020.DOI: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1109/ACCESS.2020.3014732\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2020.3014732\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. Baitha, and N. Gupta, \u0026ldquo;A comparative analysis of passive filters for power quality improvement\u0026rdquo;. In 2015 \u003cem\u003eInternational Conference on Technological Advancements in Power and Energy (TAP Energy)\u003c/em\u003e, IEEE, Kollam, India, June,2015, (pp. 327\u0026ndash;332).DOI: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1109/TAPENERGY.2015.7229640\u003c/span\u003e\u003cspan address=\"10.1109/TAPENERGY.2015.7229640\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eD.Li, T.Wang, W.Pan, X.Ding, and J.Gong, \u0026ldquo;A comprehensive review of improving power quality using active power filters\u0026rdquo;. Electric Power Systems Research, vol.199, p.107389,2021. DOI:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.epsr.2021.107389\u003c/span\u003e\u003cspan address=\"10.1016/j.epsr.2021.107389\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS.Agrawal, D.K. Palwalia, and M.Kumar, \u0026ldquo;Performance analysis of ANN based three-phase four-wire shunt active power filter for harmonic mitigation under distorted supply voltage conditions\u0026rdquo;. IETE Journal of Research, vol.68, no.1, pp.566\u0026ndash;574, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e2022.DOI:https://doi.org/10.1080/03772063.2019.1617198\u003c/span\u003e\u003cspan address=\"2022.DOI:10.1080/03772063.2019.1617198\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQ.Liu, and S.Kim, \u0026ldquo;Temperature-field-angle dependent critical current estimation of commercial second generation high temperature superconducting conductor using double hidden layer Bayesian regularized neural network\u0026rdquo;. Superconductor Science and Technology, vol.35, no.3, p.035001,2022.DOI: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1088/1361-6668/ac45a2\u003c/span\u003e\u003cspan address=\"10.1088/1361-6668/ac45a2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK.Ishaque, Z.Salam, and G.Lauss, \u0026ldquo;The performance of perturb and observe and incremental conductance maximum power point tracking method under dynamic weather conditions\u0026rdquo;. Applied Energy, vol.119, pp.228\u0026ndash;236,2014. DOI:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.apenergy.2013.12.054\u003c/span\u003e\u003cspan address=\"10.1016/j.apenergy.2013.12.054\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eH.W.Kim, S.S. Kim and H.S.Ko,, \u0026ldquo;Modeling and control of PMSG-based variable-speed wind turbine\u0026rdquo;. Electric Power Systems Research, vol.80, no.1, pp.46\u0026ndash;52,2010. DOI: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.epsr.2009.08.003\u003c/span\u003e\u003cspan address=\"10.1016/j.epsr.2009.08.003\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eC.J.O\u0026rsquo;Rourke, M.M.Qasim, M.R. Overlin, and J.L.Kirtley, \u0026ldquo;A geometric interpretation of reference frames and transformations: dq0, Clarke, and park\u0026rdquo;. IEEE Transactions on Energy Conversion, vol.34, no.4, pp.2070\u0026ndash;2083,2019.DOI: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1109/TEC.2019.2941175\u003c/span\u003e\u003cspan address=\"10.1109/TEC.2019.2941175\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA.S.N.Mokhtar, B.B.I.Reaz, M.Maruffuzaman, and M.A.M.Ali,, \u0026ldquo;Inverse Park transformation using cordic and phase-locked loop\u0026rdquo;. Rev Roumaine des Sciences Techniques-Series Electrotechnique et Energetique, vol.57, no.4, pp.422\u0026ndash;431,2012.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR.L.de Araujo Ribeiro, T.D.O.A.Rocha, R.M.de Sousa, E.C.dos Santos, and A.M.N Lima, \u0026ldquo;A robust DC-link voltage control strategy to enhance the performance of shunt active power filters without harmonic detection schemes\u0026rdquo;. IEEE Transactions on Industrial Electronics, vol.\u003cem\u003e62\u003c/em\u003e, no.2, pp.803\u0026ndash;813,2014.DOI: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1109/TIE.2014.2345329\u003c/span\u003e\u003cspan address=\"10.1109/TIE.2014.2345329\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Karuppanan, and K.K.Mahapatra, \u0026ldquo;PI and fuzzy logic controllers for shunt active power filter\u0026mdash;A report\u0026rdquo;. ISA transactions, vol.51, no.1, pp.163\u0026ndash;169,\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e2012.DOI:https://doi.org/10.1016/j.isatra.2011.09.004\u003c/span\u003e\u003cspan address=\"2012.DOI:10.1016/j.isatra.2011.09.004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eV.Deshpande, P.Modi and A.V.Sant, \u0026ldquo;Analysis of Levenberg Marquardt-ANN based reference current generation for control of shunt active power filter\u0026rdquo;. \u003cem\u003eMaterials Today: Proceedings\u003c/em\u003e, vol.\u003cem\u003e62\u003c/em\u003e, pp.7104\u0026ndash;7108,\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e2022.DOI:https://doi.org/10.1016/j.matpr.2022.02.030\u003c/span\u003e\u003cspan address=\"2022.DOI:10.1016/j.matpr.2022.02.030\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP.N.Srinivasu, \u003cem\u003eet al.\u003c/em\u003e, \u0026ldquo;Using recurrent neural networks for predicting type-2 diabetes from genomic and tabular data\u0026rdquo;. Diagnostics, vol.12, no.12, p.\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e3067,2022.DOI:https://doi.org/10.3390/diagnostics12123067\u003c/span\u003e\u003cspan address=\"3067,2022.DOI:10.3390/diagnostics12123067\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS.Chakraverty\u003cem\u003eet al\u003c/em\u003e.,, \u0026ldquo;McCulloch\u0026ndash;Pitts neural network model\u0026rdquo;. \u003cem\u003eConcepts of soft computing: fuzzy and ANN with programming\u003c/em\u003e, pp.167\u0026ndash;173,2019.DOI: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-981-13-7430-2_11\u003c/span\u003e\u003cspan address=\"10.1007/978-981-13-7430-2_11\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Power quality, Photovoltaic, wind power systems, SAPF, neural networks, multi hidden layer RNN","lastPublishedDoi":"10.21203/rs.3.rs-3971795/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3971795/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRecently, renewable energy systems are more prevalent than traditional energy systems. Especially, Photovoltaic (PV) systems and wind power systems are playing an important part in meeting the world's energy needs. The present needs of more nonlinear loads drastically create power quality issues in the grid-connected system. As a result, it causes an undesired power quality defect that takes the manner of an alteration in the amplitude and patterns of voltage and current in the power system. The shunt active power filter (SAPF) delivers the appropriate current in shunt to grids while suppressing harmonics created by grid-tied nonlinear loads. This paper proposes a multi hidden layer recurrent neural network (RNN) controller for the developing a reference signal in synchronous reference frame theory. RNN is proposed for its high efficacy compared to neural networks, due to self-loops and memory. The effectiveness of the proposed multi-hidden layer RNN-based SAPF system is compared with the performance of neural networks and traditional PI controller-based SAPF systems in terms of harmonic mitigation. Effective harmonic mitigation by proposed multi-hidden layer RNN results current THD of proposed microgrid to 1.74%.\u003c/p\u003e","manuscriptTitle":"Harmonic Mitigation of the Microgrid Using Multi Hidden Layer Recurrent Neural Network Controlled SAPF","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-02-22 09:27:18","doi":"10.21203/rs.3.rs-3971795/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ddd99875-92ae-49e7-b019-762248302ff4","owner":[],"postedDate":"February 22nd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-05-01T21:55:09+00:00","versionOfRecord":[],"versionCreatedAt":"2024-02-22 09:27:18","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3971795","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3971795","identity":"rs-3971795","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}