A New Comparative Approach Based on Features of Subcomponents and Machine Learning Algorithms to Detect and Classify Power Quality Disturbances

AbstractCurrent measurement systems based on the IEEE-1159 standard have some limitations and robustness problems under noisy and fast-changing conditions. Besides, applying different methods for each Power Quality Disturbance (PQD) to every window is required but time-consuming and not feasible. Therefore, different kinds of two-stage methods, Detection and Classification (D&C), have been improved in many studies. Then, the required measurement can be performed to define disturbance. For this purpose, a new approach based on features of subcomponents with Machine Learning Algorithms (MLAs) to detect and classify PQDs is proposed. 21-class dataset including single and multiple PQDs under different noisy conditions was prepared randomly. Of this dataset, determined features were extracted and some of these were selected. Then, selected features were trained and tested with some MLAs in a workstation. Results obtained from comparative MLAs and the other classification methods show that the best MLA with related features is Random Forest with 96.97% while LightGBM, k-Nearest Neighbors, and XGBoost 96.85%, 96.73%, and 92.82% accuracy, respectively. Because the selected features, optimized parameters, and the related MLA were obtained by investigating for features provided from the PQDs in the whole parameter space, this approach brings the advantages of high accuracy, low D&C complexity, and computing load.Keywords: machine learning algorithmsfeature selectionchi-squaredhyperparameter optimizationcross validationpower quality disturbancestime-series signalswellflickerharmonicstransientnotchspikesaginterruption AcknowledgmentThe numerical calculations reported in this article were fully performed at TUBITAK ULAKBIM, High Performance and Grid Computing Center (TRUBA resources).Disclosure StatementNo potential conflict of interest was reported by the author(s).Ethical approvalNot Applicable.DATA AVAILABILITYThe dataset generated during and/or analyzed during the current study are available from the corresponding author on upon request.Competing interestThe authors have no competing interest to declare that are relevant to the content of this article.Authors' contributionsS.A. wrote the main manuscript text. He has prepared the signal dataset synthetically and experimentally. Then Zero Crossing and Windowed Data processes have been done. After that, feature extraction is performed. Sections 2 and 6, and Figures 1, 4, 5 and 9 with Tables 1 and Citation2 have been carried out by him. Section 3, Figure 3 and Table 3 in Section 4 and Figures 7 and 8 with Table 4 and Table 5 in Section 5 are all made out by H.M.A. and E.Y. Sections 1, 4, 5 and 7 and related Figures 2, 10, 11, 12, 13 and Tables 6 and 7 have been obtained and written together.Figure 12. CM results for noise levels of the kNN. (a) for 20 dB (b) for 40 dB (c) for 60 dB.Display full sizeDisplay full sizeFIGURE 13. Experimental results. (a) for RF (b) for LightGBM (c) for kNN.Display full sizeAdditional informationFundingThis study is not funded by any facility. Notes on contributorsSıtkı AkkayaSıtkı Akkaya received the B.Sc. and M.Sc. degrees in electrical and electronics engineering from Erciyes University, Kayseri, Turkey, in 2009, and 2012, respectively. In 2018, he received Ph.D. degree from Gazi University, Ankara, Turkey. From 2009 to 2020, he was a research assistant in the Department of Electrical And Electronics Engineering, Yozgat Bozok University, Yozgat, Turkey. Since 2020, he has been an Assistant Professor with the Department of Electrical and Electronics Engineering of Faculty of Engineering and Natural Sciences, Sivas University of Science and Technology, Sivas, Turkey. His current research interests include techniques of machine learning, image processing and signal processing for power quality analysis. Department of Electrical and Electronics Engineering, Sivas University of Science and Technology, Sivas, Turkey. E-mail: sakkaya@sivas.edu.trEmre YüksekEmre Yüksek received the B.S. degrees in computer engineering from Ankara University, Ankara, Turkey, in 2017. He received M.S. degree in management information systems from Sivas Cumhuriyet University, Sivas, Turkey in 2020. He is currently a Ph.D. student in Sivas University of Science and Technology, Sivas, Turkey. Since 2020, he has been a research assistant with the Department of Computer Engineering of Faculty of Engineering and Natural Sciences, Sivas University of Science and Technology, Sivas, Turkey. His current research interests include techniques of machine learning, deep learning, reinforcement learning, indoor mapping and path planning and image processing. Department of Computer Engineering of Faculty of Engineering and Natural Sciences, Sivas University of Science and Technology, Sivas, Turkey. E-mail: eyuksek@sivas.edu.trHasan Metehan AkgünHasan Metehan Akgün received the B.S. degrees in mechatronics engineering and machine engineering from Erciyes University, Kayseri, Turkey, in 2017. He received M.S. degree in machine engineering from İnönü University, Malatya, Turkey in 2020. He is currently a Ph.D. student in Sivas University of Science and Technology, Sivas, Turkey. Since 2021, he has been a lecturer with the Department of Unmanned Aerial Vehicle (UAV) Technology and Operator of Sivas Vocational School, Sivas University of Science and Technology, Sivas, Turkey. His current research interests include techniques of reinforcement learning, image processing, machine learning, automatic control and augmented reality. Department of Unmanned Aerial Vehicle (UAV) Technology and Operator of Sivas Vocational School, Sivas University of Science and Technology, Sivas, Turkey. E-mail: hmakgun@sivas.edu.tr