Comparison of artificial neural network and multiple regression for partial discharge sources recognition.

Abstract

This paper compares the capabilities of the artificial neural network (ANN) and multiple linear regression (MLR) for recognizing and discriminating partial discharge (PD) defects. Statistical fingerprints obtained from a several PD measurement were applied for training and testing both the ANN and MLR. The result indicates that for both the ANN and MLR trained and tested with the same insulation defect, the ANN has better recognition capability. But, when both ANN and MLR were trained and tested with different PD defects, the MLR is generally more sensitive in discriminating them. In this paper, the results were evaluated for practical PD recognition and it shows that both of them can be used simultaneously for both online and offline PD detection.

Citation

MAS'UD, A.A., MUHAMMAD-SUKKI, F., ALBARRACIN, R., ARDILA-REY, J.A., ABU-BAKAR, S.H., AZIZ, N.F.A., BANI, N.A. and MUHTAZARUDDIN, M.N. 2017. Comparison of artificial neural network and multiple regression for partial discharge sources recognition. In Proceedings of the 9th IEEE-GCC conference and exhibition 2017 (GCCCE 2017), 8-11 May 2017, Manama, Bahrain. New York: IEEE [online], article number 8448033, pages 519-522. Available from: https://doi.org/10.1109/IEEEGCC.2017.8448033

• Conference Name: 9th IEEE-GCC conference and exhibition 2017 (GCCCE 2017)
• Conference Location: Manama, Bahrain
• Start Date: May 8, 2017
• End Date: May 11, 2017
• Acceptance Date: May 31, 2016
• Online Publication Date: May 8, 2017
• Publication Date: Aug 31, 2018
• Deposit Date: Jan 16, 2018
• Publicly Available Date: Jan 16, 2018
• Electronic ISSN: 2473-9391
• Publisher: IEEE Institute of Electrical and Electronics Engineers
• Article Number: 8448033
• Pages: 519-522
• Series Title: Proceedings of the IEEE GCC conference and exhibition
• Series ISSN: 2473-9391
• ISBN: 9781538627563
• DOI: https://doi.org/10.1109/IEEEGCC.2017.8448033
• Keywords: Partial discharge; Regression analysis; Artificial neural network
• Public URL: http://hdl.handle.net/10059/2671