Classification of partial discharge events in GILBS using probabilistic neural networks and the fuzzy c-means clustering approach

In this paper, we present an approach to determining classification of partial discharge (PD) events in Gas Insulated Load Break Switches (GILBS). Probabilistic neural networks (PNN) and the fuzzy C-means (FCM) clustering approach are used as the classification method. Discrete wavelet transform (DWT) is employed to suppress noises in the measured signals of high-frequency current transformers (HFCT). SF₆ is an insulation gas in Gas Insulated Switchgear (GIS) and GILBS. Three kinds of different defects are designed and placed inside three GILBS. The proposed method greatly improves the classification correctness ratios of the defect models using conventional observations of phase resolved partial discharge (PRPD). To accurately determine the different defect models, the proposed method uses feature extraction and statistical analysis of the measured signals. Finally, experimental results validate that the proposed approach can effectively classify the PD events in GILBS.