Classification of partial discharge events in GILBS using discrete wavelet transform and probabilistic neural networks

This paper proposes an approach to determining classification of partial discharge (PD) events in Gas Insulated Load Break Switches (GILBS). Discrete wavelet transform (DWT) is employed to suppress noises of measured signals by the high-frequency current transformer (HFCT). Three kinds of different defects are designed and placed inside three GILBS individually. For accurately determination of the different defect, feature extraction and statistics analysis of the measured signals are used in the proposed method. Finally, experimental results validate that the proposed approach can effectively discriminate the PD events in GILBS.