GIS Partial discharge identification based on AlexNet neural network

Gas Insulated Switchgear (GIS) is an important part of the current power system, and its safe and stable operation is directly related to people’s production and life. Partial discharge (PD) is an important reason for the decline of GIS insulation level, so the diagnosis of GIS PD faults can promptly eliminate insulation faults and guarantee the high-quality operation of the power system. This paper analyses the GIS PD PRPD(Phase Resolved Partial Discharge) pattern features of UHF signals, and perform image preprocessing with grey scale, binarization and image segmentation. In order to avoid the subjectivity of feature data extraction, this paper adopts AlexNet neural network to identify the pattern, and finally the total identification rate reaches $97.70 \%$, which achieves the purpose of timely operation and maintenance of GIS.