Detection and Analysis of GIS Discharge Defects Based on Deep Learninng Method

Gas insulated switchgear (GIS) is an important equipment of power system equipment. Partial discharge (PD) inspection is an important method for determining the internal insulation condition of GIS equipment in extra high voltage (EHV) substations. PD detection can detect equipment defects ahead of time and provide an early warning to prevent further equipment and grid failure. This paper introduces a GIS live inspection case in which abnormal signals are discovered and traced using ultra high frequency (UHF) and ultrasonic methods. A PD type identification technology based on deep learning method is proposed as an alternative to the traditional discharge type identification technology. In order to evaluate the performance of the partial discharge type recognition technology in this paper, a comparative experiment was carried out with the recognition method based on back propagation neural network (BPNN). The results show that the former has significantly better accuracy in identifying the type of discharge than the latter. This paper provides some background and experience for GIS live detection, as well as operation and maintenance.