Type Identification of GIS Partial Discharge Based on SF6 Decomposition Characteristics under Negative DC Voltage

Abstract

In order to use SF6 decomposition characteristics to identify faults of DC gas-insulated switchgear (GIS) under partial discharge (PD), the author chooses 6 experimental voltages to represent different PD stages, and carries out SF6 decomposition experiments. The experimental results show that SF6 decomposition produces include five stable components of CF4, CO2, SO2F2, SOF2 and SO2, among which SOF2 is the most important decomposed product, and the concentration of sulfur-containing components is higher than that of carbon-containing components. Finally, a feature set consisting of 21 concentration ratios is constructed, and the maximum relevance minimum redundancy criterion is used for the feature selection. BP neural network and support vector machine are used for fault diagnosis, and the accuracy rate is higher than 88%. The research work lays the foundation for the on-line monitoring and insulation state evaluation of DC GIS based on SF6 decomposition characteristics in the future.