Partial Discharge Type Identification in GIS Based on SCG Algorithm

摘要

During the insulation failure of Gas Insulated Switchgear (GIS), partial discharge will occur in the power grid. This phenomenon not only affects the operation of GIS, but also affects the security of the power grid. In order to study and identify four kinds of discharge phenomena better, four kinds of partial discharge signals were collected in the GIS partial discharge laboratory, and 13 kinds of characteristic parameters were extracted. The Scaled Conjugate Gradient (SCG) propagation algorithm based on artificial neural network was applied for training. Continuously change the number of hidden neurons in the hidden layer until the appropriate number was selected to achieve the highest partial discharge accuracy in GIS. The training results show that the recognition accuracy was 96.5% when the number of hidden neurons in the hidden layer was 18, which provided a new idea for the future study of partial discharge pattern recognition in GIS.