Dynamic Modeling and Identification of Arc Grounding Fault in Resonant Grounding System

When an arc grounding fault occurs in the medium voltage distribution network, the dynamic compensation of Petersen Coil will weaken the fault signal, making the fault transient characteristics weak and difficult to identify. This paper proposes a fault diagnosis method based on convolutional neural network-long short term memory network (CNN-LSTM). The arc mathematical modeling method based on genetic algorithm can accurately simulate the arc model that fits different fault scenarios. The time-frequency spectrum fuses the compensation quantity of Petersen Coil and the transient characteristic quantity. The simulation is able to construct different fault scenario data sets, and the hybrid network then extracts arc grounding fault features to achieve reliable identification of different faults.
¹