Grounding Fault Line Selection Data Model for Distribution Network Based on Random Forest Algorithm

Abstract

Aiming at the problem of fault line selection when arc grounding occurs in distribution network, a method based on random forest algorithm is proposed. This method uses wavelet packet analysis to extract the characteristic information of fault zero sequence current signal, uses kernel principal component analysis to reduce the complexity of sample data, and uses random forest algorithm to train fault line selector suitable for single-phase arc grounding fault in distribution network. Simulation results show that the random forest fault line selector can effectively select the feeder with single-phase arc grounding fault. The scheme has practical significance in fault line selection of single-phase arc grounding in distribution network.