Open-circuit Fault Diagnosis of Three-phase Inverter for Grid-connected Distributed Generation System

Abstract

To solve the increase cost of acquisition of multivariate information and the impact of the rapidity of diagnosis system in processing multivariate data, we proposed a novel fault diagnosis method for three-phase inverter based on wavelet packet transform and the probabilistic neural network, which can further improve the accuracy of open-circuit fault diagnosis. The wavelet packet transform is used to extract the feature vector of the output voltage of three-phase inverter. The probabilistic neural network has strong tolerance and can be employed to classify the feature vector. The simulation results show that the method has better diagnosis accuracy, which can diagnose the open-circuit fault effectively.