Fault Diagnosis for IGBTs Open-Circuit Faults in Photovoltaic Grid-Connected Inverters Based on Statistical Analysis and Machine Learning

A new fault diagnosis method for IGBTs open-circuit faults based on statistical analysis and machine learning is proposed to improve the reliability of photovoltaic power generation system. Firstly, empirical mode decomposition (EMD) is used to realize the adaptive filtering of the noise in three-phase current. Secondly, statistical analysis and generalized discriminant analysis (GDA) are used for feature extraction and feature dimensionality reduction. Then, BP neural network is used for fault pattern recognition. Finally, the rapidity and accuracy of the proposed method are verified by simulation experiments. At the same time, the proposed method is compared with the traditional feature extraction method based on fast Fourier transform (FFT) and EMD and principal component analysis (PCA)-based dimension reduction method. The results show that the proposed method has high fault recognition rate and simple neural network topology.