In order to solve the problem of fault diagnosis of switch module of V2G DC charging pile, a diagnosis method based on random forest algorithm is proposed in this paper. The method adopts wavelet packet analysis to extract the feature information of fault current signal, and uses kernel principal component analysis to reduce the complexity of sample data. The fault diagnosis device suitable for DC charging pile switch module is trained by using random forest algorithm. The experimental results show that the random forest fault diagnosis can effectively diagnose the switch module fault. The scheme is of practical significance in the fault diagnosis of DC charging pile switch module.