The traditional fault diagnosis method often depends on the complex signal processing process and rich expert experience. It needs to segment the signal accurately and the process is tedious, which is not conducive to the field use. In this paper, the deep belief network (DBN) method optimized by particle swarm optimization (PSO) is used to directly extract features from the original power data, and the restricted boltzmann machine (RBM is used to fit the data features layer by layer to realize the dimension reduction of the data at the same time. Then extreme learning machine (ELM) is used to classify each state, which improves the speed of diagnosis. The results show that the accuracy is improved by 4% to 96% and the time used is greatly reduced compared with the support vector machine (SVM) optimized by PSO.