Abstract: In order to solve the problems of false alarm rate and missing alarm rate in the traditional fault

detection method of substation primary equipment，a fault detection method of substation primary equipment based on deep convolution neural network was proposed. The substation primary equipment image and scene image were collected under a variety of lighting conditions，and the substation primary equipment data set and scene data set were set up，and then the various types of data sets were preprocessed. The time domain and frequency domain signal features of substation primary equipment were extracted by deep convolution neural network adaptive method. The deep convolution neural network was used as the feature extractor，and the unknown fault detector was formed by using immune learning characteristics to realize the fault detection of substation primary equipment.Experimental results show that the proposed method can effectively reduce the false alarm rate and missing alarm rate of substation primary equipment，and effectively improve the detection timeliness and fault identification rate.

      Key words: deep convolutional neural network；substation primary equipment；fault detection；information

collection；preprocessing