Abstract: In order to solve the problem of intelligent fault diagnosis of ZPW-2000R track circuit, a fault diagnosis model of ZPW-2000R track circuit based on deep convolution neural network is proposed. By inputting 38 real-time monitoring variable data stored by microcomputer, 29 kinds of fault types including indoor and outdoor equipment of track circuit can be automatically diagnosed, and the accuracy rate of fault diagnosis can reach 96%. It provides an effective intelligent solution for track circuit fault diagnosis.