A Fault Diagnosis Scheme for High-Speed Train Bogie based on Depth-wise Convolution

Abstract

The fault detection and isolation system is the key element for the safe long-term operation of high-speed train. The multi-channel signals provided by parallel monitoring system are usually closely coupled and highly uncertain, which are difficult to analyze. This paper proposed a depth-wise convolution modular structure for fault diagnosis with the multi-channel signal to address the complex and dynamic operating conditions of high-speed trains. A scalable modular structure is designed to provide low coupling and high transparency, which could easily configurable function-level according to the requirements. Depth-wise convolution is employed to avoid premature channel fusion. The experimental demonstrate that the proposed scheme improves the accuracy of high-speed train bogie fault diagnosis, including cases with noise and with speed-varied condition, which has practical value to industrial applications.