Incipient faults Detection Based on Hellinger Distance and Canonical Variate Analysis for Traction Systems in High-speed Trains

Abstract

The traction system of high-speed trains plays a decisive role in the safe operation of the train. This paper aims at the real-time detection of incipient faults in tractions system of high-speed trains. A new fault detection method in traction systems based on Hellinger distance and canonical variable analysis (HD-CVA) is presented. Firstly, the correlation between input variables and output variables is considered by using the HD-CVA method to obtain the parameter matrix and generate residual data. Secondly, the obtained Hellinger distance is used to build the statistics and set appropriate thresholds. Finally, the fault decision is made according to the detection logic given in the paper. The traction drive control system (TDCS) of CRH2 high-speed trains is used as the testing platform, and the experiment selects the operation data of four typical faults. The validity and feasibility of the HD-CVA are verified. The results show that this method can effectively detect the occurrence of incipient faults in the traction system.