A Traction Control System Signal-Based Method for Gear Slack Fault Detection in Electric Locomotive

Gear slack fault (GSF) is a typical fault in traction motor drive system, which will result in the inability to convert motor torque into driving force. However, it is often misdiagnosed as a wheel set idling and speed sensor faults, which can easily cause traction control unit (TCU) to perform incorrect protective actions, leading to safety accidents. Therefore, it is very important to make full use of information collection and calculation capabilities of TCU to detect this fault. In this article, a real-time fault detection method based on time series event mode (TSEM) of multidimensional information in TCU is proposed. First, based on the GSF mechanism, relevant signals are selected, and the feature indexes are extracted. Second, according to the operation condition switching law of the traction system, six TSEMs related to feature indexes are established for representing GSF. Third, the similarity between the real-time sampled window data and each TSEM is calculated. Then, the diagnostic decisions are made by using the obtained similarities. Finally, the proposed real-time fault detection method is verified by the actual operation data. The experimental results show that, comparing with the existing fault detection method, the proposed method has better performance in sensitivity and reliability.