Data-Driven ToMFIR-Based Active Incipient Fault Detection for the Suspension System of High-Speed Trains

As an important component of high-speed trains, the suspension system plays a crucial role in ensuring the operational stability, comfort, and safety of the train. This paper aims to propose a data-driven ToMFIR (total measurable fault information residual) based active fault detection scheme for incipient faults in the suspension system of CRH (China Railway High-speed) trains. Firstly, based on the train attitude data obtained from sensors, T-S (Takagi-Sugeno) fuzzy data modeling is performed. Subsequently, ToMFIR residuals are designed on the basis of LQ decomposition and least squares identification techniques. To achieve beneficial proactive amplification of the ToMFIR residuals, auxiliary signals are introduced. Simultaneously, redundant actuators are employed to prevent the auxiliary signals from interfering with normal system operation. Finally, a Hellinger distance-based evaluation function is introduced to monitor changes in the incipient fault indicator. The simulation results demonstrate that the proposed active fault detection scheme effectively identifies incipient faults in suspension actuators, springs, and dampers.