Abstract
To tackle the misdiagnosis issue arising from the resemblance in current distortion caused by open-circuit (OC) faults and current sensor (CS) faults in the permanent magnet synchronous motor (PMSM) drive system of electric vehicles (EVs), a joint fault diagnosis strategy based on a novel finite-time observer (FTO) is proposed. First, a holistic model of the EV PMSM drive system is constructed, which takes into account both OC faults and CS faults. Next, the original system is successfully decoupled into two distinct subsystems by implementing system augmentation and nonsingular coordinate transformation, where one subsystem solely encompasses OC faults and the other exclusively contains CS faults. This facilitates the decoupling of the two types of faults. Subsequently, FTOs are designed for each subsystem to accurately estimate the reconstructed states and generate residuals. Then, residuals are constructed from the error systems, and on this foundation, OC fault detection variables and adaptive thresholds are designed to effectively enhance system robustness. In addition, a localization strategy based on sliding window accumulation is proposed to achieve precise localization of single and double power switch OC faults. Meanwhile, CS fault identification variables are established based on the CS fault estimates. Finally, comparable hardware-in-the-loop (HIL) experimental results are provided to validate the efficacy and resilience of the proposed algorithm.