EKF/MM-based Fault Diagnosis Algorithm of PMSM in the Presence of Sensor, Open and Short-Circuit Faults

Recently, due to environmental concerns and the quest for energy efficiency, the use of PMSM is increasing across various fields, such as automobiles and aircraft. Therefore, the state of PMSM significantly impacts the safety of vehicles. In this paper, a diagnosis method for the detection of a stator open- phase fault and inter-turn short circuit fault is proposed. First, whether a sensor is failing is determined through the Whiteness Test. Subsequently, the phase currents are estimated using the Extended Kalman Filter, and fault detection is conducted by comparing of the measured and estimated currents. Following the fault detection, the fault is classified using a Multiple Model filter, in which possible fault states are modeled. Each filter's residuals and covariances are compared, and a model with the highest fault probability is determined as the source of the fault. Simulations were conducted to verify the proposed diagnosis method.