Current Sensor Incipient Fault Diagnosis in PMSM Drive Systems Using Novel Interval Sliding Mode Observer

This article presents a novel method for timely and accurate detection of incipient faults in current sensors of permanent magnet synchronous motor (PMSM) drive systems by utilizing an interval sliding mode observer (SMO). Initially, a hybrid logic dynamic model is developed to incorporate current sensor incipient faults, system parameter perturbation, and unknown disturbances. The current sensor incipient faults are then reconstructed using state augmentation and nonsingular coordinate transformation techniques. Subsequently, an adaptive interval SMO with a fast convergence rate and the ability to suppress high-frequency chattering is designed to estimate the reconstructed state accurately. The residual between the actual value and the observed value of the reconstructed state is employed as the detection variable, and adaptive upper and lower thresholds are designed for incipient fault diagnosis of the current sensor. The proposed method is capable of diagnosing both incipient faults and significant faults in current sensors. Experimental results validate the accuracy and robustness of the diagnostic method.

I. Introduction

Permanent magnet synchronous motors (PMSMs) have gained widespread use in new energy vehicles and high-speed trains due to their high power density and efficiency [1], [2]. However, the operating conditions of PMSMs are complex and variable, and they often operate continuously for extended periods, leading to frequent faults. One major type of fault that significantly affects the safety of PMSM operation is the current sensor fault within the drive system [3], [4]. In comparison to significant faults in current sensors, incipient faults are characterized by small amplitudes and insignificant fault signatures. They are typically caused by overload current impacts or performance degradation, making them difficult to detect during the initial stages. However, with time, the fault amplitudes gradually increase, potentially leading to faults with destructive impacts on system performance [5], [6]. Therefore, it is highly important to investigate methods for diagnosing incipient faults in current sensors to ensure the safe operation of PMSMs.