Feature Extraction and Applicability Comparisons for Fault Detection of Inter-Turn Short-Circuited PMSM

Abstract

Signal distortions in voltage, current, and speed are caused by inter-turn short-circuit faults (ISFs) in permanent magnet synchronous machines (PMSMs). At present, researchers mainly select the collected signal according to experience and it still lacks the theoretical analysis for the signal selection. For this reason, this article investigates the signal processing for the detection of the ISF in the PMSM. The signals of speed, current, and zero-sequence voltage are adopted and the fault features of the ISF are acquired. The applicability comparisons of the fault feature for different signals are conducted according to the sensitivity and the signal-to-noise ratio (SNR). Simulations and experiments are done to confirm that the theoretical analysis is correct. It is found that the second harmonic occurs in the speed signal and q-axis current iq, and the zero-sequence voltage signal adds the fundamental component under the ISF state. From the indexes of SNR and sensitivity, the zero-sequence voltage is the most effective signal for ISF detection, and the q-axis current iq is better than the speed signal.