Fault Detection and Identification for Inverter-Driven Permanent Magnet Synchronous Machines

Performing fault detection and separation in inverter-driven permanent magnet synchronous machines has the second priority after designing a precise control to ensure sustained and reliable operation. In this paper, a noninvasive fault detection and identification algorithm is proposed to detect three faults: static eccentricity, partial demagnetization, and inter-turn short circuit. Furthermore, severity of the aforementioned faults is estimated. The algorithm is evaluated in two inverter drives: field-oriented control and direct torque control. The proposed method is based on the frequency spectrum of the instantaneous reactive power obtained by the fast Fourier transform. The spectral components of reactive power are used as features for detection and classification. A three-phase fractional slot concentrated winding machine is simulated under healthy and faulty states for both drives under different load and speed conditions. Results indicate that the proposed diagnosis method detects and classifies faults. Moreover, it is robust to the controller bandwidth and noise, unlike more popular algorithms.