Permanent Magnet Synchronous Motor Faults Detection Using Current Harmonics and K-Means Clustering

PMSMs have acquired wide applications in industries because of high power density, high efficiency, and reliable operation. PMSMs possess numerous benefits, including low maintenance, smooth operation, quick dynamic response, and improved thermal management due to permanent magnets used instead of rotor windings. These characteristics render PMSMs suitable for use in electric vehicles (EVs), industrial automation, renewable energy systems, aerospace and consumer appliances. Despite the many advantages, PMSMs are also prone to a variety of faults such as stator inter-turn short circuit, rotor demagnetization, and bearing faults, which may lead to performance degradation and machine failure. For reliable PMSMs operation, several fault detection techniques are employed. One of the critical issues in fault diagnosis in PMSM is that correct detection of the nature of the fault is difficult, particularly when small fluctuations in the currents. Stator interturn short circuit faults, demagnetization faults, or bearings faults frequently create small deviations in electrical and mechanical signatures like those suppressed by usual working noise, loads fluctuation, or ambient disturbances. In this paper, we have compared the behavior of various electrical and mechanical faults associated with a PMSM motor. Mainly, Inter-turn short fault, demagnetization fault, and eccentricity faults have been analyzed using Finite Element Analysis (FEA) modeling, frequency domain analysis, and k-Means Clustering. The results bring into focus the effectiveness of these analytical approaches in classifying fault types with high accuracy and provide a robust foundation for the development of sophisticated real-time fault monitoring systems for PMSMs.