Diagnosis of Interturn Short Circuit Faults in PMSM: A Review of Multi-Physics Models and Machine Learning Algorithms

This paper reviews the Interturn Short Circuit Fault (ISCF) diagnosis of Permanent Magnet Synchronous Motor (PMSM) focusing on two primary methodological approaches: multi-physics model-based methods and Machine Learning (ML) algorithms. Multi-physics model-based methods, leveraging signals such as current, voltage, parameter variations, and magnetic field distortion, offer strong interpretability but often suffer from limited adaptability under dynamic operating conditions; conversely, ML algorithms demonstrate notable potential in enhancing fault severity classification accuracy but are constrained by their dependence on extensive training data and the computational cost associated with embedded deployment. Looking forward, key research directions include: integrating physical mechanism insights with data-driven models to enhance interpretability and robustness; developing lightweight diagnostic models for efficient embedded implementation; and establishing integrated diagnosis and prognosis frameworks to enable full lifecycle health management platforms for PMSM.