Accurate Degradation Prediction Method for PMSM Rotor Magnets Based on Multivariate Segmentation Modeling

Permanent magnet synchronous motors (PMSMs) face significant demagnetization failure risks due to temperature effects on permanent magnets (PM). Therefore, researching PM degradation state identification and lifetime prediction is crucial for motor reliability analysis. This paper proposes degradation modeling and lifetime prediction methods integrating multi-feature characterization and segmented modeling. First, the impact of uniform and localized demagnetization on motor performance is analyzed. The fundamental frequency amplitude and characteristic harmonic content of back-EMF were selected as degenerate feature, and their variation with time and temperature was investigated. Second, effective stress levels are screened through failure mechanism consistency verification. Finally, a multivariate piecewise degradation model is established by combining the piecewise Wiener process and Copula function, enabling lifetime prediction under normal-temperature stress conditions. This research expands the theoretical framework for permanent magnet degradation modeling, providing technical support for motor health monitoring and intelligent maintenance strategies.