Comparative Study of Online Loss Estimators Based on Electrical and Thermal Models for AC Permanent Magnet Machines

Loss estimation in electric machines is important for performance evaluation, condition monitoring, and optimal control. The parameters determining losses can vary with operating conditions; therefore, accurate estimation of losses is challenging. Online loss estimation is one solution, where parameters are identified and updated as they vary with operating conditions over time. For applications requiring high accuracy, estimators which are robust to model uncertainty are needed. This paper presents a comparative robustness analysis for online loss estimators of surface mount permanent magnet (SMPM) machines based on both electrical and thermal machine models. It is found that estimators based on thermal models are more robust than those based on electrical models over a wide range of operating conditions. The conditions under which the thermal model can achieve better robustness is obtained through analysis and validated through simulation.