Comparison of Model-Based Speed Estimation Methods for PMSM Drives
Abstract
This paper presents a unified comparative benchmark of four model-based sensorless speed-estimation algorithms for permanent magnet synchronous motor drives: Luenberger observer (LO), sliding-mode observer (SMO), model reference adaptive system (MRAS), and extended Kalman filter (EKF). The four estimators are implemented under the same PMSM model, operating profile, and simulation conditions. Their performance is evaluated using speed RMSE, position RMSE, steady-state error, maximum transient error, dq-axis current response, THD, disturbance rejection, and implementation complexity. The results show that the EKF achieves the highest estimation accuracy, whereas the SMO provides strong disturbance robustness. The LO and MRAS offer simpler implementation and lower computational cost, although MRAS exhibits larger transient errors. The comparative findings provide practical guidelines for selecting sensorless estimators according to the required trade-off among accuracy, robustness, harmonic behavior, and implementation cost.