A comparative study of Kalman filtering for sensorless control of a permanent-magnet synchronous motor drive

This paper presents a comparative study of the novel Unscented Kalman Filter (UKF) and the Extended Kalman Filter (EKF) for estimation of the rotor speed and position of a permanent-magnet synchronous motor (PMSM) drive. The general structure of the EKF and the UKF are reviewed. The various system vectors, matrices, models and algorithm programs are presented. Simulation studies on the two Kalman filters are carried out using Matlab and Simulink to explore the usability of the UKF in a sensorless PMSM drive. In order to compare the estimation performances of the observers, both filters are designed for the same motor model and control system and run with the same covariances. The simulation results indicate that the UKF is capable of tracking the actual rotor speed and position provided that the elements of the covariance matrices are properly selected. Since covariance tuning of the Kalman filter is often a trial-and-error process, an unconventional, asymmetric way of setting the model covariance parameters is introduced. It is shown that tuning is easier and the method gives a significant improvement in performance and filter stability.