An Adaptive Extended Kalman Filter Observer-For Permanent Magnet Synchronous Motor Position Sensorless Control Systems

Abstract

In this paper, an improved Sage-Husa adaptive extended Kalman filter observer is proposed to accurately estimate rotor speed and electrical angle in Position Sensorless Control Systems. This addresses the issue of the noise covariance matrix in the traditional Kalman filter not aligning with the actual system dynamics and being challenging to rectify. The observer is established on the Kalman-filtered extended inverse potential observer to formulate the full-order state equation of Permanent Magnet Synchronous Motor Position. This approach is then integrated with Normalised phase-locked loop and low-pass filtering techniques. By incorporating the Sage-Husa noise estimator into the conventional Kalman filter and introducing adaptive parameters to dynamically adjust the system noise matrix and process noise matrix, the observer exhibits robust adaptability and rapid convergence. However, it tends to exhibit weak stability and divergence. To counter the stability issue, the inertia weight formula in the traditional Sage-Husa noise estimator is enhanced by incorporating a stability coefficient to ensure semi-positive noise covariance matrix.In the speed loop of Permanent Magnet Synchronous Motor position sensor-less control system, Sliding Mode Control is integrated to enhance the system's immunity to interference and response speed. Additionally, the Maximum Torque Per Ampere module is employed to enable maximum torque-current ratio control for reducing motor losses. Through simulation and experimental analysis, the primary designed observer in this study demonstrates exceptional estimation accuracy and adaptability across all speed ranges.