State estimation of bearingless permanent magnet synchronous motor using improved UKF

Unscented Kalman filter (UKF) algorithm was widely used in the speed sensorless control of Motor. However, the problem of bad robustness of the model parameter change, slow convergence and lower tracking ability to abrupt state still exist. Combined with strong tracking filter, an improved UKF is proposed in this paper. The time-varying fading factor and softening factor are introduced to adaptively adjust gain matrices and the state forecast covariance square root matrix, in order to realize the residuals sequences orthogonality and force the UKF to track the real state rapidly. The speed sensorless vector control system of bearingless permanent magnet synchronous motor (BPMSM) was set up based on this estimation approach. The simulation results illustrate that, contrast to ordinary UKF, the proposed method is capable of precisely estimating the rotor speed and space position, high robustness is achieved under the conditions of step response or load disturbance.