Sensorless control of PMSM using an adaptively tuned SCKF

Abstract

This study reports the application of an adaptively tuned square‐root Cubature Kalman filter (SCKF) for the speed and position estimation of a permanent magnet synchronous motor (PMSM) drive. The proposed estimator is observed to exhibit improved noise rejection characteristics as compared to the hitherto widely applied extended Kalman filter (EKF) observer. A third degree spherical–radial cubature rule is used in the Cubature Kalman filter (CKF) to numerically compute the multivariate moment integrals of the general Bayesian estimation equation. CKF is a non‐linear filter which avoids linearisation and the associated errors. The realisation of CKF using the square‐root algorithm results in numerical stability, as with the realisation of EKF using the square‐root algorithm. Simulation results are presented for a three‐phase inverter‐fed PMSM, along with the experimental results. The estimator and the control algorithms are realised on the MATLAB real‐time environment, interfaced with the hardware using the National Instruments data acquisition system NI PCI‐6221.