Purpose – The purpose of this paper is to investigate the influence of back-EMF and current harmonics on position and speed estimation accuracy for single and dual three-phase (DTP) permanent magnet synchronous machines (PMSMs) with two fundamental-model-based sensorless control strategies which are widely utilized for AC machines, i.e. flux-linkage observer (FO) and simplified extended Kalman filter (EKF). Design/methodology/approach – The effect of distorted back-EMF is studied for sensorless vector control of single three-phase PMSM. For the influence of current harmonics, unlike the existing literature where the current harmonics are deliberately injected, in this paper, sensorless switching-table-based direct torque control (ST-DTC) strategies for DTP-PMSM which inherently suffer from non-sinusoidal stator currents in addition to the distorted back-EMF, are investigated experimentally. Findings – By employing the FO and simplified EKF-based sensorless vector control of single three-phase PMSM, it can be concluded that the rotor position estimation accuracy is less affected by the back-EMF harmonics when the simplified EKF method is utilized since it is less sensitive to such noises. When the influence of non-sinusoidal stator currents together with back-EMF harmonics is investigated for the conventional and modified ST-DTC of DTP-PMSM, it is indicated that the simplified EKF exhibits better position and speed estimation accuracy in both the conventional and modified ST-DTC strategies. In addition, its steady-state performance shows a slight superiority over that based on FO, in terms of flux and torque ripples, and THD of phase currents. For the dynamic performance, the estimated speed of simplified EKF shows less phase lag and fluctuations compared to that of FO. Originality/value – This paper introduces the influence of back-EMF and current harmonics on sensorless control performance for single and DTP PMSMs. Detailed experimental results show that the simplified EKF exhibits better rotor position and speed estimation accuracy compared to that of FO due to its higher noise-rejection ability.
Read full abstract