Robust Encoderless Control for PMSM Drives via a Revised Unscented Kalman Filter

Model-based encoderless control techniques generally rely on accurate flux linkage or back electromotive force (EMF) information, which often depends on various observation methods. However, most of these suffer performance degradation from system disturbances and sampling noise. In this work, we proposed a novel encoderless control structure based on a revised unscented Kalman filter (UKF). Via the active flux concept, we built a new flux model for Kalman filter to avoid erroneous convergence. Furthermore, the structure achieves self-correction dynamically based on stator current estimation error, which enables this technique to be more robust to parameter mismatch. Comprehensive experimental validation is performed on a 4.8 kW PMSM test platform. Experimental data confirm the effectiveness of the proposal.