An adaptive sensorless control system of PMSM based on ANN

An artificial neural network (ANN) based a real-time adaptive sensorless control system for permanent; magnet synchronous motor (PMSM) is proposed to cure the problems rising from parameter variations and system uncertainties. At first, for speed sensorless control, the stator voltages and currents are measured and an extended Kalman filter (EKF) algorithm is adopted for the speed estimation. Secondly, based on the technique of ANN a real-time adaptive controller is designed, which results in accurate speed control of the motor. The initial estimation of coefficients of the proposed ANN speed controller is obtained by offline training method, and online training has been carried out to update the ANN under continuous mode of operation. Dynamic back-propagation with the Levenburg-Marquardt (LM) algorithm is utilized for online training Purposes. Finally, the validity of the proposed sensorless PMSM control system with the proposed ANN based adaptive speed controller and EKF estimator has been illustrated through experiments.