Sensorless control of PMSM using a new efficient neural network speed estimator

In order to reduce the cost and improve the reliability of variable speed drives, sensorless techniques for estimation rotor speed from measurement of voltage and current have been the subject of intensive research. This paper proposes a sensorless control strategy for Permanent Magnet Synchronous Motor (PMSM) control using a novel neural network algorithm. The proposed observer uses a neural network trained to learn the electrical and mechanical motor models using the current prediction error. Experiments were performed, showing that the proposed network topology and training algorithm have advantages to the classical ones currently employed in sensorless control.