Full-Speed Sensorless Control Scheme for Permanent Magnet Synchronous Motor Using Artificial Neural Network

Abstract

Extended back electromotive force (EEMF) has been widely used as speed estimation of permanent magnet synchronous motors (PMSMs) due to its excellent performance at medium and high speeds. However, at low speed, EEMF method has low estimation accuracy. Several sensorless methods have combined EEMF with different accurate low-speed estimation techniques to attain a satisfactory estimation performance in the whole speed range. However, the transition between EEMF and low-speed estimation methods is a crucial step that requires two separate speed estimators with different structures. This leads to increased design complexity. Therefore, in this paper, a single sensorless full-speed control is proposed using an artificial neural network (ANN). The proposed ANN method does not require any transition method compared with hybrid full-speed sensorless methods. Simulation results show that proposed ANN can estimate rotor position for full speed compared with EEMF which estimates the rotor speed at medium and high speed.