Novel non-singular terminal sliding mode observer based on support vector regression parameters optimization for PMSM sensorless control

In order to improve the stability and dynamic performance on the sensorless control system of permanent magnet synchronous motor, a novel nonsingular terminal sliding mode observer is proposed in this paper. It discards the differential term with the integral term which avoids the noise caused by the differential signal and the singular problem that may occur in the sliding mode process. Then, a support vector regression machine is used to optimize the most important parameters which affect novel nonsingular terminal sliding mode observer's performance. Simulation results show that sensorless control system using novel nonsingular terminal sliding mode observer has high tracking accuracy and small error in rotor position and speed, and has advanced dynamic performance and robustness.