States and Disturbance Observers for Sensorless Control of PMSM with Overcurrent Protection

This paper proposes a neural network (NN)-based observer scheme and a disturbance observer scheme are proposed for the sensorless control of permanent magnet synchronous motors (PMSMs). First, the NN observer is developed to estimate the back-electromotive force (EMF), where the requirement for an exactly known model in either the rotary or stationary reference frame is not needed. To cope with the difficulties of EMF estimation under variable speed, another neural network observer is developed to estimate the rotor's position and speed tracking errors. Next, the disturbance observers are introduced to estimate unknown disturbances in real time. Then, a finite-time control law is proposed to ensure that the angular velocity accurately tracks the reference speed. Finally, the proposed control method is validated experimentally, and the results confirm its superiority over conventional techniques.