Improved Performance of Sensorless Control for PMSM Based on Neuro-fuzzy Speed Controller

Abstract

Starting from the capabilities of controllers developed using the fuzzy logic on modeling and equalization of nonlinearities, just as the mathematical model of a Permanent Magnet Synchronous Motor (PMSM) is nonlinear, this article presents a control system of a PMSM based on three neuro-fuzzy controllers ensuring the continuous adjustment of the tuning parameters Kp and Ki of the PI-type speed controller. When there is a high-dynamics PMSM rotor speed reference and a significant variation of the load torque, using a PI controller with fixed tuning parameters, even when using a Particle Swarm optimization (PSO) technique will result in inferior results as opposed to the case when the tuning parameters are adaptive. Thus, an observer of the rotor speed and of the load torque is additionally presented in the case of the sensorless control. The first two neuro-fuzzy controllers prove their superiority due to the high-dynamics speed reference, and if there are also significant variations of the load torque, a third neuro-fuzzy controller is required for additional adjustments in order to achieve superior control performances. The validation of the proposed control system is performed by numerical simulations in the Matlab/Simulink environment.