Position Sensorless Control for an Interior Permanent Magnet Synchronous Motor SVM Drive with ANN Based Stator Flux Estimator

Abstract

An Interior Permanent Magnet Synchronous Motor (IPMSM) drive fed by Space Vector Modulated (SVM) inverter without rotor position sensor is proposed in this paper. The control system estimates the motor stator flux and its position through a Real Time Recurrent Neural Network (RTRNN). The RTRNN is trained off-line to estimate stator flux from the speed error and stationary α-and β-axis current components obtained through Park's transformation. Vector control of the motor is achieved through command torque and magnetizing current components generation and necessary transformations. The estimated flux and rotor angle are found to match accurately with their corresponding actual values. A Proportional plus Integral (PI) controller is used to control the motor variables in closed loop. The effectiveness of the drive system is tested for different operating conditions. The proposed system is found to work satisfactorily under these conditions and perturbations of parameters.