ANN and RF-MRAS Sensorless Speed Control of DTC-SVM Based FPIM Drive with Improved Flux and Torque Ripples

This paper propose the artificial neural network (ANN) based sensorless speed control of five p h ase induction motor(FPIM) with improved speed error for electric vehicle(EV) application.Conventional sensorless speed control techniques like model reference adaptive system(MRAS), back emf (BEMF) and sliding mode observer (SMO) not suitable for medium and high speed regions and also very sensitive to change in motor parameters. Therefore, proposed scheme is a better sensorless speed estimator known as the ANN which not only overcomes the aforementioned drawbacks of conventional estimators but also has the ability to achieve state estimation for non-linear systems and reduced dependency towards machine parameters. ANN sensorless speed control is replacing entire MRAS block and reduce it mathematical modelling proportional and integral (PI) control too. ANN DTC-SVM achieve good response with reduced ripples in flux,torque a nd a lso b etter t ransient r esponse of the DTC-FPIM based EV. During steady state and very low speed, it can minimize acoustic noise, torque, flux, current, and speed pulsations. The proposed sensorless ANN-DTC-FPIM modelling and simulation is performed in MATLAB and done experimental validation.