Sensorless DTC IM drive for an EV propulsion system using a neural network

Abstract

This paper deals with a sensorless direct torque control (DTC) of an induction motor (IM) for an electric vehicle (EV) propulsion system using a neural network. The drawback of conventional DTC is the generation of relatively large torque ripple. In the proposed scheme, first the traditional switching lookup table of a three-level torque controller DTC is replaced with a neural network controller. For further reducing the torque ripples, a three-level torque controller is replaced by the five-level torque controller and then the switching lookup table of the five-level torque controller DTC is replaced with a neural network controller. These sensorless DTC schemes of an IM drive are simulated using Matlab/Simulink. The simulated results are compared with conventional DTC and it is found that the ripples in the torque, as well as in stator current, are reduced drastically.