Investigation of ANN tuned PI speed controller of a modified DTC induction motor drive

The snag in a DTC scheme based induction motors is the presence of high content of torque ripples in the output torque. This has been reduced by modifying the three-level torque controller to five-level torque controller. Moreover, the controllability of the torque in motor with no overshoot and minimal ripples, good transient and steady state responses forms the basis of performance analysis. This paper presents an improvement in the performance of the modified DTC induction motor drive using artificial neural network based tuning of PI Speed Controller. Since induction motor is a non-linear system therefore ANN has been preferred as it is an excellent estimator for such nonlinear system. The comparison of the simulated results of the proposed method with the conventional method of tuning based on time response specifications shows that the proposed method has improved the performance to a large extent under various operating conditions along with successfully reducing ripples in the torque for the modified DTC drive.