Neurocontroller for induction motors

Abstract

In this paper the Artificial Neural Network (ANN) is used to identify and control an induction machine with the performance of a Field-Oriented Control (FOC). The control system designed in this work is called the Nurocontroller, and it is trained to reflect the nonlinear behavior of an indirect vector controller and of the PI controller used in the control system. Once the Neurocontroller captures the nonlinear dynamics of the induction motor control, it can replace the conventional Field-Oriented Controller (FOC) and the PI controller with a similar dynamic performance. The data used for the training of ANN's are obtained from computer simulation of the Field oriented control (FOC). The methodologies used to train the ANN's are the back propagation algorithm (BP). Simulation results reveal some very interesting features of the Neurocontroller and show that the network has a good potential for use as an alternative to the conventional FOC Decoupling control of induction motor, with the further advantage of being insensitive to parametric variations.