Self-tuning control of induction motor drive using neural network identifier

This study presents a new self-tuning PI speed controller with load torque observer and feedforward compensation based on neural network identification for an induction motor. A two-layer neural estimator is also used to provide a real-time adaptive estimation of the unknown motor dynamics. The widely used projection algorithm is used as the learning algorithm for this network, to minimize the difference between the motor's actual response and that predicted by the neural estimator. The proposed neural estimator uses this learning to adjust PI speed controller with a load torque observer to generate the control signal online, thereby bringing the motor output to a desired reference trajectory. The theoretical analysis, simulation and experimental results demonstrate the proposed scheme's effectiveness.