Quantum Evolutionary Algorithm Based High Performance Control of Induction Motor Drive Using ANN Flux Estimator

Abstract

In this study, Quantum Evolutionary Algorithm (QEA) based high performance control of induction motor is proposed, whose rotor flux is estimated by chaotic learning based Artificial Neural Network (ANN). The control principle is based on Direct Torque Control (DTC) with Space Vector Modulation (SVM) technique. The SVM reduces the torque and flux ripples and improves steady state performance. Fast dynamic speed response is obtained through maintaining the rotor flux constant as in the case of field orientation control. Main flux saturation effect is also considered throughout the study for more realistic representation in the analysis. QEA based proportional-integral (P1) controller tuning is used for getting optimized gain coefficients of P1 controller which also help us to get fast speed response induction motor drive. The performance of the drive system with QEA based P1 controller is compared with Conventional Genetic Algorithm (CGA) based P1 controller. There is no speed fluctuation in the speed response of QEA based induction motor drive under steady state condition whereas a little bit speed fluctuation is present in the CGA based induction motor drive. 4 In this work, chaotic learning based Artificial Neural Networks (ANNs) such as Backpropagation (BP), Real Time Recurrent Learning (RTRL). and Correlated Real Time Recurrent Learning (CRTRL) are proposed for improved rotor flux estimation of induction motor drive which makes the control system position sensorless. Chaotic variations of learning rate are included with the learning rate of BP, RTRL, and CRTRL algorithms based ANNs which generates a chaotic time series and a rescaled version of the series is used as Learning Rate (LR) during the training process. We have shown the improvement of BP and CRTRL algorithm based ANNs in rotor flux estimation of induction motor drive due to the use of chaotic variations in learning rate. A high performance simple speed estimator is also presented in this work. It is confirmed that the proposed speed estimator is capable to estimate the speed accurately even at very low speed. Effectiveness of the proposed controller is tested by simulation for different set speed. This simple speed estimator for the induction motor drive makes the controller cost effective and sensorless. Robustness of the control system is tested by using the parameter perturbation and applying sudden load torque disturbance within very short time interval. The control system is also computationally efficient and tested by introducing 10% instrumental error to reference voltage vectors.