Intelligent control for an induction motor

Abstract

The induction motor drive is a dynamic, recurrent, and non-linear system. Despite the great efforts devoted to induction motor control, many of the theoretical results cannot be directly applied to practical systems due to the following difficulties: 1) complicated computations involved; 2) nonlinearity of the system; and 3) uncertainty in the machine parameters. This thesis addresses these problems by incorporating intelligent control in the induction motor drive system, using artificial intelligence (AI) techniques in association with conventional control methods. Using Matlab/Simulink software, three induction motor models (current-input model, voltage-input model, discrete-state model) are built for the simulation studies of the controller. An expert-system based acceleration controller is developed to overcome the three drawbacks (sensitivity to parameter variations, error accumulation, and the need for continuous control with initial state) of the vector controller. Simulation results obtained on the expert-system based controller show that the performance is comparable with that of a conventional direct self controller, hence proving the feasibility of expert-system based control. A hybrid fuzzy/PI two-stage control method is developed to optimize the dynamic performance of a current and slip controller. The performance of the two-stage controller approximates that of a field-oriented controller. Very encouraging results are obtained from a computer simulation and a digital signal processor (DSP) based hardware experiment. In order to improve the performance of a direct self controller (DSC), an ANN-based DSC is proposed. The execution time is decreased from 250 撘峴 (for a DSP-based controller) to 21 撘峴 (for the ANN-based controller), hence the steady-state control error is almost eliminated. Detailed simulation is made using Matlab/Simulink and Neural-network Toolbox. Addressing the current research trend, a speed-sensorless controller with Kalman filter is investigated. A real-coded genetic algorithm (GA) is used to optimize the noise covariance and weight matrices of the EKF, thereby ensuring filter stability and accuracy in speed estimation. Simulation studies on a constant V/Hz controller and a direct self controller demonstrate the efficacy of the proposed method. The prototype field-oriented control (FOC) drive system comprises a DSP, a data acquisition board, a three-phase induction motor, current sensors and a personal computer (PC). Experimental results are presented to validate the performance of the optimized EKF.