Induction Motor Speed Control Employing LM-NN Based Adaptive PI Controller

Abstract

Induction motors are the widely adopted electrical machines that revolutionized the industrial process due to their versatility, simplicity, reliability, ruggedness, less maintenance, quiet operation, low cost, high performance, and longevity. This paper presents a Levenberg-Marquardt neural network (LM-NN) based adaptive proportional-integral (PI) control strategy for controlling the speed of three-phase induction motor. The adaptive PI controller adjusts the voltage and frequency of the voltage source inverter (VSI) to minimize the reference speed tracking error under abrupt change of mechanical torque. It develops and tests the proposed LM-NN based adaptive PI controller model in MATLAB/SIMULINK platform. Besides, it derives the control properties of volt/hertz technique from its rotor axis oriented mathematical model. Moreover, the output parameters of the LM-NN are tuned employing a heuristic optimization technique called the back tracking search algorithm (BSA) where the objective is to minimize the integral time squared-error (ITSE). The result shows improved transient and steady state performance for the LM-NN based adaptive PI controller over the conventional PI controller that validates the efficacy of the proposed technique.