Compensation of rotor time constant variations in induction machines using artificial neural networks

M. Ebrahimi, A. Shoulaie, H. Seifi & V. Tahani

Abstract

An artificial neural network (ANN) is used in this paper as a simple on-line, rotor time constant identification scheme for an indirect vector control (IVC) of an induction motor. A three layer NN has designed and trained off line with the steady state mathematical model of the IVC control scheme in detuning operations. if the estimated flux and a supposed rotor time constant are input to the NN, then the actual rotor time constant is estimated directly from the NN. The indirect vector control used with the neural network estimator has been studied in the detuning condition. The performance of the controller is good even when the rotor time constant is increased from nominal rate to twice the nominal value.