A variable structure adaptive neural network power system static VAR s
tabilizer is developed. The static VAR compensator (SVC) controlled by
the above proposed controller is used for voltage regulation and enha
ncing power system stability. The artificial neural network (ANN) is t
rained off-line using the variable structure control system Benchmark
data at different operating conditions and external disturbances. More
over, the trained ANN parameters (weights and biases) are tuned and up
dated on-line using the synchronous machine speed deviation state as t
he ANN output error to increasingly improve the power system performan
ce. A sample digital simulation result of the power system speed devia
tion state responses when reference voltage, speed deviation state and
input power disturbances take place are obtained. The digital simulat
ion results prove the effectiveness and robustness of the present adap
tive neural network in terms of a high performance power system. (C) 1
998 Published by Elsevier Science S.A. All rights reserved.