Design and performance analysis of neural network based dynamic stability controller for multi-generator power system

Abstract

Power system is a large distributed system with large number of instruments and machines which require continuous adjustment between each part for a proper and efficient performance. This continuous adjustment is done by a combination of Controllers, Stabilizers, Regulators, etc. This paper presents a design of decentralized controller based on neural network (NN) for multi-generator power system. NNs are used here to sense dynamic fault and reduce its effect by adjusting the weights considering current circumstances. This controller deals with dynamic faults of a large decentralized power system occurred from load changes. The controller is trained by previous values of voltage, frequency and power but controller's inputs are instantaneous system signals. This ensures the dynamic and overall system stability. Simulation studies with IEEE-14 bus power system are used to display the virtue of the proposed controller designs.