DESIGN OF A NEURAL NETWORK BASED SVC CONTROLLER

Abstract

A controller to control the output of a Static Var Compensator (SVC) to damp power system oscillations is developed in this paper. The proposed SVC controller is based on the discrete time filtered direct control theory by which a multilayer neural network with the hyperbolic tangent activation function is derived. Advanced weight tuning algorithm based on a modified delta rule and projection algorithm are used to update the weights of the proposed neural network (NN) and improve the learning rate. Simulation studies with the proposed controller on a single machine-infinite bus system show that the power system stability is improved and the proposed algorithm has a better performance than the traditional controllers.