Neural Network Based Excitation Control of Synchronous Generator

Abstract

usage of Neural Networks' (NN) based excitation control on single machine infinite bus, its implementation and experimental studies have been reported earlier. The proposed feed forward neural network integrates a voltage regulator and a power system stabilizer. It is trained on-line from input and output signals of a synchronous generator. A modified error function used for training the neural network by the back propagation algorithm uses the reference and terminal voltage as controlling voltage and active power deviation to provide stabilization. The complete control algorithm is implemented on a fixed point DSP and tested in laboratory environment on an 83 kVA, 50 Hz synchronous generator connected over one transmission line to network. The experimental results described in this paper show advantages of this method over classic control method (an excitation current loop with P controller and a terminal voltage loop with PI controller).