Neural Network-Based Robust Automatic Voltage Regulator (AVR) of Synchronous Generator

Abstract

The voltage stability and power quality of the electrical system depend on proper operation of AVR. Nowadays, Design technology of AVR is being broadly improved. Nonlinearities and parametric uncertainties are unavoidable problem faced in controlling the output voltage of Synchronous Generator (SG) when working alone or with others. This paper proposes a Nonlinear Auto Regressive-Moving Average control (NARMA-L2) as a voltage controller which is one type of Neural Network (NN) plant structure. Nonlinearities due to the effect of saturation in machine between generated voltage and field current, uncertainties arise because variation of the load connected with time and the change of rotors resistance with temperature. Due to this fact, Proportional- Integral- Derivative (PID) controller cannot be used effectively since it is developed based on linear system theory. NN controller shows less over shoot and settling time than PID controller with different conditions of load. Also NN controller shows high robust characteristic than PID controller.

Keywords

• neural network controller
• pid controller
• narma
• synchronous generator
• automatic voltage regulator