Dynamic compensator design for HV AC power system using artificial neural networks

Abstract

This paper presents a method for designing a dynamic compensator based on artificial neural networks (ANN). The ANN is trained to give the proper compensator parameters (X, Tz) so as to always assign certain eigenvalues to desired locations in the output feedback system. The eigenvalues of concern are those associated with the angle /spl delta/ and speed /spl omega/. These eigenvalues are to be assigned to the specified locations under variations in several system parameters [static nonlinear load parameters A/sub p/ and A/sub q/, transmission line reactance, x/sub e/, and generated real power, P/sub G/]. The exact and ANN's results of compensator's parameters are plotted. In addition speed response is provided for the compensated and uncompensated systems. Results show that the ANN can be used on line once the off line training is performed to determine the compensator data so as to maintain the desired response of the system under variations in system parameters.