Optimal feedback control design using genetic algorithm in multimachine power system

This paper introduces an application of genetic algorithm (GA) to design weighting matrices Q and R elements in Linear Quadratic Regulator (LQR) optimization process. The weighting matrices Q and R are the most important components in LQR optimization. The performance of matrices Q and R determines the output performances of the system. Commonly, a trialand-error method has been used to construct the matrices Q and R elements. This method is very simple, but very difficult to produce good control performances. Also it takes a long time to choose the best values in its processing. Because of this, in order to improve control performances by selecting the elements of matrices Q and R, the Bryson method can be employed to give better results in shorter time than the previous method. In this paper, we use GA to construct the weighting matrices Q and R properly. We design the weighting matrices of a power system optimization by using the trial-and-error method, Bryson method, and GA method and compare the control performances. It is shown that the GA calculation is the most useful among the three methods to improve system performances via matrices Q and R design to minimize settling time of control. This idea gives a new alternative procedure in time varying feedback control to improve the stability performances.