An Adaptive Evolutionary Algorithm Combining Evolution Strategy and Genetic Algorithm (Application of Fuzzy Power System Stabilizer)

Abstract

The research of power system stabilizer (PSS) for improving the stability of power system has been conducted from the late 1960's. Conventionally lead-lag controller has been widely used as PSS. Root locus and Bode plot to determine the coefficient of lead-lag controller (Yu, 1983; Larsen and Swann, 1981; Kanniah et al., 1984), pole-placement and eigenvalue control (Chow & Sanchez-Gasca, 1989; Ostojic & Kovacevic, 1990) and a linear optimal controller theory (Fleming & Jun Sun, 1990; Mao et al., 1990) have been used. These methods, using a model linearlized in the specific operating point, show a good control performance in the specific operating point. But these approaches are difficult to obtain a good control performance in case of operating conditions such as change of load or three phase fault, etc. Therefore, several methods based on adaptive control theory (Chen et al., 1993; Park & Kim, 1996) have been proposed to give an adaptive capability to PSS for nonlinear characteristic of power system. These methods can improve the dynamic characteristic of power system, but these approaches cannot be applied for the real time control because of long execution time. Recently the research for intelligence control method such as fuzzy logic controller (FLC) and neural network for PSS has greatly improved the dynamic characteristic of power system (Hassan et al., 1991; Hassan & Malik, 1993). Fuzzy rules and membership functions shape should be adjusted to obtain the best control performance in FLC. Conventionally the adjustment is done by the experience of experts or trial and error methods. Therefore it is difficult to determine the suitable membership functions without the knowledge of the system. Recently, evolutionary computations (EC) that is a kind of a probabilistic optimal algorithm is employed to adjust the membership functions and fuzzy rules of FLC. The EC is based on the natural genetics and evolutionary theory. The results of this approach show a good performance (Abido and Abdel-Magid, 1998, 1999). EC is based on the principles of genetics and natural selection. There are three broadly similar avenues of investigation in EC: genetic algorithm (GA), evolution strategy (ES), and evolutionary programming (EP) (] Fogel, 1995). GA simulates the crossover and mutation of natural systems, having a global search capability (Goldberg, 1989), whereas ES simulates the evolution of an asexually reproducing organism. ES can find a global minimum, and by combining another EC it also could be efficient local search technique (Gong et al., 1996 ). O pe n A cc es s D at ab as e w w w .ite ch on lin e. co m