An Adaptive Genetic Algorithm Combined With Chaos Search

Population-based genetic algorithm is a kind of global optimization probability searching method, which is on the basis of Darwin's evolutionary theory and Mendel's theory of genetic mutations. Genetic Algorithm(GA) has been widely applied to many fields, such as computer science, engineering technology, management science and social science, etc. And it gradually becomes one of the key technologies of intelligent computation in the 21st century. Due to the probabilities of crossover (Pc) and mutation(Pm) are constants in the Standard Genetic Algorithm(SG- A), so it's not very efficient when applied to the optimization problems of complex multi-variable. At the same time, there exists premature convergence probably. Beca- use of the reason given above, Adaptive Genetic Algorithm(AGA)has been proposed naturally, in which Pc and Pm adaptively change according to the individual's fitness. So that the algorithm can keep the population diversity, meanwhile, guarantee the convergence of algorithm. In practice, due to the complexity of problem itself, objective function usually appears as the form of high-dimension, multi-peak, multiv- ariable and non-consecutive. At this time, under the consideration of fast convergence and global optimization, it's hard to obtain satisfactory results either by using genetic algorithm or by using adaptive genetic algorithm. Therefore, on the basis of analysis and study of characteristics of AGA, according to the hybrid optimization strategy and the rules of hybrid GAs, the paper incorporates chaos optimization algorithm into the AGA, and proposes a new Adaptive Genetic Algorithm Combined with Chaos Searching (AGACCS).The proposed algorithm not only has all the features of AGA, but also further improves the global searching ability, prevents it from premature convergence, obtains fast convergence and high computational precision. The computational results on several benchmark functions have shown that the proposed algorithm is superior to adaptive genetic algorithm.