Application of self-adaptive dynamic niche genetic algorithm in global multimodal optimization problems

Purpose: Genetic algorithm is a kind of random search method evolved from the genetic mechanism, it has strong robustness and optimization ability. However, a large number of researches indicated that the traditional genetic algorithms have many deficiencies and limitations in global multimodal optimization, such as they are prone to premature convergence, high computational cost and weak local search abilities. The purpose is to overcome these disadvantages through the creation of a new algorithm for solving global multimodal optimization problems, which is self-adaptive dynamic niche genetic algorithm (SDNGA). Methodology: By studying the GA optimization and niche theory, we combine multi-groups and niche method to traditional genetic algorithm, which is used in the solution of global multimodal optimization problems. The proposed algorithm is applied to test functions to demonstrate its effectiveness and applicability. Findings: We adopted the niche technology to divide each generation of a group into several subgroups. Then we choosed the best individual from each subgroup as the representative of such a subgroup, and then carried out the hybridization and mutation to produce a new generation within the population and between populations, thus enhancing the global optimization ability of the algorithm, and improving the convergence speed. Originality: We made a study of genetic algorithm and niche theory to apply in the global multimodal optimization problem. We discussed the ideas and the steps of proposed algorithm, made the qualitative analysis on the searching ability and the convergence speed. The research on this aspect lias not been found at present. Practical value: We proposed a self-adaptive dynamic niche genetic algorithm, which can be used in global multimodal optimization problems. The test experimental results have shown that SDNGA has good searching ability, good performance and very strong robustness, which allows for solutions of higher quality. © Zhanshen Feng, Yan Yu, 2016.