Based Searching-Valley Sequential Niche Genetic Algorithm for Detecting Multiple Optima

In this paper a new sequential niche algorithm for locating all solutions in multimodal optimization problems is introduced. The algorithm combines crowding with searching-valley function to keep the diversity of population efficiently. The sequential process proceeds to search for additional extrema. Unlike other algorithms the efficiency of this sequential niche genetic algorithm is not highly sensitive to the niche radius. This algorithm consists of global search, local search and identification of the optima. The global search ensures the diversity, the local search ensures the convergence of each local optimum. Several comparative simulation experiments with previous niche algorithms demonstrate the novel algorithm´s performance and reliability.