Niching community based differential evolution for multimodal optimization problems

This paper proposes to solve multimodal optimization problems by enhancing species-based DE (SDE) with niching community strategy and one-to-one greedy selection strategy. The proposed niching community based SDE (NCSDE) algorithm has the following three advantages when solving multimodal optimization problems. Firstly, there is no need to use prior knowledge to determine niching parameter. Secondly, the restriction in a small niching community can facilitate locating multiple optima and exploitation at a higher accuracy level. Thirdly, one-to-one greedy selection strategy can avoid losing diversity. The proposed NCSDE algorithm is evaluated on 20 multimodal test functions. The experimental results show that the proposed NCSDE algorithm can obtain very competitive performance over other advanced niching algorithms on multimodal optimization problems, and outperforms others on most of the test functions.