A Novel Differential Evolution Algorithm Based on Chaos Local Search

As an effective tool for optimization, differential evolution (DE) has aroused much interest. But the premature convergence of it often gives rise to erroneous results so should be improved. In this paper, a novel differential evolutionary algorithm (DECH) based on chaos local search (CLS) is proposed, which divides DE algorithm into two stages. Firstly, DECH runs with original DE model ´DE/best/1/bin´ to ensure ability of global search. Secondly, DECH runs with chaos search to ensure ability of local search. In experiments, it is proved that DECH can balance the ability of the local search and global search to keep the diversity of population. Compared with original DE and other modified DE by testing on five typical benchmark functions, DECH has superior global convergence and robustness, which means it´s suitable for solving global optimization problems.