A Hybrid Differential Evolution Algorithm with Opposition-based Learning

Differential evolution (DE) is a popular optimization technique, however it also tends to suffer from premature convergence. One possible way to fix this problem is adaptively to choose the right mutation strategy and control parameter setting for distinct problems. Recently, a new concept, opposition-based learning, was introduced to computational intelligent, which was experimentally proven to be effective and robust. Therefore, a new approach is proposed to combine these two means in attempt to enhance the ability of DE. In the proposed approach, one solution produced by different mutation strategies and parameter setting is used to generate the corresponding opposite one, and then these two solutions are simultaneously evaluated to make the better one as the offspring. The experiments are conducted on 13 well-known benchmark functions, and the experimental results compared with other several state-of-the-art DE variants show that the proposed approach is effective and robust.