An adaptive differential evolution algorithm with elite gaussian mutation and bare-bones strategy

Abstract

Both differential evolution algorithm (DE) and Bare-bones algorithm (BB) are simple and efficient, but their performance in dealing with complex multimodal problems still has room for improvement. DE algorithm has great advantages in global search and BB algorithm has great advantages in local search. Therefore, how to combine these two algorithms' advantages remains open for further research. An adaptive differential evolution algorithm based on elite Gaussian mutation strategy and bare-bones operations (EGBDE) is proposed in this paper. Some elite individuals are selected and then the mean and the variance of the bare-bones operation are adjusted with the information from the selected elite individuals. This new mutation strategy enhances the global search ability and search accuracy of differential evolution with parameters free. It also helps algorithm get a better search direction and effectively balance the exploration and exploitation. An adaptive adjustment factor is adopted to dynamically balance between differential mutation strategy and the elite Gaussian mutation. Twenty test functions are chosen to verify the performance of EGBDE algorithm. The results show that EGBDE has excellent performance when comparing with other competitors.

Keywords

• differential evolution (de)
• bare-bones (bb)
• gaussian mutation
• global optimization