Differential Evolution Improved with Intelligent Mutation Operator Based on Proximity and Ranking
Guogang Cao, Cong Jun Cao, Qing Zhang, Wenju Li · 2018
Differential evolution (DE) is the perfect optimization algorithm to solve real-parameter optimization problems. The most important of DE operators is the differential mutation operator. Motivated by the selection strategies of the proximity-based mutation operator and the ranking-based one, a novel intelligent selection for mutation operator is proposed, in which the selection of some individuals for mutation is based on proximity between individuals and ranking of fitness successively. Since the proposed intelligent selection affects only the mutation step of DE algorithms, it could be directly applied to other DE mutation strategies. Through extensive experimental studies on the CEC 2017 benchmark suite, the results show that it remarkably enhances the performance of most mutation strategies for all benchmark problems in the DE/best/1 and DE/current-to-best/1 evolution strategies and has good characteristics about computational complexity.