Fast Mixed Strategy Differential Evolution Using Effective Mutant Vector Pool

摘要

The mutant vector has significant influence on the performance of Differential Evolution (DE). Different mutant vector always generates different result, one outstanding mutant vector for a specify problem perhaps achieve unbearable bad result for another question. There still no one perfect mutant vector can solve all problems excellently. In this situation, mixed strategy method is proposed to improve the performance of DE by combining multi-effective mutant vectors together. This paper proposes a fast mixed strategy DE (FMDE). The new method uses two best mutant vectors selected from the mutant vector pool and applies a fast mixed method to generate better result without increase computing expense. The FMDE is evaluated by 27 benchmarks selected from Congress on Evolutionary Computation (CEC) competition. The experiment result shows FMDE is competitive, stable and comprehensive, abstract environment.