Multi-Population Coevolutionary Differential Evolution Algorithm

Abstract

The multi-population coevolutionary differential evolution (DE) based on estimation of distribution algorithm (EDA) is proposed. DE completes optimum search using the difference information between the individuals in the population, but the global population evolution information can not be used sufficiently. In this paper, the multi-population co-evolutionary is introduced, which incorporate the merits of the DE and EDA. The elite mutation is proposed in DE. To overcome the greed characteristic, the chaotic initialization and replacement are introduced in DE and the individual diversity in EDA is adjusted based on the individual density. Simulation results show the good global search ability of the proposed algorithm.