基于多策略组合的改进差分进化算法

Improved differential evolution algorithm with multi-strategies

作为求解全局数值优化及复杂黑盒问题的流行方法之一,差分进化算法的性能主要取决于其变异策略和控制参数的设置。为避免算法陷入局部最优和早熟收敛,提出了一种基于多策略组合的改进差分进化算法。首先,该算法利用个体种群适应度排名,再将种群分成3个不同子种群,并依据各子种群特征分配不同的变异算子。其次,为进一步提高算法的搜索能力,分别考虑不同差分项的作用并充分利用父代和子代间的差异信息,提出了改进的自适应参数设置方法。最后,通过在IEEE CEC2014标准测试集上进行数值实验,并与5个DE变体相比,数值实验及非参数检验结果表明所提算法具有更快的收敛速度和寻优能力。

As one of the most popular methods for solving global numerical optimization and complex black-box problems,the performance of the differential evolution(DE)algorithm is mainly depends on its mutation strategy and the setting of control parameters.In order to avoid falling into local optimum and premature convergence,an improved DE algorithm based on multi strategy combination was proposed.Firstly,the algorithm utilizes individual population fitness for ranking,divides the population into three different subpopulations,and assigns different mutation operators based on the characteristics of each subpopulation.Secondly,to further improve the search ability of the algorithm,an improved adaptive parameter setting method was proposed by considering the effects of different differences and fully utilizing the difference information between parents and offspring.Finally,through numerical experiments on the IEEE CEC2014 standard test set,and compared with five DE variants,numerical experiments and non-parametric test results show that the proposed algorithm has faster rate of convergence and optimization ability.