自适应两阶段分组求解大规模全局优化问题  

Solving Large-scale Global Optimization Problems Based on Adaptive Two-stage Grouping

作　　者：贾欣 王宇嘉[1] 聂方鑫 孙福禄 JIA Xin;WANG Yu-jia;NIE Fang-xin;SUN Fu-lu(Electrical and Electronic Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学电子电气工程学院,上海201620

出　　处：《小型微型计算机系统》2023年第1期14-23,共10页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61403249)资助.

摘　　要：协同进化是解决大规模全局优化问题的一种有效策略,但是该策略不能对存在相关性变量的大规模问题进行有效分组,最终导致算法性能下降.针对上述问题,提出一种基于自适应两阶段分组的差分协同进化算法.首先,在第1阶段分组中,根据决策变量贡献度,将其分为正促进组和负抑制组;然后,在第2阶段分组中,分别对两组内的变量进行相关性识别,根据相关变量所占比例进行自适应分组;最后,采用差分协同进化算法对分组后的组件进行优化.实验结果表明本文所提方法能够实现对大规模全局优化问题中相关变量的有效分组,提高了算法的收敛性,通过标准大规模优化测试函数集验证了算法的有效性和适用性.Coevolution is an effective strategy to solve large-scale global optimization problems,but it can not effectively group the variables with correlation,which finally leads to the performance degradation of the algorithm.To solve these problems,a differential co-evolution algorithm based on adaptive two-stage grouping is proposed.Firstly,in the first stage,according to the contribution of decision variables,the participants were divided into positive promotion group and negative inhibition group.Then,in the second stage of grouping,the variables in the two groups were respectively identified for correlation,and adaptive grouping was made according to the proportion of relevant variables.At last,differential coevolution algorithm is used to optimize the grouped components.What the experimental results show is the method that is proposed can effectively group the relevant variables in the large-scale global optimization problem and promote the convergence of the algorithm.The effectiveness and applicability of the method that is proposed are verified by the standard large-scale optimization test function set.

关 键 词：大规模优化问题 两阶段分组 贡献度 相关性 差分协同进化 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]