大规模多目标进化优化算法研究进展

Research Progress on Large-scale Multi-objective Evolutionary Optimization Algorithm

关键词:  大规模多目标优化  进化算法  协同进化  决策变量分析  变量分组

Key words:  large-scale multi-objective optimization  evolutionary algorithm  cooperative coevolution  decision variable analysis  variable grouping

摘要:

现实中存在许多大规模多目标优化问题（Large-scale Multi-objective Optimization Problem，LSMOP），它们对传统的多目标进化算法（Multi-objective Evolutionary Algorithm，MOEA）提出了挑战，有关LSMOP的研究已成为多目标优化领域的研究热点之一。本文系统分析了近年来提出的各种大规模多目标进化优化算法（Large-scale Multi-objective Optimization Evolutionary Algorithm，LSMOEA），根据这些算法的主要思想和技术特点将它们粗略地分成4种类型，即基于协同进化（Cooperative Coevolution，CC）、基于决策变量分析、基于问题重构以及其他方法，并对今后LSMOP的研究方向提出建议，以期将LSMOP的研究引向深入。

Abstract:

There are many Large-scale Multi-objective Optimization Problem (LSMOP) in reality,which pose great challenge to traditional Multi-objective Evolutionary Algorithm (MOEA).The research on LSMOP has become one of the research hotspots in the field of multi-objective optimization.This article systematically analyzes various Large-scale Multi-objective Optimization Evolutionary Algorithm (LSMOEA) proposed in the past few years.According to the main ideas and technical features of these algorithms,they are roughly divided into 4 types,namely based on Cooperative Coevolution (CC),based on decision variable analysis,based on problem reconstruction and other methods.Furthermore,some future research directions and suggestions are proposed to guide the study on LSMOP deeply in this article.