文献类型：期刊文献

英文题名：A dual-population cooperative evolutionary algorithm based on contribution degree for large-scale many-objective optimization

作者：Ding, Weichao[1];Jiang, Qinwen[1];Gu, Chunhua[1];Luo, Fei[1];Dong, Wenbo[1]

机构：[1]East China Univ Sci & Technol, Sch Informat Sci & Engn, Shanghai 200237, Peoples R China

年份：2026

卷号：176

外文期刊名：ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE

收录：;EI(收录号：20261520463206);WOS:【SCI-EXPANDED(收录号:WOS:001742492100001)】；

基金：This work was supported by National Natural Science Foundation of China (NO. 62506130 and NO. 62403201) , Shanghai Special Zone Program for Basic Research (22TQ1400100-16) , and Natural Science Foundation of Shanghai, China (24ZR1415200, 23ZR1414900, 22ZR1416500) .

语种：英文

外文关键词：Large-scale many-objective optimization; Decision variable analysis; Cooperative evolution; Contribution degree; Resource allocation

摘要：Large-scale many-objective optimization problems refer to the multi-objective problems with more than 100 variables and three conflicting objectives that need to be optimized simultaneously, which are prevalent but more challenging in engineering applications. Most large-scale multi-objective evolutionary algorithms (LSMOEAs) based on decision variable analysis focus mainly on improving the accuracy and efficiency of classifying variables, which ignores the in-depth exploitation of the properties of decision variables. This paper proposes a dual-population cooperative evolutionary algorithm based on contribution degree. First, two new metrics called convergence contribution degree and diversity contribution degree are introduced to quantify each variable group's contribution to convergence and diversity respectively. A novel decision variable analysis method is then designed to calculate these metrics. Next, two independent populations are optimized in parallel using a reference population-based strategy and an angle-based strategy to balance convergence and diversity. Meanwhile, to achieve the adaptive allocation of computational resources, the proposed algorithm dynamically assigns optimization tasks to each population based on the contribution degrees. Finally, an enhancement strategy based on maximum distance and minimum angle is designed to complement the strengths of both populations. We conduct extensive experiments comparing different LSMOEAs on 144 large-scale multi-and many-objective benchmark test instances and the practical time-varying ratio error estimation problem to demonstrate the effectiveness of the proposed algorithm. The experimental results validate that the proposed algorithm exhibits strong competitiveness and outperforms other methods in both multi-and many-objective large-scale problems.