Journal of System Simulation
Abstract
Abstract: The decision variable dimension of large-scale multi-objective optimization problems can reach hundreds or even thousands. For existing large-scale multi-objective evolutionary algorithms based on decision variable analysis, which usually consume a large amount of computational resources for grouping and fail to consider the interactions between convergence-related variables and diversity-related variables, a large-scale multi-objective evolutionary algorithm based on multi-region adaptive dynamic grouping was proposed. The algorithm employed a Gaussian mixture model to partition the decision space into multiple regions; within each region, feature vectors were constructed for each decision variable, and spectral clustering was utilized to perform grouping. To validate the effectiveness of the algorithm, it was compared experimentally with different algorithms on 140 large-scale benchmark test problems. The results demonstrate that the proposed algorithm exhibits better performance in terms of both convergence and diversity metrics.
Recommended Citation
Liang, Binhao; Wei, Jingxuan; and Liang, Fengqin
(2026)
"Large-scale Multi-objective Evolutionary Algorithm Based on Multi-region Dynamic Grouping,"
Journal of System Simulation: Vol. 38:
Iss.
4, Article 10.
DOI: 10.16182/j.issn1004731x.joss.25-0216
Available at:
https://dc-china-simulation.researchcommons.org/journal/vol38/iss4/10
First Page
974
Last Page
987
CLC
TP391.9
Recommended Citation
Liang Binhao, Wei Jingxuan, Liang Fengqin. Large-scale Multi-objective Evolutionary Algorithm Based on Multi-region Dynamic Grouping[J]. Journal of System Simulation, 2026, 38(4): 974-987.
DOI
10.16182/j.issn1004731x.joss.25-0216
Included in
Artificial Intelligence and Robotics Commons, Computer Engineering Commons, Numerical Analysis and Scientific Computing Commons, Operations Research, Systems Engineering and Industrial Engineering Commons, Systems Science Commons