A dual dynamic constraint boundary based constrained multi-objective evolutionary algorithm for small feasible regions

Addressing constrained multi-objective optimization problems (CMOPs) with small feasible regions presents a significant challenge, as existing algorithms often struggle to balance feasibility, diversity, and convergence within the population. To overcome this challenge, we propose a dual dynamic con...

Full description

Saved in:
Bibliographic Details
Published in:Expert systems with applications Vol. 275; p. 127008
Main Authors: Zhu, Cong, Yang, Yongkuan, Kong, Xiangsong, Yang, Yanxiang
Format: Journal Article
Language:English
Published: Elsevier Ltd 25.05.2025
Subjects:
Co-evolution
Constrained multi-objective optimization
Constraint handling
Dynamic constraint boundary
Multi-objective evolutionary algorithm
Small feasible region
Constraint handling
Constrained multi-objective optimization
Co-evolution
Multi-objective evolutionary algorithm
Dynamic constraint boundary
Small feasible region
ISSN:0957-4174
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Addressing constrained multi-objective optimization problems (CMOPs) with small feasible regions presents a significant challenge, as existing algorithms often struggle to balance feasibility, diversity, and convergence within the population. To overcome this challenge, we propose a dual dynamic constraint boundary-based constrained multi-objective evolutionary algorithm, referred to as TPDCB. In TPDCB, the original CMOP is transformed into two dynamic CMOPs using a dual dynamic constraint boundary strategy to better identify feasible solutions. Specifically, for the two dynamic CMOPs within the constraint relaxation boundary, the first dynamic CMOP primarily focuses on multi-objective optimization, while the second dynamic CMOP equally emphasizes both multi-objective optimization and constraint satisfaction to enhance individual diversity. Furthermore, an auxiliary problem without constraints is introduced by treating constraint violations as an additional optimization objective, which improves the algorithm’s global convergence. Finally, a tri-population co-evolution framework is proposed to simultaneously tackle all three constructed problems. The algorithm’s performance is evaluated on 22 benchmark problems and three real-world applications, and compared to seven state-of-the-art algorithms. Experimental results demonstrate that TPDCB is competitive in solving CMOPs with small feasible regions. •A MOEA for CMOPs with small feasible regions.•Dual-population dynamic constraint boundary.•(M+1)-objective environment selection.•Tri-population co-evolution.
ISSN:0957-4174
DOI:10.1016/j.eswa.2025.127008