.png){kind=icon}
.png){kind=icon}
Return
An Adaptive Constraint Relaxation Strategy Based Coevolutionary Algorithm for Constrained Multi-Objective Optimization
.png){kind=icon}
PRE
AI
DOI：10.1109/TETCI.2025.3595713.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Obtaining high-quality solutions for constrained multi-objective optimization problems (CMOPs) has been extensively researched in recent years. One popular approach is the coevolutionary framework, which incorporates an auxiliary archivethat does not consider constraints. This auxiliary archive helps quickly identify solutions on the non-constrained Pareto front and guides the evolution process to avoid local optima. However, most existing studies pay little attention to solution diversity in the decision space, resulting in poor distribution of solutions when dealing with problems that have a discontinuous Pareto front. Furthermore, using such an auxiliary archive may lead to overlooking solutions on the constrained Pareto front, which not only provides incorrect information but also wastes computational resources. In this study, we propose a novel coevolutionary constrained multi-objective evolutionary algorithm called ACREA to address these issues. ACREA incorporates the adaptive constraint relaxation method (ACR) in the leading archive to provide searching information, and an adaptive local convergence indicator to balance the diversity, constraints and convergence simultaneously in the primary archive. We evaluate the performance and search behavior of ACREA against several state-of-the-art algorithms on the selected test suites and real-world problems. The results demonstrate that ACREA is both effective and efficient in solving CMOPs.
Keywords:
Constrained multi-objective optimization
evolutionary algorithm
coevolutionary
local convergence
Journal
I
IF:
6.5
Papers:
1.4K
Citations:
4.5K