基于混合多策略约束处理的多目标进化算法

Hybrid Multi-Strategy Constrained Evolutionary Algorithm for Multi-Objective Optimization

摘要: 约束多目标优化问题(CMOPs)广泛存在于工程实践中，如何在满足约束条件的同时获得分布良好的Pareto前沿是该领域的核心挑战。本文提出了一种基于混合多策略约束处理的多目标进化算法(Hybrid Multi-Strategy Constraint Handling Based Evolutionary Algorithm for Constrained Multi-Objective Optimization, HMSC-EC)。该算法采用三档案协同机制，分别负责前向探索、多样性增强和可行性利用，并结合自适应环境选择和基于进化代数的算子选择策略，有效平衡了收敛性和多样性。实验结果表明，HMSC-EC在多个基准测试问题上表现优异，特别是在处理复杂约束时具有显著优势。

Abstract: Constrained Multi-Objective Optimization Problems (CMOPs) are prevalent in engineering practice. A core challenge in this field is how to obtain a well-distributed Pareto front while satisfying all constraints. This paper proposes a Hybrid Multi-Strategy Constraint handling based Evolutionary algorithm for Constrained multi-objective optimization (HMSC-EC). This algorithm employs a three-archive collaborative mechanism, responsible for forward exploration, diversity enhancement, and feasibility exploitation, respectively. It integrates adaptive environmental selection and generation-dependent operator selection strategies to effectively balance convergence and diversity. Experimental results demonstrate that HMSC-EC performs excellently on multiple benchmark problems, showing significant advantages particularly when dealing with complex constraints.