约束多目标进化算法研究进展

约束多目标进化算法（CMOEAs）能够同时处理多个相互冲突的目标函数和约束条件，引导种群逼向可行域的最优解，受到了研究者的广泛重视。首先介绍了约束多目标优化问题（CMOPs）的相关定义和多目标进化算法（MOEAs）的三种分类；其次，系统地分析了当前CMOEAs中约束处理机制，凝练出当前主要的四种约束处理方法；然后，从基于支配、基于指标、基于分解三个方面对CMOEAs的研究进展进行了详细综述；最后，指明了CMOEAs存在的挑战和未来研究方向。

Research progress of constrained multi-objective evolutionary algorithms

Constrained multi-objective evolutionary algorithms(CMOEAs) can deal with multiple conflicting objective functions and constraints simultaneously and guide the population towards the optimal solution in the feasible domain, which has received extensive attention from researchers. This paper firstly introduced the relevant definitions of constrained multi-objective optimization problems(CMOPs) and three classifications of multi-objective evolutionary algorithms(MOEAs). Secondly, it systematically analyzed the current constraint processing mechanisms in CMOEAs, and condensed the current four main constraint processing methods. Then, it reviewed the research progress of CMOEAs in detail from three aspects: dominance-based, index-based and decomposition-based. Finally, it pointed out the challenges and future research directions of CMOEAs.