A New Evolutionary Algorithm for Multi-objective Optimization Problems
A New Evolutionary Algorithm for Multi-objective Optimization Problems
复制标题
DOI：
--
复制
发表时间：
2003
期刊：
影响因子：
--
通讯作者：
Zhixiu Wei
中科院分区：
文献类型：
--
作者：
Zhixiu Wei
在目前成功的进化多目标算法（MOEAs），精英主义和无共享因子是两个共同的特点，并已被证明可以显着提高性能。基于这两个原则，本文提出了两种改进单目标优化算法的策略：多父体交叉策略和群体爬山策略，分别保证种群收敛到真实的Pareto最优前沿和均衡前沿，有效地防止了早熟收敛，并实现了均衡前沿的均匀分布。
Among the currently successful Evolutionary Multi-Objective Algorithms (MOEAs), elitism and no sharing factor are two common characteristics and have been demonstrated to improve performance significantly. Based on these two principles, two heuristics, with which impressive improvements were showed in single objective optimization, are introduced in a newly designed EMOA in this paper: multi-parent crossover, which ensures that the population converges to the true Pareto optimal front; and swarm hill climbing, which effectively helps prevent premature convergence and achieve a well distributed trade-off front.