文献类型：期刊文献

中文题名：一种基于超体积指标的多目标进化算法

英文题名：Hypervolume-Based Multi-Objective Evolutionary Algorithm

作者：王学武[1];魏建斌[1];周昕[1];顾幸生[1]

机构：[1]华东理工大学化工过程先进控制和优化技术教育部重点实验室,上海200237

年份：2020

卷号：46

期号：6

起止页码：780

中文期刊名：华东理工大学学报（自然科学版）

外文期刊名：Journal of East China University of Science and Technology

收录：CSTPCD;;Scopus;北大核心:【北大核心2017】；CSCD:【CSCD_E2019_2020】；

基金：国家自然科学基金(62076095,61973120,61673175)。

语种：中文

中文关键词：多目标优化;超体积;非支配排序;IBEA

外文关键词：multi-objective optimization;hypervolume;non-dominated sorting;IBEA

摘要：基于超体积指标的进化算法能够有效地解决多目标优化问题,可以获得收敛性和分布性均较好的解集,但计算复杂度高、程序运行效率低。针对二维和三维的多目标优化问题,提出了一种基于超体积指标的多目标进化算法(MOEA-HV)。利用精确计算种群中个体的独立贡献超体积来指导种群进化,在基于指标的进化算法(IBEA)前对所有种群个体进行非支配排序,提前删除被支配的个体,从而减少个体独立贡献超体积的计算量来提升运行效率,同时与NSGA-Ⅲ算法相结合来优化算法的分布性。实验结果表明,MOEA-HV算法的运行效率更高,且能够获得较好的收敛性和分布性。
Hypervolume-based evolutionary algorithms can effectively solve the multi-objective optimization problem and obtain promising solution sets with fast convergence and uniform distribution.However,this kind of algorithms have higher computational complexity and lower programming efficiency.Aiming at the twodimensional and three-dimensional multi-objective optimization problems,this paper proposes a hypervolumebased multi-objective evolutionary algorithm(MOEA-HV)so that the individuals’exclusive hypervolume contributions can be accurately calculated to guide the evolution of the whole population.Before the indicatorbased evolutionary algorithm(IBEA)being utilized,the proposed algorithm employs non-dominated sorting among all individuals to delete dominated individuals so that the amount of calculation of the individuals’exclusive hypervolume contributions can be reduced and the operational efficiency can be improved.Meanwhile,other strategies in NSGA-Ⅲare utilized to optimize the distribution of the proposed algorithm.It is shown via the experiment results that the proposed MOEA-HV has higher efficiency while maintaining the trade-off between the fast convergence and the uniform distribution.