文献类型：期刊文献

中文题名：基于分区和多子种群的多模态多目标粒子群优化算法

英文题名：Multi-modal Multi-objective Particle Swarm Optimization Algorithm Based on Zoning Search and Multi-subpopulation

作者：袁浩[1];黄文焘[2];姜庆超[3];范勤勤[1]

机构：[1]上海海事大学物流科学与工程研究院,上海201306;[2]上海交通大学电力传输与功率变换控制教育部重点实验室,上海200240;[3]华东理工大学能源化工过程智能制造教育部重点实验室,上海200237

年份：2026

卷号：33

期号：4

起止页码：744

中文期刊名：控制工程

外文期刊名：Control Engineering of China

收录：;北大核心:【北大核心2023】；

基金：教育部人文社会科学研究规划基金资助项目(23YJAZH029);上海市浦江人才计划项目(22PJD030);国家自然科学基金资助项目(61603244)。

语种：中文

中文关键词：多模态多目标优化;进化算法;分区搜索;子种群;粒子群优化算法

外文关键词：Multi-modal multi-objective optimization;evolutionary algorithm;zoning search;subpopulation;particle swarm optimization algorithm

摘要：寻找一个好的近似帕累托前沿和找到足够多的等效帕累托最优解是多模态多目标优化的两个重要目标。为此,提出一种基于分区搜索和多子种群的多模态多目标粒子群优化算法。首先,为了降低搜索难度,采用分区搜索对整个决策空间进行划分,获得多个子空间;然后,对种群进行聚类,将种群划分为不同的子种群;最后,在每个子空间中独立使用改进的多模态多目标粒子群优化算法进行搜索,从而找到更多的等效帕累托最优解。实验选取了22个多模态多目标测试函数对所提算法进行验证,并将其与7种较先进的多模态多目标进化算法进行对比。实验结果表明,相比于对比算法,所提算法可以在决策空间中找到更多的等效帕累托最优解,在目标空间中找到逼近性和多样性更好的近似帕累托前沿。
Finding a good approximate Pareto front and obtaining a sufficient number of equivalent Pareto-optimal solutions are two crucial objectives for multi-modal multi-objective optimization problems.To carry out the above objectives,a multi-modal multi-objective particle swarm optimization algorithm based on zoning search and multi-subpopulation is proposed.Firstly,in order to reduce the difficulty of searching,the entire decision space is divided into multiple subspaces by using a zoning search strategy.Then,the population is clustered and divided into different sub-populations.Finally,in each subspace,the improved multi-modal multi-objective particle swarm optimization algorithm is employed to independently search each subspace to find multiple equivalent Pareto-optimal solutions.In the experiment,the proposed algorithm is verified by 22 multi-modal multi-objective test functions,and compared with seven advanced multi-modal multi-objective evolutionary algorithms.The experimental results show that,compared with the seven algorithms,the proposed algorithm can find more equivalent Pareto-optimal solutions in the decision space,and the approximate Pareto front found in the objective space has better approximation and diversity.