基于自适应聚合距离的多目标进化算法  

A multi-objective evolutionary algorithm based on adaptive aggregation distance

作　　者：曾亮 曾维钧 李燕燕 全睿 王珊珊 ZENG Liang;ZENG Wei-jun;LI Yan-yan;QUAN Rui;WANG Shan-shan(School of Electrical and Electronic Engineering,Hubei University of Technology,Wuhan 430068,China;Hubei Key Laboratory for High-efficiency Utilization of Solar Energy and Operation Control of Energy Storage System,Wuhan 430068,China;Xiangyang Industrial Institute of Hubei University of Technology,Xiangyang 441100,China)

机构地区：[1]湖北工业大学电气与电子工程学院,武汉430068 [2]太阳能高效利用及储能运行控制湖北省重点实验室,武汉430068 [3]襄阳湖北工业大学产业研究院,湖北襄阳441100

出　　处：《控制与决策》2024年第4期1113-1122,共10页Control and Decision

基　　金：国家自然科学基金项目(51977061);湖北省重点研发计划项目(2020BAB114);湖北省教育厅科学研究计划重点项目(D20211402);襄阳湖北工业大学产业研究院2022年度项目(XYYJ2022C04)。

摘　　要：随着目标数的增多,种群收敛性与分布性的冲突愈加激烈,传统的多目标进化算法的选择算子难以平衡种群的收敛性与分布性.对此,提出一种基于自适应聚合距离的多目标进化算法.首先,采用参考点支配关系替代原有的Pareto支配关系,以增加选择压力,加强收敛性;其次,提出自适应聚合距离,通过动态变化的惩罚参数来自适应调整收敛性与分布性的比例;最后,设计一种带有淘汰算子的方法以改进小生境选择策略,根据自适应聚合距离的大小进行选择和淘汰操作.为验证算法的可行性,将所提出算法在测试问题上与其他4种优秀的多目标进化算法进行比较,并应用于两个实际应用中,仿真结果表明,所提出算法的综合性能更优,能有效平衡种群的收敛性与分布性.The conflict between convergence and distribution of population will intensify when the dimensionality of objective increases.The selection operator of the traditional multi-objective evolutionary algorithm is difficult to balance the convergence and distribution of the population.To solve this problem,a multi-objective evolutionary algorithm based on adaptive aggregation distance(MOEA-AAD)is proposed.Firstly,the reference point dominance relation is used to replace the original Pareto dominance relation to increase the selection pressure and strengthen the convergence.Then,an adaptive aggregation distance is proposed to adaptively adjust the ratio of convergence and distribution through dynamically changing penalty parameters.Finally,a method with a deletion operator is designed to improve the niche selection strategy,and the selection and elimination operation are carried out according to the value of the adaptive aggregation distance.In order to prove the optimization performance of the proposed algorithm,it is compared with four other excellent multi-objective evolutionary algorithms on the test problem and applied to two real-world applications.The simulation results show that the proposed algorithm has better comprehensive performance,which can effectively balance the convergence and distribution of the population.

关 键 词：多目标进化算法 参考点支配 自适应聚合距离 淘汰算子 收敛性与分布性 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]