基于种群关联策略和强化解集准则的高维多目标进化算法  

Many-objective evolutionary algorithm based on population association strategy and enhanced solution set criterion

作　　者：覃灏 李军华 QIN Hao;LI Jun-hua(Key Laboratory of Jiangxi Province for Image Processing and Pattern Recognition,Nanchang Hangkong University,Nanchang 330063,China)

机构地区：[1]南昌航空大学江西省图像处理与模式识别重点实验室,南昌330063

出　　处：《控制与决策》2022年第11期2808-2817,共10页Control and Decision

基　　金：国家自然科学基金项目(62066031,61866025,61866026);江西省自然科学基金项目(2018BAB202025);江西省优势科技创新团队计划项目(2018BCB24008);基于自适应参考点策略和降维技术的高维多目标进化优化研究项目(YC2020030)。

摘　　要：一般的高维多目标进化算法无法有效处理不同类型的Pareto前沿.针对这一情况,提出一种基于种群关联策略和强化解集准则的高维多目标进化算法(many-objective evolutionary algorithm based on population association strategy and enhanced solution set criterion,MaOEA/PAS-ESC).该算法在环境选择中采用种群关联策略(population association strategy,PAS)和强化解集准则(enhanced solution set criterion,ESC)协同指导种群进化.PAS利用解与参考向量的角度和欧氏距离以及种群中解之间的距离构建角度与距离联合函数(joint function of angle and distance,JFAD),选择多样性良好的解,然后ESC利用参考点与种群间的联系组成适应度函数,选择收敛性良好的解,以共同达到有效平衡多样性和收敛性的目的.实验结果表明,采用MaOEA/PAS-ESC处理高维多目标优化问题具有更强的竞争性能,而且提高了处理不同类型Pareto前沿的能力.Research shows that the general many-objective evolutionary algorithm can not effectively deal with different types of Pareto fronts.In view of the above situation,this paper proposes a many-objective evolutionary algorithm based on the population association strategy and enhanced solution set criterion(MaOEA/PAS-ESC).In this algorithm,the population association strategy(PAS)and enhanced solution set criterion(ESC)are used to guide the population evolution.The PAS uses the angle and Euclidean distance between the solution and the reference vector as well as the distance between the solutions in the population to construct the joint function of angle and distance(JFAD)and select the solution with good diversity.Then,the ESC uses the connection between the reference point and the population to form the fitness function and select the solution with good convergence,in order to balance diversity and convergence effectively.The experimental results show that the MaOEA/PAS-ESC not only has stronger competitive performance in dealing with many-objective optimization problems,but also improves the ability to deal with different types of Pareto fronts.

关 键 词：高维多目标进化算法 种群关联策略 强化解集准则 种群进化 角度与距离联合函数 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]