[1]李杭涓,崔志华,谢丽萍.基于权重学习的高维多目标进化算法[J].计算机技术与发展,2021,31(01):18-23.[doi:10. 3969 / j. issn. 1673-629X. 2021. 01. 004]
　LI Hang-juan,CUI Zhi-hua,XIE Li-ping.Research on Multi-objective Evolution Algorithm Based on Weight Learning[J].,2021,31(01):18-23.[doi:10. 3969 / j. issn. 1673-629X. 2021. 01. 004]
点击复制

基于权重学习的高维多目标进化算法()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
31
期数:
2021年01期
页码:
18-23
栏目:
人工智能
出版日期:
2021-01-10

文章信息/Info

Title:
Research on Multi-objective Evolution Algorithm Based on Weight Learning
文章编号:
1673-629X(2021)01-0018-06
作者:
李杭涓; 崔志华; 谢丽萍
太原科技大学 计算机科学与技术学院,山西 太原 030024
Author(s):
LI Hang-juan; CUI Zhi-hua; XIE Li-ping
School of Computer Science and Technology,Taiyuan University of Science and Technology,Taiyuan 030024,China
关键词:
收敛性; 权重信息; 学习策略; 小生境选择; 进化算法
Keywords:
convergence; weight information; learning strategy; niche selection; evolutionary algorithm
分类号:
TP301.6
DOI:
10. 3969 / j. issn. 1673-629X. 2021. 01. 004
摘要:
在多目标进化算法解决多目标优化问题的过程中,随着目标个数的增加,种群个体进化方向的盲目性逐渐显露出来,同时还存在着收敛性和多样性难以平衡的问题。 针对以上两个问题,以基于参考点策略的快速非支配排序遗传算法NSGA-III 为框架,分别从产生候选解和选择候选解两个角度进行算法改进,从而得到一个新的进化算法 WL-NSGAIII。在新算法的匹配选择阶段,设计了一种基于权重的个体学习策略,即利用种群信息构建代内关系为种群个体提供进化方向并增加候选解集的收敛性。 同时,在新算法的环境选择阶段,利用权重信息对小生境选择策略进行改进。 为了验证新算法的有效性,通过模拟实验将新算法与现有算法在 DTLZ 问题测试集中进行比较。 仿真结果表明,新算法在大多数基准问题上具有良好的效果。
Abstract:
In the process of the multi-objective evolutionary algorithm solving the multi-objective optimization problem,as the number of targets increases,the blindness of the individual population evolution direction gradually emerges. At the same time,the problem of converg-ence and diversity is difficult to balance. Aiming at the above two problems,using the reference point strategy-based fast non-domina-ted genetic algorithm NSGA-III as the framework, the algorithm is improved from the perspective of generating candidate solutions and selecting candidate solutions, respectively,to obtain a new evolutionary algorithm WL-NSGAIII. In the matching selection phase of the new algorithm,a weight-based individual learning strategy is designed,which uses population information to build intragenerational relationships to provide evolutionary direction for the population individuals and increase the convergence of candidate solution sets.? At the same time,in the environment selection stage of the new algorithm,weight information is used to improve the niche selection strategy. In order to verify the effectiveness of the new algorithm,it is compared with the existing algorithm in the DTLZ problem test set through simulation experiments.It is showed that the new algorithm is effective in solving most benchmark problems.

相似文献/References:

[1]刘慧婷 姜晓涛 陈健.基于遗传算法的网格任务调度方法研究[J].计算机技术与发展,2012,(04):69.
　LIU Hui-ting,JIANG Xiao-tao,CHEN Jian.Research of Grid Task Scheduling Strategy Based on Genetic Algorithm[J].,2012,(01):69.
[2]张代远[].一类新型改进的广义蚁群优化算法[J].计算机技术与发展,2012,(06):39.
　ZHANG Dai-yuan.A New Improved Generalized Ant Colony Optimization Algorithm[J].,2012,(01):39.
[3]于舒娟,张昀,杨磊.基于改进蚁群优化的盲检测算法[J].计算机技术与发展,2013,(11):74.
　YU Shu-juan,ZHANG Yun,YANG Lei.An Blind Detection Algorithm Based on Improved Ant Colony Algorithm[J].,2013,(01):74.
[4]彭军,刘振,徐学文. 新型紧致遗传算法及其性能分析[J].计算机技术与发展,2015,25(11):120.
　PENG Jun,LIU Zhen,XU Xue-wen. Novel Compact Genetic Algorithm and Its Performance Analysis[J].,2015,25(01):120.

更新日期/Last Update: 2020-01-10