计算机工程 ›› 2019, Vol. 45 ›› Issue (5): 161-168.doi: 10.19678/j.issn.1000-3428.0051232

• 人工智能及识别技术 • 上一篇    下一篇

基于自适应邻域策略的改进型MOEA/D算法

耿焕同,韩伟民,丁洋洋,周山胜   

  1. 南京信息工程大学 计算机与软件学院,南京 210044
  • 收稿日期:2018-04-16 出版日期:2019-05-15 发布日期:2019-05-15
  • 作者简介:耿焕同(1973—),男,教授、博士生导师,主研方向为多目标优化、智能计算、数据挖掘;韩伟民、丁洋洋、周山胜,硕士研究生。
  • 基金项目:

    国家重点研发计划(2017YFC1502104);江苏省自然科学基金(BK20151458)。

Improved MOEA/D Algorithm Based on Adaptive Neighborhood Strategy

GENG Huantong,HAN Weimin,DING Yangyang,ZHOU Shansheng   

  1. College of Computer and Software,Nanjing University of Information Science and Technology,Nanjing 210044,China
  • Received:2018-04-16 Online:2019-05-15 Published:2019-05-15

摘要:

为避免传统MOEA/D算法使用固定领域规模易造成种群进化效率降低的情况,提出一种基于自适应邻域策略的改进算法。设计一种能够反映子问题进化幅度和种群进化状态的判断机制。针对进化过程中的收敛性和分布性需求,提出基于进化状态判断的自适应邻域策略,从而根据种群和子问题的进化状态设定不同的邻域规模。使用WFG系列测试函数进行实验,结果表明,该算法能有效平衡进化过程中种群的收敛性与分布性,提高解集的整体性能。

关键词: 基于分解的多目标进化算法, 邻域更新能力, 进化状态, 判断机制, 自适应邻域策略

Abstract:

Traditional Multi-objective Evolutionary Algorithm based on Decomposition(MOEA/D)uses the fixed neighborhood scale,which reduces the population evolution efficiency.To solve this problem,an improved algorithm based on the Adaptive Neighborhood Strategy(ANS) is proposed.This paper designs a kind of judgment mechanism which can reflect the evolution magnitude of the sub-problems and the evolutionary state of the population.Based on evolutionary state judgments,an ANS is proposed to address the requirements of convergence and distribution in the evolutionary process,and then it can set different neighborhood sizes according to the evolutionary states of the population and sub-problems.Use the WFG series of functions to do the test,and the results show that this algorithm can effectively balance the convergence and distribution of population in the evolution the process,and improve the overall performance of the solution set.

Key words: Multi-objective Evolutionary Algorithm based on Decomposition (MOEA/D), neighborhood updating capability, evolutionary state, judgment mechanism, Adaptive Neighborhood Strategy(ANS)

中图分类号: