高维多目标进化算法研究综述
DOI:
CSTR:
作者:
作者单位:

东北大学自动化研究中心

作者简介:

孔维健

通讯作者:

中图分类号:

TP301

基金项目:


Large-Dimensional Multi-Objective Evolutionary Algorithms
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    目前的多目标进化算法主要是基于Pareto支配关系实现个体解之间的比较与选择.这类方法能够有效地解决2目标的优化问题,然而随着目标个数的增加,当优化目标超过4维即具有高维目标时,它们的搜索能力与优化效果将大大下降.针对这一问题,首先对目标个数增加对优化问题的影响给予了分析,并验证了基于Pareto排序的NSGA2算法在高维目标优化问题中的表现,然后对当前提出的高维目标进化算法进行了分类综述,最后总结了这些方法的特点与缺陷,并给出了进一步可能的研究方向.

    Abstract:

    The current Multi-Objective Evolutionary Algorithms (MOEAs) are mostly based on Pareto dominance relation to perform compare and ranking of individuals in the population. This kind of methods can solve two-objective optimization problems successfully, but their search ability and performance will deteriorate badly when the number of objectives exceeds four. Aiming to the problem, the influence is analyzed the number of objectives bring on optimization problems. And it is examined that NSGA2 performs in large-dimensional multi-objective optimization problems by a set of experiments, which is known as the representative of MOEAs baesd on Pareto ranking. Then the large-dimensional multi-objective evolutionary algorithms are surveyed by categories. Finally the proposed methods are evaluated and topics for future research are suggested.

    参考文献
    相似文献
    引证文献
引用本文

孔维健.高维多目标进化算法研究综述[J].控制与决策,2010,25(3):321-326

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2009-05-26
  • 最后修改日期:2009-08-27
  • 录用日期:
  • 在线发布日期: 2010-03-20
  • 出版日期:
文章二维码