电子学报 ›› 2016, Vol. 44 ›› Issue (11): 2639-2645.DOI: 10.3969/j.issn.0372-2112.2016.11.011

• 学术论文 • 上一篇    下一篇

基于权值向量的偏好多目标优化方法

张兴义1,2, 蒋小三2, 张磊1,2   

  1. 1. 安徽大学计算机与科学技术学院, 安徽合肥 230039;
    2. 安徽大学计算智能与信号处理教育部重点实验室, 安徽合肥 230039
  • 收稿日期:2015-05-11 修回日期:2015-08-24 出版日期:2016-11-25
    • 通讯作者:
    • 张磊
    • 作者简介:
    • 张兴义,男,1982年生,2009年毕业于华中科技大学自动化学院,获得博士学位,现为安徽大学计算机科学与技术学院副教授、博士生导师.主要研究方向包括非传统计算模型与算法、多目标优化算法及应用、膜计算.E-mail:xyzhanghust@gmail.com;蒋小三,男,1988年出生,安徽安庆人,现为安徽大学计算机应用技术专业硕士研究生.研究方向为偏好多目标进化算法.E-mail:westsxj88@126.com
    • 基金资助:
    • 国家自然科学基金 (No.61272152,No.61502001); 安徽大学学校学术与技术带头人引进工程 (No.J10117700050)

A Weight Vector Based Multi-objective Optimization Algorithm with Preference

ZHANG Xing-yi1,2, JIANG Xiao-san2, ZHANG Lei1,2   

  1. 1. School of Computer Science and Technology, Anhui University, Hefei, Anhui 230039, China;
    2. Key Lab of Intelligent Computing and Signal Processing of Ministry of Education, Anhui University, Hefei, Anhui 230039, China
  • Received:2015-05-11 Revised:2015-08-24 Online:2016-11-25 Published:2016-11-25

摘要:

偏好多目标优化方法是多目标优化领域的一个重要分支,其主要目的是仅搜索Pareto前沿面上部分区域内决策者感兴趣的解.基于MOEA/D算法根据预先设定的均匀分布的权值向量搜索Pareto最优前沿面的思想,本文提出了一种基于权值向量的偏好多目标优化方法,该方法通过引入具有偏好信息的权值向量,使算法仅搜索偏好点附近的解.仿真实验结果表明,与现有偏好多目标优化算法相比,本文方法具有支持多偏好点、偏好区域大小可控、偏好点位置无特别要求及偏好解具有更好收敛性的优势.

关键词: 多目标优化, 偏好多目标优化算法, 权值向量, 偏好解

Abstract:

Multi-objective optimization algorithms with preference are an important branch of multi-objective optimization.Its main aim is to find the Pareto optimal solutions in local regions interested by Decision Makers.Based on the idea of MOEA/D algorithm to search the Pareto front according to uniformly distributed weight vector,this paper proposes a weight vector based multi-objective optimization algorithm with preference.In the proposed method,the weight vector with preference is designed,by which the solutions around the preferred point interested by Decision Maker are found.Compared with existing algorithms,the simulation results verify that the proposed method can support multiple reference points,flexibly control the extent of preferred region,have no special requirement of the position of preference points and achieve better converge.

Key words: multi-objective optimization, multi-objective optimization algorithms with preference, weight vector, preferred solution

中图分类号: