[1]张春江,高亮,吴擎,等.基于分解的多目标进化算法在工程优化中的应用[J].郑州大学学报(工学版),2015,36(06):38.[doi:10.3969/j.issn.1671-6833.2015.06.008]
　ZHANG Chunjiang,TAN Kay Chen,GAO Liang,et al.Multi-Objective Evolutionary Algorithm Based on Decomposition for Engineering Optimization[J].Journal of Zhengzhou University (Engineering Science),2015,36(06):38.[doi:10.3969/j.issn.1671-6833.2015.06.008]

点击复制

基于分解的多目标进化算法在工程优化中的应用()

分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:

36

期数:

2015年06期

页码:

38

栏目:

出版日期:

2015-12-25

文章信息/Info

Title:

Multi-Objective Evolutionary Algorithm Based on Decomposition for Engineering Optimization

作者:

张春江¹; 2; 高亮¹; 吴擎³; Kay Chen Tan²

1.华中科技大学机械科学与工程学院,湖北武汉430074; 2. Department of Electrical and Computer En-gineering，National University of Singapore，Singapore 117583;3.华中农业大学工学院,湖北武汉430070

Author(s):

ZHANG Chunjiang1; 2; TAN Kay Chen2; GAO Liang1;  wU qing3

1. School of Mechanical Scinece & Engineering，Huazhong University of Science and’Technology，Wuhan 430074，China，2.Department of Electrical and Computer Engineering，National University of Singapore，Singapore 117583; 3.School of Engineer-ing，Huazhong Agricultural University ，Wuhan 430070，China

关键词:

多目标进化算法;  MOEA/D; 归一化; 工程优化; 差分进化; 约束处理

Keywords:

multi-objective evolutionary algorithm; MOEA/D;  normalization;  engineering optimization;  dif-ferential evolution;  s-constraint handling

DOI:

10.3969/j.issn.1671-6833.2015.06.008

文献标志码:

A

摘要:

将基于分解的多目标进化算法(Muli-objective Evolutionary Algorithm Based on Decomposition,MOEA/D)应用于工程优化问题时,由于各目标函数在数量级及量纲上的不同,需要对目标函数进行归一化处理.首先,采用一种自适应﹖约束差分进化算法( e Constrained Differential Evolution, gDE)寻找各个目标在Pareto前沿上的最大值和最小值,利用这些值对各目标进行归一化处理;然后,用MOEA/D进行求解,并在算法中加入了自适应s约束处理技术;最后,采用一个标准测试问题和一个焊接梁设计优化问题对该算法进行测试,并与其他两种归一化方法进行了比较.根据提出的方法,MOEA/D能对Pare-to前沿的一端进行集中优化,因而能处理一些Pareto前沿两端难以优化的问题.

Abstract:

In order for effective application of Multi-Objective Evolutionary Algorithm based on Decomposition(MOEA/D) in engineering optimization，normalization of the range of objective values is needed. A self-a-daptive s constrained Differential Evolution ( gDE) algorithm is proposed to obtain the minimum and maximumvalues of each objective on the Pareto Front ( PF). After normalization，MOEA/D can then be effectively ap-plied. In addition ,the self-adaptive s constraint method is combined with MOEA/D for constraint handling. Abenchmark problem and a weld bean design problem are used to evaluate the performance of the algorithm a-gainst two other normalization methods. One main advantage of the proposed method is the selective concen-trated optimization on some regions on the Pareto front which allows handling of problems where regions of Pa-reto front are difficult to be optimized.

常用功能