[1]汪慎文,王佳莹,张佳星,等.SI7:应用精英档案和反向学习的多目标差分进化算法[J].郑州大学学报(工学版),2020,41(06):40-45.[doi:10.13705/j.issn.1671-6833.2020.06.011]
　Wang Shenwen,Wang Jiaying,Zhang Jiaxing,et al.A Multi-ob＜x＞jective?Differential?Evolution Algorithm with Elite-Archive and Opposition-ba＜x＞sed Learning[J].Journal of Zhengzhou University (Engineering Science),2020,41(06):40-45.[doi:10.13705/j.issn.1671-6833.2020.06.011]

点击复制

SI7:应用精英档案和反向学习的多目标差分进化算法()

分享到：

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

卷:

41卷

期数:

2020年06期

页码:

40-45

栏目:

出版日期:

2020-12-31

文章信息/Info

Title:

A Multi-ob＜x＞jective?Differential?Evolution Algorithm with Elite-Archive and Opposition-ba＜x＞sed Learning

作者:

汪慎文; 王佳莹; 张佳星; 王峰; 王晖

河北地质大学信息工程学院,河北石家庄050031, 河北地质大学人工智能与机器学习研究室,河北石家庄050031, 武汉大学计算机学院,湖北武汉430072, 南昌工程学院信息工程学院,江西南昌330099

Author(s):

Wang Shenwen;  Wang Jiaying;  Zhang Jiaxing;  etc.

School of Information Engineering, Hebei University of Geosciences, Shijiazhuang 050031, Hebei University of Geosciences Artificial Intelligence and Machine Learning Research Office, Hebei Shijiazhuang 050031, Wuhan University School of Computer, Hubei Wuhan 430072, School of Information Engineering, Nanchang School of Engineering, Nanchang 330099, Jiangxi 330099

关键词:

多目标优化;  精英档案;  反向学习;  差分进化算法;  网格约束分解

Keywords:

Multi -goal optimization;  elite archives;  reverse learning;  differential evolution algorithm;  grid constraint decomposition

DOI:

10.13705/j.issn.1671-6833.2020.06.011

文献标志码:

A

摘要:

针对多目标优化问题日渐复杂，受集成算法思想的启发，提出一种应用精英档案和反向学习的多目标差分进化算法。该算法通过建立一个外部档案，来保存种群进化过程中的非支配解，采用精英保留策略维持档案规模，提高算法收敛速度。在进化过程中根据反向学习代跳跃概率，使用一般反向学习策略生成反向解，扩大搜索范围，提高种群多样性。利用网格系统确定解的坐标，并根据一定的约束生成交叉池，在交叉池中选择父代个体，利用差分进化算法产生新个体，通过网格约束分解排序算法选择下一代种群。将此算法与其他几个代表性算法在UF测试函数上进行实验，结果表明，所提出的算法多样性及收敛性表现更优。

Abstract:

The multi-ob＜x＞jective optimization problem is becoming more and more complex. Inspired by the ensemble algorithm, a multi-ob＜x＞jective?differential?evolution algorithm with elite-archive and opposition-ba＜x＞sed learning is proposed in this paper. In this algorithm, an external archive is created to save the nondominated solutions in the evolutionary process of the population, and an elite strategy is adopted to maintain the size . Use the preset opposition-ba＜x＞sed generation jumping and the general opposition-ba＜x＞sed learning strategy to generate the different solutions of the individual and stored in the elite archive, to expand the search scope and improve population diversity. The grid is used to determines the coordinates of the solutions, and the restricted mating pool is generated according to certain constraints. The parent solutions are selected in the restricted mating pool to produce the new individual by using differential?evolution algorithm, then generate the next iteration population by constrained decomposition with grids sorting. The experimental results show that the proposed algorithm is superior to some state-of-the-art multi-ob＜x＞jective algorithms in diversity and convergence on UF test problems

常用功能