计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (25): 60-64.DOI: 10.3778/j.issn.1002-8331.2008.25.019

• 理论研究 • 上一篇    下一篇

基于两种进化模式的双种群协作差分演化算法

王培崇1,2,贺毅朝2,钱 旭1   

  1. 1.中国矿业大学 机电与信息工程学院,北京 100083
    2.石家庄经济学院 信息工程学院,石家庄 050031
  • 收稿日期:2007-10-31 修回日期:2007-12-25 出版日期:2008-09-01 发布日期:2008-09-01
  • 通讯作者: 王培崇

Cooperation differential evolution algorithm with double populations and two evolutionary models

WANG Pei-chong1,2,HE Yi-chao2,QIAN Xu1   

  1. 1.School of Mechanical Electronic & Information Engineering,China University of Mining & Technology,Beijing 100083,China
    2.Information Engineering School,Shijiazhuang University of Economics,Shijiazhuang 050031,China
  • Received:2007-10-31 Revised:2007-12-25 Online:2008-09-01 Published:2008-09-01
  • Contact: WANG Pei-chong

摘要: 提出了一种基于两种进化模式的双种群协作差分演化算法(DPDE)。在DPDE中,两个种群通过协作共同进化。首先,各种群以不同的进化模式,通过个体竞争实现自身进化;其次,种群之间基于局部信息传递和共享机制,通过随机交换个体方式相互协作、共同进化,既实现了不同进化模式间的优势互补,又可以改善种群的多样性。对于5个典型Benchmark测试函数,通过与DE和DEfirDE算法的比较表明:DPDE具有更好的全局收敛性和鲁棒性,特别适合求解高维多模态函数的最优化问题。

关键词: 差分演化, 进化模式, 协作进化, Benchmark函数

Abstract: Propose a cooperation differential evolution algorithm with double populations and two evolutionary models (DPDE).There are two populations in DPDE.First,each population has its evolutionary model,and they finish evolution by different evolutionary model independently.Secondly,they implement coevolution based on local information transfer and share between populations.It not only realizes developing advantages and avoiding disadvantages,but also adjusts population diversity.Comparing with DE and DEfirDE using five typical Benchmark functions,the results show that DPDE has superior global convergence and robust,especially fit for solving multimode and high dimension function optimization problems.

Key words: differential evolution, evolutionary model, cooperate evolution, benchmark function