计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (33): 33-36.

• 学术探讨 • 上一篇    下一篇

智能体遗传算法空间维数的比较研究

周 頔1,李勇明2,曾孝平2   

  1. 1.四川文理学院,四川 达州 635000
    2.重庆大学 通信工程学院,重庆 400030
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-11-21 发布日期:2007-11-21
  • 通讯作者: 周 頔

Research of space dimension of agent genetic algorithm

ZHOU Di1,LI Yong-ming2,ZENG Xiao-ping2   

  1. 1.Sichuan University of Arts and Science,Dazhou,Sichuan 635000,China
    2.College of Communication Engineering of Chongqing University,Chongqing 400030,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-11-21 Published:2007-11-21
  • Contact: ZHOU Di

摘要: 将智能体引入到遗传算法构成一个局部环境,可有效保持种群的多样性从而获得优良的优化性能。但是这个局部环境的空间维数一直未得到研究。根据智能体遗传算法的工作原理,空间维数越小,越能避免过早收敛现象发生。基于此,提出一种维数为1的链式智能体遗传算法(CAGA),并针对函数优化问题将其与文献[4]提出的维数为2的网络式智能体遗传算法(MAGA)进行了比较。实验采用了多个多维复杂函数进行优化实验,结果表明,该遗传算法比二维网格式遗传算法可获得更优的优化结果。

关键词: 维数, 智能体, 遗传算法, 数值优化

Abstract: The introduction of agent into genetic algorithm can keep the diversity of population to obtain good optimization performance.According to the principle of agent algorithm,the smaller the dimension is,the better the early convergence can be avoided.Based on it,this paper proposes Chainlike Agent Genetic Algorithm(CAGA) with one dimension.Besides,according to numerical optimization problems,this algorithm is compared with MAGA with two dimensions proposed in the paper [4].The experimental results show that CAGA can obtain better optimization performance than MAGA in paper [4].

Key words: dimension, agent, genetic algorithm, numerical optimization