作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2007, Vol. 33 ›› Issue (09): 22-24,6.

• 博士论文 • 上一篇    下一篇

基于搜索空间信息的新型遗传算法

李 航,李敏强,寇纪淞   

  1. (天津大学系统工程研究所,天津 300072)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-05-05 发布日期:2007-05-05

Improved Genetic Algorithm Based on Information Searching Space

LI Hang, LI Minqiang, KOU Jisong   

  1. (Institute of Systems Engineering, Tianjin University, Tianjin 300072)
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-05-05 Published:2007-05-05

摘要: 针对多模态函数优化问题,提出了一种遗传算法。用正交设计法对搜索空间进行探索,用差分法确定适应函数关于各变量的灵敏性,对灵敏性较高的几个变量,按其所在维度对搜索空间进行划分。用遗传算法分别对各子空间进行单次搜索,根据各子空间当前最优解的大小依概率对其应用遗传搜索。根据情况决定是否对各子空间递归调用该算法。仿真实验证明:在同等计算量的情况下,该算法相对于其他遗传算法能明显提高全局最优解的精度,并获取更多的局部最优解和其他有效信息。

关键词: 多模态问题, 正交设计, 空间划分, 遗传算法

Abstract: A genetic algorithm is presented for the multi-modal landscapes. It explores the searching space using orthogonal design and partitions it by the information acquired. According to the probability function based on their current optimum, it applies the standard genetic algorithm to these subspaces in order. It can recursively apply this algorithm accordings to the need. The experiments show that this algorithm, compared with other traditional genetic algorithms, converges to the global optimum more accurately within equal time. It can offer more local optimum solutions and other useful information of the searching space.

Key words: Multi-modal landscapes, Orthogonal design, Partition of solution space, Genetic algorithms

中图分类号: