• 论文 •    

求解电路划分问题的混合式遗传算法

南国芳,李敏强,寇纪淞   

  1. 天津大学系统工程研究所,天津300072
  • 出版日期:2004-10-15 发布日期:2004-10-25

Hybrid genetic algorithm for solving circuit partitioning problem

NAN Guo-fang, LI Min-qiang, KOU Ji-song   

  1. Inst. of System Eng., Tianjin Univ., Tianjin300072, China
  • Online:2004-10-15 Published:2004-10-25

摘要: 为了进一步降低超大规模集成电路设计的复杂性、增强图形可读性,提出了一种求解电路划分问题的混合式遗传算法。与传统遗传算法不同的是,该算法采用了顺序交叉和单元交换变异方式,同时对交叉概率和变异概率进行了优化设计。与K-L算法及传统遗传算法得出的结果比较,该混合式遗传算法在计算时间和划分结果上显示出其优越性。

关键词: 电路划分, 神经网络, 遗传算法, 模拟退火

Abstract: In order to further reduce the complexity of super large- scale integration design and enhance graphical readability, a hybrid genetic algorithm for solving circuit partitioning problem was presented. The difference between this algorithm and traditional genetic algorithm lied in the fact that this algorithm adopted order crossover and cell exchange mutation methods; at the same time crossover probability and mutation probability were optimized. In computing time and partitioning results this hybrid genetic algorithms superiority was appeared compared with results derived by K-L algorithm and genetic algorithm.

Key words: circuit partitioning, neural network, genetic algorithm, simulated annealing

中图分类号: