混合遗传算法自适应策略研究

Research on Adaptive Strategies for Hybrid Genetic Algorithm

在针对性设计使得混合遗传算法可处理大规模组合优化问题的基础上,分析问题解空间的特征,研究相应自适应策略。提出和采用了交叉全局探索单个模因构造、候选解接受、局部搜索和种群多样性保持等自适应策略,大幅减少了混合遗传算法运行时间。以超大规模集成电路标准单元布局问题为测试实例,实验结果表明了这些自适应策略的有效性。

By the problem of specific designs,the hybrid genetic algorithms can deal with the large-scale combinatorial optimization problems.On this basis,we analyze the characteristics of the problem solution space,and then research on the corresponding adaptive strategies.We propose and adopt a set of adaptive strategies including single meme construction for cross global exploration,candidate solution acceptance,local exploitation,and population diversity maintenance,which can greatly reduce the runtime of the hybrid genetic algorithm.Taking the VLSI standard cell placement problem as a test case,the experimental results show that these adaptive strategies are both efficient and effective.