摘要
由于遗传算法解决问题时容易陷入局部极值点,根据遗传算法全局搜索能力强和模拟退火算法局部搜索能力优的特点,将它们混合使用,同时改进初始群体产生方法,使随机产生的初始群体之间有较明显的差别,能均匀分布在解空间,并采取与进化代数相关的多精英保留策略及改进的自适应选择与变异操作.模拟退火算法的结束条件改进为当连续五代个体与前一代适应值无变化或当前温度小于结束温度.仿真实验表明新算法在求解多峰值问题时改善了遗传算法的局部搜索能力,有效地解决了遗传算法的早熟现象,显著提高了遗传算法求得全局解的概率.
As it is easy for genetic algorithm to fall into local extreme point,based on genetic algorithm to have strong global searching power and simulated annealing algorithm to have powerful local searching,if hybriding genetic algorithem and simulated annealing algorithm,it will help improve the method for getting initial groups and make clear differences between the randomly generated initial groups,it can be evenly distributed in the solution space.Simulated experiments show that in solving multimodal mroblems,this new hybrid genetic algorithm effectively improves the local searching power of hybrid genetic algorithms and prevents its premature phenomenon.It greatly increases the probability of obtaining the global solution as well.
出处
《甘肃联合大学学报(自然科学版)》
2011年第4期14-17,21,共5页
Journal of Gansu Lianhe University :Natural Sciences
关键词
多峰值优化
改进遗传算法
改进模拟退火算法
混合算法
optimization of multimodal problem
improvement of genetic algorithm
improvement of simulated annealing algorithm
hybrid algorithm