基于均匀设计的遗传算法及其应用  

Uinform Design Based Genetic Algorithm and Its Application

作　　者：张常利[1] 杜永贵[1] 

机构地区：[1]太原理工大学信息工程学院,山西0300240

出　　处：《软件》2010年第11期68-72,共5页Software

摘　　要：本文为克服遗传算法解决有关优化问题时,对遗传算法的各操作参数及初始种群的选取往往凭借经验带有盲目性,还需要大量反复的试验才能获得比较理想的操作参数和初始种群的缺点,引入均匀试验设计的思想将遗传算法参数设定描述为一个多因素多水平的优化设计问题,应用均匀设计方法选取遗传算法操作参数和初始种群,对全维状态观测器设计的实例应用表明该方法简单有效,具有工程实用价值,为遗传算法的实际应用提供了科学指导。This paper devotes to overcome the defect when we use genetic algorithm to solve some optimum problems, we always select the genetic operator parameters and initial populations with experiences and great blindness.After that we also have to do many trials to get a ideal operator parameters combination and populations.Now we resort to the thought of uniform trial design and describe the genetic operator parameters and initial populations selection as a optimum design problems with multi-factor and multi- level and adopt the uniform design method to get the advisable genetic operator parameters and initial populations.We practise this improved genetic algorithm to design a full state observer and the simulation results show that the method is simple and effective and have practical value in engineering ,and indicate that uniform design provides a scientific instruction to the genetic algorithm in practical application.

关 键 词：遗传算法 均匀设计 遗传操作参数 初始种群 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]