共享遗传算法的搜索精度

共享机制赋予了遗传算法在一次搜索过程确定目标函数多个极值点的能力。本文着重研究共享遗传算法的搜索精度问题。通过对共享遗传算法建立概率模型 ,本文得到了共享遗传算法搜索精度不一定会随群体的不断进化而任意减小的结论。对于只进化有限代数的算法 ,本文分析了一类常用共享函数中算法参数和目标函数形状对算法搜索精度的影响 ,提出了有利于提高搜索精度的算法设计准则

THE SEARCHING ACCURACY OF GENETIC ALGORITHMS WITH SHARING

Sharing methods give genetic algorithms the ability to locate several extremities of an objective function through only a single search This paper studies the searching accuracy of genetic algorithms with sharing According to a model of sharing GA, the conclusion is pointed out that the accuracy of searching results sometimes cannot be made arbitrarily high even when the number of generations become infinite For the population that evolves finite generations, discussions are made about the influence of the shape of share functions and objective functions on the accuracy of the searching results Based on this, some design rules are proposed which will benefit the accuracy of sharing methods