一种改进的求解约束函数优化问题的演化算法  

AN IMPROVED EVOLUTIONARY ALGORITHM FOR OPTIMIZATION OF CONSTRAINED FUNCTION

作　　者：谢柏桥[1] 戴光明[1] 石红玉[1] 

机构地区：[1]中国地质大学计算机学院,湖北武汉430074

出　　处：《计算机应用与软件》2008年第7期48-50,共3页Computer Applications and Software

基　　金：湖北省自然科学基金资助项目(2003ABA045)

摘　　要：带约束的函数优化是函数优化中最多,也是较难的问题。针对这个问题提出一种改进的算法,它是基于遗传算法的非参惩罚函数的函数优化。通过改进广义的目标函数,对不可行解恰当地进行惩罚,并引进柯西组合、柯西变异以及高斯变异,极大地提高了算法的全局搜索和局部搜索能力,克服传统遗传算法"爬山能力差"的弱点。The optimization problems of constained function are very common in function optimization, and they are difficult to handle. An advanced algorithm is proposed for this problem, which is a non-parameter penalty function optimization based on GAS. The generalized objec- tive function is improved, and Cauchy combination, Cauchy mutation and Gauss mutation are introduced. The abilities of global search and local search of this algorithm are enhanced, and the shortage of the traditional GAS in mountain climbing is overcome.

关 键 词：约束优化 柯西变异 高斯变异 惩罚函数 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] O224[自动化与计算机技术—计算机科学与技术]