求解约束优化问题的不可行比率差分算法  

A Deferential Evolution Based on No Feasible Solution Rate for Constrained Optimization Problems

作　　者：陈德祥[1] 文诗华[2] 宋武[1] 

机构地区：[1]海南琼州学院电子信息工程学院 [2]湘潭大学教务处

出　　处：《科学技术与工程》2013年第31期9408-9412,共5页Science Technology and Engineering

基　　金：国家自然科学基金项目(61070088)资助

摘　　要：利用差分进化算法求解约束优化问题时存在两个关键问题,一是种群个体根据约束处理准则从不可行区域中快速接近可行区域;二是在可行区域内根据约束处理准则如何更好地进行全局搜索。提出了一种基于不可行解比率的差分进化算法求解约束优化问题,算法最主要的特点是利用了种群不可行比率的信息,使种群快速地接近可行区域,并且通过不可行比率选择不同的差分算子,从而完成对约束优化问题的求解,通过对13个benchmark的测试,结果表明所提出的算法是有效的。There are two key points inusing deferential evolution （DE） to solve constrained optimization prob- lems （COPs）. The one is the individual of deferential population quickly get near to the feasible area from the non- feasible area. The other involves how to make a global searching in the feasible area according under the constrain- ed condition. Deferential evolution based on the non-feasible rate to solve constrained optimization problems is pro- posed in this paper, constrained optimization principle, which not only instruct to selection operator of DE but also is a guidance to choose different type of deferential operator are designed according to the individual non-feasible rate in the population. 13 benchmark constrain functions are tested in the experiment and it shows that our algo- rithm proposed in this paper is more effectively than other algorithms.

关 键 词：差分算法 不可行解 比率 约束优化问题 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]