带有单亲遗传特征的蚁群算法  

Ant colony algorithm with partheno-genetic features

作　　者：周鹏[1] 

机构地区：[1]湖北汽车工业学院,湖北十堰442002

出　　处：《计算机工程与设计》2007年第9期2001-2002,2099,共3页Computer Engineering and Design

基　　金：湖北省教育厅科学技术研究基金项目(B200623002)

摘　　要：蚁群算法是一种新型的模拟进化算法,具有许多优良的性质,但同时也存在着计算时间过长和易导致早熟收敛等缺点。单亲遗传算法不使用基本遗传算法常用的交叉算子,简化了遗传操作过程,且不要求初始群体具有广泛多样性,计算速度较快,不存在早熟收敛现象。将这两种算法结合,提出一种具有单亲遗传特征的蚁群算法。将蚁群算法每次搜索结果作为初始种群,进行单亲遗传优化改良,求得最短路由。在旅行商问题上的实验证明了该算法的有效性。Ant colony algorithm （ACA） is a novel simulated evolutionary algorithm, which shows so many promising characters. But it has some shortcomings such as needing long computing time and it is easy to bring immature convergence. Partheno-genetic algorithm （PGA） does not use crossover operator of simple genetic algorithm. In PGA the genetic operator is simple and initial populations need not be varied, PGA makes the speed of convergence faster than ACA and there is not immature convergence. A new ant colony algorithm with partheno-genetic features is proposed based on the combination of ACA and PGA, It improves the search result of ACA as primal group of PGA to come at the best solution each time. The optimization computing of some examples on traveling salesman problem is made to show that it is useful and simple.

关 键 词：蚁群算法 单亲遗传算法 模拟进化算法 早熟收敛 旅行商问题 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]