基于动态聚类邻域分区的并行蚁群优化算法  

Parallel Ant Colonies Optimization Algorithm Based on Nearest Neighbor Classify Used to Dynamic K-Means Cluster

作　　者：丁建立[1] 陈增强[1] 袁著祉[1] 

机构地区：[1]南开大学信息学院,天津300071

出　　处：《系统工程理论与实践》2003年第9期105-110,共6页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(60174021);天津自然科学基金重点项目(013800711);河南科技攻关项目

摘　　要：　本文算法体现"分而治之"的思想,首先采用动态K均值聚类快速邻域分解,其次应用蚁群算法同时对分区并行优化计算,最后基于分区重心进行邻域全局连接,得到大规模TSP问题的满意解.For a large number of TSP of combinatorial\|explode NP\|Hard problems, Any only one algorithm will be face challenge in optimization characteristic and CPU run\|time. The paper thinking is 'divide and rule'. First, we fast nearest neighbor classify used to \$k\$\|means cluster. Second, we are parallel computing used to ant colonies optimization for every group. Finally, we get good results used to globally optimizing use of between\|group linkage and between\|group Centro baric distance.

关 键 词：动态K均值聚类 邻域分区搜索 并行蚁群优化算法 大规模TSP问题 

分 类 号：O232[理学—运筹学与控制论]