基于群智能SOM算法的特征造型搜索技术  

Searching technology for feature module based on self-organizing clustering

作　　者：李峰[1] 孙立镌[1] 

机构地区：[1]哈尔滨理工大学计算机科学与技术学院,哈尔滨150080

出　　处：《计算机工程与应用》2011年第13期1-3,共3页Computer Engineering and Applications

基　　金：国家自然科学基金 No.60173055~~

摘　　要：为解决语义特征化后复杂造型协同设计中造型共享库中的造型检索问题,提出了一种基于群智能自组织聚类算法。该算法首先将语义特征造型信息向量化,通过语义造型特征树得到语义特征造型特征集,以蚁群算法做为自组织准则,并以粒子群算法做为蚁群移动模型,将特征语义群分布在一个平面上进行聚类,递归收集聚类结果。试验证明,采用此种方法,可以对特征造型完成准确率很高的聚类,使特征库组织性得到了很大提高。To solve the problem of searching complex feature model in the shared module library for collaborative design,this paper proposes a self-organizing clustering algorithm to re-arrange the structure of modules.Firstly,the module tree information is converted to vector represented.Then ant colony is used as the rule of self-organizing,and particle swarm optimization algorithm is used to calculate the next position.Finally,the objects are clustered and the result is collected by a recursive algorithm.After the self-organizing cluster,the structure of module library becomes better arranged.

关 键 词：特征语义造型 自组织聚类 蚁群算法 粒子群 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]