基于模糊邻近关系的结构聚类  

Structural clusters based on fuzzy proximity relations

作　　者：唐旭清[1] 方雪松[2] 朱平[1] 

机构地区：[1]江南大学理学院,无锡214122 [2]中国建筑科学研究院地基基础研究所,北京100013

出　　处：《系统工程理论与实践》2010年第11期1986-1997,共12页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(60475017);国家重点基础研究发展计划(973计划)(2004CB318108);高等学校博士学科点专项科研基金(200403057002);国际科技合作研究项目(2007DFA706700);江南大学创新团队发展计划

摘　　要：在有序粒度空间理论的基础上,提出了基于模糊邻近关系的结构聚类分析理论和方法.首先,给出了依据距离的一致聚类的概念,提出了模糊粒度空间的一致聚类特征;其次,给出了模糊邻近关系结构聚类的粒度表示,并获得了基于模糊邻近关系结构聚类的快速算法;进而,提出了获取最佳聚类的新方法,并且这一方法是全局最优的;最后,给出了通过两个模糊邻近关系的交运算获取结构聚类融合的方法.为复杂系统结构分析研究提供了一整套理论工具和方法.On the basis of fuzzy granular space,structural clustering analysis based on fuzzy proximity relations is proposed.Firstly,the concept of consistent clustering according to metric is proposed,and the research on consistent clustering of a fuzzy granular space is given.Secondly,the granular representation on structural clustering of a fuzzy proximity relation is studied,and the algorithm to obtain its structural clustering is given.Thirdly,the research to determine the optimal clustering based on a fuzzy granular space from a fuzzy proximity relation is proposed,and a new method to obtain it is given,which the new method is overall.Finally,the fusion technology is attained by applying the intersection operation of two fuzzy proximity relations.These conclusions provide a comprehensive theoretical tool and methodologies for studying structural analysis of complicated systems.

关 键 词：模糊邻近关系 模糊粒度空间 一致聚类 最佳聚类 聚类融合 

分 类 号：O159[理学—数学] TP18[理学—基础数学]