基于凝聚层次聚类的K均值结构面产状分组  

Partitioning discontinuities of rock mass based on K-means and hierarchical clustering analysis algorithms

作　　者：李运生 宋金龙 李煜[1] 靖盼盼 LI Yunsheng;Song Jinlong;LI Yu;JING Panpan(School of Earth Science and Engineering,Nanjing University,Nanjing 210046,China)

机构地区：[1]南京大学地球科学与工程学院,江苏南京210046

出　　处：《人民长江》2018年第6期44-49,共6页Yangtze River

基　　金：国家重点基础研发发展计划"973"项目(2013CB036001);国防科技重点项目(科工二司[2015]297号)

摘　　要：结构面产状数据分组是进行岩体力学性质分析及稳定性评价的基础工作,是工程地质领域一个重要的研究课题。K均值聚类方法从提出到现在,在结构面分组中得到了很好的应用,但是它需要事先给定聚类中心及分组数。为此,提出了改进算法,采用凝聚层次聚类法作为前处理,得到间距较大的几个聚类中心,随后以此聚类中心为初始聚心进行K均值聚类,并同时根据相关目标函数来确定最优分组数。将此改进方法应用于野外实测结构面分组中,得到的结果可靠,分类合理,可以准确确定结构面的优势产状。Partitioning of rock mass discontinuities,as an important research subject in engineering geological field,is a basic work for study on mechanical and stability analysis of rock mass.K-means cluster analysis has been well applied in discontinuities partitioning since it was put forward.However,the initial clustering centers and the number of clusters have to be determined in advance.Therefore,we adopt hierarchical clustering analysis to get the centers of large spacing.Then these centers are used as the initial centers for K-means clustering.Thereafter,the optimal number of groups is finally determined according to the relevant objective function.Being applied to a practical project,this proposed method shows a reliable result and reasonable partitioning to determine the dominant partitioning of discontinuities.

关 键 词：结构面分组 产状数据 K均值聚类 凝聚层次聚类 岩体力学 

分 类 号：P642[天文地球—工程地质学]