聚类方法在电气石颜色成因研究中的应用  

An Application of the Clustering Analysis in the Study of the Tourmaline Color Genesis

作　　者：王进军[1] 陶晓风[1] 王武军 

机构地区：[1]成都理工大学,四川成都610059 [2]济南新型建筑材料厂,山东济南250100

出　　处：《东华理工学院学报》2006年第1期38-42,共5页Journal of East China Institute of Technology

摘　　要：天然电气石颜色多种多样,颜色成因非常复杂。为了更好了解它的颜色成因,采用统计学中的聚类方法对其颜色成因进行了分类。该分类方法以Matlab软件为工作平台,从测试数据(电子探针分析结果)中选出对电气石颜色形成起重要作用的氧化物,如TiO2,FeO,Fe2O3,MnO作为聚类分析的特征变量。以这些变量作为距离聚类分析的基础,对所研究各种颜色的电气石进行聚类。通过聚类分析研究,把电气石颜色成因的结果与其它测试方法的结果进行对比,证实本聚类方法真实可靠。The natural tourmalines＇ color is diversity and its genesis is complexity. In order to learn its color genesis, we study their color genesis on the ground of research from us and predecessor by the clustering analysis of statistics. This method bases on the Matlab software. We select some oxides which mainly effect tourmalines＇ color from the testing data. These oxides are used as particular variables of clustering analysis, such as TiO2, FeO, Fe2O3 and MnO. Various color tourmalines are classified by means of former variables. We contrast color genesis result by clustering analysis with result from others test methods. Obviously, this clustering analysis is credibility.

关 键 词：电气石 聚类方法 颜色成因 氧化物 特征变量 

分 类 号：P581[天文地球—岩石学]