古代玻璃文物的成分分析及类型鉴别  

Composition Analysis and Type Identification of Ancient Glass Artifacts

作　　者：韩丽[1,2] 王筠尹 HAN Li;WANG Junyin(School of Mathematics and Physics,Yuxi Normal University,Yuxi,653100,China;School of Statistics and Mathematics,Yunnan University of Finance and Economics,Kunming 650221,China)

机构地区：[1]玉溪师范学院数理学院,云南玉溪653100 [2]云南财经大学统计与数学学院,云南昆明650221

出　　处：《玉溪师范学院学报》2024年第3期78-86,共9页Journal of Yuxi Normal University

基　　金：云南省教育厅科学研究基金项目(项目编号:2023J0997);玉溪师范学院2022年校级教学研究与改革实践项目(项目编号:202215);玉溪师范学院2022年校级一流课程(项目编号:2022kc11);玉溪师范学院2022年课程思政示范项目(项目编号:2022sz15)。

摘　　要：根据古代玻璃文物的化学成分数据,运用主成分分析和灰色关联分析对高钾玻璃和铅钡玻璃的化学成分进行研究,发现对于两种玻璃制品其化学成分关联性较高的是氧化锶(SrO)和氧化铁(Fe_(2)O_(3)),关联系数分别为1.0和-1.2;运用决策树算法对古代玻璃的化学成分进行数据挖掘并建立随机森林模型,将编号为A1,A6,A7的未知文物划分为高钾玻璃,将编号为A2,A3,A4,A5,A8的文物划分为铅钡玻璃.对该分类结果进行检验,发现该结果的信息损失不到1%,分类结果的BrierScore为0.007 8,这表明所建立的随机森林模型非常可靠,从而为考古工作者对古代玻璃制品的研究提供了依据.Based on the chemical composition data of ancient glass artifacts,the chemical components of highpotassium glass and leadbarium glass were studied with Principal Component Analysis(PCA)and Grey Relational Analysis.After the chemical components highly correlated to both types of glass artifacts were found to be strontium oxide and iron oxide,with correlation coefficients of 1.0 and 1.2,respectively,the algorithmic approach of decision trees was used to analyze the chemical data associated with ancient glass,which then informed the formulation of a Random Forest model.Ultimately,artifacts numbered A1,A6,and A7 were classified as highpotassium glass,while artifacts numbered A2,A3,A4,A5,and A8 were classified as leadbarium glass.The verification of these results showed that the information loss was less than 1%,with a Brier Score of 0.0078 for the classification results.This indicates that the newly established Random Forest mode,confirmed to be very reliable,provides a basis for archaeological researchers to study ancient glass artifacts.

关 键 词：相关性 灰色关联分析 随机森林模型 主成分分析 

分 类 号：P157[天文地球—天文学]