基于马氏距离算法区分玻璃样品的研究  

Distinguishing the Glass Samples by Mahalanobis Distance Algorithm

作　　者：刘慧娟 姜华[1] 王璐[1] LIU Huijuan;JIANG Hua;WANG Lu(Beijing Municipal Institute of Forensic Science,Beijing 100054,China)

机构地区：[1]北京市公安局刑侦总队,北京100054

出　　处：《刑事技术》2018年第3期218-221,共4页Forensic Science and Technology

基　　金：公安部应用创新计划项目(No.2014YYCXBJSJ003)

摘　　要：目的借助计算机编程来区分玻璃样品。方法利用激光剥蚀电感耦合等离子体质谱(LA-ICP/MS)方法对200种玻璃样品的42种元素浓度进行测定,通过计算机编程对玻璃样品进行区分。结果随机抽取70种玻璃样品通过马氏距离算法进行归类,68组数据正确归类,测试结果的准确率达到97%。结论依据马氏距离算法能够有效区分玻璃样品。Objective To distinguish glass samples by an automated algorithm. Methods The concentrations of 42-kind elements were analyzed from 200-category tested glass samples with LA-ICP/MS method. Mahalanobis distance was used to establish a computer-operated programming algorithm for distinguishing the glass samples. Results Among the 70-catogery glass samples that were randomly selected from the newly-built sample database to classify by mahalanobis distance algorithm, 68 sets of data were sorted out accurately, revealing the distinguishing accuracy up to 97%. Conclusion Glass samples can be effectively distinguished based on Mahalanobis distance algorithm.

关 键 词：激光剥蚀电感耦合等离子体质谱 玻璃 马氏距离 

分 类 号：DF794.3[政治法律—诉讼法学]