文献类型：期刊文献

中文题名：近红外透反射光谱用于掺假牛奶的快速识别初探

英文题名：Primary Study on Rapid Identification of Counterfeit Milk Based on Near Infrared Transmission Spectroscopy

作者：郭美兰[1];孙正鹏[1];张超[1];倪力军[1];张立国[1]

机构：[1]华东理工大学化学与分子工程学院,上海200237

年份：2010

卷号：51

期号：5

起止页码：270

中文期刊名：化学世界

外文期刊名：Chemical World

收录：北大核心:【北大核心2008】；CSCD:【CSCD2011_2012】；

基金：上海市大学生创新实验计划项目(S0811)

语种：中文

中文关键词：牛奶;掺假奶;近红外透反射光谱;快速识别

外文关键词：pure milk;counterfeit milk;near infrared spectroscopy;rapid identification

摘要：为快速识别掺加违禁物质的假牛奶,在市售纯奶基础上按不同比例添加水+糊精、水+糊精+三聚氰胺(硝酸铵、尿素)的溶液,配制掺假牛奶,测试真、假牛奶样品的近红外透反射光谱,分别采用PLS-DA、KNN、LDA、改进与简化的KNN(IS-KNN)模式识别方法和不同光谱预处理方法进行真、假牛奶的识别。各模式识别方法在不同预处理光谱下对真、假两类牛奶的识别正确率最高为95.12%和96.34%并且在最高正确率下对所有掺假牛奶均能正确识别。表明采取适宜的近红外光谱预处理方法和模式识别方法,可以良好地对纯奶和掺假牛奶进行区分。
To rapidly identify counterfeit milk containing some prohibited substances,such as melamine,we prepared some counterfeit milk by adding dextrin water solution,dextrin water solution and melamine(ammonium nitrate,urea)in different mix proportions to finished pure milk.The transmission near-infrared(NIR)spectra of the pure and adulterate milk samples were collected.PLS-DA,KNN,LDA and improved and simplified KNN(IS-KNN)methods based on different pretreated NIR spectra were applied to discriminate milk sample.The highest prediction accuracy of the models built by these pattern recognition methods is 95.12%and 96.34%,respectively,and almost all the adulterated milk samples could be correctly discriminated.The result indicates that if appropriate spectrum pretreatment method and pattern recognition method were used,pure and counterfeit milk could be well discriminated.