基于近红外光谱的四元调和食用油定量分析  

Quantitative Analysis of Quaternary Edible Blended Oil Based on Near Infra-red Spectroscopy

作　　者：李淑娟 卞希慧[1] 李倩[1] 范清杰 陈娇娇[1] LI Shujuan;BIAN Xihui;LI Qian;FAN Qingjie;CHEN Jiaojiao(State Key Laboratory of Separation Membranes and Membrane Processes,School of Environmental and Chemical Engineering,Tianjin Polytechnic University,Tianjin 300387,China;Tianjin Green Security Technology Co.,Ltd.,Tianjin 300384,China)

机构地区：[1]省部共建分离膜与膜过程国家重点实验室,天津工业大学环境与化学工程学院,天津300387 [2]天津格润赛福科技有限公司,天津300384

出　　处：《天津科技大学学报》2018年第3期18-24,共7页Journal of Tianjin University of Science & Technology

基　　金：国家自然科学基金资助项目(21405110);天津市科技特派员资助项目(16JCTPJC44700)

摘　　要：配制51个含有玉米油、大豆油、稻米油和芝麻油的四元调和油样品,采用近红外光谱(near infra-red,NIR)技术直接扫描其光谱,考察不同预处理方法对近红外光谱的预处理效果.在最佳预处理方法基础上,结合偏最小二乘回归(partial least square regression,PLS)建立各组分定量分析模型.将未知样品光谱代入模型中,预测各组分的含量.结果显示,玉米油、大豆油、稻米油和芝麻油的相关系数分别为0.980,0、0.988,7、0.984,7、0.988,9.因此,近红外光谱技术结合化学计量学可以实现四元调和油组分的准确快速定量分析.51 samples of quaternary edible blended oil containing corn oil,soybean oil,rice oil and sesame oil were prepared.The spectra of these samples were collected with near infra-red（NIR）transmission spectroscopy.The effects of different preprocessing methods on NIR spectroscopy were investigated.Based on the best preprocessing method,a calibration model for quantitative analysis of each component was established with partial least square regression（PLS）.The model used spectra to predict the content of each component of the unknown quaternary edible blended oil samples.The results showed that the correlation coefficient of corn oil,soybean oil,rice oil and sesame oil is 0.980,0,0.988,7,0.984,7,0.988,9,respectively.Therefore,accurate quantitative analysis of quaternary edible blended oil can be achieved with NIR spectroscopy combined with chemometrics.

关 键 词：近红外光谱 食用调和油 预处理方法 偏最小二乘回归 定量分析 

分 类 号：O65[理学—分析化学]