近红外光谱法主成分分析6种植物油脂的研究  

PRINCIPAL COMPONENT ANALYSIS OF 6 KINDS OF VEGETABLE OILS AND FATS BY NEAR INFRARED SPECTROSCOPY

作　　者：李娟[1] 范璐[1] 邓德文 周展明[3] 吴存荣[3] 唐怀建[3] 

机构地区：[1]河南工业大学化学化工学院,河南郑州450001 [2]瑞士步琪实验室仪器公司中国市场部,上海200030 [3]河南工业大学粮油食品学院,河南郑州450001

出　　处：《河南工业大学学报（自然科学版）》2008年第5期18-21,共4页Journal of Henan University of Technology：Natural Science Edition

基　　金：河南工业大学专项基金项目(07XJC001)

摘　　要：采用近红外光谱技术,结合主成分分析方法,研究区分6种植物油脂的测定方法.用不同品种和不同产地的油料制备植物油,其中32种大豆、34种花生、28种菜籽和12种棉籽,采集不同厂家生产的20种棕榈油和12种米糠油样品.测定6种植物油脂138个样品的近红外透射光谱.以植物油脂的光谱信息作变量,应用NIRCal5.2软件进行光谱预处理及主成分分析,随机取2/3的样品作定标集,1/3作验证集,选取负荷量差别较大的7个主成分进行得分比较.结果显示,138个样品被识别为相互独立的6组,分类精度100%,验证准确率100%.An analytical procedure was developed to discriminate and analysis 138 vegetable oils using near-infrared （NIR） spectroscopy and method of multivariate analysis that was principal component analysis （PCA）. In the samples, 34 peanut oil, 32 soybean oil, 12 cottonseed oil and 28 rapeseed oil were extracted from seeds, which were from different catalog and growing areas. 12 rice bran oil and 20 palm oil came from different oil plants. PCA and spectra pretreated using the cluster technique of NIRCal5.2 was able to classify the samples as pure oil and fat based on their transmission spectra. The spectra information as variance was used in the processing of spectra pretreated. PCA using 7 principal components was able to classify the samples which 2/3 of samples were selected for calibration, the others for validation. The results show that the classification accuracy and validation accuracy were yielded about 100%. It means that all the samples were complete clustered in six distinct groups in the principal component analytical plot. Therefore, the NIR/PCA method could be used for discriminate different vegetable oils and fats.

关 键 词：近红外光谱 主成分分析 植物油脂 

分 类 号：TS225.1[轻工技术与工程—粮食、油脂及植物蛋白工程]