近红外光谱技术结合变量选择方法定性检测食用植物油中的腐霉利

发布时间：2018-05-15 22:17

本文选题：近红外 + 腐霉利　；参考：《光谱学与光谱分析》2016年12期

【摘要】：利用近红外光谱技术对食用植物油中的腐霉利进行定性检测研究。以国家标准规定的腐霉利最大残留限量为界线,将不同腐霉利含量的食用植物油样本分为合格组和不合格组。采用QualitySpec台式近红外光谱仪采集两类样本的光谱,利用无信息变量消除(UVE)和子窗口重排分析(SPA)方法进行波长变量筛选,并应用线性判别分析(LDA)、偏最小二乘-线性判别分析(PLS-LDA)及判别偏最小二乘(DPLS)方法建立两类样本的分类模型。结果表明,近红外光谱技术可以对两类样本进行分类。UVE方法可以有效筛选有用波长变量,提高分类模型的性能。UVE-DPLS所建立的分类模型性能最优,其预测集样本的分类正确率、灵敏度及特异性分别为98.7%,95.0%和100.0%。
[Abstract]:The qualitative detection of Pythium in edible vegetable oil by near infrared spectroscopy (NIR) was studied. The samples of edible vegetable oil with different Pythium content were divided into two groups: qualified group and unqualified group. The spectrum of two kinds of samples was collected by QualitySpec desktop near infrared spectrometer, and the wavelength variables were screened by using the methods of eliminating UVEwithout information variable and analyzing the subwindow rearrangement. The classification models of two kinds of samples were established by using the methods of linear discriminant analysis (LDA), partial least squares linear discriminant analysis (PLS-LDAs) and discriminant partial least squares (DPLSs). The results show that Near-Infrared Spectroscopy can classify two kinds of samples. UVE can effectively screen the useful wavelength variables, improve the performance of the classification model. UVE-DPLS has the best performance, and the classification accuracy of the samples can be predicted. The sensitivity and specificity were 98.7% and 100.0%, respectively.
【作者单位】：江西农业大学生物光电技术及应用重点实验室;
【基金】：国家自然科学基金项目(31271612) 江西省自然科学基金项目(20151BAB204025) 江苏省农产品物理加工重点实验室开放基金项目(JAPP2013-6)资助
【分类号】：TS227;O657.33

【参考文献】