运用模拟UV数据的KICA-SVR模型同时测定饮料中多种色素的含量

Simultaneous determination of multiple food colorants content in beverages based on KICA-SVR model of simulated UV data

在模拟水性饮料和实际饮料样品中加入不同种类和含量的色素,运用核独立成分分析(KICA)法提取样品的紫外-可见(UV)光谱数据中独立组分(IC)信息,以IC的系数矩阵进行支持向量回归(SVR)处理并建立UV-KICA-SVR模型,用于直接预测各种色素的含量.用此方法测定水性饮料中柠檬黄、日落黄、诱惑红、苋菜红、胭脂红和亮蓝6种色素的含量,相对标准偏差分别为1.5%,2.2%,2.0%,2.5%,2.6%和1.2%,检测限分别为0.5 mg·L-1,0.5 mg·L-1,0.5 mg·L-1,1.0 mg·L-1,1.0 mg·L-1和0.5 mg·L-1.

By extraction of the independent components（ IC） from the ultraviolet-visible（ UV） spectrum of simulated mixtures of food colorants disolved in water and beverage using kernel independent component analysis（ KICA）,the UV-KUA-SVR model were established by support vector regressing（ SVR） using the IC’s coefficients. This model can be used to simultineously determine multiple colorants,i. e. lemon yellow,sunset yellow,red temptation,amaranth,carmine,and brilliant blue,with the relative standard deviations（ RSDs） are 1. 5%,2. 2%,2. 0%,2. 5%,2. 6%,and 1. 2%,respectively,the limits of detection（ LOD） are 0.5 mg·L-1,0.5 mg·L-1,0.5 mg·L-1,1.0 mg·L-1,1.0 mg·L-1 and 0. 5 mg·L-1,respectively.