偏最小二乘多元校正法用于三组分食用色素的同时测定

A Multivariate Calibration Method of Partial Least Square Technique for Simultaneous Determination of Pigments in Ternary Mixtures

将偏最小二乘法(PLS)与可见分光光度法相结合,对胭脂红、苋菜红、日落黄三种食用色素进行不经分离的同时定量分析。使用交叉验证法选择主成分数建立了校正模型,预测结果令人满意。采用随机抽样方法研究了校正模型中样品集和样本容量对预测能力的影响,结果表明,在文章所研究的体系中,偏最小二乘法的预测能力不因样品集和样本容量的不同而有明显差异,从而说明了文章所用方法具有较大的可靠性和适用性。

A partial least square （PLS） method combined with spectrophotometry was applied to simultaneous determination of ternary mixtures of carmine, amaranth and sunset yellow. A cross -validation method leaving one out calibration sample at a time, was used to select the optimal number of component for the calibration models and the test results were satisfactory. A random sampling procedure was accomplished to investigate the validation of the calibration models using different sample sets and of its size. It has been shown that, in this ternary mixtures, the PLS calibration method is insensitive to the samples selected. This work can provide an appropriate and reliable way to determine the pigments in mixtures.