偏最小二乘法同时测定食醋的有效成分和防腐剂的含量

Simultaneous determination of main composition and additive in vinegar by NIR and partial least squares

偏最小二乘方法建立近红外光谱与组分浓度的多元校正模型,用于同时快速测定食醋的有效成分(总酸)和防腐剂(苯甲酸)含量。采用透射模式,样品不经任何处理测定近红外光谱。用17个食醋样品建立偏最小二乘法(PLS)模型,用3个样品作为预测样品以评估PLS模型。结果:食醋中的总酸和苯甲酸在2 125～2 325 nm区域之间,与光谱有良好的线性关系,总酸的PLS模型中,隐变量为3时,预测均方根误差0.038 7 g·L^(-1),总酸含量与光谱的线性相关系数达到0.999 7,相对预测误差5.89%;苯甲酸的PLS模型中,隐变量为6时,预测均方根误差降至0.013 8 g·L^(-1),苯甲酸含量与光谱的线性相关系数达到0.999 8,相对预测误差降至4.29%。

Method of quick and simultaneous determination of main compositions （total acid） and additive （benzoic acid） in vinegar by near infrared （NIR） spectroscopy coupled with partial least squares was discussed. Transmission NIR spectra of vinegar samples were measured without any pretreatments for the samples. Spectra of 17 samples in the range of （2 125 -2 325） nm were used to build partial least squares （PLS） model, and those of 3 samples were predicted to evaluate the model. The results showed that in the range of （2 125 - 2 325） nm, there were good linear relationships between absorbance and the content of total acid and benzoic acid. NIR spectra of vinegar were regressed against the concentrations of total acids and benzoic acid solutions. The minimum value of root-meansquare error of prediction （RMSEP） of total acids was 0. 038 7 g · L-1 when the number of latent variables was 3, and the correlation coefficient was 0. 999 7, while the minimum value RMSEP of benzoic acid was 0. 013 8g ·L-1 when the number of latent variables was 6. and the correlation coefficient was 0. 999 8.