摘要
应用傅立叶变换近红外光谱(FT-NIR)法和化学分析法测定了400个陈化烤烟烟叶样品的NIR光谱数据和淀粉、多酚、色素含量,采用偏最小二乘法(PLS)法将光谱数据和化学测定值进行拟合,建立了淀粉、多酚、色素的NIR预测模型,进行了模型的内部交叉验证,并用未参与建模的100个样品进行了外部验证。结果表明,淀粉、多酚、色素的NIR预测模型的决定系数(R2)分别为97.64%、86.97%、87.08%,交互校验均方差(RMSECV)分别为0.338、0.283、0.011;模型的预测结果与化学分析法无显著性差异,其RSD均<5%。FT-NIR光谱法可用于批量陈化烟叶样品淀粉、多酚和色素含量的快速检测。
The NIR spectrum data of 400 aged flue-cured tobacco samples were obtained by FT-NIR spectrometry, and the contents of starch, polyphenols and pigments in the samples were determined with chemical analysis. The NIR prediction models were established by fitting the spectrum data to the chemical analysis data of 300 samples with partial least square （PLS） , and the models were cross verified within the data set and the other 100 samples not involved in the establishment of the models. The results showed that the coefficients of determination （R2） of the models for starch, polyphenols and pigments were 97. 64%, 86.97% , 87.08% , and their root mean squares error of cross validation （RMSECV） were 0. 338, 0. 283, 0. 011, respectively. There was no significant difference between prediction values and chemical analysis values, and their relative standard deviations （RSD） were all below 5%. The FT-NIR spectrometry is fit for the rapid determination of pigment, starch, polyphenol in batch of aged tobacco samples.
出处
《烟草科技》
EI
CAS
北大核心
2009年第8期38-41,55,共5页
Tobacco Science & Technology
基金
国家烟草专卖局基金资助项目“以成熟度为中心配套生产技术试验示范与推广”(110200401012)
