A method was developed for the rapid quantitative analysis of routine chemical components in reconstituted tobacco raw materials ( flaky and powdery ) based on the combined application of near -infrared diffuse reflection spectroscopy and chemo-metrics methods .The near-infrared spectrums of 354 representative raw material samples were collected by Fourier transform near-infrared analyzer .The models describing the relationships between near -infrared spectrum and content of nicotine , total sugar , reducing sugar and chlorine were established by using partial least squares method , and the predicted effect , stability and repeat-ability of these models were evaluated .The results revealed that all prediction models had semblable and low root mean square error of calibration and root mean square error of prediction , and their correlation coefficients were all above 0.98.Moreover, the average relative deviations of the predicted values of nicotine , total sugar, and reducing sugar were all less than 4%; while the chlorine model showed an average absolute deviation of 0.04% and an average relative deviation of 6.09% in predicting 30 different sam-ples.The internal and external verification of these prediction models indicated that this quantitative analysis by near -infrared spectroscopy could obtain accurate and reliable result , which was not significantly different from that obtained by using flow analysis method .%提出一种基于近红外漫反射快速定量分析再造烟叶原料（废烟片及烟末）的常规化学组分方法，应用傅里叶变换近红外分析仪扫描原料光谱并结合偏最小二乘法，对具有代表性的354个再造烟叶原料建立了近红外光谱同原料烟碱、总糖、还原糖及氯的含量关系模型，并对模型的预测效果、稳定性及重复性进行了评价。结果表明：烟碱、总糖、还原糖及氯模型的校正均方差与预测均方差接近且均较小，模型的相关系数均在0．98以上。烟碱、总糖、还原糖预测平均相对偏差均低于4％，Cl模型预测30个样品的平均绝对偏差为0．04％，预测平均相对偏差6．09％。模型通过内部及外部验证，表明近红外定量分析结果准确、可靠与流动分析法无显著性差异。
展开▼
