近红外光谱法快速定量检测烤烟中的13种非挥发性有机酸
Rapid quantitative determination of 13 nonvolatile organic acids in flue-cured tobacco by near-infrared spectroscopy
-
摘要:
基于傅里叶变换近红外漫反射光谱技术，通过采集220个烤烟样品的近红外光谱和多元散射校正、一阶微分及SG平滑光谱处理，采用偏最小二乘（PLS）法建立烤烟中13种非挥发性有机酸的定量预测模型，用部分样本作为独立验证集进行预测，并比较了不同香型典型烟叶样品中13种非挥发性有机酸质量分数的差异。结果表明：①校正集与验证集的相关系数（r）约为0.9；②各化学指标的预测均方差（SEP）与校正均方差（SEC）的比值均为 0.8~1.2；③验证集样本标准偏差（SD）与SEP的比值均大于2.0；④中间香型烟叶中13种非挥发性有机酸的质量分数最高，清香型次之，浓香型最低。建立的近红外定量预测模型准确、可靠，可用于烤烟中13种非挥发性有机酸的快速定量检测。 Abstract:Fourier transform near-infrared （FT-NIR） spectroscopy was used to collect spectra of 220 flue-cured tobacco leaf samples. After multiplicative signal correction （MSC）, first derivation and Savitzky-Golay （SG） smoothing processing, quantitative prediction models of 13 nonvolatile organic acids were established by using partial least square （PLS） method and using some samples as independent validation set. The mass fractions of 13 nonvolatile organic acids in the typical tobacco leaf samples of different flavor styles were compared. The results showed that: 1） The correlation coefficient （r） of calibration set and validation set in each model was about 0.9. 2) The ratios of standard error of prediction （SEP） to standard error of calibration (SEC） of each chemical index were 0.8-1.2. 3） The ratios of standard deviation （SD） to SEP were greater than 2.0. 4） The mass fractions of 13 nonvolatile organic acids in tobacco leaves of medium flavor style were the highest, followed by fen-flavor style, and those of full flavor style were the lowest. The established quantitative near-infrared prediction models were accurate and reliable and could be used to rapidly quantify the mass fractions of 13 nonvolatile organic acids in flue-cured tobacco.
下载: