光谱学与光谱分析, 2017, 37 (10): 3088, 网络出版: 2017-12-25  

初烤完整烟叶总植物碱的近红外光谱测量方法研究

The Study of Near Infrared Spectroscopy Measurement Method for Total Alkaloids of Flue-Cured Complete Tobacco Leaves
作者单位
1 中国农业大学现代精细农业系统集成研究教育部重点实验室, 北京 100083
2 云南省烟草农业科学研究院, 云南 昆明 650021
摘要
为了探讨近红外光谱分析技术检测完整烟叶化学成分的可行性, 利用近红外光谱分析技术, 对初烤完整烟叶的光谱采集方式及总植物碱定量分析建模方法进行了研究。 以云南省昆明市不同乡镇、 不同品种的初烤烟叶为研究对象, 分别采用烟叶的叶尖、 叶中、 叶基光谱及其平均光谱建立初烤完整烟叶总植物碱近红外偏最小二乘法(PLS)定量分析模型以选择出代表完整烟叶信息的建模光谱; 分别用KS和SPXY方法对样品的校正集进行选择, 采用向后区间偏最小二乘法(BiPLS)、 无信息变量消除法(UVE)、 竞争适应性重加权采样法(CARS)等选择特征变量, 对模型进一步优化。 研究结果表明, 采用叶尖、 叶中、 叶基3个部位的平均光谱建立的模型相比单独每个部位光谱所建立模型的预测精度提高了8.5%~9.5%, 与全光谱建模相比, 用KS-BiPLS建立模型能明显改善模型的预测能力, 模型的预测精度约提高了10%, 模型的校正集决定系数和均方根误差分别为0.917 4和0.226 1, 检验集决定系数和预测均方根误差分别为0.902 0和0.200 7。 本研究方法适用于完整的初烤烟叶, 无需对样品进行预处理, 对于大量的初烤烟叶, 能够快速、 无损测定烟叶总植物碱含量, 可以节省大量的时间。 同时, 该研究为初烤烟叶分级、 提高原料的品质提供技术支持, 也将为卷烟生产的过程控制提供科学依据。
Abstract
In order to discuss the feasibility of chemical composition of complete tobacco by near infrared spectroscopy analysis technique, we used near infrared spectroscopy analysis technique to study spectrum acquisition of the complete flue-cured tobacco leaves and total alkaloid quantitative analysis modeling methods. The samples of the flue-cured tobacco in this research were collected from different towns and different varieties in Kunming city of Yunnan province. Respectively with tobacco leaf tip, the middle part of leaf, leaf base spectrum and the three parts of the average spectrum to establish quantitative analysis model for total alkaloids of flue-cured complete tobacco leaves with Partial Least Squares (PLS); respectively use KS and SPXY methods to divide flue-cured tobacco samples into calibration set and validation set and use back interval partial least squares (BiPLS), no information on variable elimination method (UVE) and competition adaptive re-weighted sampling method (CARS) to select characteristic variables to optimize the model. Research results showed that, the prediction accuracy of the model established with tobacco leaf tip, the middle part of leaf, leaf base spectrum the three parts of the average spectrum improves 8.5%~9.5% compared to the model established with the spectra of the individual parts. Compared with full-spectrum modeling, using KS-BiPLS to establish the model can significantly improve the model’s predictive ability and the prediction accuracy of the model is improved by about 10%. The correction coefficient and root mean square error of the model are 0.917 4 and 0.226 1 respectively. The determination coefficient and root mean square error of the validation set are 0.902 0 and 0.200 7 respectively. This method is applied to flue-cured complete tobacco leaves and it can be used to estimate total alkaloids of flue-cured complete tobacco leaves content rapidly and non-destructive. It will save a lot of time for a large number of flue-cured tobaccos. And it will also help to provide technical support for the classification of the flue-cured tobacco leaves and improve the quality of raw materials and provide scientific basis for the process control of cigarette production.

何春荣, 杨宇虹, 李军会, 劳彩莲. 初烤完整烟叶总植物碱的近红外光谱测量方法研究[J]. 光谱学与光谱分析, 2017, 37(10): 3088. HE Chun-rong, YANG Yu-hong, LI Jun-hui, LAO Cai-lian. The Study of Near Infrared Spectroscopy Measurement Method for Total Alkaloids of Flue-Cured Complete Tobacco Leaves[J]. Spectroscopy and Spectral Analysis, 2017, 37(10): 3088.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!