近红外光谱分析技术结合化学计量学方法快速判别品牌卷烟烟丝  

Near-infrared Spectral Analysis Technology Combined Chemometrics Methods for Fast Discriminating Different Brands of Cut Tobacco

作　　者：袁大林[1,2] 汤建国 杨晨龙[3] 乔丹娜 孟昭宇 牟定荣 吴海龙[1] 

机构地区：[1]湖南大学化学化工学院,长沙市麓山南路410082 [2]红塔烟草(集团)有限责任公司技术中心 [3]昆明理工大学化学工程学院

出　　处：《光谱实验室》2013年第3期1447-1450,共4页Chinese Journal of Spectroscopy Laboratory

摘　　要：简便、快速、准确地鉴别不同品牌的卷烟烟丝,对品牌卷烟的开发与维护具有重大的意义。本研究利用近红外光谱分析技术快速、直接测定品牌卷烟烟丝,结合光谱预处理、逐步变量筛选法及判别分析法等化学计量学方法,建立3个不同品牌烟丝的判别模型。基于筛选得到的24个变量,所建立模型的训练与交互检验平均正确率分别为98.4%和95.2%,成功实现了对不同品牌成品烟丝的判别预测。研究结果表明,近红外光谱分析技术结合化学计量学方法用于不同品牌烟丝的快速识别,方法简便,结果可靠。Simply, quickly and accurately discriminating different brands of cut tobacco is very important for development and maintenance of tobacco brand. In present study ~ the NIR spectra of cut tobacco samples of three different brands were obtained by NIR spectral analysis technology,and then the discrimination model based on NIR spectra for these three brands was established by using several chemometrics methods （spectral preprocessing,stepwise variables selection and discrimination analysis methods）. As the results, the final discriminating model based on 24 selected variables gave average discriminating correct rate of training and cross validation as 98.4% and 95.2%respectively. This study showed that NIR spectral analysis technology combined chemometrics methods could be a fast and reliable way for discriminating different brands of cut tobacco.

关 键 词：近红外光谱分析技术 卷烟烟丝 化学计量学 

分 类 号：O657.33[理学—分析化学]