作者：赵科文¹, 陈实¹, 尹中尉¹, 邱昌桂², 刘静², 朱叶梅², 周文忠², 杨盼盼²

作者单位：1. 贵州中烟工业有限责任公司毕节卷烟厂, 贵州 毕节 551700;
2. 云南同创检测技术股份有限公司, 云南 昆明 650106

关键词：近红外光谱;卷烟;润滑油污染;模式识别

摘要：

为实现对油渍污染卷烟的快速识别及溯源。采集卷烟生产过程中使用的6种润滑油污染在4种卷烟纸上样品的近红外光谱，结合未污染样品的近红外光谱，优选出原始样品原始光谱及二阶导数光谱中共有的差异性光谱波段范围，采用模式识别算法（principal component analysis- mahalanobis distance, PCA-MD）分别建立污染及未污染样品模式识别模型和不同卷烟上6种润滑油的模式识别模型。结果表明：1）基于污染及未污染样品原始近红外光谱及二阶求导光谱差异性分析，优化出的模式识别模型建模波数范围是：6000~5300 cm^-1和4500~4000 cm^-1；2）建立的污染及未污染模式识别模型前3个主成分累计得分贡献率97.826%，模型分类效果明显，建模集及外部验证集样品的识别准确率均为100%；3）分别建立的6种油渍在4种卷烟纸上的溯源类模型，前3个主成分得分累计贡献率均大于96%，模型分类效果明显，建模集及外部验证集样品的识别准确率均为100%。所建立的基于近红外光谱分析方法结合模式算法（PCA-MD）可实现卷烟生产过程出现的“黄斑烟”中油渍烟的快速识别及污染油渍溯源。

Identification and tracing of cigarette paperoil stains based on near infrared spectroscopy and pattern recognition algorithm
ZHAO Kewen¹, CHEN Shi¹, YIN Zhongwei¹, QIU Changgui², LIU Jing², ZHU Yemei², ZHOU Wenzhong², YANG Panpan²
1. Bijie Cigarette Factory, China Tobacco Guizhou Industrial Co., Ltd., Bijie 551700, China;
2. Yunnan Comtestor Detection Technology Co., Ltd., Kunming 650106, China
Abstract: In order to realize the quick identification and tracing of the smoke contaminated with oil. Collecting cigarette production process used in 6 kinds of oil pollution on the four kinds of cigarette paper samples of near infrared spectrum, combined with uncontaminated samples of near infrared spectrum, with the original sample the original spectrum and second order derivative spectra of difference spectrum band range, using pattern recognition algorithm (Principal component analysis-mahalanobis short, PCA-MD) pollution and no pollution sample pattern recognition model is set up respectively and different cigarette on pattern recognition model of 6 kinds of lubricating oil.The results show that:1) Based on the analysis of the difference between the original near infrared spectra and the second derivative spectra of the contaminated and non-contaminated samples, the wave-number range of the optimized pattern recognition model is

6000-5300 cm^-1 and 4500-4000 cm^-1. 2) The first three principal components of the established pollution and unpollution pattern recognition model contribute 97.826%, and the model classification effect is obvious. The identification accuracy of the modeling set and the external verification set is 100%. 3) The traceability models of 6 kinds of oil stains on 4 kinds of cigarette paper were respectively established. The cumulative contribution rates of the first 3 principal components were all greater than 96%, and the model classification effect was obvious. The identification accuracy of the samples in the modeling set and the external verification set was 100%. The established near infrared spectroscopy analysis method combined with the mode algorithm (PCA-MD) can realize the rapid identification and the source tracing of oil stains in the "yellow spot smoke" in the cigarette production process.

Keywords: near infrared spectroscopy (NIRS);cigarettes;lubricating oil pollution;pattern recognition
2020, 46(6):51-55  收稿日期: 2019-10-20;收到修改稿日期: 2019-12-06
基金项目: 贵州中烟工业有限责任公司科技项目（GZZY/KJ/BJ/2016 D Y 003-1）
作者简介: 赵科文(1971-),男,贵州毕节市人,工程师,研究方向为卷烟工艺和卷烟设备
参考文献
[1] 卷烟第3部分包装、卷制技术要求及贮运：GB5606.3—2005[S]. 北京：中国标准出版社，2005.
[2] 杨蓝, 周沅桢, 刘泽, 等. 烟支黄斑污染物的追溯与鉴别[J]. 贵州农业科学, 2019, 47(5): 132-137
[3] 司晓喜, 向能军, 马宁, 等. 黄斑烟卷烟纸中糖和保润剂的测定及微观形貌分析[J]. 烟草科技, 2018, 51(5): 46-54
[4] 秦诗棋, 周沅桢, 刘泽, 等. 利用FAIMS法鉴别黄斑烟污染物的来源[J]. 中国烟草学报, 2018, 24(4): 7-15
[5] 王菊香, 李伟, 瞿军. 基于红外光谱的在用润滑油衰变信息的快速确定和关联性分析[J]. 分析仪器, 2017(2): 58-63
[6] 田广军, 吕盼杰, 谷栎娜. 反射光谱法快速检测润滑油污染浓度的回归模型[J]. 光谱学与光谱分析, 2017, 37(7): 2152-2157
[7] YANG C, YANG Z U, ZHANG G, et al. Characterization and differentiation of chemical fingerprints of virgin and used lubricating oils for identification of contamination or adulteration sources[J]. Fuel, 2016: 163
[8] 王菊香, 邢志娜, 韩晓, 等. 基于FTIR的在用润滑油燃油污染的定量检测[J]. 润滑与密封, 2014, 39(8): 77-80,98
[9] 周岩, 陈晓梅. 近红外分析仪在快速测定正构烃碳分布的应用[J]. 中国仪器仪表, 2010(S1): 189-190
[10] CORTÉS V, BLASCO J, ALEIXOS N, et al. Monitoring strategies for quality control of agricultural products using visible and near-infrared spectroscopy: A review[J]. Trends in Food Science & Technology, 2019
[11] ZHANG C, SU J H. Application of near infrared spectroscopy to the analysis and fast quality assessment of traditional Chinese medicinal products[J]. Acta pharmaceutica Sinica. B, 2014, 4(3): 182-192
[12] 王毅, 马翔, 温亚东, 等. 近红外光谱与多元统计方法用于生产过程实时分析[J]. 光谱学与光谱分析, 2013, 33(5): 1226-1229
[13] 李伟, 冯洪涛, 周桂圆, 等. Hotelling T²结合多组分NIR校正模型在卷烟生产过程质量监测中的应用[J]. 烟草科技, 2014(7): 5-9
[14] 段焰青, 孔祥勇, 李青青, 等. 近红外光谱法预测烟草中的纤维素含量[J]. 烟草科技, 2006(8): 18-22
[15] 黄扬明, 田旷达, 张吉雄, 等. 中红外技术结合PCA-MD用于烟用香精香料定性研究[J]. 光谱学与光谱分析, 2018, 38(S1): 136-137
[16] 王家俊, 邱启杨, 刘巍. FTIR-ATR指纹图谱的主成分分析-马氏距离法应用于烟用香精质量控制[J]. 光谱学与光谱分析, 2007(5): 65-68