天然产物研究与开发 ›› 2016, Vol. 28 ›› Issue (增刊2): 323-326.doi: 10.16333/j.10016880.2016.S.032

• 开发研究 • 上一篇    下一篇

近红外光谱技术检测烟草薄片主流烟气中的焦油和烟碱

张鼎方1,于静2*   

  1. 1福建中烟工业有限责任公司,厦门 361021;2厦门烟草工业有限责任公司,厦门 361022
  • 出版日期:2016-12-31 发布日期:2017-03-03

Detection of Tar and Nicotine in Mainstream Smoke of Reconstituted Tobacco Based on Nearinfrared Spectroscopy

ZHANG Dingfang1,YU Jing2*   

  1. 1China Tobacco Fujian Industrial Co.,LTD,Xiamen 361021,China;2Xiamen Tobacco Industrial Co.,LTD,Xiamen 361022,China
  • Online:2016-12-31 Published:2017-03-03

摘要: 为了建立快速检测烟草薄片主流烟气中焦油和烟碱的方法,实验利用红外光谱采集了204个烟草薄片样品光谱,将光谱与相应薄片样品焦油与烟碱释放量数据对应,建立了烟草薄片主流烟气中焦油和烟碱释放量的预测模型,并用22个校正集样品对模型采用留一(leaveoneout)交叉验证法进行了验证和优化。结果表明:建立的焦油和烟碱释放量预测模型的交叉验证相关系数(R)分别为0.94和0.95、均方差(RMSECV)分别为055和0.023,均达到要求,烟气中烟碱释放量预测值与标准值的平均相对偏差为4.1%,烟气中焦油释放量的平均相对偏差为4.59%,确定了焦油和烟碱的最佳预测模型,该近红外数学模型预测薄片烟气中烟碱和焦油的释放量都具有很好的准确性。

关键词: 烟草薄片, 近红外光谱, 主流烟气

Abstract: In order to establish rapid detection method of tar and nicotine content in mainstream smoke of reconstituted tobacco,nearinfrared spectroscopies of 204 samples were collected and the near infrared prediction model was established.By using the leaveoneout cross validation method for modeling.The results showed that sample correlation coefficient(R)of tar and nicotine was 0.94 and 0.95and the mean square error(RMSECV)was 0.55 and 0.023 respectively,nearinfrared spectroscopy was compared with the conventional standard testing methods,the best prediction model was thus determined.The results showed that nearinfrared spectrum technology can more accurately determine the release quantity of tar and nicotine in mainstream smoke of tobacco slice.

Key words: reconstituted tobacco, nearinfrared spectroscopy, mainstream smoke

中图分类号: 

TS452