论文部分内容阅读
为快速有效进行烟叶原料烟气释放量分析,通过近红外光谱预处理、光谱特征谱段选择和模型参数优选等方面的研究,构建了对加工前和加工后烟叶原料近红外光谱均适用的稳健的偏最小二乘烟气预测模型,实现了对烟叶原料单位焦油、一氧化碳和烟气烟碱的有效分析。结果表明:焦油模型的交叉验证均方根误差和相关系数分别为1.01和0.87,一氧化碳模型的交叉验证均方根误差和相关系数分别为1.16和0.81,烟气烟碱模型的交叉验证均方根误差和相关系数分别为0.25和0.93。单位焦油、一氧化碳和烟气烟碱的有效预测范围分别为:(18.6~27.01)mg/g,(16.89~24.41)mg/g,(1.5~3.41)mg/g。研究结果能够为烟草企业更好评价烟叶原料品质、稳定卷烟配方烟气释放量、提升产品质量一致性提供技术支撑。
In order to quickly and effectively analyze the release of smoke from tobacco leaves, the near-infrared spectroscopy pretreatment, the selection of spectral characteristic spectrum and the optimization of model parameters were established. The results showed that the NIR spectra of tobacco leaves before and after processing were stable Partial least squares flue gas prediction model to achieve an effective analysis of tobacco tar units, carbon monoxide and fumes nicotine. The results show that the root mean square error of cross validation and the correlation coefficient of the tar model are 1.01 and 0.87 respectively, the cross validation root mean square error and the correlation coefficient of the carbon monoxide model are 1.16 and 0.81 respectively, and the cross validation root mean square The error and correlation coefficients are 0.25 and 0.93, respectively. The effective prediction range of tar, carbon monoxide and nicotine per unit area were (18.6-27.01) mg / g, (16.89-24.41) mg / g and (1.5-3.41) mg / g, respectively. The results can provide technical support for tobacco enterprises to better evaluate the quality of tobacco raw materials, stabilize the release of cigarette smoke and improve the consistency of product quality.