为了细致阐述烟叶原料和卷烟产品中烟碱量与烟气烟碱量之间的相互关系.以云南省17个县市收集的148个单料烟质量检测数据和全国14家卷烟厂中2个卷烟品牌各自的107、209个检测数据为材料,对烟碱量和烟气烟碱量2个指标进行相关、回归分析并通过回归方程进行预测.结果表明,A品牌中烟气烟碱量与烟碱量之间的相关系数为0.858,以烟碱量作为反应变量得到回归方程为:y=0.741+1.186x;以烟气烟碱作为反应变量得到回归方程为:γ=-0.134+0.621x.2项指标预测偏差绝对值分别为:烟碱量0.042%～3.04%;烟气烟碱量0.22%～6.51%.B品牌中烟气烟碱与烟碱量之间的相关系数为0.754,以烟碱量作为反应变量得到回归方程为:γ=1.106+0.946x;以烟气烟碱量作为反应变量得到回归方程为:γ=-0.204+0.601x.2项指标预测偏差绝对值分别为:烟碱量0.29%～6.22%,烟气烟碱量0.91～9.70%;单料烟中烟气烟碱与烟碱量之间的相关系数为0.940,以烟碱量作为反应变量得到回归方程为:γ=0.283+1.032x;以烟气烟碱作为反应变量得到回归方程为:γ=0.028+0.856x.2项指标预测偏差绝对值分别为:烟碱量0.016%～20.37%;烟气烟碱量0.084%～18.5725%.说明烟叶原料和卷烟成品都可以从烟气烟碱量与烟碱量两者间建立较好的回归模型,且能满足预测精度要求.%For more carefully explain the relationship between nicotine content and nicotine content in smoke of raw tobacco and cigarette products. We collected 148 uablended cigarette quality detection data in 17 countries of Yunnan Province and 107, 108 between two cigarette brands quality detection data in 14 cigarette factories of the whole country as materials. Then the correlation analysis and regression analysis is made between nicotine content and nicotine content in smoke and the value of two indexes are predicted by the regression equation. The results showed that the correlation coefficient is 0.858 between nicotine content and nicotine content in smoke for A brand, we put nicotine contentas a response variable, and get the regress equation:(y) =0. 741 + 1. 186x. We put nicotine content in smoke as a response variable, and get the regress equation :(y) = -0. 134 + O. 621 x, the absolute value of prediction error in two indexes is respectively: nicotine content 0.042％ -3.04％; nicotine content in smoke O. 22％ -6.51％. The correlation coefficient is O. 754 between nicotine content and nicotine content in smoke for B brand, we put nicotine content as a response variable, and get the regress equation:(y) = 1. 106 +0.946x;we put nicotine content in smoke as a response variable, and get the regress equation:(y) = -0. 204 +0.601x, the absolute value of prediction error in two indexes is respectively: nicotine content 0.29％ - 6.22％; nicotine content in smoke 0.91％ - 9.70％. The correlation coefficient is 0.940 between nicotine content and nicotine content in smoke for unblended cigarette. We put nicotine content as a response variable, and get the regress equation :(y) = O. 283 + 1. 032x;we put nicotine content in smoke as a response variable, and get the regress equation: (y) = O. 028 + O. 856x, the absolute value of prediction error in two indexes is respectively: nicotine content 0.016％ -20.37％; nicotine content in smoke 0. 084％ -18.57％. That we can establish the better regression model from both with nicotine content and nicotine content in smoke of raw tobacco and cigarette products. And it can meet the requirement on prediction accuracy.
展开▼
