含水率对丙二醇近红外预测值及加料均匀性系数的影响

为探究烟片含水率对丙二醇近红外模型预测效果及加料均匀性系数的影响，制备不同丙二醇含量且每一含量均包含不同含水率的烟片样品，并预测其丙二醇含量，分析不同含水率以及不同丙二醇含量对预测结果的影响，且建立基于含水率的丙二醇预测结果的校正模型，并利用该模型分析样品校正前后的加料均匀性系数变化规律。结果表明：（1）待检样品与建模样品的含水率差异越大，预测值的绝对偏差也越大。当待检样品含水率大于建模样品含水率时，预测值偏高；反之，预测值偏低。（2）含水率与预测值的绝对偏差间呈极显著线性关系(R2=0.938)，因此，可以根据含水率校正预测值。（3）对某牌号、批次卷烟的加料均匀性评价结果表明，待检样品含水率与建模样品相比平均偏高约1.70%，校正前、后的加料均匀性系数相差2.0%。综上所述，控制含水率对丙二醇含量近红外预测值的影响，有利于提高加料均匀性评价结果的准确度。

Effect of Moisture Content on NIR Predicted Value of Propylene Glycol in Tobacco Leaf and Casing Uniformity Coefficient

The paper had studied the effect of moisture content on the NIR predicted value of propylene glycol in tobacco leaf, and its effect on casing uniformity coefficient. Results showed that (1) the greater the difference in moisture content between testing sample and model sample, the greater the absolute deviation of the predicted value, when the testing sample’s water content is larger than it of the model samples, the predictive value is high; conversely, the predictive value is low. (2) There is a significant linear relationship between the water content and the absolute deviation of predicted value (R₂=0.938), therefore, the forecast value can be corrected according to the moisture content. (3) It showed that the evaluation result of the casing uniformity from a same batch cigarettes, because the water content of testing sample was about 1.70% than it of model sample, there is difference of 2.0% between the casing uniformity coefficient calculated according to the predicted value and adjusted value of propylene glycol. In a word, it is helpful to improve the evaluation results of casing uniformity by controlling the influence of moisture content on the NIR predicted value of propylene glycol.