以在2003年—2012年向国内17个主要烤烟产地, 五大烟叶生态产区, 收集的4 733个烟叶样品的近红外光谱为试验对象, 其中浓香型烟叶光谱1 580条, 清香型2 004条, 中间香型1 149条; 应用基于主成分及Fisher准则(PPF)的方法建立烟叶生态产区和风格特征的投影分析模型, 以生态产区模型的分析结果阐释了烟叶香型风格划分的合理性, 并依据风格特征模型得到的预测样品投影值与模型中各类投影均值之间的欧氏距离, 对预测样品给出最近和次近类别及描述风格特征程度的量化分值, 结合模型中各类投影值的离散度以及设定阈值, 将预测结果细化为典型浓、 浓偏中、 中偏浓、 典型中、 中偏清、 清偏中、 典型清、 清透浓、 浓透清等9类或超模型范围样品; 应用2102在实际复烤生产加工及各烟叶原料基地中收集的不同部位、 不同产地的35个烟叶样品验证了分析结果的合理性。 该种分析方式, 不仅可以实现预测样品的判别分析, 并且可得到关于预测样品更细化的风格特征信息, 可对烟草工业企业实现原料的跨区组合及叶组配方等提供指导。

In the present paper, a total of 4 733 flue-cured tobacco samples collected from 2003 to 2012 in 17 provincial origins and 5 ecological areas were tested by near infrared spectroscopy, including the NONG(Luzhou) flavor 1 580 cartons, QING(Fen) flavor 2004 cartons and Intermediate flavor 1 149 cartons. Using projection model based on principal component and Fisher criterion (PPF), Projection analysis models of tobacco ecological regions and style characteristics were established. Reasonableness of style flavor division is illustrated by the model results of tobacco ecological areas. With the Euclidean distance between the predicted sample projection values and the mean projection values of each class in style characteristics model, a description is given for the prediction samples to quantify the extent of the style features, and their first and second close categories. Using the dispersion of projected values in model and the given threshold value, prediction results can be refined into typical NONG, NONG to Intermediate,Intermediate to NONG, typical Intermediate, Intermediate to QING, QING to Intermediate, typical QING, QINGto NONG, NONG to QING, or super-model range. The model was validated by 35 tobacco samples obtained from the re-dryingprocess in 2012 with different origins and parts. This kind of analysis methods not only can achieve discriminant analysis, but also can get richer feature attribute information and provide guidance to raw tobacco processing and formulations.