聚类分析与模糊数学在烟叶品质综合评价中的应用

生产上烟叶的质量评价主要通过外观品质(烟叶分级)来进行。由于主观性的影响,这种方法常常导致同样的烟叶被不同的人分成不同的等级。随着近红外线分析方法在烟叶质量评价中的普遍应用,以化学成分为依据的烟叶质量评价方法将成为主流趋势。为此,本研究以河南省南阳烟区的烟叶样品为例,通过层次聚类的方法对烟叶主要化学成分进行分析,并求得各类的平均值,之后根据聚类结果用模糊数学对其化学品质进行综合评价,由于先用层次聚类分析对样本进行了归类,使得大样本的综合评价变得简单清晰,因此该方法特别适用于大量样本的综合评价。应用该方法对南阳烟区的烟叶样品进行综合评价,结果表明,若把50份样品分为2类,则腰叶和上二棚叶几乎各自聚为一类。若把50份样品分为4类,则模糊评价的结果显示,第Ⅲ类样品的得分最高,表明其烟叶质量最佳。该方法的评价结果与腰叶优于上二棚叶的传统看法一致,与南阳烟区优质烟叶分布的实际情况也基本一致,表明用本研究的方法来评价烟叶的质量是比较客观的,也是可行的。

Comprehensive Evaluation of Flue-cured Tobacco Leaves Base on Cluster Analysis and Fuzzy Mathematics

The quality of cured tobacco leaves was mainly assessed by sense of smell,taste,touch and sight in the past long time,and this assessment always led to different results of the same cured tobacco leaves.So a method combining cluster analysis with fuzzy mathematics was established to access the flue-cured tobacco leaves' quality.First,cluster analysis was used to classify the samples,after that fuzzy mathematics was easily used to access the tobacco leaves' quality because the numbers of samples had been greatly decreased after cluster analysis.Then the method was used to evaluate the tobacco leaves quality in Nanyang area and the result showed that the cutters and the upper leaves from 25 sites were almost individually classified to different cluster when all samples were classified to 2 clusters.The result also showed that leaf quality of the the third cluster's tobacco was the best when all samples were classified to 4 clusters.The results coinsided with the fact that the quality of cutters were better than upper leaves and the distribution of high grade tobacco leaves in Nanyang area,which showed that the method to assess the tobacco leaves' quality was feasible.