基于GECA的成品烟叶内在化学成分聚类分析

　　摘要：为解决多因素内在化学成分对不同品种烟叶的影响分析，为实现生产过程质量均匀化奠定基础，通过ANTARIS II近红外光谱仪采集不同等级的成品烟叶烟碱、总糖、还原糖、总氮、K和CL六种化学成分含量。采用灰熵关联分析方法，分析分为七个步骤：通过对原始数据进行规格化处理；计算出不同等级烟叶的各评价指标相对于最优参考序列、最差参考序列的关联系数；计算不同等级烟叶不同指标关联系数密度分布值；根据每种指标所占的权重计算权重密度分布值；计算灰色关联熵；计算灰色熵关联度；形成不同等级成品烟叶样品的相对关联度，根据灰色熵关联度进行不同等級烟叶的聚类分析。以麒麟复烤厂的23个等级成品烟的六个化学成分为分析对象，最后聚为五个类别，每个类别在六个指标含量上总是最接近的。通过把灰色熵关联方法建立的烟叶聚类模型与传统感官品吸结论进行比较，证实通过灰色熵关联方法建立的烟叶聚类模型与传统品吸结论比较吻合，可以用此方法对选后烟叶进行分析，建立复烤加工配方的参考模型，有利于提高配方的科学性，在均质化生产模型建立上迈进重要的一步。
　　Abstract： In order to solve the influence of multi-factor intrinsic chemical composition on different varieties of tobacco leaves， and lay the foundation for achieving the homogenization of the production process the nicotine， different levels of finished tobacco tobacco nicotine， total sugar， reducing sugar， total nitrogen， K and CL six chemical composition content were obtained by ANTARIS II near infrared spectroscopy. Using the gray entropy correlation analysis method， the analysis is divided into seven steps： normalizing the raw data； calculating the correlation coefficient of the evaluation index of different grades of tobacco leaves relative to the optimal reference sequence and the worst reference sequence； calculating different grades and the relative entropy degree of the samples with different grades of finished tobacco leaves， according to the gray entropy correlation degree， the gray entropy correlation degree， the gray entropy correlation degree， the gray entropy correlation degree， the gray entropy correlation degree， the gray entropy correlation degree， Clustering analysis of different grades of tobacco leaves. The six chemical constituents of the 23 grade grades of the Kirin re-drying factory were analyzed for the five categories， each of which was always the closest to the six indicators. he tobacco leaf clustering model established by the gray entropy correlation method was compared with the traditional sensory absorption theory. It was confirmed that the tobacco leaf clustering model established by the gray entropy correlation method was consistent with the traditional method. This method can be used to analyze post-election tobacco leaves， the establishment of re-roasting processing formula reference model is conducive to improving the scientific formula， and in the homogenization of the establishment of an important step in the production model.
　　关键词：烟叶内在化学成分；加权；灰熵关联分析；聚类分析
　　Key words： tobacco leaf internal chemical composition；weighting；gray entropy correlation analysis；cluster analysis

相关热词搜索：烟叶 成品 内在 化学成分 分析