基于主成分分析的烟草质量一致性评价  

Evaluation of tobacco consistency quality based on principal component analysis

作　　者：张光辉[1] 李俊丽[1] 阴耕云 

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650504 [2]云南烟草研究所,云南昆明650106

出　　处：《计算机与应用化学》2014年第4期461-465,共5页Computers and Applied Chemistry

基　　金：云南省应用基础研究计划项目(2011FZ036)

摘　　要：研究了以烟叶化学成分为基础,利用主成分分析和聚类分析方法对烟草质量一致性进行评价的方法。该方法采用主成分分析方法,剔除了原始指标之间的重复信息,将原有的17种常规化学成分指标变量综合为3个主元成分变量代替。依照烟草样本主成分特征数据,以类内离散度为衡量指标评价同一批次烟草质量的一致性。用此方法对5个批次、不同类型的烟草进行了一致性评价,评价结果表明,成品烟丝的一致性最好,原料烟叶的一致性相对较差,该结果与传统评价结果一致。本文提出的评价方法基础数据易得,实现简单,评价结果贴合实际。This paper discusses a new method that evaluates the consistency quality of tobacco based on chemical components and uses principal component analysis and cluster analysis. The method uses principal component analysis to eliminate duplicate information, so that original 17 chemical components are reduced to three principal component variables. In accordance with the characteristic data of the tobacco samples, the parameter of class dispersion is calculated, and as an indicator to evaluate consistency quality of the same batch tobacco. Using this method to evaluate the consistency quality of five batches tobacco, the results indicate that the consistency of the finished tobacco is best, and the raw tobacco is relatively poor. The results are consistent with the traditional evaluation results. The proposed method in this paper has advantages that the basic data are easy to obtain and evaluation is simlole, convenient and fitting to the actual situation.

关 键 词：主成分分析 化学成分 类内离散度 质量评价 

分 类 号：TQ015.9[化学工程] TP391.9[自动化与计算机技术—计算机应用技术]