基于主成分分析和聚类分析的烤烟质量评价模型构建

Model of the Quality Evaluation of Flue-cured Tobacco Based on Principal Component Analysis and Cluster Analysis

为了建立更为科学、客观的烟叶质量评价方法,采用外观质量评价与常规化学成分评价相结合的方法,研究8个外观质量指标和8个化学成分指标,建立了基于主成分分析和聚类分析的Fisher判别函数的烟叶质量评价模型,并对河南省平顶山市42个烟叶样品进行了分类。采用主成分分析方法共提取了6个主成分,并对各个样品6个主成分得分进行聚类分析,同时根据6个主成分得分和聚类分析结果进行判别分析,对比Fisher判别函数和聚类分析的结果,两者有较好的一致性。因此,可以利用这个烟叶评价模型对烟叶进行质量评价。这使得烟叶质量评价方法更加客观,也开辟了烟叶质量评价的新途径。表5,参16。

To be more scientific and more external, combined with appearance evaluation method and conventional chemical evaluation method, 8 appearance quality characteristics and 8 chemical quality characteristics were researched. And a flue-cured quality evaluation model was es- tablished based on principal component analysis, cluster analysis and Fisher discriminant functions. 42 samples were classified with this model in Ping ding-shan city He＇ nan Province. 6 main components were extracted with principal component analysis. 42 samples were clustered by 6 main components scores with cluster analysis. Discriminant analysis used based on 6 main components scores and the result of cluster analysis. The model was acceptable because the results of Fisher discrimiuant functions and cluster analysis had a well consistent with each other. Accordingly, this study suggested that the model be a new method with more external evaluation for tobacco quality.