论文部分内容阅读
本文对10个不同产地的5种黄芩借TLC、HPLC、UV和IR进行化学测量,用计算机模式识别技术对数字化的化学特征进行归类,得到与植物分类学一致的结果。全部数据经逐次主成分分析法提取特征,用马氏(Mahalanobis)距离做为量度指标进行聚类分析,得到聚类谱系图;用Bayes意义下的多组判别分析法建立了判别函数。用三个样品进行验证,得到与实验符合的结果,表明用模式识别法鉴定黄芩是可行的。本法可作为鉴定其它中药的参考。
In this paper, chemical measurements of five species of jaundice from 10 different origins were carried out by TLC, HPLC, UV and IR. Computerized pattern recognition techniques were used to classify the chemical characteristics of the digitized species and the results were consistent with plant taxonomy. All data were extracted by successive principal component analysis (PCA), Mahalanobis distance was used as a measure to cluster analysis, cluster pedigree maps were obtained, and discriminant functions were established using multi-group discriminant analysis in Bayesian sense. Three samples were used for verification and the results agreed with the experiment showed that it is feasible to identify jaundice using pattern recognition. This method can be used as a reference for the identification of other traditional Chinese medicines.