模式识别技术在不同来源甘草药材分类中的应用

Application of pattern recognition technology in classification of Radix et Rhizoma Glycyrrhizae of different sources

目的:建立不同来源甘草药材的模式识别分类方法。方法:通过 HPLC 快速获得19个不同来源甘草药材样品的色谱指纹图谱,色谱条件固定相为 Kromasil KR100-5 C_(18)(150mm×4.6mm),流动相为乙腈-0.05%三氟乙酸溶液,梯度洗脱为乙腈20%(0min)→40%(5min)→50%(10min),流速为0.8mL·min^(-1),检测波长为254nm,从每个色谱指纹罔中提取出325个特征变量,采用聚类分析对甘草药材样品进行相似性评价,并应用 PLS 方法建立起不同植物种、不同种植方式甘草的2个分类回归模型。结果:采用分类模型对不同植物种、不同种植方式甘草进行判别,2个分类回归模型的分类准确率均为100%。结论:此方法可用于不同来源甘草药材的分类研究,本文以 HPLC 指纹图谱结合化学计量学知识,为不同中药样品的质量评价提供了一种分类方法。

Objective：To establish a method for pattern recognition classification on Radix et Rhizoma Glycyrrhizae of different sources. Methods： HPLC system was applied to obtain the chromatograms of the 19 Radix et Rhizoma Glycyrrhizae samples. The Kromasil KR100 -5 C18 （ 150 mm ×4. 6 mm） column was used. A mixture of acetonitrile 0.05% trifluoroacetic acid solution was used as the mobile phase with a flow rate of 0. 8 mL · min^-1 and the gradient mode was acetonitrile 20% （0 min） →40% （5 min）→50% （ 10 min）. The detection wavelength was 254 nm. The 325 characteristic variables were obtained from each chromatogram. The clustering analysis was applied to evaluate the similarity of the chromatograms. Then partial least - squares regression method was applied to build the prediction models of classification which were used to distinguish Radix et Rhizoma Glycyrrhizae of different species and cultivation modes. Results：The classification accuracy of the two prediction models of classification was 100% for Radix et Rhizoma Glycyrrhizae of different sources. Conclusion：This method is practicable in the pattern recognition classification on Radix et Rhizoma Glycyrrhizae of different sources. The developed technique could be further applied to the classification of different traditional Chinese medicine（TCM）.