基于支持向量机的黄连饮片产地识别研究

目的建立基于支持向量机的黄连饮片产地区分识别模型。方法采集4个产地多批黄连样本,量化外部特征值(包括形状、气味、味道等),并按《中国药典》2010年版方法测定内部特征值(包括水分、总灰分、醇浸出物,指标成分表小檗碱、黄连碱、巴马汀、小檗碱质量分数),在Matlab 7.0平台进行数据降维和融合,建立黄连产地区分模型。结果单独分析各项数据不能较好区分各产地黄连饮片,而采用支持向量机建模后所有特征叠加识别率达到97.1%,能准确区分各黄连饮片产地,内外特征的高识别率说明各特征子集间具有一定的互补性,可综合辨识不同产地黄连饮片的差异性。结论建立的基于支持向量机的识别模型,实现产地的区分,为黄连产地区分提供研究思路和基础。

Discrimination of Cotidis Rhizoma from different habitats by support vector machine

Objective To establish the identification model based on support vector machine for distinguishing the origin of Cotidis Rhizoma. Methods Collection of Cotidis Rhizoma samples from four habitats, quantification of the external characteristics (including shape, smell, taste, etc.), and according to the current Pharmacopoeia method, determination of characteristic internal values (including moisture, ash, alcohol extract, index composition epiberberine, coptisine, palmatine, berberine content, etc.); In the Matlab 7.0 platform data dimensionality reduction and fusion, to create the habitats for Cotidis Rhizoma for the discrimination among the models. Results The separate analysis of each datum can not distinguish the habitats of Cotidis Rhizoma, while after the use of support vector machine modeling, all superposition feature recognition rate reached 97.1%, which can accurately distinguish the habitats of Cotidis Rhizoma; The internal and external feature of high recognition rate is a certain complementarity between each feature subset, the difference of Cotidis Rhizoma from different habitats could be comprehensively identified. Conclusion The identification model based on support vector machine and the distinction of producing habitats could provide the research ideas and basis for the distinction among the habitats of Cotidis Rhizoma.