瑶药“老班药”特殊分类的化学模式识别方法研究  

Study on chemical pattern recognition method for the special classification of the Yao's medicine

作　　者：杨天鸣[1] 苏蕊[1] 付海燕[1] 王蒙[1] 梁玉彬 张璐[1] 黄国贞 范尧[1] 

机构地区：[1]中南民族大学药学院,湖北武汉430074

出　　处：《时珍国医国药》2015年第9期2169-2172,共4页Lishizhen Medicine and Materia Medica Research

基　　金：科技部国家"十二五"科技支撑计划(No.2012BAI27B00);国家自然科学基金(No.21205145);湖南大学化学生物传感与计量学国家重点实验室开放基金(No.201111)

摘　　要：目的运用近红外漫反射技术结合主成分-马氏距离判别分析方法,对瑶药"老班药"的传统分类进行科学表征。方法采集二十种瑶药"老班药"的近红外漫反射光谱,经预处理后,通过主成分分析-马氏距离判别分析方法构建瑶药"老班药"模式识别模型。结果所得模型对二十种瑶药"老班药"四种类别"虎、牛、钻、风"的正确识别率达到90%以上,表明该模型具有较好的预测分类能力。结论运用简便、快速、无损的近红外漫反射技术结合主成分分析-马氏距离判别分析方法,对瑶药"老班药"的传统分类进行了科学表征,结果表明瑶药"老班药"-"虎、牛、钻、风"的特殊分类有比较可靠的物质基础。近红外光谱漫反射技术结合马氏距离判别分析是一种适用于瑶药"老班药"特殊分类研究的可靠的模式识别方法。Objective To scientific and digitized characterize the traditional categories of the 20 kinds of the Yao＇s medicine by near- infrared diffuse reflectance spectroscopy （NIRDRS） combined with PCA- Mahalanobis distance pattern recognition method. Methods The near infrared original spectra of 20 kinds of Yao＇s medicine were pre - processed by muhiplicative scatter correction （MSC） and followed by PCA - Mahalanobis distance discriminant analysis treatment. Results Perfect classification ratio with 90% was obtained for the MSC fingerprint information of the NIRDRS based on PCA - Mahalanobis distance model. Conclu- sion PCA - Mahalanobis distance discriminant analysis can as a simple, rapid, accurate and green method extract otherness of MSC fingerprint information to characterize the traditional categories of the 20 kinds of the Yao＇s medicine.

关 键 词：瑶药 老班药 近红外光谱 多元散射校正 主成分-马氏距离判别分析方法 

分 类 号：R282.5[医药卫生—中药学]