基于红外光谱数据的中药药性识别研究  

Study on the Property Recognition of Traditional Chinese Medicines Based on Infrared Spectrum

作　　者：刘进[1] 邓家刚[2] 覃洁萍[2] 冯旭[2] 王丽丽[2] 罗祖良[2] 韦日伟[2] 

机构地区：[1]广西师范学院,广西南宁530023 [2]广西中医学院,广西南宁530001

出　　处：《时珍国医国药》2010年第3期561-563,共3页Lishizhen Medicine and Materia Medica Research

基　　金：国家"973"计划资助项目(No.2007CB512602)

摘　　要：目的探讨中药红外光谱与药性的相关性。方法以红外光谱数据为分类指标,用主成分分析提取主成分,然后采用交叉验证法对不同药性中药进行交叉训练,建立中药药性的预测模型。结果该模型对预测集中平性药的识别正确率为83.33%,对非平性药的识别正确率为82.5%,总正确率82.89%,结论中药红外光谱与药性具有一定的相关性。Objective To explore the relationship between the properties of traditional Chinese medicine（TCM） and infrared spectrum data.Methods The infrared spectrum data of TCM were selected as indexes,and the principal components of the data were found out through PCA and trained with cross validation method through SVM,so as to establish the neutral and non-neutral property of TCM prediction model.Results The model recognition accuracy for the neutral medicine in prediction set was 83.33%,and 82.5% for non-neutral medicine,which made up an accuracy of 82.89% in average.Conclusion The infrared spectrum and properties of TCM have some relevance.

关 键 词：主成分分析 支持向量机 红外光谱 中药药性 

分 类 号：R282.7[医药卫生—中药学]