摘要
用红外指纹图谱和主成分分析法 ,对来自 8省的 74个山药样本的道地与非道地性进行了聚类分析 ,以主成分 2和 3作二维线性投影图 ,均具有较好的聚类和鉴别作用。利用已建模型预测了 11个样本的道地性 ,正确率达 90 .9%。结果表明 :所建模型合理 ,具有明显的分类和鉴别作用 ,具有一定的可行性和可靠性 ,为山药道地性鉴别提供了一种新的方法。
Based on the fingerprint infrared spectra and the principal component analysis, 74 dioscoreae samples from 8 provinces of China were analyzed to determine their trueborn quality. The 2-dimontional projection plot was drawn with second and third principal components. The 2D plot indicates it is a good clustering analysis method for the classification and discrimination of dioscoreae. 11 Samples were predicated by this model,. the correctness is 90.9%. This model is reliable and practicable. So it can be applied to the trueborn quality analysis of dioscoreae.
出处
《分析化学》
SCIE
EI
CAS
CSCD
北大核心
2002年第10期1231-1233,共3页
Chinese Journal of Analytical Chemistry
基金
国家中医药管理局科技重大项目 (中国医药科 2 0 0 1ZDZX0 1)
河南省科委自然科学基金资助课题
关键词
红外光谱
山药
主成分分析
药材鉴别
Fourier transform infrared spectroscopy
dioscorea opposita thunb
trueborn herbal medicine
principal component analysis
