摘要
以赤芍的红外指纹图谱作为聚类分析的对象 ,在建立主成分模型的基础上 ,采用SIMCA聚类分析法对赤芍进行了快速的分类研究。结果表明 :由于野生与栽培赤芍的红外指纹图谱变异度较大 ,其光谱聚类的结果较理想 ,盲样检测的正确率可达 90 % ;由于不同产地的赤芍的红外指纹图谱变异度较小 ,进行聚类分析后用盲样检测 ,正确率为 75 % ,对于变异度较小的样品 ,若要提高识别的置信度 ,需增加样品的数目和采样的代表性。总之 ,红外指纹图谱与聚类分析法相结合可以快速鉴别药材的道地性。
On the basis of building a model of main components and with the fingerprint infrared spectrum as the object, Chishao was rapidly classified and closely studied by using the method of clustering analysis. Because of the greater difference between the spectra of wild Chishao and those of the cultivated, the outcome of the clustering is ideal. When testing with the blind samples (which were picked out from the chosen samples) the accuracy of clustering reaches up to 90%, while, because of the lesser difference among the spectra of Chishao collected from different producing areas in China, the accuracy of clustering is only 75%. In order to increase the clustering accuracy, the number of the samples should be increased and also these samples should be more characteristic. On the whole, with the combination of the fingerprint infrared spectrum and the clustering analysis, the genuineness of herbal medicines can be discriminated rapidly.
出处
《光谱学与光谱分析》
SCIE
EI
CAS
CSCD
北大核心
2002年第2期232-234,共3页
Spectroscopy and Spectral Analysis
关键词
红外光谱
聚类分析法
赤芍
无损鉴别
盲样
中药材
fingerprint infrared spectrum
clustering analysis
undamaged determination
Chishao
blind sample
