基于双指标分析法和聚类分析法的鸡骨草红外指纹图谱比较研究  

Comparative Study on the Infrared Fingerprint of Abrus Cantoniensis Based on the Methods of Sequential Analysis of Dual-Indexes and Cluster Analysis

作　　者：孔德鑫[1,2] 黄庶识[2] 黄荣韶[1] 陈秀丽[2,3] 王一兵 陈植成[2,3] 

机构地区：[1]广西大学农学院,广西南宁530005 [2]广西科学院生物物理实验室,广西南宁530003 [3]广西师范大学物理与电子工程学院,广西桂林541004

出　　处：《光谱学与光谱分析》2010年第1期45-49,共5页Spectroscopy and Spectral Analysis

基　　金：国家自然科学基金项目(30860010);广西自然科学基金项目(0640011);广西大型仪器协作共用网项目资助

摘　　要：利用双指标分析法和系统聚类分析法,鉴别了不同年限不同产地鸡骨草的红外指纹图谱。结果显示,(1)应用双指标分析法时,只做13点平滑就可以将6种鸡骨草样品完全分开,可以在六维空间中考察不同中药样品之间的相互关系,并可用数量将这些关系精确地表示出来;应用系统聚类分析时只做13点平滑的初始数据聚类效果较差,但13点平滑后再做一阶导数处理,聚类效果很明显,6类鸡骨草样品可以完全分开。(2)应用双指标分析法,在样本量大时,共有峰率变异峰率统计较麻烦,但精确度较高;聚类分析法,精确度相对较低,但是运作比较方便。(3)运用以上两种方法均表明,同一产地栽培鸡骨草化学成分较相似,同一产地不同年限栽培鸡骨草化学成分差异较大。The methods of sequential analysis of dual-indexes and cluster analysis were utilized to investigate the infrared fingerprints of A. cantoniensis planted in different years and different places in Guangxi, China. The results showed that 6 samples were able to be completely separated only through 13 point smoothing, when the dual-indexes analysis was applied in the present research, and the accurate relationship between these samples could be inspected and expressed by quantitative relationships under 6-dimensional spaces; however, the effect of cluster was bad only through 13 point smoothing of raw spectra, and it was very difficult to find out the regular sequences while the cluster analysis was applied. Furthermore; the 6 samples were able to be completely separated if raw spectra were dealed with 1st derivative after 13 point smoothing, and the clustering effects were more obvious and 6 samples of A. cantoniensis were completely separated. The above two methods could be used to evaluate the quality of Chinese medicinal materials easily when the sample was not excessive quantitatively, but the method of dual-indexes analysis was more difficult than the clustering analysis if the sample size was too large, since a mass of data such as common peak ratio and variation peak ratio of the IR fingerprint spectra were processed and analyzed statistically, while this method could accurately find out the closest relationship between any samples through comparing the quantitative relationships of common peak ratio and variation peak ratio of each sample under 6 dimensional space; the precision of cluster analysis was less than dual-indexes analysis, but it was more convenient than dual-indexes analysis when large sample data were analysed. Finally the above two methods all showed that the chemical composition of the A. cantoniensis was similar in the same cultivated area, but the difference in chemical composition of A. cantoniensis in different years was distinct even they were in the same place.

关 键 词：鸡骨草 红外光谱 双指标法 聚类分析 共有峰率 变异峰率 序列分析 

分 类 号：O657.3[理学—分析化学]