基于主成分分析和聚类分析的野生战骨红外指纹图谱比较研究

Comparative study on FTIR fingerprint of wild Premna fulva based on principal component analysis and cluster analysis

关键词:  战骨  主成分分析  聚类分析  红外指纹图谱  化学成分

Key words:  Premna fulva  principal component analysis  cluster analysis  infrared fingerprint  chemical composition

摘要:

利用傅里叶变换红外光谱仪检测不同产地战骨的红外光谱，并通过主成分分析模型，聚类分析以及主成分载荷因子等化学计量学方法，对不同产地样本进行鉴别，找出不同产地样本化学组成的差异。结果表明：（1）在主成分分析模型检测产地中广西平果县和扶绥县产地相距最近，在三维图上的距离也最近，广西天峨县与扶绥县和田阳县相距较远，在散点图上的距离也较远；云南和贵州地理位置处于广西境外，在散点图上的距离则偏离广西各地；（2）对主成分分析模型结果通过聚类分析验证表明二者的分析结果较一致，与主成分三维散点图相比，聚类分析图能在各距离水平上对研究对象进行准确的划归；（3）通过提取PCA模型中PC1的载荷因子，发现不同产地战骨样品化学组成的差异主要体现在柚皮素、香草酸、丁香酯素、多糖、苷类5种物质成分含量的不同。

Abstract:

In order to identify and in depth excavate the differences of chemical compositions in Premna fulva from different regions，inrared spectra of multi region sampled individuals were detected using the fourier transform infrared spectroscopy. After that，chemometrics methods including principal component analysis，clustering analysis and principal component loading factors were utilitied to pretreat original spectra data. The results showed that:(1)in 3 dimensional scatter maps，among the different samples，the distance between Pingguo and Fusui of Guangxi was nearest，so the natural consequence，the distances of the samples from these two places were nearest，while the real distance from Tiane to Fusui and Tianyang was further in Guangxi，the distances of the samples among these three places were also further in 3 dimensional scatter maps. In addition，geographical position of Yunnan and Guizhou were far from Guangxi，the results showed that scatter plot chart on distance were also deviated from Guangxi all around. (2)The principal component analysis model results was verified by clustering analysis and had similar consequences，while，as far as the two methods were concerned，they had individual characteristic，compared with the method of principal component analysis，clustering analysis chart could be accurately classified the research objects on each distance level，but it was difficult to directly reflect the interrelationship among research objects. (3)Through extracting the loading factor based on principal component analysis，the result indicated that the differences of quality of P.fulva among different regions mainly represented in the different contents of chemical components including naringenin，vanillic acid，syrinaresiol，polysaccharide and glycosides.