基于近红外光谱技术建立不同产地骆驼蓬药材定性判别模型.docx

基于近红外光谱技术建立不同产地骆驼蓬药材定性判别模 型 [ 摘要 ]该研究将光纤传感技术与近红外漫反射光谱相 结合，通过主成分分析、 聚类分析、 SIMCA 等方法直接对骆 驼蓬药材进行检测，用于识别不同产地的骆驼蓬，建立快速 无损鉴别骆驼蓬产地的新方法。研究中建立全波长原始光谱 图，通过主成分分析前 2 个主成分得分图基本可以区分 4 个 产地的骆驼蓬 ;波长 866～ 2 507 nm，MSC 为预处理方法建立 最佳聚类分析模型对预测集预测的正确率达到 91.67%，可对 4 个产地骆驼蓬基本区分 ;波长为 1 085～2 507 nm，预处理方 法为归一化法建立最佳 SIMCA 模型，对样品进行预测，只产地昌吉的样品没有被识别，其余样本判别正确，总识别率97.22%。研究结果表明利用近红外漫反射光纤光谱结合 SIMCA 法识别能力最佳，可用于骆驼蓬的鉴别。 [ 关键词 ]骆驼蓬 ;近红外光纤传感技术 ;主成分分析 ;聚类分析 ;SIMCA 分析 Establishment of different origin Peganum harmala qualitative discrimination model based on near-infrared spectroscopy LI Li* ， LI Ying ， WANG Ting-yuan （Collgeg of pharmacy ， Xinjiang Medical University ， Urumqi 830011 ， China） [Abstract]The optic-fiber sensor technology combined with near-infrared diffuse reflection spectroscopy was applied to directly analyze Peganum harmala and identify different origin of P. harmala on the basis of principal component analysis ， clustering analysis ， SIMCA method ， which resulted in the establishment of a new method to rapidly and nondestructively identify the origin of P. harmala. The original full wavelength spectrum for principal component analysis and the score of first two principal components can distinguish four origins of P. harmala basically. In the wavelength range of 866-2 507 nm ， MSC as pretreatment method to establish the best model of clustering analysis to forecast the samples with the accuracy of 91.67%， can distinguish the four origins of P. harmala while in the wavelength of 1 085- 2 507 nm ， normalization method as pretreatment methods to establish a best model of SIMCA to forecast the sample， all the samples except for the changji sample have been identified with a total recognition rate of 97.22%. The results show that using near infrared diffuse reflectance spectroscopy combined with SIMCA is the best method that can be effectively used to identify the P. harmala. [Key words]Peganum harmala; near infrared spectroscopy and fiber-optic sensor technology; principal component analys