基于近红外光谱的快速、无损分析技术在肉苁蓉药材质量控制中的应用研究  

Application of rapid and nondestructive analysis technology based on near-infrared spectroscopy in quality control of cistanche deserticola

作　　者：焦慧平 莲花 李润甜 渠弼 白文明 JIAO Hui-ping;LIAN Hua;LI Run-tian;QU Bi;BAI Wen-ming(College of Pharmacy of Inner Mongolia Medical University,Inner Mongolia,010110,China)

机构地区：[1]内蒙古医科大学药学院,内蒙古呼和浩特010110

出　　处：《化学研究与应用》2025年第2期439-448,共10页Chemical Research and Application

基　　金：内蒙古医科大学面上项目(YKD2021MS022)资助。

摘　　要：本文基于近红外光谱结合化学计量学方法建立预测肉苁蓉药材水分、浸出物、灰分、苯乙醇苷、多糖和甘露醇含量的快速无损分析方法。采集各批次肉苁蓉样品的近红外漫反射光谱,利用传统分析方法测定各批次肉苁蓉样品中各成分含量的化学分析值,采用偏最小二乘回归法和支持向量机回归法结合各种光谱前处理手段建立由肉苁蓉近红外光谱预测水分、浸出物、灰分、苯乙醇苷、多糖和甘露醇含量的定量模型,并通过校正组样品和验证组样品评价模型的预测精度和稳定性。采用偏最小二乘法建立的预测苯乙醇苷、多糖、甘露醇和灰分含量的最优定量模型的预测值和化学分析值间的相关系数rc和rv分别为0.950 7和0.916 5,0.980 6和0.908 9,0.937 5和0.941 7,0.938 2和0.919 4。采用支持向量机回归法建立的预测水分和浸出物含量的最优定量模型的预测值和化学分析值间的相关系数rc和rv分别为0.960 2和0.934 8、0.901 8和0.910 0。本研究建立的含量测定方法分析速度快、不破坏分析样品的药用价值。To establish a rapid and nonhazardous analytical method for predicting the contents of water,extract,ash,mandeloside,polysaccharide and mannitol in Cistanche deserticola based on near-infrared spectroscopy combined with stoichiometry.The near-infrared diffuse reflectance spectra of each batch of Cistanche deserticola samples were collected,and the chemical analysis values of each component content in each batch of Cistanche samples were determined by traditional analytical methods.Partial least squares regression method and support vector machine regression method combined with various spectral pretreatment methods were used to establish a quantitative model for predicting the contents of moisture,extract,ash,mandeloside,polysaccharide and mannitol from the near-infrared spectrum of Cistanche deserticola.The prediction accuracy and stability of the model were evaluated by the samples of the calibration group and the validation group.The correlation coefficients r_c and r_v between the optimal quantitative models established by partial least squares method and the chemical analysis values were 0.9507,0.9165,0.9806,0.9089,0.9375,0.9417,0.9382 and 0.9194,respectively.The correlation coefficients r_c and r_v between the predicted values of the optimal quantitative model established by support vector machine regression method and the chemical analysis values were 0.9602 and 0.9348,0.9018 and0.9100,respectively.The content determination method established in this study is fast and does not destroy the medicinal value of the analyzed samples,which can provide technical guidance for the application of rapid and non-destructive analysis technology in the quality control of Cistanche deserticola.

关 键 词：肉苁蓉 近红外光谱 含量测定 化学计量学 

分 类 号：O657.3[理学—分析化学]