基于近红外光谱技术的蜂胶总黄酮含量分析

目的: 应用可见/近红外光谱技术结合特征波长提取方法研究快速无损测定蜂胶毛胶中主要活性成分总黄酮含量的方法。方法: 对112个蜂胶样本进行光谱扫描,并采用紫外分光光度法测得蜂胶总黄酮的化学值。经过平滑、变量标准化、一阶及二阶求导、去趋势等预处理,选择最优预处理方法。采用3种有效波长提取方法:连续投影算法(SPA)、载荷系数法(x-LW)和回归系数法(RC),将提取的有效波长作为最小二乘-支持向量机(LS-SVM)、偏最小二乘法(PLS)和多元线性回归(MLR)的输入变量,分别建立了各种线性和非线性模型。结果: 以预测集的预测相关系数(R_p),预测标准误差(RMSEP)和相对分析误差(RPD)作为评价指标比较各模型。结果表明,蜂胶毛胶中总黄酮含量预测的最优模型为SPA-PLS模型,R_p=0.954,RMSEP=9.32,RPD=4.02。结论: 基于光谱技术和化学计量学方法的检测模型具有较好的检测效果,可作为蜂胶毛胶中有效成分的快速测定新方法。

Objective: To develop a new method for the determination of total flavonoids in Chinese propolis by means of visible and near infrared spectroscopy(Vis-NIR)combined with characteristic wavelength selection. Methods: Spectral data of 112 Chinese propolis samples were obtained by using a Vis-NIR spectroradiotometer.The content value of total flavonoids of samples was determined by ultraviolet spectrophotometry(UV).The optimum preprocessing method was selected through preprocessing by smoothing,variable standardization,first and second derivation and detrending.The effective wavelengths were selected by successive projections algorithm(SPA),x-loading weights(x-LW)and regression coefficient(RC)which were employed as the inputs of partial least squares(PLS),multiple linear regression(MLR)and least square-support vector machine(LS-SVM)to develop different linear and non-linear models. Results: The correlation coefficient of prediction(R_p),the root mean square error of prediction(RMSEP)and the relative percent deviation(RPD)in the prediction set were applied as evaluation indicators to compare the performances of these models.The best results were achieved by the SPA-PLS model established with the raw spectra,and the prediction results were R_p=0.954,RMSEP=9.32 and RPD=4.02. Conclusion: The results showed that the model based on spectroscopic technology and chemometrics which was studied in the present paper played a good role in the detection and offered a new method for the rapid content determination of total flavonoids in Chinese propolis.