近红外光谱技术应用于中草药口服液在线质量控制的化学计量学建模

摘要

以一种中草药口服液中多糖含量、可溶性固形物含量及pH为对象,利用近红外光谱技术对其进行实时在线检测,重点研究了化学计量学建模方法.建模过程采用了一种系统工程的方法,综合考虑了光谱预处理、特征波段选择、数据选择和分组、近红外原始光谱和化学分析参考值的误差分析,并把工具箱集成在一起.还考察了光谱采集的重复性、采样点取样条件,并对实验室测定的参考值进行了重复性及测量误差的考察.通过对比最终选择的是偏最小二乘法(PLS)结合遗传算法自动选波段.除此之外,以模型的交叉验证均方根误差(RMSECV)、预测均方根误差(RMSEP)、质量指标化学参考值与预测值间的相对误差为指标筛选最佳建模参数.建模结果显示,多糖模型预测值与化学参考值间的相对误差可控制在10%内,可溶性固形物误差在5%内, pH误差在4%内.生产使用表明,所建立的模型可有效应用于在线、快速对质量指标做出准确的预测.%Chemometric model development methods were investigated for the application of the near infrared spectroscopy(NIR)to on-line real-time monitoring of the content of polysaccharides,the content of soluble solids as well as pH of a liquid tonic. A system engineering approach was employed in building chemometric models via integration of spectra preprocessing,feature wave number selection,grouping and data selection for training and validation,as well as examination of the raw NIR measurements and laboratory analytical reference values. The final selected methods for data pre-processing,feature wave number selection and model building were the first order derivative, genetic algorithm,and PLS. The model performance was evaluated by RMSEC,RMSEV,and RMSEP through cross validation and assessed using test data as well as real data from the industrial process. It was shown that the relative error between the predicted value and the reference value could be controlled within 10% for the content of polysaccharide. The relative errors for predicting soluble solid and pH were controlled within 5% and 4%,respectively. The models were successfully used in commercial applications for timely estimation of the product quality related properties.