[关键词]
[摘要]
目的 应用拉曼光谱技术结合多种预处理算法和多种特征波段筛选方法建立数学模型,对甘草配方颗粒提取过程中甘草苷和甘草酸含量实时监测。方法 以甘草配方颗粒为研究对象,收集提取过程中各个时间点的提取液样本,进行拉曼光谱检测。采集得到的光谱与液相色谱结果对应,分别建立3种甘草苷和甘草酸定量校正模型,考察不同预处理方法对模型的影响,优选出最佳变量筛选方法。结果 甘草苷模型中标准正态变换(SNV)预处理方法和甘草酸模型中Savitzky-Golay 13点平滑方法对模型性能参数提升幅度最大。通过连续投影算法筛选的甘草苷和甘草酸模型分别只需要4个和3个光谱变量即可达到全光谱变量模型水平,通过竞争性自适应重加权算法(CARS)筛选后建立的贝叶斯岭回归(BRR)甘草苷和甘草酸定量模型具有全局最优性能。结论 拉曼光谱技术应用于甘草配方颗粒提取过程中所建立模型性能良好,为实现中药配方颗粒提取过程实时监测和快速分析提供了研究基础。
[Key word]
[Abstract]
Objective To establish the mathematical models for a real-time Raman spectral analytical method combined with a variety of pretreatment algorithms and a variety of feature bands to monitor the content of liquiritin and glycyrrhizic acid during the extraction of Gancao (Glycyrrhizae Radix et Rhizoma, GRR) formula granules (GRRFG). Methods Taking GRRFG as the research object, different samples of extracts at various time points were collected in the extraction process and detected by Raman spectroscopy. The results of HPLC were used as a reference to build the quantitative calibration models for liquiritin and glycyrrhizic acid, respectively. A variety of spectral pre-processing methods and variable selection methods were applied to the establishment of models. Results The results showed that the standard normal variate transformation (SNV) pre-processing method in the liquiritin model and the Savitzky-Golay 13 point smoothing method in the glycyrrhizic acid model were suitable to improve the performance parameters of the mathematical models. Successive projection algorithm with only four bands and three bands were almost the same as the results of raw spectrums of liquiritin and glycyrrhizic acid. The bayesian ridge regression (BRR) model of these two compounds processed by the competitive adaptive reweighted sampling (CARS) showed the global optimal performance. Conclusion The Raman spectroscopy technology is applied to the GRRFG extraction process, and the established mathematical models have good performance, which provide a research foundation for realizing the real-time monitoring and rapid analysis of the extraction process of traditional Chinese medicine formula granule.
[中图分类号]
R283.6
[基金项目]
浙江省自然科学基金项目(LY20H280014);中国学位与研究生教育学会研究课题(2019YX09)