[关键词]
[摘要]
目的 探索采用多源信息融合建模技术提高中药提取工艺模型的校正和预测性能。方法 以丹参脂溶性成分乙醇提取过程为研究对象,收集不同来源丹参饮片模拟原料波动,采用实验设计(DOE)模拟工艺参数变化,以提取过程近红外光谱(NIRS)作为过程状态变量。采用HPLC法分析提取液中丹参酮ⅡA、隐丹参酮和丹参酮I的含量。将原料属性、工艺参数和过程状态变量组合为自变量,以提取液有效成分含量为因变量,采用偏最小二乘(PLS)法建立提取液质量预测模型。结果 建模结果为丹参酮ⅡA交叉验证均方根误差(the root mean squared error of cross validation,RMSECV)为0.172 8 mg/g,预测均方根误差(the root mean squared error of prediction,RMSEP)为0.031 7 mg/g,性能偏差比(ratio of performance to deviation,RPD)为6.91;隐丹参酮RMSECV为0.153 4 mg/g,RMSEP为0.024 2 mg/g,RPD为4.02;丹参酮I RMSECV为0.117 1 mg/g,RMSEP为0.043 2 mg/g,RPD为4.76。结论 多源信息融合模型的校正和预测性能均优于常规模型,可有效提升丹参提取液质量可预测性和可控性。
[Key word]
[Abstract]
Objective The potency of multi-source information fusion technology was explored to improve the model calibration and prediction performance of Chinese medicine extraction process. Methods The ethanol extraction process for isolating fat-soluble components from Salvia miltiorrhiza was taken as the research carrier. S. miltiorrhiza from different sources were collected to simulate the fluctuation of materials. The changes of process parameters were simulated by design of experimental (DOE), and the process near infrared spectra (NIRS) were used as the process state variables. The contents of tanshinone ⅡA, cryptotanshinone, and tanshinone I were determined by HPLC. The raw material properties, process parameters and process state variables were combined as independent variables. The content of effective components in the extract was taken as the dependent variable. The partial least squares (PLS) algorithm was used to establish the quality prediction model of the extracts. Results The modeling results respectively showed that the RMSECV was 0.172 8 mg/g, RMSEP was 0.031 7 mg/g, RPD was 6.91 (tanshinone ⅡA); RMSECV was 0.153 4 mg/g, RMSEP was 0.024 2 mg/g, RPD was 4.02 (cryptotanshinone); RMSECV was 0.117 1 mg/g, RMSEP was 0.043 2 mg/g, RPD was 4.76 (tanshinone I). Conclusion The calibration and prediction performance of multi-source information fusion model are better than the conventional model, which can effectively improve the quality predictability and controllability of S. miltiorrhiza extract.
[中图分类号]
[基金项目]
国家自然科学基金资助项目(81403112);国家中药标准化项目(ZYBZH-C-QIN-45)