基于信息熵理论的BP-ANN结合CCD-RSM优化黄连-黄柏提取工艺研究  

Optimization of extraction process of Coptidis Rhizoma-Scutellariae Radix based on BP-ANN and CCD-RSM in information entropy theory

作　　者：王宝才[1] 李俊江[1] 徐志伟[1] Bao-Cai WANG;Jun-Jiang LI;Zhi-Wei XU(Center for Scientific Research Preparation,Gansu Provincial Hospital of TCM,Lanzhou 730050,China)

机构地区：[1]甘肃省中医院科研制剂中心,兰州730050

出　　处：《药物流行病学杂志》2023年第11期1267-1274,共8页Chinese Journal of Pharmacoepidemiology

基　　金：白银市科技计划项目(2021-2-25Y)。

摘　　要：目的基于熵权法结合星点设计-效应面法(CCD-RSM)和误差反向传播人工神经网络(BP-ANN)建模优化黄连-黄柏的提取工艺。方法采用HPLC测定表小檗碱、黄连碱、巴马汀、小檗碱的含量,应用UV测定总生物碱的含量,并计算提取量。采用信息熵理论对上述5个指标的的提取量与干膏得率进行综合评分,以CCD-RSM的13组数据作为训练数据,采用BP-ANN进行建模及分析,以综合评分作为考察指标,仿真模拟预测黄连-黄柏最佳提取工艺参数。结果最佳提取工艺为加11倍量水煎煮2次,每次煎煮95 min,此时综合评分达到最大值106.41。结论BP-ANN建立的数学模型具有良好的预测性,优化的提取工艺高效、稳定、可行。Objective To optimize the extraction process of Coptidis RhizomaScutellariae Radix.Methods The contents of epiberberine,coptisine,palmatine and berberine were determined by HPLC,the contents of total alkaloids were determined by UV,and the quantity of water extracted was calculated.The above five indexes and dry extract yield were comprehensively scored based on information entropy theory.13 sets of data from the central composite design-response surface methodology(CCD-RSM)were used as training data,modeled and analyzed using back propagation artificial neural network(BP-ANN),and simulated to predict the optimal extraction process parameters of Coptidis RhizomaScutellariae Radix using the composite score as the index of investigation.Results The best conditions were 11 times of water,boiling 95 minutes each time,twice for boiling,the maximum comprehensive score is 106.41 at this point.Conclusion The mathematical model established by BP-ANN has good predictability,and the optimized extraction process has the characteristics of high efficiency,stability,and feasibility.

关 键 词：黄连-黄柏 误差反向传播人工神经网络 星点设计-效应面法 提取工艺 信息熵理论 

分 类 号：R284.2[医药卫生—中药学]