基于Box-Behnken响应面法与BP神经网络优化雪莲药渣化学成分提取工艺  

Optimization of extraction of dregs of Saussurea involucrata after ethanol extraction based on Box-Behnken response surface methodology and BP neural network

作　　者：艾尼玩尔·买提库尔班 马桂芝[1] 滕亮[2] Ainiwaner Maitikuerban;MA Gui-zhi;TENG Liang(College of Pharmacy,Xinjiang Medical University,Urumqi 830011;Department of Pharmacy,The First Affiliated Hospital of Xinjiang Medical University,Urumqi 830054)

机构地区：[1]新疆医科大学药学院,乌鲁木齐830011 [2]新疆医科大学第一附属医院药学部,乌鲁木齐830054

出　　处：《中南药学》2024年第4期923-929,共7页Central South Pharmacy

基　　金：新疆维吾尔自治区重点研发计划项目(No.2017B03013-2)。

摘　　要：目的 优化雪莲药渣提取工艺。方法 在测得提取液中绿原酸、芦丁和多糖含量的基础上采用层次分析法(AHP)赋权得到综合评分。以料液比、提取时间和提取次数为考察因素,以综合评分为优选指标,运用Box-Behnken响应面设计和BP神经网络优化雪莲药渣化学成分的提取工艺。结果 Box-Behnken响应面法优化的提取工艺为加9倍量水回流提取3次,每次提取42 min。选择响应面试验中的17组数据作为训练数据和验证数据,BP神经网络预测的优选的提取工艺为加6倍量水回流提取3次,每次煎煮30 min。验证试验结果表明,BP神经网络优化工艺实际综合评分高于响应面法实际综合评分,提示BP神经网络预测的优选的提取工艺更为合理。结论 BP神经网络预测的优选的提取工艺合理、稳定、可行,可为雪莲药渣的二次开发利用提供参考。Objective To optimize the extraction of dregs of Saussurea involucrata after ethanol extraction.Methods Based on the content of chlorogenic acid,rutin and polysaccharides in the extract,the analytic hierarchy process(AHP)was used to assign weights to obtain a comprehensive score.The extraction for the chemical constituents of dregs was optimized with a composite score as a preference index.Material liquid ratio,extraction time,and extraction times served as factors.To accomplish this,the Box-Behnken response surface design and BP neural network were used.Results The extraction conditions were optimized by the Box-Behnken response surface methodol as follows:reflux extraction being conducted with 9 times of water for 42 min,repeating 3 times.Seventeen sets of data were selected from the response surface test for training and validation.The BP neural network predicted the preferred extraction to be adding six times of water to reflux the extraction for 3 times,followed by decoction for 30 min each time.The validation test demonstrated that the BP neural network optimization achieved a higher actual comprehensive score than that of the response surface method,indicating a more reasonable extraction progress by the BP neural network.Conclusion The extraction progress by the BP neural network outlined in this paper is reasonable,stable and viable.It serves as a useful reference for the secondary development and utilisation of dregs of Saussurea involucrata.

关 键 词：雪莲 Box-Behnken响应面法 BP神经网络 层次分析法 提取工艺 

分 类 号：R283[医药卫生—中药学]