文献类型：期刊文献

中文题名：合欢皮多糖提取工艺优化及BP神经网络模型

英文题名：Optimization and BP neural network model of extraction of polysaccharide from Albizia julibrissin Durazz

作者：韩伟[1];孙晓海[1];罗文峰[1]

机构：[1]华东理工大学中药现代化工程中心,上海200237

年份：2013

卷号：35

期号：5

起止页码：57

中文期刊名：南京工业大学学报（自然科学版）

外文期刊名：Journal of Nanjing Tech University(Natural Science Edition)

收录：CSTPCD;;北大核心:【北大核心2011】；

语种：中文

中文关键词：合欢皮;总多糖;Box—Behnken;BP神经网络

外文关键词：Albizzia julibrissin Durazz ; total polysaccharide ; Box-Behnken design ; BP neural network（BPNN）

摘要：优化合欢皮多糖微波辅助提取工艺和建立BP神经网络模型。以多糖得率为指标,选取微波辐射时间、液固比和溶剂pH为自变量,采用Box-Behnken试验设计,结合响应面分析法(RSA),建立回归方程;在此基础上,建立BP神经网络(BPNN)模型。优化条件下的总多糖实际得率为3.648%,BPNN模型相对误差为1.042%,小于Box-Behnken设计模型相对误差(1.452%)。该BPNN模型预测性能良好,对工艺研究的开发具有一定的实用价值。
With polysaccharide yield as a index, microwave radiation time, liquid/solid ratio and pH as variables, Box-Behnken design and the response surface analysis （RSA） method were adopted to establish the regression equation. Furthermore, BP neural network （BPNN） model with good predictive perform- ance was established. The experimental yield of total polysaccharide was 3.648%, while the relative er- ror of BPNN was 1. 042%, thus it was less than that of Box-Behnken design model （ 1. 452% ）. The BPNN model was reliable and could be used in extraction process research and development.