作　　者： ; ; ;

机构地区： 广东食品药品职业学院

出　　处： 《海峡药学》 2014年第2期15-17,共3页

摘　　要： 目的探讨多元非线性回归与BP神经网络在香菇多糖提取工艺研究中的应用;方法使用香菇多糖提取工艺中提取时间、提取温度、料液比、醇析乙醇量和多糖提取率的实验结果,分别建立多元非线性回归方程和BP神经网络;结果多元非线性回归拟合mse为0.1483,预测误差为1.44%;BP神经网络拟合mse为0.1474,预测误差为1.29%;结论多元非线性回归与BP神经网络均可用于多因素的非线性模型建立。 OBJECTIVE To discuss the application of multi-nonlinear regress analysis and back-propagation neural network in the research of extracting lentinan. METHODS The multi-nonlinear regress analysis and BP neu- ral network with the data of extracting time, extracting temperature, ration of lentinan to water and amount of ethanol on alcohol precipitation were builded up. RESULTS The mes of multi-nonlinear regress analysis was 0. 1483 and predictive encoding was 1.44%, BP neural network was 0. 1474 and predictive encoding was 1.44% ; CONCLU- SION Multi-nonlinear regress analysis and BP neural network are feasible in building up the nonlinear model.

关 键 词： 多元非线性回归 神经网络 提取工艺 香菇多糖

领　　域： [化学工程]