利用神经网络预测头孢菌素C的生物合成  

Prediction of the cephalosporin C biosynthesis by a back propagation neural network model

作　　者：冀志霞[1] 储炬[1] 庄英萍[1] 张嗣良[1] 

机构地区：[1]华东理工大学生物反应器工程国家重点实验室,上海200237

出　　处：《中国抗生素杂志》2007年第1期17-21,共5页Chinese Journal of Antibiotics

基　　金：上海市科委重点科技攻关项目(034319220)

摘　　要：利用头孢菌素C发酵过程积累的数据,建立BP神经网络预估模型,实现以发酵前期的菌浓和pH对效价的预测,将此模型应用于生产实际,分别通过倒种和改变培养基中碳源组成的方法,使头孢菌素C的合成水平分别提高了11.8%和15.7%,表明模型具有较好的预测功能。The back propagation （BP） neural network model was set up by cephalosporin C fermentation data, and the productivity was forecasted by PMV and pH changing tendency of early fermentation phase. The model was proved to be effective and with good prediction capacity. By increasing inoculum and changing medium carbon source, the productivity of cephalosporin C fermentation was further increased by 11.8% and 15.7%, respectively, which was well predicted by the established model.

关 键 词：头孢菌素C 反向传播 神经网络 预测 

分 类 号：R978.11[医药卫生—药品]