基于BP神经网络和遗传算法的面包酵母高密度发酵培养基优化  

Optimization of baker's yeast high density fermentation medium by optimized BP neural network based on genetic algorithm

作　　者：谈亚丽[1] 李啸[1,2] 邹嫚[2] 张江[1] 姚娟[2] 李知洪[2] 俞学锋[2] 

机构地区：[1]三峡大学化学与生命科学学院,宜昌443003 [2]安琪酵母股份有限公司,宜昌443003

出　　处：《工业微生物》2013年第1期64-68,共5页Industrial Microbiology

摘　　要：为实现面包酵母的高密度发酵培养，构建一个BP神经网络模型，用于回归面包酵母高密度发酵培养基中显著影响因子与茵体密度之间的非线性关系，并在此基础上结合遗传算法进对此模型进行全局寻优，得到关键因子最佳浓度分别为：葡萄糖52．3g／L，酵母浸出粉10．4g/L，（NH；）2S041．9g／L。采用此优化配方进行摇瓶培养，所得茵体密度为3．95×10^8个／mL，比对照提高了61．2％。结果证实了人工神经网络的模拟和预测功能在微生物培养基优化方面有一定应用价值。In order to fulfill the high density cultivation of baker＇s yeast, the back-propagation neural network was adopted to construct a nonlinear predictable model which suggested the relationship between the key factors of the culture medium and the biomass of baker＇s yeast. And then the global optimization on this model with the genetic algorithm was conducted. Finally the optimal dose of these significant factors was obtained： glucose 52.3 g/L, yeast extract powder 10.4 g/L, （NH4 ）2S041.9 g/L. Using this optimal medium, the biomass of the baker＇s yeast cultivated in shake flasks was as high as 3.95 10S/mL, increased by 61.2% compared with that of the primitive culture medium. It demonstrated that the application of artificial neural network in the optimization of microbiological culture media was feasible and efficient.

关 键 词：面包酵母 高密度培养 BP神经网络 遗传算法 发酵优化 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TQ926.1[自动化与计算机技术—控制科学与工程]