基于BP神经网络的pH对解淀粉芽孢杆菌Q-426发酵影响的研究

Effect of pH on the batch fermentation process of Bacillus amyloliquefaciens Q-426 based modeling with BP neural network

为控制解淀粉芽孢杆菌Q-426发酵过程,研究了在3.7 L发酵罐中pH对菌株Q-426发酵过程的影响。选取发酵时间、pH作为自变量,菌体浓度、底物浓度、发酵产物活性作为目标量,通过构建自变量矩阵和参考序列,构建了基于BP神经网络的菌株Q-426发酵过程的预测模型。通过运用MATLAB神经网络工具箱进行训练,得出优化网络模型,并根据建立的模型进行预测。将预测值与实测值对比,拟合及预测的平均相对误差均在4%以内,说明该模型有较好的适用性。

The paper aims to control the batch fermentation process of Bacillus amyloliquefaciens Q-426. To this end, the effect of pH on the fermentation of B. amyloliquefaciens Q-426 was investigated in 3.7 L fermentation tank. We present a model based on BP neural network to predict the cell concentration（C）, substrate concentration（S）, and product activity （P） according to the fermentation time and pH data. The BP neural network prediction model is trained using the fermentation time and pH value as inputs and the C, S and P as output. The network model was performed by the BP neural network toolbox in MATLAB. By comparing the fit/predicted value and experimental value, the relative errors （mean value） of C,S and P in our prediction model is within 4 %.Therefore, our study shows the feasibility of the prediction model for C,S and P using BP neural network.