青霉素发酵过程建模研究

青霉素发酵过程是一种具有非线性、时变性的复杂生化反应系统,由于一些生物参数在线检测困难,许多生化过程的代谢途径尚不明确,难以建立精确数学模型.而神经网络具有非线性、多变量、自学习、并行处理等特点,用于非线性系统的建模具有无可比拟的优势.因此,以青霉素发酵过程生化机理模型产生的数据为样本,训练RBF神经网络,建立了基于RBF神经网络的发酵过程模型.该模型可用于发酵过程中状态变量的估算与预测,并且可估计底物、产物、菌体浓度的变化趋势,对实际工作具有指导意义.

Modeling of the Penicillin Fermentation Process

The penicillin fermentation process is a complex biochemical system with time varying, nonlinear and uncertainty feature. It is difficult to obtain an accurate mathematics model of the penicillin fermentation process, since some of the life-form parameters can not be measured on-line and the metabolizing approaches of many biochemical process are not well known. Neural networks have many features such as nonlinear, multivariable and self-learning ability, so they are effective in modeling of nonlinear system. The model of the penicillin fermentation process has been established by using data created by the penicillin fermentation process mechanism model to train neural network, and simulations were made to prove this fermentation model. Using this model, the tendency of state change of the fermentation process can be known in advance. This is very useful in practice.