智能预测模型在多粘芽孢杆菌发酵中的应用

Research on paenibacillus polymyxafermentation of intelligent forecasting model

多粘芽孢杆菌发酵是一个极其复杂的生物反应过程,其诸多参数具有动态非线性。传统的发酵实验都是在不断地尝试中进行的,如果能预测发酵过程中的主要参数则会大大提高实验效率。用一种智能预测模型,即将Kohonen网络、Elman神经网络和粒子群优化算法有机结合,可以预测发酵过程中的主要参数。该模型的仿真实验结果符合多粘芽孢杆菌发酵的动力学特点,实现了部分参数的有效预测。研究结果表明该智能预测模型不但能够综合各种单一预测模型的优点,而且能够随时间的推移其内在结构不断变化,适用于多粘芽孢杆菌发酵过程的参数预测和特性优化。相对于传统预测方法,提高了预测效率。

PaenibaciUus polymyxa fermentation is complex process of bioreaction. Its parameters are dynamic and non-linear. The tradi- tional experiment of fermentation was studied with the Kohonen network, Elman network and the particle swarm optimization algorithm combined to establish the intelligent forecasting model in this paper. The simulation results of the model were accord with kinetic fea- ture of paenibacillus polymyxa fermentation, and the predicted values of some parameters were reasonable. The experimental results showed that not only can the intelligent forecasting model sum up the merit of kinds of single model, but also can change the interior configuration. Compared with the traditional prediction method, the intelligent forecasting model enhanced the forecasting efficiency.