论文部分内容阅读
用神经网络描述未知的反应动力学参数，结合反应器物料平衡方程，提出了生化过程的神经网络组合模型。并提出了特别适合微生物发酵过程的Ｍｏｎｏｄ饱和型和基质抑制型的神经元传递函数。在Ｈｅｂｂ学习的基础上，引入教师指导信号，提出了神经网络误差一次反向传播的快速学习算法。将此组合模型用于某流加发酵过程状态变量和动力学参数的在线估计，仿真研究获得了满意的结果。组合模型具有训练速度快、预测精度高等优点，为动力学结构未知的生化过程模型化提供了一条新的途径。
Using neural network to describe the unknown reaction kinetics parameters, combined with the reactor material balance equation, a biochemical neural network combination model was proposed. The monod saturated and matrix-suppressed neuronal transfer functions that are particularly suitable for microbial fermentation processes are proposed. On the basis of Hebb learning, the teacher guidance signal is introduced and a fast learning algorithm of neural network error once backpropagating is proposed. The combined model was applied to the on-line estimation of state variables and kinetic parameters in a flow-fed fermentation process. The simulation results obtained satisfactory results. The combined model has the advantages of fast training speed and high prediction accuracy, which provides a new approach for the modeling of biochemical processes with unknown kinetic structure.