基于误差补偿的谷氨酸发酵过程模型预测控制研究  

Model Predictive Control Based on Error Compensation for Glutamic Acid Fermentation Process

作　　者：关长亮[1] 王贵成[1] 

机构地区：[1]沈阳化工大学信息工程学院,辽宁沈阳110142

出　　处：《沈阳化工大学学报》2016年第1期70-75,共6页Journal of Shenyang University of Chemical Technology

基　　金：国家自然科学基金资助项目(61104093)

摘　　要：谷氨酸发酵过程是一种具有高度的非线性、强烈的时变性和大滞后等因素的生化过程.本文将神经网络和预测控制算法相结合,用于解决谷氨酸发酵过程非线性控制问题.首先,通过神经网络实现非线性模型的预测能力;然后,引入动态矩阵控制,建立基于误差补偿的谷氨酸神经网络预测系统,通过滚动优化﹑反馈校正﹑误差补偿进行多步预测,最终构建谷氨酸发酵过程多变量非线性控制系统.仿真结果表明:该方法能够实现对谷氨酸发酵非线性过程的有效控制.Glutamic acid fermentation process is biochemical process with high nonlinearity, strong time- varying and large lag,its internal mechanism is very complex. Nonlinearity exists in most of real subjects for fermentation process industryAnd the nonlinearity characters do not allow for linear approximation. Since some features of high nonlinear system cannot be instead of linear equation approximately, it must a- dopt nonlinear analysis technology. The paper combines both neural network and predictive control, which is to solve the problem of nonlinear control. Firstly, it realizes the predictive ability of nonlinear model by neural network. Then, it brings in dynamic matrix control to establish the prediction system of neural net- work based on error compensation glutamic acid. The multistep prediction is carried out by roiling optimi- zation, feedback compensation and error compensation. Finally, it is constructed the multivariable nonlinear control system for glutamic acid fermentation. Simulation results show that the method can realize the val- id control for for the nonlinear process of glutamic acid fermentation.

关 键 词：谷氨酸发酵过程 非线性系统 神经网络 多变量 动态矩阵控制 

分 类 号：TP23[自动化与计算机技术—检测技术与自动化装置]