管式催化反应分布参数过程建模与控制仿真研究

发布时间：2018-04-28 21:02

本文选题：分布参数系统 + PCA主元分析　；参考：《太原科技大学》2015年硕士论文

【摘要】：管式反应器体积庞大、内部温度高以及工况复杂,传统的基于集总参数模型的控制方法无法满足现代工业的高精度温控要求。如何基于分布参数系统模型精确控制催化反应的温度成为满足管式催化反应要求的关键问题。本文首先给出了催化棒-管式催化反应过程的PDE方程,针对分布参数系统无限维本质特点,研究了基于数据驱动的时空分离方法,采用PCA主元分析法进行时空分解同时降低模型维数,得到降维后的系统时间输出。然后,运用递推最小二乘法辨识出系统的ARX耦合模型,并采用多变量广义预测控制方法实现了控制器的设计,达到对温度的有效控制。由于多变量广义预测控制设计的复杂性,论文进一步对ARX耦合模型进行简化,辨识出解耦的ARX模型,将MIMOARX模型解耦为多个SISOARX模型,并设计了单变量广义预测控制器,更好地达到了设定的温度控制目标,简化了预测控制器的设计。仿真结果表明,对于弱非线性的分布参数系统,采用3个以上主元的PCA降维方法能够得到系统99%以上的能量；同时,采用广义预测控制进行ARX耦合模型和解耦模型的控制；能够实现分布参数系统的控制。基于多个SISO解耦模型,有利于提高控制精度,简化控制器的设计。
[Abstract]:Because of its large volume, high internal temperature and complex operating conditions, the traditional control method based on lumped parameter model can not meet the requirement of high precision temperature control in modern industry. How to accurately control the temperature of catalytic reaction based on distributed parameter system model has become the key problem to meet the requirements of tubular catalytic reaction. In this paper, the PDE equation of catalytic rod-tube catalytic reaction process is first given. According to the nature of infinite dimension of distributed parameter system, a data-driven spatio-temporal separation method is studied. The PCA principal component analysis method is used to decompose time and space and reduce the dimension of the model, and the output of system time after dimension reduction is obtained. Then, the ARX coupling model of the system is identified by recursive least square method, and the multivariable generalized predictive control method is used to realize the design of the controller, which can effectively control the temperature. Because of the complexity of the design of multivariable generalized predictive control, the paper further simplifies the ARX coupling model, identifies the decoupled ARX model, decouples the MIMOARX model into multiple SISOARX models, and designs a univariable generalized predictive controller. The temperature control target is better achieved and the design of predictive controller is simplified. The simulation results show that for weakly nonlinear distributed parameter systems, more than 99% of the energy can be obtained by using the PCA dimensionality reduction method with more than three principal components, and the generalized predictive control is used to control the ARX coupling model and the decoupling model. The distributed parameter system can be controlled. Based on multiple SISO decoupling models, the control accuracy is improved and the controller design is simplified.
【学位授予单位】：太原科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TQ032.4

【相似文献】