原油换热网络多变量预测控制

Multi-variable model predictive control of crude oil heat exchanger networks

针对原油换热网络输入输出变量多、传统的控制方法控制难度大等问题,本文建立了用于仿真的Simulink辨识模型.利用模型预测控制原理,采用旁路控制的方法,借助预测控制工具箱(MPC toolbox)实现对换热网络Simulink模型的多变量预测控制,实现了输入变量平稳、输出变量控制稳定、控制精确及动态响应时间短等优点.在本机控制的基础上,借助OPC工具箱实现不同PC机问MATLAB与PHD数据库的实时交换,达到实时控制的目的,进而实现了整个换热网络的实时多变量预测控制.

Aiming at the problems such as so many inputs and outputs,hard control using traditional control method in crude oil heat exchanger networks,an identified simulink model of heat exchanger networks is proposed for simulation in this paper.On the theory of model predictive control using MPC toolbox bypass control method is taken to realize the model predictive control of a multi-input and multi-output simulink model of crude oil heat exchanger networks.There are many advantages that it makes the input smoother,the output more stable,the control more accurate and the dynamic response time shorter.On the base of controlling in one computer,using OPC toolbox this paper realizes the exchange of database between MATLAB and PHD database system on time and completes the aim of controlling on time in different computers. At last it realizes the multi-variable model predictive control of all heat exchanger networks.