一种实用的模型自适应MPC策略  

A practical model adaptive strategy for multivariable model predictive control

作　　者：张惠良[1] 曲德伟[2] 殷卫兵[2] 罗雄麟[1] 

机构地区：[1]中国石油大学(北京)自动化研究所,北京102249 [2]北京安稳优科技有限公司,北京102249

出　　处：《计算机与应用化学》2007年第4期483-488,共6页Computers and Applied Chemistry

摘　　要：在丙烯精馏塔的控制中,由于被控对象的非线性特性,采用线性模型的模型预测控制器难以保持良好的控制性能。本文提出基于系统稳态模型的模型自适应MPC策略,利用稳态模型在不同操作点上被控变量对操纵变量及扰动变量的相对变化率的变化,来刷新RMPCT控制器中各通道的模型增益。在模型输出对输入的相对变化率的计算中,使用主操作区间内的变化率以替代实际操作点的变化率,并采用设定模型变化域和控制模型变化频度的方法,以解决模型变化过大,与模型变换周期和RMPCT控制周期不协调而引起的系统不稳定等问题。实际投运效果表明:采用该控制策略,塔顶、塔底温度控制偏差与传统RMPCT比下降了一个数量级,有利于稳定与提高产品质量。It is difficult to keep favorable controlling performance of model predictive control （MPC） by linear model due to controlled plant nonlinearity in propylene rectifier. The Model Adaptive Strategy for Muhivariable Model Predictive Control based on system steady model is proposed, by the relative gain variation of controlled variables to manipulated variables and disturbance varia.bles based on the system steady model, the channel＇s gain of transfer functions are modified. In the gain calculation of the output to input parameters of the steady model, the gain＇s variations of dominating operating point is applied instead of that of current operating point, and by setting the modification size of the control models and limiting the control model modification frequency, the system unsteady controlling prob- lem is avoided induced by the large gain change of controlled model or the impact of the RMPCT controlling cycle and model modifying cycle. The results of online application show： the variations of controlled temperature are decreased one level by this strategy, and the product quality is improved.

关 键 词：模型自适应 模型预测控制 鲁棒模型预测控制技术 稳态模型 丙烯精馏塔 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程] TP273[自动化与计算机技术—控制科学与工程]