Abstract:

The model predictive control is widely used in real process because of its decoupling and strong robustness. In the past studies, the effects of disturbances on model predictive control were mostly ignored. For the measurable disturbances whose dynamic rules are known, the influence of measurable disturbances on controlled variables can be predicted, and thus the measurable disturbances can be used as feed forward variables in model predictive control. The introduction of feed forward variables will affect the control performance of the system. The result solving directly under no analysis cannot reflect the actual effect of model predictive control. From the viewpoint of feasible region, this paper studies the feasible region changes caused by feed forward variables through the geometric forms. The convex space method is used to determine the size of the feasible region by solving the feasible region of the vertex set and obtain the effects of feed forward variables on the feasible region. The simulations verify the validity of the method in this paper.

Key words: process control, model predictive control, feed forward, feasible region

摘要：

实际生产过程中，预测控制因其解耦性能和强鲁棒性得以广泛使用。在预测控制的研究中大都忽略控制过程中的干扰作用。对于控制过程中存在的可测且变化规律已知的干扰作用，干扰对输出的影响具有一定的可预见性，可通过在预测控制器中引入前馈的方法加以利用。前馈变量的引入会对系统的控制效果产生影响，如果不先对其影响进行分析而直接求解优化，最终结果不能反映预测控制的实际效果。本文从可行域的角度出发，通过几何表现形式，直观分析前馈变量的引入造成的可行域变化；进一步使用了凸空间的思想，通过求解可行域的顶点集合来确定可行域的大小，进而得出前馈变量对系统可行域的影响效果，通过仿真验证了本文方法的有效性。

关键词: 过程控制, 预测控制, 前馈, 可行域