预测控制动态过程对控制目标可达性影响分析

Impact of Dynamic Process on the Reachability of Control Target of Model Predictive Control

预测控制作为应用最广泛的先进控制策略,在复杂的工业过程控制中占据重要地位。预测控制可行域的确定是系统实施控制的先决条件。为保证控制效果,系统输出变量要一直保证在可行域内,且要求在系统稳定的情况下可行域内均可达。在实际控制过程中,可行域内系统输出的稳态值与目标设定值存在稳态误差的情况时有发生。系统实际的可达区域与可行域存在差异。可行域的变化,势必对控制效果产生重要影响。因此,需要在系统稳态条件决定的可行域的基础上,分析预测控制的动态过程对系统可行域的影响。为避免影响,需要设定新的目标值以保证目标可达。本文将以映射的思想充分分析预测控制算法的动态过程,证明输出变量动态响应对系统可行性的影响,提出满足约束条件下充分可达的设定值选取规则。

As the most widely used advanced control strategy, model predictive control strategy occupies an important position in complex industrial process control. Determining the range of the feasible region is a prerequisite to control the system. To ensure the control effect, the output variables need to be controlled running in the feasible region, while all the value of the feasible region need to be reachable all the time. However, it happens easily in the actual industrial process that there are stable errors between output variables and set point values when the system comes to stable state. It means that the feasible region differs from the reachable region of the system. Changing of the feasible region has an effect on the control result certainly. Therefore, it＇s necessary to analyze the effect of dynamic process of the MPC to the feasible region, which is decided based on the constraints of the manipulated variables and output variables. For avoiding the influence of the dynamic route, a new reasonable set point need to be set to make sure that the output values are reachable, and the system states are controllable as well. In this paper, on the basis of the spatial geometry mapping, the influence is fully analyzed about the P steps prediction formula derivation for state space model. While, it＇s also proved that the dynamic response of output variables have effect on the feasibility of the system. Based on all above, aiming at the situation for target values unreachable in feasible region, a new method of selecting set point is proposed to ensure the reachability and feasibility on the condition of satisfying constraints.