Economic Optimization and Control Based on Multi Priority Rank RTO and Double Layered MPC

Abstract

A prevailing hierarchical structure in industrial process optimization and control includes three levels, i.e., a real time optimization (RTO) level, a double layered model predictive control (MPC) level (which is composed of a steady-state target calculation (SSTC) layer and a dynamic control layer), and a distributed control level. In this paper, a multi priority rank RTO algorithm, in which a new variable is introduced to uniformly express the set points, is presented to get optimal set points according to their importance levels. In order to guarantee the feasibility of the dynamic control layer during tracking the steady-state targets calculated in SSTC layer, the region of attraction is added into the SSTC layer as additional constraints, hence, the steady-state targets can be calculated online and transmitted to the dynamic control layer at each instant to guide the state to achieve the steady-state state gradually. The effect of the above methods are illustrated through an example.

Enhancing dynamic operation optimization feasibility for constrained model predictive control systems

“…As standard in MPC for tracking set-points literature [1,23], two requirements including set-point tracking and relaxing constraints are considered in MPC. In the present work we extend the formulation presented in [11] to take into account the dynamic constraint as well as the terminal invariant constraints to relax the terminal constraints and ensure dynamic constraints violation minimization for the aim of ensuring feasibility.…”

“…The constraints related to the terminal state and transition steady-state is (8i) and (11) for i ∈ I 1∶p x , resulting in the following condition: (12) which corresponds to (8i) with Optimization problem considers dynamic operation to ensure at least one feasible solution in dynamic control under the aims of target tracking and constraint violation minimum. In addition, compared to the terminal equality constraint in [11], the terminal set constraint proposed in this paper relaxes the terminal constraints and provides larger feasible region. We define the set of feasible control sequence  s N (x, z r ) ∶= {u|∃x (k) s.t.…”

“…Nevertheless, the stability of the literatures [20][21][22], which adopts hierarchical control framework, have not been established. The terminal equality constraint has been applied in [11] without stability analysis, which is extremely restrictive compared to the single-layered MPC with the adoption of terminal invariant set constraints.…”

A novel modeling and controlling approach for high‐order nonlinear systems

“…Now we consider the tracking problem of nonlinear system (4). First rewrite the characteristic model (10) in terms of tracking error dynamics.…”

“…To gain better control performance, many solutions have been advanced in recent years. Model predictive control [2,3,4], optimal control [5,6,7], and output feedback control [8,9] have been successfully employed to tackle the stabilizing or tracking problems of complex nonlinear systems.…”