Robust model predictive control via multi-scenario reference trajectory optimization with closed-loop prediction. (April 2021)

• Record Type: Journal Article
• Authors: Li, Hao; Swartz, Christopher L.E.
• Is Part Of: Journal of process control. Volume 100(2021)
• Page Start: 80
• Page End: 92
• Publication Date: 2021-04
• Subjects: Dynamic optimization -- Model predictive control -- Robust MPC -- Hierarchical control -- Closed-loop prediction; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2021.02.006
• Languages: English
• ISSNs: 0959-1524

Abstract

Abstract: This paper presents a two-layer control structure to address parameter uncertainty within a plant. The lower layer is formulated as a nominal MPC that computes control actions to regulate the underlying plant, and the upper layer computes optimal set-point trajectories for the lower level to track. The upper layer is formulated as an optimization problem that takes into account the closed-loop behavior of uncertain plant scenarios under the action of nominal MPC. The upper layer facilitates the lower layer in avoiding constraint violations and producing less conservative control actions by assigning time-varying set-point trajectories to the nominal MPC. The benefits of this approach are illustrated through application to linear single-input–single-output transfer function case studies, and a nonlinear multi-input multi-output evaporator process. Highlights: The paper presents a novel two-layer robust MPC paradigm. An upper level optimization problem computes set-point trajectories for nominal MPC. Utilizes predicted closed-loop response of plant scenarios under constrained MPC. Performance is demonstrated through application to linear and nonlinear case studies.