The importance of cooperation and consistency in two-layer Model Predictive Control

• Published in: 2016 17th International Carpathian Control Conference (ICCC) pp. 825 - 830
• Main Authors: Zanoli, Silvia Maria, Pepe, Crescenzo
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2016
• Subjects: Computer architecture; constraints; Control systems; dynamic optimization; Economics; energy saving; environmental impact; Model Predictive Control; Optimization; Predictive control; Softening; steady state optimization; Steady-state
• DOI: 10.1109/CarpathianCC.2016.7501210

In this paper a two-layer linear Model Predictive Control strategy is proposed. The problem of establishing a consistency between the mathematical formulations of the two layers is analyzed. At the lower layer, a basic Model Predictive Control module computes the future control moves taking into account predictions on the manipulated and controlled variables. In the quadratic optimization problem that characterizes this module, constraints over the prediction and control horizons and steady state optimal targets need to be properly set. At this purpose, at the upper layer of the control system architecture, a Targets Optimizing and Constraints Softening module has been introduced. This module is characterized by the minimization of a linear cost function, subject to linear inequality constraints. Its objective is the search of optimal steady state targets, assuring, for example, an economic and/or chemical optimal operative condition, while meeting steady state process constraints. A solution that guarantees the consistency between the formulations of the two layers is proposed, defining a cooperation between them. Practical results deriving from the introduction of the proposed architecture in a control system that drives a pusher type reheating furnace located in an Italian steel plant have proven the soundness and the reliability of the proposed approach.