∞-norm formulation of model predictive control problems

A general mathematical programming framework for multivariable model predictive control problems is presented. Existing model predicitive formulations, based on 1 and 2-norms (Quadratic Dynamic Matrix, Linear Dynamic Matrix, and Model Algorithmic Control) are discussed within the general framework. A new formulation, based on the ∞-norm, is introduced and developed. The characteristics of this formulation, which make it particularly attractive for the control of chemical processes, are discussed. A simple procedure for the design and implementation of constrainted control systems, based on the Internal Model Control structure, is proposed. A simple example is included to demonstrate this procedure.