Set Point Optimization in Multivariable Constrained Predictive Control

Abstract In this paper a procedure is presented to compute, at each sampling time, optimal future value of the controlled variables that can be effectively reached by the process in a given time horizon, taking into account the process dynamics and constraints on inputs and outputs. The optimality is considered in relation to an economic index involving the process variables. These values of the controlled variables are taken as the desired optimum set points, applying a rolling horizon policy. To illustrate the behaviour the system, an application example with a two-inputs, three outputs simulated non-linear chemical reactor is presented.