Some Feasibility Issues Related to Constrained Generalized Predictive Control

Abstract This paper analyzes the feasibility of the generalized predictive control law under constraints on the input, output or other auxiliary signals that depend linearly on the system variables. These constraints are formulated as sets of linear equalities or inequalities; the control sequence is therefore elaborated based on a quadratic optimization problem. The feasibility issues are related on one hand to the well posedness feature, and on the other hand to the compatibility with the set-point constraints. The prediction of the feasibility is of great interest from this point of view and necessary feasibility conditions are presented. Two possible approaches are followed, one strictly related to the specific set-point and the second, more general, examines the geo-metrical description of the optimization domain. The main practical advantage is that all the results are based on off-line numerical procedures offering qualitative information prior to the effective implementation.