Hierarchical model predictive control

This paper deals with the problem of designing stable hierarchical control schemes with Model Predictive Control (MPC). Specifically, at any layer of the considered structure, a robust MPC regulator is used to compute the control variables. In turn, these variables are produced by the subsystems at lower layers and must fulfill the robustness constraints coming from the higher layers. By means of a negotiation procedure between the different layers and a suitable switching strategy, it is possible to achieve the overall closed-loop stability. Although the results reported here are still preliminary, the devised synthesis procedure is very general and opens the way to the treatment of many control structures of practical relevance.