Distributed Satisficing MPC With Guarantee of Stability

This technical note develops conditions for the closed-loop stability of a new class of cooperative distributed model predictive control, called satisficing, whereby the relative importance of each controller changes dynamically depending on the prevailing conditions instead of being adjusted a priori. In the satisficing approach, the less satisfied controller will gain more importance. The main properties of the proposed approach are illustrated with an unstable dynamic system composed of three subsystems.