Stability conditions for decentralized model predictive control under packet drop communication

Abstract

We propose a decentralized model predictive control (MPC) design approach for possibly large-scale processes whose structure may not be dynamically decoupled. The decoupling assumption only appears in the prediction models used by the different MPC control agents. In [1] we presented a sufficient criterion for analyzing a posteriori the asymptotic stability of the process model in closed-loop with the set of decentralized MPC controllers. The communication model among neighboring MPC controllers was supposed faultless, so that each MPC could successfully receive the information about the states of its corresponding submodel. Here we present a sufficient condition for ensuring closed-loop stability of the overall closed-loop system when a certain number of packets containing state measurements may be lost.