Distributed Model Predictive Control for Complex Interconnection Systems

This paper presents a new communication-based distributed model predictive control (DMPC) scheme for spatially coupled systems based on linear matrix inequalities (LMI). The large-scale system in discussion is formed by several interconnected subsystems with mixed structures connected in parallel, serial and cycle configurations. Each subsystem is regulated by a dedicated model predictive controller (MPC). These local con-trollers communicate with their interconnected partners to up-date the coupling information for their online optimisation problems. An interconnection stability condition based on the dissipative systems theory is derived, taking the coupling effects into consideration. The stability condition is then converted into dynamic stability constraints for individual MPCs in a distributed manner. Upon their satisfaction, the closed-loop large-scale system with these distributed MPCs is stabilised accordingly. The main results are illustrated in simulation for a three-unit system.