A Distributed Version of Han's Method for DMPC of Dynamically Coupled Systems with Coupled Constraints*

Most of the literature on Distributed Model Predictive Control (DMPC) for dynamically coupled linear systems typically focuses on situations where coupling constraints between subsystems are absent. In order to address the presence of convex coupling constraints, we present a distributed version of Han's parallel algorithm for a class of convex programs. The algorithm we propose relies on local iterative updates only, instead of using system-wide information exchange as in Han's original algorithm. The new algorithm is then used to develop a new distributed MPC method that is applicable to sparsely coupled linear dynamical systems with coupled linear constraints. Convergence to the global optimum, recursive feasibility, and stability can be established using only local communications between the subsystems.