Augmented Lagrangian Method-based Distributed Model Predictive Control

Distributed Model Predictive Control (DMPC) approach is viewed as one of the most efficient methods for solving a class of discrete-time linear systems with local and coupled global constraints. For the dual problem of Model Predictive Control, [1] proposed an algorithm based on the Alternating Direction Methods of Multipliers (ADMM), which utilizes some tightening rules of the coupled constraints and a finite-time consensus algorithm. Along this research direction, this paper considers solving the dual form of overall MPC optimization problem involving all systems by a set of augmented Lagrangian-based parallel splitting methods. Under some reasonable assumptions, the convergence of these algorithms, the performance of the closed-loop system including the recursive feasibility and stability can be guaranteed. Furthermore, we illustrate the performance of both algorithms via a numerical example.