Distributed Learning Model Predictive Control with Coupled Constraints and Communication Noise

This paper presents a novel distributed learning model predictive control (DLMPC) scheme for linear systems with coupled constraints under communication noise. In contrast to conventional distributed MPC schemes, the proposed scheme combines data-driven learning techniques with a novel noisy alternating direction multiplier method (NADMM). By learning from previous iterative trajectories, the proposed scheme enhances performance over time. Additionally, it addresses coupling constraint problem between subsystems and mitigates the impact of communication noise. The effectiveness of the proposed scheme is verified through a simulation example.