Cloud-Edge Cooperative Distributed NMPC for Disturbance Rejection in Heterogeneous Vehicle Platoons

This paper presents a novel cloud-edge cooperative disturbance rejection framework for heterogeneous vehicle platoons, addressing the critical challenges posed by multifaceted disturbances and the requirement for string stability. The proposed architecture strategically decomposes the control task: at the vehicle edge, a finite-time disturbance observer (DOB) is designed to enable fast and accurate estimation and compensation of matched disturbances. Concurrently, a distributed nonlinear model predictive controller (DNMPC) is deployed in the cloud, which incorporates a robust constraint tightening strategy to systematically manage uncompensated unmatched disturbances and residual uncertainties. To proactively suppress error propagation along the platoon, a disturbance string stability (DSS) constraint is explicitly embedded within the DNMPC formulation. Furthermore, an adaptive event-triggered mechanism (AETM) is developed to dynamically schedule control updates, significantly reducing computational and communication loads while ensuring safety and avoiding Zeno behavior. Rigorous theoretical analysis establishes the recursive feasibility and regional input-to-state practical stability (ISpS) of the closed-loop system, as well as the string stability of the platoon. Simulation results validate the proposed framework's superior performance in robust tracking and operational efficiency under heterogeneous disturbance conditions.