Energy-efficient distributed model predictive control with communication delay compensation for vehicle platooning

Maintaining string stability and energy efficiency in vehicle platoons under stochastic V2X (Vehicle to Everything) communication delays presents a significant challenge. Communication latency often leads to inaccurate reference trajectories, which degrade tracking precision and fuel economy. To address this, this study develops an ecological communication delay compensation distributed model predictive control (ECO-CDMPC) strategy featuring a separate trajectory prediction architecture. Unlike existing methods, the proposed framework independently reconstructs the trajectories of the leader and neighboring vehicles using long short term memory (LSTM) networks to synchronize asynchronous data streams. A distinctive innovation of this work is the theoretical coupling of the learning module with the control law, where the LSTM prediction error bound is explicitly integrated into an input to state stability (ISS) analysis. This integration establishes a rigorous safety boundary, allowing the multi objective cost function to maximize fuel savings without compromising platoon stability. Simulation results across diverse network topologies validate that the proposed approach effectively mitigates the impact of stochastic delays. The ECO-CDMPC achieves fuel savings ranging from 0.7% to 64.9%, where the latter represents a scenario dependent upper bound achieved under severe communication delays, while maintaining strong tracking performance and improved driving comfort.