Fuel Economy-Oriented Parameter Tuning for Platoon Distributed Model Predictive Control Based on Bayesian Optimization

The platoon control of the longitudinal motion enables vehicles to achieve system-wide benefits for energy efficiency and road throughput. The fuel consumption analysis of vehicle platooning either only considered how to set the vehicle spacing strategy, or only reduced the fuel consumption locally in the given spacing strategy from the controller's perspective. There is a separation between vehicle distance target setting and vehicle distance target tracking. Hence, it is necessary to consider the fuel consumption analysis in both the steady-state process and the dynamic response process. The former is related to the spacing strategy parameters and the latter is about the weights and the prediction horizon length in model predictive controller. Hence, oriented by fuel economy, we utilize Bayesian optimization to determine the appropriate inter-vehicle spacing policy and distributed model predictive control parameters for the platoon under different road conditions. The parameter tuning is performed offline from a global perspective when the tracking target of the platoon changes. Simulation results indicate that the proposed method shows superiority in economic performance.