Hierarchical eco-driving of connected hybrid electric vehicles: Integrating predictive cruise control and cost-to-go approximation-guided energy management

This study introduces a real-time hierarchical eco-driving control strategy specifically designed for hybrid electric vehicles (HEVs). The approach incorporates a predictive cruising control technique emphasizing economy (Eco-PCC) for the purpose of vehicle speed trajectory planning, in conjunction with an energy management strategy relying on approximations of optimal cost-to-go values (ACG-EMS) to guarantee effective power allocation. Within this framework, the velocity of the host vehicle is established utilizing predictive models, which encompass forecasts of preceding vehicle speeds derived from a conditional linear Gaussian (CLG) model, statuses of traffic lights, and an economic cost estimation indirectly derived through the least absolute shrinkage and selection operator (LASSO). The design of ACG-EMS predominantly encompasses real-time power allocation employing a neural network model trained with offline optimization outcomes derived from dynamic programming (DP). This is supplemented by proportional-integral (PI) correction to align with the actual operational dynamics of the engine. Simulation and hardware-in-the-loop (HiL) experiments conducted in virtual interconnected environments within CarMaker validate the efficacy of the proposed control strategy. Comparative analysis against benchmark strategies based on predictive cruising control (PCC) and equivalent consumption minimization strategy (ECMS) showcases enhancements in energy efficiency ranging from 7.7% to 6.7% and 6.5% to 2.6% under different driving conditions, respectively.