Traffic-aware hierarchical eco-driving approach for connected hybrid electric vehicles at signalized intersections

Eco-driving is considered one of the promising techniques for enhancing vehicle energy and traffic efficiency. However, most research efforts have focused on predefined traffic conditions, overlooking the impact of time-varying traffic states while being constrained by the computational capabilities of onboard controllers. To bridge this gap, leveraging the vehicle-to-cloud (V2C) technique, this paper proposes a traffic-aware hierarchical eco-driving approach for connected hybrid electric vehicles (HEVs) with joint speed planning and energy management strategy. In the cloud server, future traffic flow speed, serving as dynamic traffic constraints, is predicted based on a dynamic mode decomposition (DMD) algorithm. Subsequently, a global eco-speed planner employs chance-constrained programming to account for traffic prediction uncertainty and determines the ego vehicle's speed trajectory over a moving distance horizon. In the onboard controller, power allocation between the engine and electric motor is optimized using model predictive control (MPC) while tracking the global planned speed trajectory. The effectiveness of the proposed traffic-aware eco-driving approach is evaluated in both free-flow highway and signalized intersection scenarios. The numerical results demonstrate that, compared to the modified intelligent driver model (modified IDM), the proposed eco-driving approach significantly enhances fuel economy and driving comfort. Furthermore, in a multi-intersection scenario with traffic constraints, the proposed approach substantially improves the control robustness, achieving energy savings of 75.09 % relative to the modified IDM strategy.