An Eco-driving Control Strategy for Connected Electric Vehicles at Intersections Based on Preceding Vehicle Speed Prediction

At intersections equipped with traffic lights, the control strategy for eco-driving vehicles must encompass not only information about the road environment but also the constraints of the preceding vehicle and the corresponding impact on the target vehicle. First, a hybrid preceding-vehicle speed prediction method integrating a simple recurrent unit (SRU) network with physical information and a back-propagation (BP) network based on historical data is proposed. This method enables the fast and accurate prediction of the speed of the preceding vehicle with minimal data requirements. Second, prediction information is incorporated into the eco-driving strategy that considers both economy and safety at traffic intersections. A rolling approximate dynamic programming method with prediction features is proposed to solve the aforementioned time-domain optimization problem. Finally, an intelligent connected hardware-in-the-loop (HIL) simulation platform is constructed to verify the method formulated in this study. The simulation results illustrate that the proposed method significantly reduces energy consumption by 11.3% to 41.67% compared to the Adaptive Cruise Control (ACC) algorithm, and it also achieves energy savings ranging from 4.02% to 33.51% in comparison to the Model Prediction control (MPC) algorithm. Moreover, the algorithm is capable of satisfying real-time requirements, thereby significantly enhancing the practical application potential of the proposed system.