Multistage Prediction-Based Eco-Driving Control for Connected and Automated Plug-In Hybrid Electric Vehicles
Pengxing Zhu, Jianjun Hu, Jiajia Li, Feng Xiao, Zhicheng Sun, Hang Peng
Abstract
Eco-driving technology is an effective way for connected and automated plug-in hybrid electric vehicles (CAPHEVs) to achieve energy conservation and emission reduction. However, the constraints of traffic flows and signal lights at intersections pose serious challenges to eco-driving of CAPHEVs. Therefore, an eco-driving control method based on dynamic consumption prediction is proposed. A hierarchical control scheme is designed that divides eco-driving problem into global velocity planning and energy management. Firstly, a multistage time-series prediction model is constructed based on two long short-term memory networks, which is applied to predict optimal energy consumption cost. Secondly, the velocity trajectory in global speed planning is optimized by combining the multistage consumption prediction model with dynamic programming. Finally, an energy management strategy considering energy consumption and pollutant emission is proposed. The real-world environment information is applied to construct the virtual simulation environment and verify the proposed method. The simulation results demonstrate that the proposed multistage prediction model has a better prediction performance than the existing popular models. Meanwhile, the total cost savings by the proposed scheme range from 13.5% to 48.7%. Particularly, the proposed method not only significantly reduces emissions, but also guarantees driving safety, and satisfies the different demands for travel time.