A cooperative energy management strategy based on region-based traffic grade prediction

The progression of the Internet of Vehicles (IoVs) has cultivated a comprehensive information environment, laying the groundwork for vehicle-environment adaptive control. Region-based traffic condition prediction offers valuable insights into overall traffic trends, aiding urban applications and vehicle-environment cooperation. In this study, an adaptive energy management strategy based on region-based traffic grade prediction (TGP-EMS) is proposed for plug-in hybrid electric vehicles (PHEVs) based on IoVs, strengthening the adaptability via accurately obtaining future region-based traffic conditions. Firstly, traffic data collected from volunteering vehicles are processed to generate representative traffic graphs and the traffic conditions are separated into several grades with distinct traffic attributes. Secondly, a differentiable pooling is integrated into a hierarchical deep learning framework to establish a region-based traffic prediction model termed Graph Pool. Thirdly, an optimal explicit solving method based on an instantaneous optimization algorithm is proposed and successfully applied to energy management. Moreover, the robustness of the introduced explicit solving algorithm is optimized based on the beetle antennae search (BAS) algorithm. Simulation results, accompanied by hardware-in-the-loop (HIL) tests, suggest that the introduced Graph Pool effectively captures spatial features across the entire traffic network, ensuring accuracy and consistency in predicting traffic conditions. Moreover, the TGP-EMS adeptly adjusts power distribution based on these predictions, showing an improvement of roughly 13.5 % compared to conventional rule-based energy management strategies.