Deep Q‐Network Algorithm‐Based Optimal Energy Management Strategy of Fuel Cell Bus Considering Intelligent Network Information

In this study, load reconstruction, speed planning methods, and Deep Q‐Network (DQN)‐based optimal energy management strategies (EMSs) are developed for an intelligent networked fuel cell bus. First, the EMS can be developed using the load reconstruction information, while the You Only Look Only (YOLO)v2 algorithm is used for passenger counting to achieve the load reconstruction. Then, the speed planning‐based energy‐saving control is proposed using traffic signals and distance information. Based on the developed MATLAB‐Prescan co‐simulation model and virtual driving experiment platform, simulation and experiment are performed to validate the proposed EMSs. To further reduce the hydrogen consumption of the target vehicle, the DQN algorithm‐based optimal EMS is developed. Furthermore, a real Linux board card system is applied to verify the traffic signal and distance information acquisition. Compared with the DQN‐based EMS with fixed load and manual driving, the EMS with the combination of load reconstruction and speed planning achieves 29.4% economic improvement.