Hierarchical optimization of SAC-driven speed planning and energy management in intelligent fuel cell hybrid vehicle platoons under complex traffic environments

Eco-driving and connected automation technology-driven platoon cooperative control techniques show significant potential for improving energy efficiency and traffic flow optimization. However, vehicle state fluctuations and road condition uncertainties in dynamic traffic environments pose serious challenges to fuel cell hybrid electric vehicle (FCHEV) platoon control. To this end, this study proposes a hierarchical co-optimization architecture that integrates speed planning and energy management to cope with complex working conditions. In the high-level speed planning layer, the ISSA-CNN-BiLSTM-ATT network is used to achieve accurate prediction of the speed of the lead vehicle, and the Soft Actor-Critic (SAC) algorithm is used for speed planning to dynamically adjust the inter-vehicle spacing in order to balance the safety, ride comfort, and economy; in the bottom energy management layer, the L2 regularized double-delay depth deterministic strategy gradient (L2-TD3) algorithm is introduced to optimize the fuel cell and battery power allocation, simultaneously considering the energy degradation cost and state-of-charge (SOC) stability. The simulation results show that the prediction error of the speed prediction strategy is reduced by 67.32 %, the speed planning strategy reduces the platoon demand power by 19.25 % and the global cost by 16.96 % under the prediction condition, and the energy management strategy reduces the global cost by 15.67 % and the SOC fluctuation by 1.69 % under the FTP75 condition.