Multi-agent-based energy management strategy for a novel electric-flywheel hybrid electric vehicle

To address the high motor power requirements and complex spatial arrangement of traditional new energy commercial vehicles, this paper proposes an electric-flywheel hybrid electric vehicle configuration. The configuration features a dual-power source and dual-motor design, facilitating power synergy among the electric flywheel, control motor and drive motor through a dual planetary gear system. Considering the multiple power sources and working modes of the vehicle, this study offers a comprehensive analysis of its power system. A multi-agent-based energy management strategy is proposed, employing the soft actor-critic algorithm to regulate torque distribution between the drive motor and control motor, and the proximal policy optimization algorithm to optimize mode switching. The agents collaborate through a centralized training and decentralized execution strategy to achieve multi-objective optimization. Results indicate that the proposed method improves velocity tracking and enhances flywheel energy utilization, achieving up to a 14.37% reduction in battery energy consumption compared to conventional EV, Rule-min EMS, PPO-min EMS, and Rule-SAC strategies. Furthermore, the real-vehicle driving cycle further validates the effectiveness of this strategy. The electric-flywheel hybrid powertrain and multi-agent strategy demonstrate significant potential for application and promotional value in vehicle control.