New system coordinates electric and hydrogen vehicles with energy use

A dual-agent framework manages vehicle scheduling and energy dispatch in electric-hydrogen-transport systems.

A new dual-agent reinforcement learning framework manages electric-hydrogen-transport systems (EHTSs). This system coordinates vehicle scheduling and multi-energy dispatch.

The vehicle-scheduling (VS) agent uses proximal policy optimisation (PPO) and a Plackett-Luce policy. It learns vehicle-priority permutations. These are mapped to charger and refueller assignments through a greedy allocation.

The energy-dispatch (ED) agent uses twin delayed deep deterministic policy gradient (TD3). This agent controls energy-flow and continuous storage.

Upstream vehicle scheduling determines service allocation and order. This shapes the electricity and hydrogen demands for downstream energy dispatch.

A three-stage training mechanism uses layer-wise pretraining and concurrent fine-tuning. This helps mitigate distribution shifts caused by scheduling.

Year-long case studies show this framework achieves the lowest vehicle-service delay and energy cost among 11 evaluated methods.

Key facts

  • The framework uses a dual-agent reinforcement learning approach.
  • The system coordinates vehicle scheduling with multi-energy dispatch.
  • The framework achieved the lowest vehicle-service delay and energy cost in year-long case studies.

All sections

Edition