In future urban power grids, hydrogen microgrids are expected to grow rapidly, providing ancillary services (frequency regulation, voltage regulation, demand-side response), while electric vehicles (EVs) will also become increasingly prevalent. However, the charging services of EVs introduce peak loads and other negative impacts on the grid. How to regulate the energy flows within clusters of hydrogen microgrids and manage the demand response for EV charging is key to improving the operation of both the power grid and transportation network. This paper first develops a price-based mechanism model to regulate hydrogen-electricity-transport energy flows. It then focuses on investigating the effectiveness of decision-making prices using Graph Neural Networks (GNN) and Physics-Informed Neural Networks (PINN) in regulating energy flows, comparing them with Deep Deterministic Policy Gradient (DDPG). The results show that the AI-driven pricing mechanism can simultaneously regulate grid voltages and traffic flows in the transportation network, offering new insights for developing practical control solutions in the future.