Currently, electric vehicle charging stations in urban power grids are experiencing rapid large-scale growth. Integrating renewable energy and hydrogen energy into charging stations to create hydrogen charging stations has become a future trend. These hydrogen charging stations are interconnected via power converters to form microgrid clusters. Coordinating the hydrogen-electricity-transportation energy flow within the coupled distribution network-microgrid cluster system is challenging due to the system’s complex topology. Regulating energy flow under dynamic topology reconfiguration scenarios is particularly difficult. This paper proposes a control algorithm based on deep reinforcement learning for topology reconfiguration in distribution network-microgrid cluster systems, enabling integrated decision-making for topology reconfiguration, energy flow control, and new energy vehicle dispatch. First, a physical model of the distribution network-microgrid cluster-transportation network is established. Second, a deep reinforcement learning control model is developed, with actions including topology reconfiguration. Finally, results show that the integrated control strategy can not only regulate energy flow but also physically alter connection patterns through topology reconfiguration, thereby further controlling energy flow.