To address the renewable energy accommodation challenge caused by the large-scale integration of distributed renewable energy into rural distribution networks, a data-knowledge hybrid-driven electricity-hydrogen coordinated dispatch strategy is proposed. Firstly, XGBoost is used to extract key features, which are then input into a BiLSTM model to forecast the hydrogen refueling demand of hydrogen fuel cell vehicles, providing accurate data boundaries for coordinated dispatch. Then, a bi-level rolling optimization framework coupling the power and transportation networks is developed. The upper level establishes an energy hub model considering hydrogen transportation delays to determine the optimal dispatch strategy of the distribution network and the inter-station hydrogen mutual-aid quantities. The lower level constructs a dynamic route planning model based on macroscopic traffic flow to update the optimal hydrogen delivery routes according to county-township road traffic states, and feeds the transportation delay information back to the upper level in real time to trigger rolling dispatch revisions. Comparative results show that the proposed strategy effectively improves the renewable energy accommodation rate of rural distribution networks.