Abstract: To address the multi-dimensional impacts of rainstorm weather on the safety and efficiency of train diagrams in crossing and merging scenarios, this paper constructed a dynamic optimization model of the train diagram based on a track state spatiotemporal matrix and proposed an adaptive adjustment strategy for the train diagram by incorporating a Meta-reinforcement Learning algorithm. Simulation results show that the model reduces train delay rates, improves train operation organization efficiency, ensures operational safety, and effectively addresses the optimization of the train diagram under rainstorm conditions, thereby providing an intelligent solution for railway dispatching and command.