Explore the path selection behaviour of rail transit passengers to provide a basis for rail transit vehicle scheduling and train operation scheme development. Construct a rail transit network model, according to the path length of passenger travel, the number of transfers and effective path screening, based on the DFS algorithm and Dijistra algorithm to propose a path search algorithm to improve the DFS. Build a passenger travel path selection model. Construct the total travel time function and set the penalty coefficient. Correct the function by the number of transfers and the degree of congestion in the carriages. The MSA algorithm is used for passenger flow allocation, and the results are verified with the Suzhou rail transit line network, which show that the results of the model interval passenger flow allocation and the passenger flow in and out of the station are consistent with the actual observations, and the model is feasible.