Abstract: In order to meet the requirements of mixed operation of trains with high and middle speeds on a passenger-dedicated line in China and increase the stochastic robustness of the train line planning system, a multi-objective expected value model for stochastic programming subjected to the carrying capacity of the train line system was built based on the graph theory, the probability theory, and the basic programming idea of expected value modeling. Then, a multi-parameter genetic algorithm with random simulation was designed for solving the model. The result of a computational example shows that the algorithm can obtain the satisfactory solution after 600 evolution generations when the detour rates for both the trains and passengers are 1.4, the period is 60 min, the K-shortest path parameter for the trains are 3 and that for passengers are 8. Therefore, a robust solution for train line plan and passenger flow assignment scheme can be generated by the proposed algorithm under the constraints of the carrying capacity of the railway network, the transportation capacities of the trains operated with different speeds, and the travel demands of passengers, etc.