Abstract: A two-layer optimization model for train line planning was proposed based on the 4 modes of train stop-schedule. The purpose of the top-layer was to optimize the stop-schedule set and the service frequencies with the minimum of the total operation cost and unserved passengers. The purpose of the bottom-layer was to establish the specific stop-schedule and a mixed integer program for passenger flow assignment with the maximum of the served passenger volume and minimum of the total travel time. A GA-based line planning optimization algorithm was implemented to realize the feedback optimization of stop-schedule generation and passenger flow assignment. The validity of the model and algorithm was verified by the case study on Taiwan high speed railway.