Abstract: The interactions between the high-speed train and the service structure/environment obviously strengthen with increase of the vehicle speed, which exhibit in the whole motion of the train. It is currently a research hotspot to establish a refined train-track dynamic model with considering real service environment and perform a more accurate simulation, but may encounter many challenges, e.g., the difficulty in modeling coupled dynamics of the train and long track structure. Modern trains can run the distance of several hundred kilometers across the track in a short time. The classical methods often deal with the motion and deformation of the coupled system in Lagrangian description, and impose an intrinsic requirement of an invariable mechanical research object of the system. Hence, they must model an entire track structure and the computation efficiency needs to be improved. Aiming at the key mechanical model in the field of high-speed rail transit, based on the model reduction and the theory of time- varying system dynamics, an efficient method is proposed to model and analyze the train-track dynamics in long distance continuous travel. For its purpose, the moving control volume is introduced here to describe dynamic of the long track in an effective region, which forms a truncated system for the analysis. The arbitrary Lagrangian- Eulerian formulation is used to establish the time-varying dynamic model of the truncated rail beam in the moving control volume. The adaptive Runge-Kutta-Fehlberg method is used to numerically solve the motion equations of the coupled system. At each time step, the integral information at current step should be preconditioned to obtain right initial value conditions for predict dynamic response of the system at next step. By this way, the discrete time-varying dynamics of the lump mass system in the moving control volume is included, representing the sleeper and ballast system. Dynamic analysis is carried out with the numerical example and obtained results are compared with those from the traditional modal superposition method. The current method is verified.