Optimal control of train movement is an important way to reduce transport cost, enhance service level, and realize sustainable development. In this paper, based on traditional optimal velocity car-following model, an improved simulation model is presented, it is used to optimize the velocity control of train movement in urban railway system. The proposed model is established by introducing a new function of objective optimal velocity into the classical optimal velocity model (See Phys. Rev. E 51, 1035, Bando et al, 1995) to realize the optimal control of train movement in complicated conditions. Numerical simulation takes the Beijing City Metro Yi Zhuang line as an example. Here some reality measurement data is used. Results show that the proposed model can well describe the dynamic characteristics of train movement under the complex limited condition. Simulation results are close to reality measurement data. This demonstrates that the proposed model is valid. Further, by analyzing the space-time graph, the change of train velocity and travel time, the evolution characters of train flow under complex conditions are discussed.