Aiming at deep horizontal displacement problem in the process of excavation construction of the second District Provincial Museum Station in 3 east section of Kunming Rail Transit Line, this paper sets up the displacement prediction curve on basis of MATLAB simulation and the establishment of BP neural network model through the input displacement data. The experimental data shows that the BP neural network model has good predictive power for the deep horizontal displacement. This method provides a new technical means for the displacement control of underground engineering.