The subway tunnel's deformation monitoring can provide the basis for judging itself safety, and the deformation within the allowable range is the requirement of subway tunnel's safety operation. Aiming at the common problems of modeling difficulties and low prediction accuracy in the prediction of deformation monitoring data based on statistical methods, in order to predict deformation data more accurately, a PSO-BP neural network model was built to analyze and predict the lateral and vertical displacement monitoring data of the subway tunnels. The results show that the prediction accuracy of PSO-BP neural network model is obviously higher than that of BP neural network model, and it can make more accurate prediction for deformation data.