As the new rail transit lines or newly opened stations are put into use, the network topology has changed, and the passenger flow on the network has become increasingly complex. How to accurately predict the passenger flow demand of the newly opened stations is a key problem that needs to be solved in the current rail transit system. Therefore, the characteristics of land use, topology, connection and residence around the newly opened station were fully considered, and ANN (nonlinear neural network) model based on multi factor constraints was constructed to predict the demand for the newly opened station of rail transit. Then, Beijing rail transit was taked as an example, the prediction accuracy of this model reached 96.6%. The results show that the ANN model, which comprehensively considers constraints of site land use, topology, connection, residence and other factors, can better capture the irregular changes in passenger flow demand under new station opening, and the prediction accuracy is higher than that of the previous methods. The research results can provide new methods for passenger flow prediction in the construction of new stations and feasibility study of rail transit system.