genetic algorithms;  neural networks;  tunnels;  Surface subsidence

The main influencing factors of surface settlement in urban shallow buried tunnel excavation were analyzed, and a neural network surface settlement prediction model for shallow buried tunnel excavation based on genetic algorithm was established. The finite element numerical simulation forward algorithm is used to obtain the neural network model learning samples and train the model. The prediction model is used in the prediction of surface settlement of light rail tunnel in a city, and the results show that the prediction of surface settlement of tunnel excavation by neural network based on genetic algorithm is feasible, and the prediction results are relatively accurate, which can better guide tunnel construction and ensure the safety of surface buildings.