The construction of underground spaces and tunnels has long been considered by human beings. Qanat excavation, which has thousand years of background in Iran, as well as the exploitation of mines and underground exploration, all represent the historical interest of mankind in excavating tunnels. Nowadays, population growth and rising standards of living in urban areas have caused traffic problems and created and expanded urban train networks. As a result, we are forced to dig the tunnels in urban areas, which mainly consists of earth materials. Generally, the excavation of tunnels and other underground structures leads to the removal of mass from the soil and rock and causes significant changes in the distribution system of tensions around the tunnel, so movements and deformation of land are the inevitable results of tunnel excavation and construction. Study of effects of Surface motion due to the shallow tunnel excavation in urban and residential areas due to possible effects they might have on surface structures are of particular importance. Therefore, in this study, it has been tried to use the numerical methods and soil simulations in line 2 of the Isfahan subway to predict the amount of the Soil settlement generated by tunnel excavation the mentioned line. Line 2 of Isfahan subway with around 23 km length and 22 stations from Zinabeeh Depot in the northeast of Isfahan to Shahid Beheshti Square of Khomeini are foreseen and it is planned to be excavated by tunnel boring machine (TBM). The simulations were performed by FLAC 3D software and the effect of factors such as depth and diameter of the tunnel on the amount of the Soil settlement generated by excavation. In the other part of this study, by using the results of numerical analysis and with the aid of neural networks method, a model for predicting the amount of the amount of the Soil settlement generated by excavation was proposed. This model is tested with the existing data with the error rate of under 6 percent is obtained. Finally, the sensitivity analysis on the effect of geomechanical properties of the soil on the amount of the Soil settlement was carried out by this model. Finally, by comparing the data from numerical analysis and the neural network method with empirical relationships, it was observed that empirical relations predict more amounts for the Soil settlement, which can be attributed to the failure of empirical relations to consider some factors, affecting the Soil settlement.