The Monitoring And Prediction Analysis Of Ground Settlement Of Deep Foundation Pit Of Subway

With the development of urban construction in China, there has been an increasing number of deep foundation pits. Especially in recent years the climax of subway construction in major cities, the size and difficulty of the deep foundation pit in subway station are growing. The deep foundation pit inevitably lead to a variety of deformation, so deep pit deformation monitoring and prediction as the key of the information construction, is one of important research topics in deep foundation pits.The paper based on a subway station deep foundation pits carry out the safety monitoring and the study of foundation deformation law. Based on the sunmmy and induction of foundation deformation and monitoring methods, the main research content is as follows:1,Elaborated in detail the impact of the adjacent ground in deep pit, summied the ground settlement deformation characteristics , analyzed the factors of the adjacent ground deformation and deformation mechanisms. Combining with engineering cases to develop a reasonable monitoring program. On the base of introducing the monitoring project implementation of deep foundation pit, determined the control volume of deformation, analyzed and processed the original data, get the time-space law of the ground settlement.2,Deformation prediction of deep foundation pits is a complex nonlinear problem, artificial neural network has advantages in dealing with nonlinear problem. Aim to achieve the deformation prediction of the deep foundation pit excavation, based on the characteristics and its self-learning algorithm of the BP network, to analysis the time-space law of deep foundation pit deformation, to establish the network model, to compile the training program, to utilize the BP network architecture, the choice of parameters, network training, to achieve the time series prediction of deep pit ground settlement. At the same time, comparing the models by adopting different number of the hidden layer node and different training function in BP network, selecting one of the best models to compare with RBF network model aim to find the ideal network model, providing reference for deep foundation monitoring .The paper relies on a deep subway foundation pit and analysis the deformation feature and monitoring status quo of the adjacent ground in deep foundation pits, carries out the reasearch of neural network model for ground settlement prediction, compares the simulation results of different parameter with the measured data, probes into the error factor of prediction result. The achievement can provide some reference for the monitoring of adjacent ground and buildings in the urban subway deep foundation pit .