The Analysis Of Peripheral Building Settlement Forecast In Deep Foundation Pit Construction

In recent years, the full use of underground space has promoted the deep foundation pit development. The construction of deep foundation pit will have the tremendous influence on its peripheral environment. Peripheral building settlement is one of typical examples.In the building settlement process, many kinds of methods can be used to predict and analyze the settlement. It mainly divides into two kinds. One is based on theoretical calculation method such as soil mechanics. Another is the forecast settlement according to actual date. Actual date comes from science observation. This paper presents crucial factor in settlement observation and lists observation plans of used date. This paper summaries the present several commonly utilities of subsidence forecast methods. In the process of predicting settlement, the effect of the settlement has many uncertainty factors, therefore it is difficult to determine the value of these factors by the theoretical calculations. The effect of various factors to the settlement reflected changes in data. Thereby, according to the measured data, the prediction method that we estimate the amount of the settlement has been applied. In this paper, grey system theory and artificial neural network methods are used to establish predictive models, analyze and discuss the model and the results.BP neural network is used to establish the accumulation settlement forecast model of the observation point. 14 observation points in 4 buildings around deep foundation pit typical position of Shenzhen subway's a deep foundation pie is selected . The forecast is made by the finally 6 observation data. Through the analysis of the predicted value and the measured data, according to specified circumstance of observation point, data are unified. The grey system theory is used to establish the settlement prediction GM(1,1) model. According to principle which the predicted value perigee contains the richest information in the model, the first spot of previous step forecast is rejected. The predicted value perigee data are added when rebirth becomes sequence. After making 6 steps forecast to the above same observation point, the rationality of this model is judged. The forecasting results of two measures are analyzed. The applicability of model is summarized and the conclusion is arised.Using BP neural network model and grey system forecast model, Carries on the settlement forecast to the select building. The forecasting results prove that BP neural network model and grey system forecast model in this paper is feasible in forecasting peripheral building settlement.