Deformation Prediction Of Deep Foundation Pit Based On BP Neural Network

With the rapid development of economic construction in China, there are large amount of deep foundation engineering, the size and the difficulty of them are increasing. The deformation which is caused by excavation of deep foundation pit is one of the important problems of deep foundation pit engineering. Monitoring and prediction of deformation of deep foundation pit is an important link in the design and construction of deep foundation pit engineering. According to the displacement monitoring data of deep foundation pit, a prediction system of horizontal displacement of foundation pit can be built, and the prediction results to the construction will feedback the design unit timely, and conducive to adjust the construction scheme, optimization design timely, so the potential safety problems of foundation pit construction can be reduced, it is important engineering significance for foundation pit engineering construction. Because of the outstanding advantages of artificial neural network in dealing with nonlinear problems.so, in this paper, the BP neural network which is most widely used is used to study deep foundation pit deformation prediction problems. Based on the preliminary research, a lot of research work and achieved certain results were made.Firstly, the characteristics of deep foundation pit engineering and application of prediction of the artificial neural network in deep foundation pit deformation were systematically summarized. Through the elaboration and analysis of the general principle of neural network, summarizing the basic model and algorithm of artificial neural network and analysing the advantage of the artificial neural network method; Secondly, the main factors which affect the deep foundation pit were comprehensively analysed and summarized, then the analytic hierarchy process to establish evaluation index system of the deep foundation pit was used, and the quantification standard of evaluation system was listed. After that, inclinometer displacement data which is corresponding monitoring date of the inclinometer hole from a deep excavation engineering in Qianjiang New City in Hangzhou was selected, and these data will be treated as the training sample data in this paper. According to the evaluation system has been established and the quantified standard,16indexes of the main factors which affect the deformation of deep foundation pit were selected and assigned, then they were taken as input unit of the network, and the deep horizontal enclosure with different depth displacement were used as the output of the network, Matlab toolbox and application of the momentum gradient descent method was used to debug each parameter of the neural network,48network models to predict the level of displacement in different depth of Retaining Deep Soil were confirmed, and the effectiveness of the models were proved. At last, two different conditions of inclinometer hole of another deep foundation pit was selected, the various factors were assigned according to the evaluation index system and quantitative standard, the BP neural network models built were used to predict soil horizontal displacement under different conditions of deep foundation pit, providing a useful reference for the engineering construction.