Study On Backing Analysis Of The Silty Parameters In Reclamation Area And Predicting The Deformation Of Foundation Pit

With the development of our economy, the shortage of land resources and structure have become an important bottleneck which restricting the economic development of coastal areas,reclamation has become a effective way to add the land and avoid constraints by national policy. But large-scale reclamation has caused a series of problems. Qianhai in ShenZhen was a beach a few years ago, after artificial reclamation, it becomes an area that has 15 square kilometers. However, when the reclamation project construction, the silt was crowded around from the original position by irregular way, which the reclamation area is covered with silt. It led to a series of geotechnical problems.The data of the expert meeting indicates that the problem of geotechnical engineering in the area is related to the insufficient understanding of the design parameters of the silt layer.Due to the slow consolidation process of silt, the geotechnical parameters are still in change,and even if the parameters of the silty layer can be obtained by laboratory or in-situ test, the experimental results and the real value are different because the size of the soil and the test results are not representative; the excessive cost or imperfect technical conditions in some special cases, which result in the soil parameters get by experiment is different form the real.Basis on a foundation pit in Qianhai, we can invert the silty parameters in the paper (inversion can be also called back analysis, we named it back analysis in the following paper), so we can get the real parameters of the project to guide the construction.Neural networks can be used to realize the nonlinear mapping between complex objects,which is suitable for solving the problems that seem to be chaotic. It deals with information in a way similar to the human brain, which simulates a number of simple cells into neurons, so that the nonlinear relationship between the input and the output can be established. In the field of civil engineering, artificial neural network can be used in backing analysis of the soil parameters, process monitoring, predicting the deformation, and so on. Thus, the neural networks is used to back analysis of the silty parameters in this paper. That is, on the basis of the existing monitoring data, the neural network toolbox in MATLAB and GTS are used to analysis the parameters of the silty layer, the last, we can use the parameters of soil layer to predict the deformation of the foundation pit of next stage. The work which had been done in this paper is as follows:1. The back analysis model of BP neural network is established, and the training times,training error, learning rate, momentum factor and the number of hidden units are determined.The number of hidden layer units is related to the correctness of the back analysis results,so we can calculation the number of hidden layer nodes according to the number of input and output layers.2. Two kinds of experimental design methods are introduced. Finally, the uniform design method is used to deal with the silt parameters, so, we can obtain 10 groups of elastic modulus E, cohesion C, internal friction angle ? and poisson's ratio ?.3. Establish the GTS model of foundation pit. 10 different combinations of parameters under uniform design will be input into GTS to calculation the displacement of the wall, then,normalize the displacement.4. The normalized data should be input into the MATLAB program and the data processed by simulation function -Sim is the soil parameters of the foundation pit. Then, each stage of the silty layer parameters should be calculated by GTS, the last, we can compare the horizontal displacement values which calculated by GTS with the measured values under different conditions to check the accuracy of the back analysis results.5. Simulate the horizontal displacement of the pile when the foundation pit is excavated to the bottom based on dynamic back analysis. Another,a method is presented to predict the deformation of foundation pit in this paper, that is, using the BP neural network to predict the horizontal deformation of the pile in the specific date of the foundation pit, so, we can use the theory to guide the construction.