Back Analysis And Displacement Prediction Of Mechanical Parameters Of Foundation Pit Based On Genetic Neural Network

Foundation pit engineering has a strong variability and uncertainty,in which the value of soil mechanics parameters are accurate,Whether the data of the pit displacement monitoring is realistic and so on are very critical to the design and construction of the foundation pit,and it is also directly related to the safety and economy of the whole project.Based on the actual engineering monitoring displacement data,combined with the genetic neural network and Midas / gts NX finite element software,the soil mechanics parameters of the foundation pit are analyzed by back analysis and the surrounding surface subsidence displacement.The main work is as follows:Firstly,the Midas / gts NX finite element software is used to model the foundation pit.The orthogonal design table is used to design the sensitivity analysis samples of the mechanical parameters of each soil.By calculating the sensitivity of the mechanical parameters of each soil layer,To determine the deformation of the foundation pit support the largest soil parameters,to carry out anti-analysis.Secondly,the learning samples and test samples of the genetic neural network are designed with the orthogonal design table and the uniform design table respectively.The mapping relationship between the displacement of the foundation pit and the mechanical parameters of the soil is established by the learning of the neural network,and then the actual monitoring displacement In the input genetic neural network model,the back analysis of the soil parameters is obtained.Finally,the parameters obtained by the inverse analysis are input into the finite element model.The calculated displacement is compared with the measured displacement.It shows that the genetic neural network is more likely to use BP neural The network is more accurate when the displacement back analysis,more suitable for practical engineering needs,with reference.Finally,this paper uses the BP neural network and the genetic neural network to predict the surface subsidence displacement based on the surface subsidence displacement of the pit excavation project in Xiamen City,and finally shows that the two methods Can be applied to the prediction of ground subsidence displacement of the foundation pit,but the genetic neural network is more accurate.We can use the genetic neural network on the late pit deformation monitoring data to make a preliminary judgment.