Study On Soil Parameters Inversion And Settlement Prediction Of Foundation Pit Based On Genetic Neural Network

Nowadays,China has become a country with the world's largest number of subway lines and stations.The mechanism of the surrounding settlement caused by the foundation pit project is very complicated.With the development of computer technology and numerical theory,the numerical analysis method is gradually applied to the excavation analysis of the foundation pit,but the numerical results are more dependent on the selection of model parameters,ignoring the specific conditions of foundation engineering,and the model parameters will be difficult to determine accurately.With the rise of modern intelligent algorithms,intelligent algorithms are beginning to be applied to the excavation of the foundation pit.The GA-ANN has the characteristics of nonlinear fitting and nonlinear optimization.Therefore,it is very practical to use GA-ANN to make back analysis to the parameters.Based on the foundation pit excavation project of Yan'er Island station of Qingdao rail transit line R2,this paper studied on the settlement deformation law of surrounding observation points during the excavation process of the subway station through measured analysis,numerical simulation,parameter inversion,settlement prediction and.also made back analysis to strength parameters.The primary research contents and achievements of this paper are as follows:(1)Measured analysis.This paper analyzed the measured data of the groundwater level,the lateral displacement of the retaining piles and the surface settlement around the station foundation pit and summarized the changing rules of these monitoring projects with the foundation pit excavation process,which provided data support for the numerical simulation scheme.(2)The numerical simulation scheme of retaining piles,concretesupport,steel support,anchor cable,contact surface of retaining piles and soil of foundation pit was determined.The elastic modulus of the soil was calculated according to the experience,and the settlement value of the excavation points with the foundation pit excavation was calculated under this working condition.Furthermore,the rationality of the model simulation scheme was verified.(3)In order to design the test samples rationally and reduce the number of tests,this paper introduced the uniform design method and designed the learning samples.Based on the advantages of GA-ANN,the multi-parameter inversion analysis method based on GA-ANN was proposed.Based on MATLAB software,the GA-ANN parameter inversion program was developed.Finally,the elastic modulus of the soil was obtained by inversion through the example of a foundation pit project,which verified the reliability and accuracy of the program and analysis.(4)The dynamic prediction of the settlement was proposed.Based on the measured data of foundation pit outside and GA-ANN,a dynamic forecasting program was developed.The nonlinear mapping relation between the pre-settlement and the post settlement was obtained and the settlement value was predicted based on a certain measurement point.The result showed that the dynamic settlement prediction method based on GA-ANN has highly reliable.