Research On The Deformation And Prediction Of Deep Foundation Pit Excavation In One Station Of Wuhan Metro Based On Inversion Method

The deep foundation pit support structure would deform during the process of the deep foundation pit excavation of a metro station.And then,it would result in ground surface settlement around the foundation pit,even endanger safety of foundation pit directly.At last,it would cause serious economic losses and social impact.Analyzing and predicting the deformation of the deep foundation pit support structure of a metro station could provide reference for the engineering analysis under the condition of similar soil layer.It played an early warning role in the construction of project.And then,it could guide to proceed the engineering project safely and orderly.On the background of the whole construction process of a station of the No.6 subway in Wuhan,this paper used bionic intelligence algorithms and numerical simulation methods to analyze and predict the deformation of the deep foundation pit support structure of one station combining with monitoring data,and then,it verified the accuracy of analysis results by the actual monitoring data.The main research contents were as follows:(1)One part should be selected as research object during the excavation process of a subway station deep foundation pit in Wuhan,the excavation process of the deep foundation pit should be divided into five working conditions,the GA-BP algorithm was used to do the back analysis of elastic modulus of soil layers which were in range of the excavation under each working condition.At last,every soil elastic modulus inversion value under each working condition should be recorded.(2)The variation tendency of every soil elastic modulus was analyzed under the condition of excavation combined with related research achievements of soil lateral unloading theory,and the theoretical equation was used to deduce and verify the accuracy of the variation tendency.(3)The accuracy of selected geo-material constitutive model was verified.The genetic algorithm was used to back-analyze the parameters in the yield function.Then,the inversion values of the parameters were contrasted with the theoretical values.At last,the most appropriate constitutive model should be determined.(4)D-P model and M-C model should be selected to calculate in the process of AutoCAD-ANSYS-flac3 D three-dimensional modeling.The calculated horizontal displacement values of simulation should be contrasted with the measured results under every condition.(5)GA-SVM algorithm,BP neural network and SVM algorithm were used to do the prediction analysis of the maximal horizontal displacement value of support structure.The main conclusion could be obtained in this paper as follows:(1)The bionic intelligence algorithms were used to back-analyze the soil parameters around the deep foundation pit of subway station and identify the constitutive model.This method was proved to be reasonable by engineering project.(2)By using the soil lateral unloading theory,modulus of elasticity of each soil layer was proved to have a descending trend with the excavation of foundation pit.This trend was consistent with the inverse results of the same soil layer which GA-BP algorithm calculated under different conditions.(3)Reasonable three-dimensional modeling method had been explored,the inversion values of geotechnical parameter and D-P constitutive model were used to establish the practical three-dimensional model of foundation pit excavation.At last,the calculating results of D-P constitutive model were consistent with the measured results.(4)GA-SVM algorithm,BP neural network and SVM algorithm were used to do the prediction analysis of the maximal horizontal displacement value of support structure.It is found that the prediction result of GA-SVM algorithm had minimum error with measured data.