Investigation On Horizontal Displacement Of Deep Foundation Pit Of Subway Based On PSO-LSSVM Models

Based on the sorting and analysis of the monitoring data of the foundation pit of Beijing metro line 19 XinFaDi station,this paper intends to study the applicability of PSO-LSSVM model(the particle swarm algorithm and the least square support vector machine)to the deep foundation pit displacement analysis.The main research contents and achievements herein are as follows:1.To organize the field monitoring data,based on Midas/GTS,the soil layer and the structure model were established,and then it was imported into FLAC 3D to carry out the excavation simulation calculation for soil layer and structural unit assignment.Based on the monitoring data,the model fitting and verification was conducted to establish a reasonable foundation pit model of XinFaDi station;2.According to the article and the fitting experience,the upper and lower limit of the elastic modulus of the soil layer was determined.The orthogonal experiment was used to design 25 different test combinations,and the established FLAC 3D finite element model was used for calculation,and reliable learning and training samples were obtained;3.The experiment has determined that the PSO-LSSVM model applies to the important parameters of this project: the inertial weight selection of the linear differential reduction strategy,the iterative frequency is = 500,and the learning factors are = = 2,and the particle has to take the = 30,and the LSSVM has the best radial core functions that are adaptive and effective.4.The PSO-LSSVM model is applied to the practicality and accuracy of displacement back analysis of deep foundation pits;By inverse analysis and calculation of the optimized pos-lssvm model,the result is relatively good,the prediction result is more consistent with the measured result,and the relative error is only 6.46%,which indicates that the application has engineering reference value5.Based on the results of the anti-analysis process,the elastic modulus of the anti-analysis is returned to the interval,which is the optimization of the proposed value to the actual value,and the corresponding layer interval,which provides a finite element simulation for the future of similar geological conditions...