| Foundation pit projects are rapidly developing,and deep foundation pit projects’excavation depth and scope are expanding.However,deep foundation pit projects are usually located in urban and other construction space-limited and densely populated areas,which may adversely affect the surrounding environment.Especially in the construction process of soft soil areas,if the pit deformation cannot be effectively controlled,it may lead to engineering accidents such as pit destabilization and rupture of surrounding buildings.Therefore,in the process of foundation pit excavation,it is not only necessary to master the deformation law of the pit but also to effectively predict the deformation law of the surrounding soil during the excavation of the pit to ensure the safety of the pit itself and the surrounding environment.This thesis takes a deep foundation pit in the form of double-row pile support as the research object.Based on the engineering example,a three-dimensional finite element model is established to study the deformation law of a double-row pile support foundation pit,and the rationality of the numerical model is verified by combining it with the field monitoring data.On this basis,by changing a certain parameter of the double-row pile supporting structure,the influence of it on the stress of the supporting structure and the deformation of the foundation pit is studied.Finally,various prediction models are used to predict the deformation of the double-row pile supporting the foundation pit to explore the models suitable for different construction stages and ensure construction safety.The main research contents and conclusions are as follows:(1)A comprehensive analysis of the monitoring data of a foundation pit project in the soft soil area of Huzhou City was carried out.On this basis,a three-dimensional finite element model was established and compared with the actual engineering monitoring data for verification,and the deformation law of the double-row pile-supported foundation pit was investigated.The research shows that:the pile moment values of the front and rear piles are distributed in an S-shape;the horizontal displacement of the front pile is characterized by a "bulging belly" with a large middle and small ends,while the horizontal displacement of the rear pile shows a rapid decrease in the upper section,indicating that the inter-pile reinforcement plays a good role in limiting the development of horizontal displacement;an initial settlement of the pile characterizes the surface settlement curve around the foundation pit.The surface settlement curve shows an overall rising trend after the initial partial disturbance,similar to the "spoon" curve;the soil in the pit shows an overall uplift and the amount of uplift is larger,and the farther away from the edge of the pit,the larger the amount of uplift is;the deep horizontal displacement of the soil body shows the maximum at 1.5m below the top of the pile,and gradually decreases with the increase of depth.(2)The influence law of seven factors,such as pile diameter,pile length,and pile spacing,on the force and pit deformation of double-row pile support structure was studied by finite element numerical simulation.The analysis shows that:the increase of the row distance,the depth of the passive zone solidification,and the appropriate decrease of the pile distance can effectively resist the horizontal displacement of the pile body,the surface settlement behind the pit and the horizontal displacement of the deep soil body,while the influence of the pile diameter,pile length,the amount of solidified cement and the width of the passive zone solidification is smaller;the increase of the pile diameter and row distance will increase the pile bending moment,and the influence is larger,while the amount of solidified cement,the width and depth of the solidification has a smaller influence on the pile bending moment.The effects of increasing the pile diameter and row distance increase the pile bending moment and have a greater effect.In contrast,the amount of cement addition,reinforcement width,and depth affect the pile bending moment less.The method of increasing the row distance combined with inter-pile reinforcement or the depth of passive zone solidification is the most effective in controlling the foundation displacement and deformation in this area while increasing only the pile diameter,pile length,cement mixture of reinforcement or the width of passive zone solidification is less effective.(3)In order to study the influence of soil parameters on deep soil horizontal displacement of foundation pits,multiple optimization models based on support vector machine(SVM)are constructed,combined with deep foundation pit monitoring data in Huzhou,the BP-Olden method is used to select the important influencing factors affecting deep soil horizontal displacement of deep foundation pits.The important influencing factors are taken as the characteristic input parameters.Deep soil horizontal displacement is taken as the output parameter,integrating A total of 24 days of measured data from 8 deep horizontal displacement monitoring points were integrated as the data set,and six prediction models,namely SVM,LSSVM,PSO-SVM,COA-SVM,PSO-LSSVM and AO-LSSVM,were constructed and trained.The results show that the best prediction accuracy and stability is the AO-LSSVM model,whose error amount is mainly distributed in the range of-0.5mm to 0.5mm and a small amount in the range of-1mm to 1mm,with an average absolute error of 0.57mm,root mean square error of 0.63mm,average absolute error percentage of 1.54%,and R2 of 0.9991.The predicted and measured values match each other.The practicality and reasonableness of the model are verified,and it is feasible to predict the horizontal deformation of the deep soil in the double-row pile-supported foundation pit.(4)In order to study the prediction models suitable for different construction stages,single-factor analysis prediction was performed using time variables based on the monitoring data of each stage at the CX17 monitoring site.The results show that:the AO-LSSVM prediction model has the characteristics of high prediction accuracy and good stability,the average absolute error is 0.03mm～0.13mm,the root mean square error is 0.04mm～0.15mm,the average absolute error percentage is 0.09%-0.15%,and the R2 is 0.9851-0.9987.The AO-LSSVM model has a good prediction accuracy for the horizontal deformation of deep soil in each construction stage of the double-row pile supporting foundation pit.Similarly,subsection prediction is carried out based on the monitoring data of DL16 monitoring points at each stage.The results show that the AO-LSSVM model has the best prediction effect in the excavation’s first and second stages.The PSO-SVM model shows better optimization ability and accuracy in the excavation and backfill stage. |
PDF Full Text Request