Study On Spatial Effect And Prediction Of Excavation Deformation In Pebble Stratum In Chengdu

With the rapid progress of urbanization in China,the construction of urban rail transit has entered a stage of rapid development.With the expansion of the construction scale of rail transit,the number of deep and large foundation pits in subway projects is increasing day by day.The safety and stability of deep foundation pit construction is one of the most important links in subway project.It is not only necessary to carry out accurate and efficient monitoring and timely warning in all directions,but also to analyze the development trend of the deformation caused by the construction disturbance of the soil,so as to make a judgment on the construction safety.Therefore,mastering the law of soil deformation and the space-time effect caused by excavation during the construction of deep foundation pit is the theoretical basis and prerequisite for ensuring the safe construction of subway deep foundation pit.However,due to the complex environment of the subway deep foundation pit project,there are many factors affecting the construction safety around,and the construction difficulty and risk often increase.In recent years,subway deep foundation pit accidents have occurred from time to time,but the consequences are very serious,causing widespread concern in the industry.This paper centers on analyzing the spatial deformation and settlement of the support structure,along with the internal support and surrounding soil of a deep foundation excavation for a metro station situated in a sandy gravel stratum.It delves into the dewatering techniques that could be employed for such a foundation excavation and examines the pattern of changes in the deformation of the foundation pit’s support structure,internal support axial force and surface settlement,furthermore,it unveils the distinctive deformation traits of the foundation pit’s support structure and the resultant surface settlement at various points during the excavation of the deep foundation pit in the sandy gravel stratum,The foundation pit deformation prediction model based on particle swarm optimization,a hybrid model combining GA and LSSVM is formulated,and the prediction results are verified according to the actual project monitoring data.The findings of this study are primarily manifested in the following areas:(1)For the dewatering design of foundation pit of considerable depth in close proximity to a river,the permeability coefficient is calculated repeatedly by using the soil settlement and the settlement monitoring results,and then the final dewatering design parameters are obtained,which is used as the basis for the dewatering design parameters of Visual Modflow foundation pit.Different from the conventional method of fixing the dewatering parameters,this method fully considers the influence of the seepage recharge of the adjacent river channel on the dewatering,improves the accuracy of the selection of the dewatering design parameters of the foundation excavation site located next to the river in the sandy gravel stratum,and verifies the feasibility of the dewatering effect through the numerical simulation results,which provides a theoretical reference for the dewatering design and construction of the foundation excavation site located next to the river in the sandy gravel stratum.(2)Using MIDAS finite element analysis,The deep foundation pit space effect theory is used to analyze the ground settlement data of foundation pit excavation.Numerical simulation analysis was conducted on the structural changes or deformation of the support structure for deep foundation pit excavation in sandy gravel stratum,and it was found that the surface settlement range on the shorter side of the foundation excavation is less than the settlement range on the longer side.There are significant spatial effects and pit angle effects in narrow and long foundation pits,and there is a strong correlation between the excavation area and depth of the foundation pit and the horizontal displacement of the support pile.Once the excavation is complete,the settlement on the shorter side of the pit is 35% less than that on the longer side.During the third section of excavation,a spatial effect becomes evident on the longer side.The displacement of the retaining pile is greater at the corner farthest from the pit along the longer side,and 38% less at the corner nearest to the pit compared to the far corner.The surface settlement at the settlement point in the middle of the long side gradually increases from far to near,presenting a triangular variation pattern as a whole;The pattern of surface settlement near the corner along the longer side of the foundation pit is largely similar to that observed at the middle of the longer side.However,compared to the middle of the foundation pit,the maximum cumulative settlement is slightly smaller,and the surface settlement is reduced by about 35%.The axial force of the concrete support at the corner of the foundation pit is higher than the axial force of the internal support at the mid-point of the long side,The axial force of the steel support at the corner of the foundation pit is found to be 30% lower than that of the internal support at the midpoint of the longer side.It has been observed that segmented excavation can efficiently manage the soil disturbance around the foundation pit under different working conditions,thereby reducing the foundation pit support pressure,and also verify the effectiveness of the model.As the excavation depth and area of the foundation pit increase,the axial force of the internal support also increases.(3)The GA is embedded into the particle PSO for parameter optimization,and the time series deformation prediction model based on least squares support vector machine is established by combining the time series model.The model integrates the feature extraction of particle swarm algorithm on the data space dimension,the ability of mutation operator in genetic algorithm to deal with data mutation,and the real-time cyclic update of time series to the measured data.By comparing the prediction accuracy with the model optimized by genetic algorithm(GA-TLSSVM),particle swarm optimization(PSO-TLSSVM)and BP neural network prediction model,it is found that GA-PSO-TLSSVM model increases the information time-varying performance of the training sequence,realizes the dynamic and accurate prediction of the measured data,and provides a theoretical basis for the prediction and control of the deformation of the subway deep foundation pit.