APPLICATION OF SOIL PARAMETERS INVERSION BASED ON PSO-MLSSVR IN DEEP FOUNDATION PIT ENGINEERING

In order to ensure the safety of foundation pit construction,numerical simulation is used to predict the displacement of support structure. The selection of geotechnical parameters has the greatest influence on the numerical simulation results. A back analysis method based on multioutput least-squares support vector regression machine(MLSSVR) and particle swarm(PSO) is proposed to estimate multiple geotechnical parameters. This paper uses PSO-MLSSVR method to invert geotechnical parameters based on horizontal displacement monitoring data from the top of the supporting piles in a deep foundation pit in Shenzhen. Based on the orthogonal design method,representative combinations of geotechnical parameters are generated. Using these combinations,finite element method(FEM) is used to calculate the displacement of the measured points. PSO algorithm is used to optimize the parameters of MLSSVR model. Using MLSSVR to construct mapping relationships between inversion parameters and displacements to invert soil parameters in rock-fill,silt and sandy clay layers. The inverse parameters are substituted into the finite element model to calculate the measured point displacements. The results show that the MLSSVR model takes less time than the single output least squares support vector regressor(LSSVR) in inverting the parameters. When the inversion parameters of both models are substituted into the finite element model for calculation,the results of MLSSVR are closer to the actual monitoring values than LSSVR. The comparative results validate the superiority of the study methodology. In different stages of construction,the inversion parameters obtained by MLSSVR are used for numerical simulation. The simulation results are in good agreement with the monitoring data,which verified the accuracy and practicability of this method. The results show that the method in this paper is helpful to the selection of soil parameters and improve the accuracy of numerical simulation results.