Global Sensitivity Analysis for CNN Based Deformation Prediction of A Cohesive Structure of A Sluice and Pumping Station
Aiming at the problem of large computational cost of traditional structural deformation sensitivity analysis in the design stage,this paper constructed a CNN model to predict the structural deformation of the sluice and pumping station cohesive structure under different material parameters and water levels,and conducted global sensitivity analysis based on the proposed model,which greatly saved the computational cost.Firstly,200 sets of structural deformation data were calculated by the PDS module of ANSYS,and the CNN structural deformation prediction model was constructed and trained on the basis of these data.Then,the sensitivity relationship of the design parameters to the structural deformation was analysed by using the Sobol method.The results show that the structural deformation of the sluice and pumping sta-tion co-composite structure is dominated by settlement,and the displacements in vertical direction are greatly affected by the elasticity modulus of composite foundation;The displacement in the upstream and downstream directions is affected by the upstream and downstream water levels;The maximum stress of the structure is affected by the upstream and downstream water depths and the modulus of elasticity of the concrete.The results of this analysis provide a reference ba-sis for the related design work.
cohesive structure of sluice pumping stationCNNsensitivity analysisSobol sequence samplingfinite element analysis
NETL
NSTL
万方数据
