Finite Element Settlement Analysis And Neural Network Prediction Of Box Culvert

Under the action of external load,the box culvert structure will produce uneven settlement of the foundation,and excessive uneven settlement will easily lead to large deformation and even damage of the structure.Therefore,ABAQUS is used to simulate and analyze the internal force deformation and displacement of box culvert structure under different working conditions,and the Improved Grey Wolf algorithm is used to optimize the BP neural network prediction model(GWO-BP)to predict the settlement value of box culvert foundation.The specific research work of this paper is as follows:(1)Introduce the advantages of ABAQUS software structure analysis,easy to operate,easy to implement and fit with actual engineering results,and use ABAQUS as an analysis tool for box culvert structure.According to the actual conditions of the external conditions of the box culvert,the constitutive model used for the main materials such as the box culvert structure and the soil body was studied and determined,and the contact definition and the ground stress balance were carried out.(2)The three-dimensional finite element model of box culvert structure was established by using ABAQUS software.The internal model of the box culvert structure model was carried out under three different working conditions(no water in the box culvert,half water in the hole,and full in the hole).The calculation and the result analysis show the stress and displacement generated by the box culvert structure: through the displacement analysis of the interaction between the box culvert and the soil body,according to the soil settlement results at the simulated box culvert foundation,the box under different working conditions is obtained.The value of settlement of culvert foundation.(3)The Grey Wolf algorithm is introduced and improved.It is found that the Improved Grey Wolf algorithm improves the global search efficiency.After combining grey wolf with BP neural network,three kinds of settlement prediction models are established: grey wolf optimized BP neural network prediction model(GWO-BP),Improved Grey Wolf algorithm optimized BP neural network prediction model(improved GWO-BP),and BP neural network prediction model.Three prediction models are used to predict the settlement of box culvert foundation.The prediction results show that the improved GWO-BP settlement prediction model has good generalization ability,fast convergence speed and high accuracy.It can effectively predict the settlement of box culvert foundation,and can play the role of monitoring and early warning.