Simultaneous Inversion And Layered Iterative Inversion Of Multilayer Soil Parameters On Shore Slope

The river bank slope project is built in order to ensure the safety of the people's production and life along the river.The economy along the river and the coastal area is more developed in our country,and the role of the bank slope engineering is particularly important.Seepage,deformation and instability are frequently encountered problems in bank slope engineering.The change of river water level causes the seepage and deformation of bank slope to aggravate.Therefore,during the construction of the project,it is necessary to monitor the seepage and deformation of the bank slope and take different measures to prevent its instability and failure.In the use of numerical simulation analysis of bank slope engineering,the simulation of numerical model and the accuracy of calculation parameters play an important role in obtaining accurate calculation results.Using the monitoring data,the inversion method can effectively acquire soil parameters,save time and effort,and has high accuracy and reliability.On the basis of the previous research results,this paper uses the ANSYS finite element software programming language APDL to develop two times to simulate the displacement and seepage fields of the bank slope under different water level conditions,and to extract the displacement and seepage pressure of the measured points,and to build the foundation for the inversion of the soil parameters.In this paper,artificial neural network is used as the inversion tool,and the artificial neural network model and principle are introduced in detail.The BP neural network program is compiled,and the BP neural network is used to invert the elastic modulus and permeability coefficient of the slope soil.For bank slope engineering with layered soil,simultaneous inversion method and hierarchical iterative inversion method are used to invert the mechanical parameters of multi-storey soil on the bank slope.When using the simultaneous inversion method,the finite element analysis model is established by ANSYS,and the displacement and seepage field of the bank slope under the action of different water levels are simulated.The displacement or seepage pressure of the soil test points of each layer is extracted.The training samples are formed and the neural network is trained.Then the measured results are replaced by the neural network,and the elastic modulus of each layer is retrieved or the soil modulus of soil is retrieved.Permeability coefficient.When the layered iterative inversion method is used,starting from the bottom layer,the elastic modulus of the above layers is assumed to be fixed first,and the BP neural network is used to invert the lowest soil modulus of soil mass,and the inversion result is replaced by the initial elastic modulus of the soil layer,then the above step by layer is repeated,until the back and back results are retrieved two times.The maximum difference is less than a certain admissible value.In order to verify the accuracy of the inversion results,the inversion results of the two methods are analyzed and compared with the measured results,and the results are ideal.The inversion results obtained by the hierarchical iterative inversion method are more accurate than the measured ones,and the results are better.In order to save the workload of the layered iterative inversion method,the layered iterative inversion method needs repeated cyclic inversion,and the hierarchical iterative inversion method is simplified by using the results of simultaneous inversion and the stepwise scanning method.The results of the simultaneous inversion method are better to simplify the stratified iterative inversion method,which can obviously reduce the workload of the stratified iterative inversion method and the results are reasonable and accurate.