Analysis Of Deep Anti-sliding Stability Based On BP Neural Network For Gravity Dams

Deep anti-sliding stability analysis is an important part of gravity dams seismic calculation.The deep anti-sliding stability calculation of finite element method needs to set the sliding surface in the calculation model in advance.When there are multiple gentle inclination angles and weak structural surfaces in the deep foundation of a gravity dam,not only will the entire finite element model modeling and meshing face greater difficulties,but also the quality of local elements will be reduced.This paper proposes a finite element analysis method for deep anti-sliding stability of gravity dams based on BP neural network.This method does not need to set the sliding surface in advance in the finite element calculation model.The BP neural network is used to fit the stress of the sliding surface according to the deep spatial stress relationship of the dam foundation,avoiding the modeling and meshing of complex sliders.This method not only has good accuracy and usability,but also reduces the workload of anti-sliding stability calculation.At the same time,this paper compares the influence of different viscoelastic artificial boundary conditions on the calculation results of the gravity dam’s deep anti-sliding stability.The main research contents of this paper are as follows:(1)According to the basic theory of BP neural network,the calculation formula of multi-layer feedforward and error back propagation process is deduced.The program of BP neural network fitting sliding surface stress and predicting the safety coefficient of anti-sliding stability is compiled.The program approximates the function relationship between coordinates and stress based on the deep spatial stress field of the dam foundation without the slider finite element model,combined with the BP neural network algorithm.As a result,the program can fit the stress value at any position of the sliding surface,thereby predicting the safety coefficient of the sliding surface stability.(2)Two-dimensional and three-dimensional finite element models of gravity dams without and with sliding blocks are established respectively.Under static conditions,the BP neural network method is used to predict the safety coefficient of anti-sliding stability of the model without sliding block.Compared with the traditional method of presetting the sliding surface and extracting the sliding surface stress,it verifies the prediction accuracy of the BP neural network method under static conditions.The influence of the number of neurons in the BP neural network program and the quality of sample data on the prediction performance of the BP neural network is analyzed.It is proposed to measure the accuracy of neural network program prediction with neural network performance parameters,and the feasibility of this method is verified.(3)The common formulas of viscoelastic artificial boundary and equivalent load are summarized.Two-dimensional and three-dimensional viscoelastic boundary and ground motion input programs have been compiled.The effects of different viscoelastic artificial boundary conditions in the free field calculation example and the anti-sliding stability analysis of the gravity dam are compared.(4)The BP neural network method is used to calculate the anti-sliding stability under dynamic conditions.In order to reduce the randomness of the BP neural network results,the method adopts multiple repeated predictions for the anti-skid coefficient prediction at each time and takes the average value as the predicted value at that time.The influence of the number of repetitions at each moment on the stability and accuracy of the prediction results is compared.