Slope Stability Analyses Based On Big Data And Convolutional Neural Networks

Slopes are often used as the load carrier in engineering construction.If the slope engineering design or management is improper,it may bring adverse effects on the construction and use of the building(structure),resulting in slope deformation,and even cause geological disasters.Slope stability often becomes a key engineering geological problem in mountain engineering construction,which determines the feasibility,safety and economy of engineering projects.Through the prediction of slope stability,the state of slope can be quickly and intuitively judged,which can provide an important basis for slope engineering design,construction and management,so as to ensure the safety and stability of slope engineering and smooth construction.Based on the summary of slope deformation and instability mechanism and on the basis of previous research results,a convolution neural network method is introduced into the evaluation of slope stability.Based on big data and convolution neural networks,a predictive model of slope stability analysis is constructed and applied in Qingyuan city Evergrande silver lake city second phase of the slope engineering.The results of the neural network method and finite element analysis are compared to verify the effectiveness of the proposed method.The main research work and achievements are as follows:(1)Construction of slope database based on limit equilibrium methodCombined with limit equilibrium method and numerical analysis technology,a slope database with rich types is constructed.A method of constructing slope stability analysis database based on limit equilibrium principle is proposed.A slope database is constructed by using numerical analysis technology for batch calculation,which contains 20000 groups of slope data,provides a large number of samples for convolutional neural network model training,which solves the problem of insufficient samples,and improves the performance and generalization ability of the model.(2)Study on slope stability model based on convolutional neural network.The convolutional neural network model of slope stability was constructed by taking the coordinates of slope key points and physical and mechanical parameters of soil layers(c??????_sat)as the input,and the slope safety factor F_s and the slope stability state as the output.Totally 18,000 slope samples from the slope database were used to train the model,and the other 2,000 samples were used to verify the network performance.The research shows that the output value of the convolutional neural network method is basically consistent with the expected value,and the performance is stable.and the calculation accuracy is high.(3)Comparative study of slope engineering stability analysis by neural network method and finite element method.Taking the slope of Evergrande Silver Lake City phase ? in Qingyuan city as the engineering background,combined with the slope engineering geological data and field exploration results,the site environment and engineering geological conditions in the study area are analyzed.Then,the stability of 16 sections of the slope site is analyzed by using the convolutional neural network and finite element numerical calculation respectively.The trained convolutional neural network model is applied to the slope of evergrande Silver Lake City Project in Qingyuan City,and the calculation errors of the convolutional neural network method and the finite element strength reduction method are compared.The results show that the accuracy and speed of the convolutional neural network method are better than those of the finite element numerical calculation method.