Prediction On Peak Shear Strength Of Rock Joints Based On BP Neural Network

In this paper,BP neural network is used to predict the shear strength of rock joints.As an information processing method,BP neural network can comprehensively predict various influencing factors and has the characteristics of self-study and high efficiency.Shear strength of rock joints is the main factor that affects the stability of rock mass.After investigation,data of direct shear test results of rock joints were collected by some scholars,among which Grasselli's direct shear test data were used to establish a prediction model after network training.Through comprehensive analysis,the shape parameters of rock joints and experimental conditions as the main factors are considered as the input parameters of the BP neural network.After a lot of network training,the prediction model can quickly predict peak shear strength of rock joints.The predicted results are compared with the peak shear strength calculated by other shear models.Then the feasibility of the method is verified by the laboratory direct shear test as a new data.The main research results obtained in this study are as follows:?1?The factors influencing the peak shear strength of rock structure face are analyzed comprehensively.The contact area ratio A₀,surface roughness parameters ?_max^*/?C+1?,rock tensile strength ?_t as as well as the normal stress ?_n on the basic parameters are considered as the main index,and constructed BP neural network prediction model for predicting peak shear strength of structural plane.?2?Based on the direct shear test data of 37 groups of Grasselli's 7 type of rocks,the peak shear strength of the rock joints can be predicted quickly by using the prediction model that trained for several times,and the expected results are consistent with the experimental results,which verifies the applicability of the method in application.?3?BP neural network prediction method reduces the complexity of shear strength formula and can quickly and effectively predict the peak shear strength of rock joints according to the influencing factors,it can be used as a tool to predict the shear strength of rock joints and provides a new research method for studying the mechanical properties of rock joints.?3?Compared with the peak shear strength model proposed by Grasselli et al.,Tang et al.,and Yang et al.,the overall prediction result error based on BP neural network model shows smaller error and higher accuracy.?4?In order to verify the feasibility of BP neural network prediction,the rock joints morphological parameters,basic mechanical properties and shear mechanical properties of granite joints samples were obtained through the test,and the data obtained from the test were calculated and sorted out to obtain the samples data used in BP neural network prediction for prediction.?5?With BP neural network prediction method to reduce the complexity of the shear strength formula,and can quickly and effectively according to the influencing factors to predict peak shear strength of structural surface,as a tool of direct shear test in predicting shear strength of rock joints can reduce the repeated work of the test on mechanical properties of the rock joints.Provides a new method to save cost and improve the efficiency of research.