Study On Spatiotemporal Evolution Characteristics Of Tension Compression Deformation Of Anchor Body Based On Convolution Neural Network

The creep characteristics of rock are the hotspot and focus of rock mechanics research.Anchoring engineering plays an important role in the field of surrounding rock support in deep underground space.With the passage of time,the creep effect of anchoring surrounding rock under complex stress environment is increasing day by day.Obviously,it has a huge impact on the safety of deep underground anchoring works.In this paper,the creep characteristics of anchors under tension-compression bidirectional stress loading environment are systematically studied by using a combination of laboratory tests,digital image correlation and deep learning.Firstly,the instantaneous mechanical properties of similar materials of rock mass were studied,the strength characteristics of the anchor under tension and compression were obtained,and the deformation law and failure mechanism of the surface of the anchor under tension and compression were analyzed;then the design and development improved A set of test systems and methods for the strain change characteristics of rock bolt surrounding rock under long-term compression-pulling action.The creep characteristics of similar materials in rock mass are studied,and three methods of visual analysis,spatial analysis and digital image correlation are used to focus on the discussion.The effect of different tension levels on the axial deformation of the anchor under the condition of tension-compression coupling was studied.Finally,a speckle image data set on the surface of the anchor was made,and the VGG convolutional neural network was used to classify the data set of creep characteristics.Training.The research results show that:(1)The uniaxial compressive strength of the anchored body is higher than that of the non-anchored body,and its compressive strength increases with the increase of the bolt size;the long-term strength of the anchored body is higher than that of the non-anchored body.Compared with the non-anchored body,the anchored body has a good ability to restrain its creep deformation;(2)The cumulative elastic strain increment,cumulative creep strain increment and cumulative total strain increment of anchor solid under pressure and tension bidirectional loading condition all increase gradually,and with the increase of load grade,the elastic strain increment of anchor solid at each stage gradually decreases,and the decreasing amplitude also decreases gradually.The increment of creep strain in each stage of anchor solid increases gradually,and the increment amplitude also increases gradually;(3)Tensile has a significant effect on the strength and deformation of specimens under compression-tension biaxial loading.With the increase of constant tensile load,the failure of specimens will accelerate to a certain extent.When the lateral tension is small,it has the ability to restrain the vertical deformation of the sample to a certain extent.When the lateral tension is large,it has the ability to promote the development of the vertical deformation of the sample.The overall vertical deformation amplitude will also increase;when the lateral tensile force is small,the lateral deformation of the sample is basically in a stable state,and the deformation amplitude is less obvious;when the lateral tensile force is large,the lateral deformation of the sample is relatively small.Obviously,there is a certain ability to promote the development of lateral deformation of the sample,and its lateral deformation capacity will increase with the increase of the vertical load level;(4)Different from the splitting failure without anchor body,some splitting failure occurs in the anchor body under pressure and tension bi-directional loading,but the tensile shear failure is the main failure,and the crack development trend of strain cloud map is consistent with the crack distribution on the surface of the specimen after failure.(5)11 categories of creep processes under five-level hierarchical compressive loads are designed,an image dataset of anchor creep stages under tension-compression coupling environment is made,and an anchor creep based on convolutional neural network is proposed.The image classification and recognition method of the changing process,and through network training,it can quickly and accurately predict the new image category of the same type,that is,the creep stage of the prediction test piece.The research results further enrich the relevant theories in the field of rock mass creep and provide a reference for the prevention and control of anchored rock mass creep.