Study On Intelligent Identification Method Of Shear Deformation Of Constant Resistance Energy Absorption Bolt

The northwest of China is mountainous,so the construction of roads,railways,airports and urban construction in this area will inevitably encounter a large number of slope projects.Anchor bolt is simple in construction,stable in structure,economical and efficient,and has been widely used in the field of slope reinforcement engineering.Constant Resistance Energy Absorption(CREA)bolt can effectively solve the problem that traditional bolt cannot adapt to large geological deformation.Nondestructive testing can take advantage of time domain signal characteristics to detect whether there are defects in the anchor bolt anchoring system,and give the size and location of defects.Micro NPR bolt is the most representative of constant resistance energy absorption bolt.Due to its non-magnetic characteristics,the shear deformation prediction of this kind of bolt under the condition of slope support can be realized by the combination of stress wave nondestructive testing technology and neural network.This thesis predicts the shear deformation of Micro NPR anchors by means of convolutional neural networks(CNN),and designs multi-modular convolutional neural networks to improve the accuracy of the prediction of the shear angle and shear position of microscopic NPR anchors.Finally,the prediction results and simulation results of the anchoring platform are verified and analyzed.The main work accomplished is as follows:(1)To understand the propagation characteristics and principles,energy attenuation characteristics and dispersion phenomena of stress waves in microscopic NPR anchors.The study of dispersion phenomena facilitates the selection of excitation signals for suitable modes based on dispersion curves;the propagation and attenuation characteristics facilitate the assessment of shear deformation of anchor rods based on the time point of the reflected waves.(2)Using ANSYS simulation software,the micro-NPR bare rod and anchor bolt anchoring models of different shear deformation were built.Excitation signal and receiving probe are applied to collect waveform signals of various shear models,and the original signal is filtered and analyzed numerically,so as to prove the feasibility of stress wave nondestructive detection method to analyze the shear condition of bolt.(3)A modular convolutional neural network(CNN)based anchor shear type recognition model was developed for both the shear angle and shear position of microscopic NPR anchors,with sub-module 1 used to identify the shear angle of anchors and sub-module 2 used to identify the shear position of anchors.The modular convolutional neural network can improve the prediction accuracy of microscopic NPR anchor shear deformation by fusing the high-precision classification recognition results of different modules.(4)The anchoring experiment platform of Micro NPR bolt with different shear deformation was built,and the shear experiment process and nondestructive testing experiment principle of the bolt were introduced.The experimental platform was used to extract the actual shear acceleration signal and input it into the modular convolutional neural network for prediction.The prediction analysis showed that the modular CNN had higher prediction accuracy,indicating that the proposed method could be applied to the prediction of shear deformation of Micro NPR bolt.