Study On Mechanism And Characteristic Perception Of Cracking And Water Seepage Of Highway Tunnel With Void Lining

Lining void is one of the common structural hidden diseases of highway tunnels,which can easily lead to uneven stress and cracking of the structure,and then the water leakage on the lining surface is formed under the action of groundwater,which seriously affects the health status of tunnel structure and road operation safety.Therefore,the study on the formation mechanism and disease characteristics of lining cracking and seepage has important theoretical significance for tunnel health monitoring,state evaluation,disease prevention,pre-maintenance and traffic control.This paper follows the technical route of disease occurrence conditions→formation mechanism→consequence influence→feature perception,and adopts the methods of field investigation,model test,machine learning and numerical simulation to systematically study the formation process and disease characteristics of lining cracking and seepage phenomenon of highway tunnel lining under cavity conditions.The main results are as follows.(1)Through the investigation of the structural health status of 41 operating tunnels on the national and provincial highways in Chongqing,it is found that there is a strong correlation between the cavity behind the lining and the apparent cracks and water leakage diseases,and the frequency of occurrence is: vault > arch waist > side wall.The form,size,location and filling type of void have a significant indigenous influence on the degree of lining damage.The 1:1 indoor model test and finite-difference time-domain(FDTD)numerical simulation based on ground penetrating radar(GPR)sensing were carried out,and the quantitative difference characteristics of GPR reflection spectrum images under different void parameters were obtained.The artificial intelligence algorithm based on support vector machine(SVM)for lining void filling identification was established,which can effectively distinguish the types of common void fillings,and the recognition accuracy is more than 90%,providing a data visualization reference for the targeted prevention and treatment of tunnel structures.(2)The cracking process of concrete lining under bending load is simulated by four-point bending test.The dynamic crack propagation process and failure mode of concrete lining with defects are obtained by digital speckle imaging(DIC)non-contact observation technology.The control effect of initial structural defects such as insufficient thickness and initial cracks on cracking behavior is also obtained.Cracking failure process of specimens can be divided into linear elastic stage,micro crack initiation stage,macro crack propagation stage and complete fracture stage.Some parameters such as crack initiation strain,fracture energy,bearing capacity is positively correlated to defect degree of specimen.Based on the model of traction-separation cohesive behavior and the criterion of maximum tensile stress,the results of numerical simulation by extended finite element method(XFEM)agree with model test well while the results are sensitive to damage stability coefficient.(3)Two-dimensional and three-dimensional analysis models were established by using the extended finite element method(XFEM)and the stress characteristics and cracking mechanism of tunnel lining under the conditions of initial voids and cracks were studied.It is found that the size effect of viod is prominent,and the arch waist position is more obvious than the vault position.The bearing capacity of the lining decreases with the increase of the initial crack depth.The initial crack has little effect on the failure mode of the lining.With the decrease of the initial lining thickness,the lining is more prone to oblique cracks,and the brittle failure characteristics are more obvious.The cracking mechanism of lining is related to the shape of void space.The propagation speed of circumferential crack is less than that of longitudinal crack.When the length and width of void are more than 4.5 m and 3.0 m respectively,longitudinal and circumferential cracks appear bending phenomenon.(4)Based on transient fluid-solid coupling theory and stiffness reduction principle,a finite element analysis method suitable for simulating local water leakage of cracked lining is established.Through the analysis of the response law of pore water pressure,surrounding rock stress and lining internal force,it is found that the influence period of seepage water is about 40 days,and the area of about 30° on both sides of seepage position is a serious influence area.Local leakage has the most obvious influence on the internal force at the arch waist of the structure,which is prone to form bias effect,stress concentration and cracking on lining structure.There is a superposition effect of leakage water effect.When leakage occurs at both sides of the arch waist,both sides of the wall and the vault,it is close to the result of uniform leakage of the lining.The sensitivity of initial defect of voided lining to seepage effect is: cracking parameter < seepage location < void combination effect.(5)Aiming at the point and line leakage of cracked lining,an indoor test device was developed to simulate the humidity,temperature and illumination environment of tunnel operation.Infrared thermal radiation characteristics of lining surface under different seepage velocities were observed by infrared thermal imaging technique.Infrared images are processed by gray processing,morphological filtering and binary processing to realize the recognition and calculation of leakage area on concrete surface.According to the variation law of temperature field in the leakage area and the geometric shape characteristics of thermal image,the basic crack types of the lining at the outlet of the leakage water can be roughly inverted,which provides a new idea for identifying the structural defects of the leakage lining in practice.(6)Based on the deep learning theory of full convolution neural network,the Bottleneck CSP structure was improved by adding attention mechanism in the feature extraction layer,and the fourth detection layer was added in the feature detection layer.An improved tunnel leakage hydrothermal imaging detection algorithm based on YOLOv5 s network structure was proposed.The improved tunnel seepage detection algorithm has the accuracy of 96.15%,the recall rate of 93.03%,and the average precision of 95.19%.The detection time reaches 0.027 seconds/frame,and the weight is only 16.8MB.It can meet the accuracy,real-time performance of tunnel seepage detection,and the lightweight requirements of mobile devices for the algorithm at the same time.Especially,it has the advantages of accuracy and convenience in detecting the tunnel vault area with high position.