Tunnel Rock Mass And Lining Structure Defects GPR Forward Simulation And Dielectric Parameter Inversion Research

With the development of the economy,China’s transportation industry has made historic progress and undergone historic changes,the modernization level of transportation has leapt to the forefront of the world.Due to the factors such as terrain,geology,climatic conditions,and construction techniques,tunnels may experience varying degrees of defects during the construction process and even in the later stages of use.The types of common defects include crack,cavity,void,uncomplete,etc.These defects can reduce the safety performance and service life of tunnels,or even cause traffic accidents.In practical engineering,the interpretation of ground penetrating radar data often relies on personal practical experience,which can easily lead to misjudgment.Moreover,radar wave reflection information can only qualitatively analyze the overview of the target body and cannot accurately depict the structure.Therefore,accurate detection of the spatial distribution of defects has become an urgent need for practical engineering inspection.The main work of this thesis includes:(1)Based on the principle of finite difference time-domain method,forward modeling is carried out with Gprmax3.0 software.Different forward models were set up to analyze the influencing factors of radar.At the same burial depth,the larger the target shape,the greater the difference in electrical properties and the more regular the shape,making the reflected signal clearer and detection better;the higher the centre frequency of the antenna,the higher the resolution and the shallower the detection depth,making it suitable for quantitative measurements.Based on the actual situation,it is necessary to select an appropriate antenna center frequency for detection;at the same time,the selection of antenna spacing is equally important,and selecting a target buried depth of 20% is more appropriate.(2)Based on the influencing factors of ground penetrating radar and the actual situation of tunnel rock mass and lining structure defects,typical defects of the tunnel were designed and the response characteristics of radar waves were analyzed.Under the interference of no rebar,the reflected signal of the defect presents an obvious hyperbola shape.When the length and volume of the defect are larger,the amplitude of the electromagnetic wave is larger,the reflected signal is stronger,and the identification model information is more accurate;under the interference of rebar,the strong reflection signal of rebar will mask the defects below.By analyzing typical radar features,theoretical guidance can be provided for the widespread application of radar forward modeling.(3)Reference a parameter inversion network with a structure of Pix2 pix,and verify the performance of the parameter inversion network through a dataset.A geologically meaningless dielectric parameter model was established,forward simulations based on the dielectric parameter model were performed,the dataset was trained and the trained parameter inversion network was validated.The accuracy of the parameter inversion network was verified by comparing the inversion results of geological meaningless datasets using genetic algorithms.Considering the actual situation,radar data that conforms to geological significance was inverted,and dielectric parameter images were reconstructed to further verify the feasibility of the parameter inversion network.(4)Based on the YK347+840～YK347+865 engineering section of Qinyu Tunnel of Lanzhou-Haikou National Expressway(G75),tunnel model test of marl tunnel is conducted through similar materials,and defects are set for radar detection.The processed real radar data are put into the dielectric parameter inversion network for training,and the defect dielectric parameter distribution map is successfully predicted,which proves the applicability of the parameter inversion network to the measured radar data.