Research On Tunnel Lining Signal Processing And Cavity Detection Method Based On Ground Penetrating Radar

With the increasing number of tunnel projects in China year by year,the detection demand for tunnel projects is increasing day by day.However,the micro damage detection method does great damage to the tunnel lining structure and can not ensure the operation safety of the tunnel.GPR can detect underground targets quickly,efficiently and non destructively.It is a means suitable for disease survey and detection in the tunnel lining structure.However,the detection environment of tunnel lining structure is easy to be disturbed and there are many types of underground cavities,there are often missing and misreading when manually interpreting and interpreting the targets in GPR echo images,and the recognition efficiency is too low.In view of the above problems,taking the cavity disease in the tunnel lining structure as the object,this paper analyzes the echo characteristics of the cavity target,retains the target echo signal and filters out other types of noise at the same time.At the same time,based on support vector machine classification algorithm and F-K migration imaging algorithm,the automatic identification and detection of cavity disease area is realized.The main research contents of this paper are as follows:(1)Research on Denoising Algorithm of GPR echo signal of tunnel lining.According to the demand of decoupled wave in GPR echo,SVD transform is used to replace KL transform,improve the current mainstream KL filtering algorithm in wavelet domain,and propose a more universal and accurate direct coupled wave filtering method,namely wavelet domain SVD filtering algorithm,based on forward simulation technology,The denoising performance of wavelet domain KL filtering algorithm and wavelet domain SVD filtering algorithm is quantitatively analyzed to verify the superiority of this method.(2)Construction of tunnel lining structure cavity disease data set.Aiming at the lack of cavity disease data set,the typical identification characteristics of cavity disease signal and non cavity disease signal are analyzed,and the unbalanced data set of cavity disease is constructed.Aiming at the traditional unbalanced data set processing methods,ACC-SMETO is proposed to solve the defects of the traditional sampling algorithm.The effectiveness and superiority of this algorithm are verified by comparing support vector machine model with other traditional sampling algorithms.(3)Research on detection method of tunnel lining cavity disease area.Aiming at the problem of cavity disease area detection in tunnel lining structure,the cavity disease area detection is divided into two steps.Firstly,based on the cavity disease data set,the optimal parameter combination of support vector machine classification model is determined to improve the recognition accuracy of the horizontal range of cavity disease area.Secondly combined with F-K migration algorithm,the horizontal range of cavity disease area is determined and processed by image morphology,the cavity disease detection is realized on the simulated GPR echo and measured data.The feasibility and effectiveness of the cavity detection method proposed in this paper are verified by the intersection merge ratio index and visual inspection.(4)Application of typical GPR signal processing and cavity detection methods in tunnel lining.In view of the deficiency of the current mainstream software in preprocessing the GPR echo data,denoising and automatic detection of cavity diseases,an application software for GPR echo signal denoising and automatic interpretation processing of cavity diseases in tunnel lining structure is designed.Based on the GUI platform of MATLAB,the GPR echo data noise is processed,and the automatic identification and detection method of cavity disease area is integrated with support vector machine and F-K migration imaging algorithm.Combined with the measured data of China Academy of Railway Sciences,the software automatic interpretation function is realized.