Research On GPR Forward Modeling And Signal Processing Of Urban Drainage Pipeline Leakage

Leakage of urban underground drainage pipes brings serious safety hazards.Ground penetrating radar(GPR)has bright application prospects in underground pipeline leakage due to its unique non-destructive,non-contact,high efficiency and high resolution detection characteristics.However,there are few relevant research achievements at present,and a systematic interpretation system of GPR spectrum has not yet been formed.At the same time,signal processing is also a long-standing research direction in the field of GPR.Therefore,based on theoretical research,physical model test,forward simulation and other methods,this thesis studies the characteristics of GPR spectrum and methods of GPR signal denoising for underground UPVC pipelines without leakage and leakage.The main research contents and achievements are as follows:(1)The physical model test of UPVC pipeline under two conditions of no leakage and leakage is conducted by using GPR,and the characteristics of GPR spectrum are summarized and analyzed from the perspective of 2D and 3D.The research shows that when the survey line is perpendicular to the pipeline,the echo is hyperbola;When the survey line is parallel to the pipeline,the echo is in a straight line.After the leakage occurs,there is a oscillation signal in the echo image,and the three-dimensional image can identify the pipeline direction and preliminarily determine the location of the leakage.(2)The forward simulation program of 3D finite difference time domain for GPR based on the absorption boundary of uniaxial anisotropic perfectly matched layer is compiled,and the two models of no leakage and leakage of the pipeline established are simulated.Research has shown that the results obtained from 3D forward modeling are basically consistent with physical model experiments.(3)An improved wavelet threshold method is proposed and compared with hard threshold denoising method and soft threshold denoising method through designed simulation denoising experiments.The research shows that the signal waveform after improved wavelet threshold denoising is the most similar to the original signal waveform,with the signal noise ratio of about 11.66% and 11.27% higher than the hard threshold method and the soft threshold method,respectively,and the root mean square error of about 20.16% and 19.61%lower than the hard threshold method and the soft threshold method,respectively.(4)Combining the improved threshold denoising method with the complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)method,the denoising effects of CEEMMDAN,hard threshold denoising algorithm,soft threshold denoising algorithm and CEEMDAN-improved wavelet threshold denoising algorithm are analyzed and evaluated objectively by simulation data and measured data of GPR.Research has shown that the joint algorithm performs better in denoising simulation data than other algorithms.The peak signal-to-noise ratio of the measured signal after denoising using the joint algorithm is46.6592 d B,which is about 35.56%,15.70%,and 13.24% higher than the CEEMDAN denoising algorithm,hard threshold denoising algorithm,and soft threshold denoising algorithm,respectively.(5)GPR is used to detect leakage of drainage pipe network of Xihua University based on joint algorithm denoising and GPR spectrum characteristics summarized above.The research shows that the combined algorithm is effective in the application of GPR detection engineering,and the characteristics of GPR spectrum summarized in this thesis are correct and effective.