Research On Signal Processing Method Of Ground Penetrating Radar For Dam Leakage Point

The quality of dams is vital and closely related to the safety of people's lives and property.At present,a considerable part of Chinese earth-rock dams built in the 1970 s and 1980 s have a long history and have been out of repair,and there are significant potential safety hazards.Among them,the dam leakage point is one of the hidden dangers,and the harm is enormous,even causing dam breakage and other major disasters.The main causes of seepage are the erosion of plant roots and the digging of animals.Ground Penetrating Radar(GPR),as a non-destructive testing method,can effectively and quickly detect dams and detect hidden dangers and eliminate hidden dangers in advance.Because of the complexity of instrument and underground medium,GPR feedback signal often has noise,clutter and artifact,and has attenuation effect on the feedback signal.Therefore,GPR signal processing is of great significance to identify dam leakage point more effectively.GPR detects objects by transmitting electromagnetic waves underground and receiving feedback signals.Therefore,it is vital to study the propagation of electromagnetic wave in underground media.Based on electromagnetic theory,a forward simulation method of ground penetrating radar(GPR)based on finite difference time domain(FTDT)is proposed.The dam leakage point model is built.The dam pore leakage model and air crack model are simulated,which provides data support for subsequent GPR signal denoising and image enhancement.Aiming at the characteristics of ground penetrating radar signal including noise,clutter and artifacts,and the signal is composed of several different morphological components,a GPR signal denoising algorithm based on morphological component analysis(MCA)is proposed.The choice of the dictionary in the MCA algorithm is vital.In view of the fact that a single dictionary cannot meet the requirements of complex GPR signals,an over-combined combination dictionary based on unsampled wavelet dictionary and curvelet dictionary is proposed.In addition,a new adaptive linear correction threshold strategy is proposed,which realizes the adaptive iteration of MCA algorithm and reduces the number of iterations.In view of the attenuation effect of electromagnetic wave when it encounters small-scale heterogeneous bodies in the propagation process of underground hybrid media and the poor image resolution of GPR signal denoising.A ground penetrating radar image enhancement algorithm based on modulus maxima is proposed.The modulus maximum algorithm firstly detects the edge of GPR image and enhances the image according to the detected edge points.For edge detection,the threshold strategy of modulus maxima algorithm is improved,which not only realizes the effective detection of image edge,but also suppresses artifacts.