Research On Three-dimensional Ground Penetrating Radar Detection Method And Intelligent Comparison Method Of Time-lapse Data Applied To Urban Roads

Three-dimensional ground penetrating radar(3D GPR,3-Dimensional Ground Penetrating Radar)is an omnidirectional,high-resolution nondestructive detection method.It has great potential in monitoring and investigating the buried voids under urban roads.This dissertation studied 3D GPR data acquisition method,data processing method,and intelligent comparison method of time-lapse GPR data in urban roads.The main contents and innovations are as follows:First,combined with the characteristics of urban roads and the structural characteristics and working principles of the 3D GPR data collection system,straight survey lines are preferred.Experimental methods to determine the key collection parameters were put forward.By analyzing experimental data,I work out that optimal antenna height was 4 cm,the maximum trace space was 2.5 cm,and there had to be at lease 3 pairs overlap profiles of adjacent survey lines.Second,the 3D GPR data pre-processing and data registration methods were studied.Time position and spatial position registration method based on the crosscorrelation algorithm was proposed,which obtains good registration results in the field data.Third,the intelligent comparison method of 3D GPR time-lapse data was studied.U-Net was trained end-to-end from data to prediction labels,and a network model with an accuracy of 0.83044 was obtained.By using this trained network,backfilled area in the field data from May 2019 to September 2019 were segmented.By comparing the prediction labels from time-shifted data,a qualitative and a quantitative judging method were given,and the relative increasing rate,relative decreasing rate of the predict labels were 0.28 and 0.07,respectively.Finally,combined with the weather records,it was speculated that the backfill area was refilled after deformation in August due to precipitation,resulting in a change in the distribution of the backfill area.In summary,this dissertation proposed a complete framework including This dissertation studied 3D GPR data acquisition method,data processing method,and intelligent comparison method of time-lapse GPR data in urban roads and laid a foundation for dynamic three-dimensional database under the urban road.