Detection Of Grouting Defect In Shield Tunnels Using Ground Penetrating Radar

Grouting behind shield tunnel wall directly affects the settlement of surrounding strata and the stability of tunnel structure.At present,there are no reliable grouting defect detection technology and equipment.It is of great practical significance to study the grouting detection technology and method to improve the identification of grouting defects.Based on the principle of ground penetrating radar(GPR),this dissertation focuses on the key technologies of signal processing and information acquisition in the application of GPR in grouting detection of shield tunnels.Firstly,the application effect of six common radar signal preprocessing methods,such as data merging,eliminating the inefficient data,desaturation truncation and digital filtering,on the actual grouting detection image of GPR,is presented.Secondly,the characteristics of four typical reflected curved wave coefficients in grouting detection images of ground penetrating radar are analyzed in detail.According to the difference of curved wave coefficients between interference signals and target signals in different directions,a target signal enhancement method based on curved wave transform directional filtering is adopted to remove the interference in radar images.Experiments on simulation data and measured data show that the method is effective.The strong hyperbolic reflection interference of steel bar in ground penetrating radar image is removed,and the reflection signals of cavity defect are enhanced.Finally,according to the different waveform characteristics of normal and diseased GPR data traces,the time-frequency features of data traces are extracted by the adaptive Gabor sub-dictionary orthogonal matching pursuit algorithm.The comparison with the orthogonal matching pursuit algorithm based on genetic algorithm shows the efficiency of this method.The extracted time-frequency features are used as input of support vector machine.The defect recognition accuracy of simulate data is above 99%,and that of actual data is above 89%.The results of simulation and experiment show that the target signal enhancement method based on Curvelet transform directional filtering can effectively remove the strong hyperbolic reflection interference of steel bar in grouting detection ground penetrating radar image and enhance the reflection signals of cavity defect.Features are extracted as input data of support vector machine by the adaptive Gabor sub-dictionary orthogonal matching pursuit algorithm.The method realizes fast recognition of defect data in a large number of radar data,and improves the interpretation efficiency of radar image.