Research On The Technology Of Road Subsurface Disease Recognition Based On Ground Penetriting Radar Image

The urban road is the most important infrastructure of a city,and the urban road collapse caused by the underground cavity is the main problem of road maintenance.Ground penetrating radar(GPR)is an effective method for highway subgrade detection.It has many advantages,such as high precision,short acquisition time,less manpower consumption and low detection cost.The defects of GPR in road underground detection are as follows:1.Underground pipelines and underground structures are numerous,and the road damage environment is lack of priori knowledge given by a physical model.The analysis of underground damage is lack of standards,and the results are mainly based on human experience.2.Underground damage analysis can only be carried out on the results of a single survey.The results are often in error because of the surrounding environment disturbance.In view of the above,a physical model of urban road underground damage was established,and it will provide a priori knowledge of damage environment.The aim of the underground damage recognition algorithm is to reduce the dependence of experience.A periodic underground damage detection algorithm is proposed to mitigate the error caused by surrounding environment.The research contents and innovations are as follows:1.In order to study the characteristics of the GPR data in each stage of the dynamic evolution,a physical model was established in the urban road.Through the disturbance of workers,the evolution process of damage under urban road was simulated.Based on the data of density,precipitation and road surface settlement,combined with the data of GPR,the interaction and influence of various factors in the physical model were studied.By comparing the change of GPR data and the parameters,the law of urban road damage evolution can be found.2.In order to solve the problem of underground damage survey mainly relies on experience,a new algorithm based on Hilbert marginal spectrum was proposed.First,the empirical mode decomposition of GPR data was carried out,and the marginal spectrum of each data was calculated.Then the total marginal spectrum of GPR data was obtained.Then,the characteristics of the GPR signals were extracted from the Hilbert marginal spectrum,and the relationship between the characteristics and the density was studied.The empirical functions of the characteristics and the density can be found.Finally,through the change of density,and the underground damage of urban road can be found.Results also show that the algorithm can be used not only for finding underground damage,but also for finding metal pipelines.3.In order to solve the problem of underground damage survey mainly relies on experience,a new algorithm based on kernel matching pursuit was proposed.The algorithm uses modern signal processing techniques.The basic idea is to distinguish the lower dimensions features by increasing the signal dimensionality.The key of increasing the dimensionality is to construct the kernel function dictionary.First,the signal and its characteristics were the parameters of the kernel function,and the redundant kernel functions dictionary was generated.Then,the linear combination of kernel functions is used to represent the signals,and the wavelet kernel function is used as the character.Finally,the empirical function of density was found and urban roads underground damage was found by density change.Results also show that the algorithm is not disturbed by metal pipelines and has a good prospect for detecting underground damage.4.In order to measure the change of urban roads underground damage,an anomaly detection algorithm based on periodic detection was proposed.By detecting the same section of the urban road,we found the change in the GPR images.Because the constant of the surrounding interference,the change in the image was caused by the underground damage,and the change area was the damaged area.In order to compare the changes of GPR images,GPR images should be preprocessed.The preprocessing algorithms include Myriad filtering algorithm,Kirchhoff migration algorithm,meticulous registration algorithm and marker-inner registration algorithm.In order to reduce the background noise of ground penetrating radar(GPR)data,an iterative Myriad filtering algorithm was proposed based on the myriad filtering.This algorithm was based on alpha stable distribution.In the iterative Myriad filtering process,the sliding window was used to calculate the data inside the window.Through iterative filter,the results were more accurate.The experimental results show that the iterative myriad filtering algorithm was better than the myriad filter algorithm,and the signal to noise ratio increased by about 3.5d B.To achieve accurate position of underground cavity,the GPR data needed to do migration.According to the characteristics of city roadbed,the Kirchhoff integral migration algorithm was chosen.The spatial position was found by reflected signal.Based on the Huygens principle,each reflection point on the abnormal medium can be used as a twice reflector source.The ground wave field is the result of the superposition of a large number of reflection points on the ground.The formula of Kirchhoff scattered energy from the same diffraction point.In the process of ground penetrating radar data acquisition,the difference of the data consistency was usually caused by the lost trace,the non-uniform trace interval,the precipitation and the software settings.In order to find out the anomaly of city roadbed accurately,a precise registration algorithm and a standard registration algorithm was proposed.The precise registration algorithm toke the outer trace registration firstly,and the data are supplemented or deleted by the search method to ensure the consistency of the survey line direction.Then,the inner trace registration algorithm based on the conditional entropy was used to ensure the consistency of the radar echo direction.The standard registration algorithm is the simplified of the precise registration algorithm.This algorithm made interpolation between markers,to achieve the goal of image registration.The correlation coefficient between the repaired data and the original data can reach more than 0.9 after reduce 90% data.After data preprocessing,the noise of GPR image was reduced and the resolution was improved.The GPR images were mutually registered.To compare the differences between the GPR data,and to find the location of the anomalies,a radar image comparison algorithm based on the sliding window was proposed.First,set the type and parameters of the sliding window,and then calculate the threshold by the otsu algorithm.The experiments show that the algorithm can get the depth,range of the roadbed anomalies.Through this underground damage recognition and processing measure algorithm of Beijing four ring road,Chaoyang District road and Shijiazhuang municipal road detecting data,underground drilling verification results and damage judgment results are basically consistent.