Research On Nondestructive Testing Technology Of Asphalt Pavement Surface Thickness And Density Based On 3D Ground Penetrating Radar

The thickness and density of the asphalt surface are usually obtained by direct measurement.Although the results are intuitive,they can only be measured at limited points.At the same time,pavement damage will produce weak points of performance,leaving hidden dangers of the disease.Three-dimensional ground penetrating radar(GPR)can transmit and receive electromagnetic waves in a wide range,fast and continuous manner,which is consistent with the detection requirements of road engineering.At present,there are some shortcomings in the measurement of asphalt pavement thickness and density by using 3D Ground Penetrating Radar: the pavement damage is caused by the thickness measurement by core calibration method;the dielectric constant model is used for density prediction,which is verified by the laboratory,lacking practical engineering application,and the model analysis accuracy is not high and the parameters are complex.Therefore,this paper seeks a high-precision,radar-based asphalt surface thickness,and density calculation method,combined with the measurement characteristics of three-dimensional ground-penetrating radar,to achieve rapid,non-destructive,continuous,and accurate measurement of new asphalt surface thickness and density.First of all,the core drilling sampling method and common center point method are used to measure the thickness of new asphalt pavement,and the measurement results are compared with the direct coring method.The error changes and error sources of the two methods are analyzed,and the measurement accuracy and stability of the two methods are evaluated.Secondly,the density prediction accuracy and prediction trend of different theoretical density prediction models in indoor molding specimens are compared;based on the radar scanning signal and volume parameters of indoor forming specimens,the density prediction model of BP neural network is constructed;the accuracy and stability of each density prediction model are evaluated through the field test results,and the prediction model with the best comprehensive effect is selected.Finally,the repeatability test of the same detection section is carried out to evaluate the repeatability of the common center point method and neural network density model in practical engineering.The test results show that the average error of the common center point method is 1.3%higher than that of the core taking calibration method,and the average error of the two methods is less than 4%;the theoretical density prediction model has a similar trend of density and has poor universality for different mixtures.The average prediction error of the neural network density prediction model is 0.184%,which is much higher than other theoretical density prediction models;the common center point method and neural network density prediction model combined with three-dimensional ground-penetrating radar have achieved high accuracy and reproducibility in practical engineering.