Ground penetrating radar forward modeling of roads based on random media model

Abstract

Road structure is composed of different pavement materials. These materials contain a large number of particles and pores with different sizes, shapes, dielectric properties and spatial locations, which determines the electromagnetic properties of roads. These feature multi-scale and discontinuous characteristics between layers, together with geometric irregularity and random non-uniformity characteristics within layers, therefore random structures will undoubtedly have a negative impact on GPR detection and data interpretation. As a supplement to the experimental observation, the forward modeling based on random media model can provide an economical and effective way for GPR detection of road hidden diseases. In this paper, discrete random media model and continuous random media model are established respectively by using digital image processing technology and stochastic process theory according to the structural characteristics of different layers of media in road structure. On the basis of the established random media model, the GPR response of holes and pipelines are simulated and analyzed by the Finite Difference Time Domain method, and the GPR signal is processed by the synthetic aperture focused imaging method. By comparing with homogeneous layered models, the results show that the forward modeling based on random media model can reflect the characteristics of ground penetrating radar signal of road structure more accurately. PVC pipe is accompanied by obvious multiple waves in the case of water filling. The polarity relationship between reflected wave and direct wave is the key to distinguish whether it is iron pipe or cavity. Synthetic aperture focused imaging algorithm can enhance the recognition of spatial location, size and dielectric properties of the target. The comparison with the field results shows that the simulated results are in reasonable agreement with the measured ones.