Study Of Forward Modeling And Inversion Algorithm Of GPR In Layered Media Based On TLM Method

Ground penetrating radar(GPR),as a kind of fast, effective and non-destructive tool, hasbeen extensively applied to non-destructive testing of road quality. The thickness, watercontent, dielectric properties of structural layers and the void existing in pavement structurecan be obtained using to inverse-analysis of real GPR signals, Construction of forward modelof GPR wave propagation in layered structure and the choice of high efficient optimizationalgorithm are keys for the inverse-analysis. In this dissertation, aiming at improvement ofspeed and accuracy in the forward and inverse algorithms of ground penetrating radar, theforward model based on Transmission Line Matrix (TLM) method and the inverse-analysisbased on1D-TLM stripping layer method is presented. The methods will help in thedevelopment of GPR theory and applications. The main achievements and conclusions of thispaper are summarized below:(1) The model of GPR electromagnetic wave propagating in lossy layered media hasbeen constructed based on Transmission Line Matrix(TLM) method. The simulated signalsreflected from multi-layered structure are obtained by TLM method with the perfectlymatched layer boundary condition, and then compared with that obtained by FDTD method.The comparison results indicate that the forward model is accurate and adaptive forsimulation of GPR wave propagation in layered media structure.(2) Starting from the essence of the inverse problem, the parameter adjustment algorithmis presented based on the layer stripping method and TLM model.(3) Simulated signals and practical radar signals are processed by spectrum estimationtechnology. The results of simulations and experiments shows that signal’s power spectrumwhich obtained by the MUSIC method can accurately estimate the number and frequency ofsinusoidal signals, and the amplitude and phase can be calculated by using the MUSICmethod cooperated the least square method. Above all, noise of signals was effectivelyreduced by using spectrum estimation technology.(4) Real layer structure model is analyzed by using radar signal processing technologyand1D-layer stripping method. The results indicated that errors are lower than8%and can satisfy the precision requirement of the project. The relation of water content and dielectricconstant is discussed in standard sands utilizing GPR method. The new fitting Toppedformula and Alharath formula were obtained by analyzing the experimental data, whichfurther validated that using GPR method can estimate the water content of layer media.