MCMC Method of Inverse Problems Using a Neural Network—Application in GPR Crosshole Full Waveform Inversion: A Numerical Simulation Study

Ground-penetrating radar (GPR) crosshole tomography is widely applied to subsurface media images. However, the inadequacies of ray methods may limit resolution images, since method a type high-frequency approximation. To solve this problem, full waveform introduced for GPR inversion. inversion computationally expensive. In paper, we introduce trained neural network that can be evaluated very quickly replace intensive forward model. Additionally, error statistically analyzed. We demonstrate methodology ground-penetrating data using Markov chain Monte Carlo (MCMC) method. An accurate model based on Maxwell’s equations replaced by network. This achieves high computation efficiency, which four orders magnitude faster than The result synthetic shows good performance network, greatly improves calculation efficiency.