Model-Driven GPR Inversion Network With Surrogate Forward Solver

TL;DR

This paper introduces UA-Net, a neural network architecture for GPR full-waveform inversion that integrates a deep learning-based forward solver and unfolds an optimization process, achieving improved accuracy and better generalization to field data.

Contribution

The paper presents UA-Net, a novel fully neural-network-based GPR FWI framework combining a DL forward solver with an unfolded ADMM optimization, enhancing reconstruction accuracy and generalization.

Findings

UA-Net outperforms classical FWI and data-driven methods in accuracy.

The integrated deep learning forward solver enables rapid predictions and gradient computations.

Transfer learning allows effective application to field data.

Abstract

Data-driven deep learning is considered a promising solution for ground-penetrating radar (GPR) full-waveform inversion (FWI), while its generalization ability is limited due to the heavy reliance on abundant labeled samples. In contrast, Deep unfolding network (DUN) usually exhibits better generalization by integrating model-driven and data-driven approaches, yet its application to GPR FWI remains challenging due to the high computational cost associated with forward simulations. In this paper, we integrate a deep learning-based (DL-based) forward solver within an unfolding framework to form a fully neural-network-based architecture, UA-Net, for GPR FWI. The forward solver rapidly predicts B-scans given permittivity and conductivity models and enables automatic differentiation to compute gradients for inversion. In the inversion stage, an optimization process based on the Alternating Direction Method of Multipliers (ADMM) is unfolded into a multi-stage network with three interconnected modules: data fitting, regularization, and multiplier update. Specifically, the regularization module is trained end-to-end for adaptive learning of sparse target features. Experimental results demonstrate that UA-Net outperforms classical FWI and data-driven methods in reconstruction accuracy. Moreover, by employing transfer learning to fine-tune the network, UA-Net can be effectively applied to field data and produce reliable results.