Integrating Ground-Penetrating Radar and FEM Modeling for 3-D Reconstruction and Mechanical Simulation of Road Hidden Defects in Asphalt Pavement
Abstract
Ground-penetrating radar (GPR) provides a fast and nondestructive method for pavement inspection. However, reliably extracting the 3-D geometric structure of hidden voids from radar images and correlating it with structural performance remains a challenging task. This study proposes a 3-D reconstruction and mechanics framework integrating migration-based imaging, signal characteristic analysis, and finite-element (FE) simulation techniques. First, preprocessed radar data undergo processing using the Kirchhoff migration algorithm to regress hidden voids in radar images to their actual locations. Subsequently, time–frequency attributes such as energy, instantaneous amplitude, and instantaneous phase are computed. Defect regions are delineated using normalized attribute volume thresholds derived from K-means clustering, enabling surface reconstruction into a 3-D morphology. Triangulation and convex hull integration yield dimensional and volumetric estimates, with intersection over union (IoU) ratio validation confirming reconstruction accuracy. It can be found that phase-based reconstruction achieves the highest precision across all attributes. The effectiveness of this method was validated using field data. Finally, a 3-D FE analysis of an asphalt pavement under standard moving axle load was conducted based on the reconstructed geometric structure of typical voids from GPR data. Results indicate that void burial depth, horizontal span, and height have a significant influence on stress and deformation fields. Shallow, wide voids enhance pavement tensile response and base layer compressive strength, while increased void height exacerbates stress concentration and weakens structural buffering capacity. This workflow enables consistent nondestructive characterization of hidden defects and mechanical risk assessment, providing actionable support for asphalt pavement maintenance and full-life-cycle management.