Electromagnetic wave response and intelligent recognition of defects at reconstructed and expanded roadbed junctions

During the reconstruction and expansion of expressways, defects at the roadbed junction can compromise driving safety and significantly reduce the service life of the road. Based on engineering cases, a generalized model of the defective reconstructed and expanded roadbed junction was developed, and the propagation simulation of electromagnetic waves in the defective roadbed junction was performed using the finite-difference time-domain (FDTD) method. The simulation results demonstrated that the electromagnetic waves formed two sets of parallel convex hyperbolas at the circular cavity defects. The presence of non-compactness defects caused the overall reflected wave signal to exhibit an imaging characteristic with a clear upper section and a blurred lower section. In addition, electromagnetic waves manifested as multiple nearly parallel convex hyperbolas near the vertical cracks. On this basis, by integrating numerical simulation results with field-measured data, a comprehensive dataset encompassing various types of defects was established. Following the optimization of the YOLO algorithm training model, the identification accuracy rates for void, non-compactness, and crack defects reached 97%, 99%, and 99%, respectively. The new method proposed in this study has universal reference value and application potential for road defect detection under different geological conditions and construction standards.