Corrosion is a leading cause of structural degradation in civil infrastructure, posing serious risks to both safety and long-term performance. Detecting such damage at an early stage is essential to support preventive maintenance. This study proposes a method for identifying corrosion in a truss bridge through vibration-based signal analysis. The approach integrates wavelet scattering for robust feature extraction and introduces a novel damage indicator (DI) derived from the resulting scattering coefficients. These features feed a decision tree model to automatically classify structural conditions. The methodology was validated experimentally on a nine-bay truss bridge model subjected to artificial corrosion at different severity levels and positions, including a healthy baseline. Results indicate that the proposed DI achieves 100% classification accuracy under the tested conditions and is robust to changes in damage location and severity. This makes the method a practical tool for early-stage corrosion assessment, with potential applications in structural health monitoring and maintenance planning.
Lujan-Olalde et al. (Thu,) studied this question.