Damage localization in bridges using curvature profiles identified from acceleration data via continuous wavelet transform

Damage identification methods based on traffic-induced vibration data have gained significant attention in structural health monitoring of bridges, driven by the need for cost-effective sensing solutions. Recent studies have demonstrated that bridge curvature profiles can be identified from sparse acceleration measurements collected during vehicle passages using standard accelerometers. However, existing approaches for estimating curvature from acceleration data often struggle to suppress dynamic effects induced by moving vehicles. These methods typically rely on low-pass filters with a rigid cutoff threshold, which can compromise accuracy, especially during high-speed vehicle passages. To overcome this limitation, this study introduces a novel approach based on the continuous wavelet transform to isolate the quasi-static curvature profile and effectively remove dynamic components. The method is tested on a model that incorporates vehicle-bridge interaction effects and road roughness. Sensitivity analyses show that the proposed method outperforms standard filtering techniques across various sensor configurations, damage locations, severities, and multiple damage scenarios, even at relatively high vehicle speeds. Validation using field data further confirms the effectiveness and generality of the proposed approach.