Data Driven Based Bridge Damage Identification Using a Statistical Approach Utilizing Vehicle‐Induced Acceleration Responses

The integration of data‐driven methodologies in bridge health monitoring is gaining increasing attention due to their ability to overcome conventional approaches’ limitations. This study proposes a bridge damage identification method for simply supported bridges, leveraging output‐only vehicle‐induced vibration responses, eliminating the need for controlled excitation. The proposed methodology comprises four key phases: data acquisition (DAQ), feature extraction, feature discrimination, and development of a damage localization index. Acceleration responses from multiple vehicle passages are collected from different bridge states. Initially, baseline data is collected from the bridge in its current condition, which serves as a reference for comparison with future measurements. The method utilizes frequency domain higher‐order spectral moments as damage‐sensitive features (DSFs) to detect structural changes. The Mahalanobis distance ( Δ ) metric is employed as a feature discriminator between healthy and damaged states, leveraging its robustness to variable scaling and correlation structures. A new damage localization index, is introduced to localize the damage within the bridge. The method is evaluated using numerical simulations of a simply supported bridge where vehicle–bridge interaction (VBI) modeling is utilized to generate acceleration response data under different damage scenarios. Further, an experimental validation is conducted on a lab‐scale bridge. The results demonstrate that the proposed method successfully detects and localizes damage in both numerical and experimental settings.