BRIDGE DAMAGE DETECTION UNDER TRAFFIC LOADING AND ENVIRONMENTAL VARIABILITY USING AI: A REVOLUTION IN INFRASTRUCTURAL ASSET MANAGEMENT

Bridges are vital components of civil infrastructure that must operate safely and reliably. Traditional methods for assessing structural health rely on the concept that changes in a structure's dynamic response may indicate potential damage. However, in the case of bridges, variations due to operational factors (like traffic loads) and environmental factors (such as temperature and humidity) can also contribute to these changes. This makes damage detection more challenging, as a bridge may still be safe while exhibiting changes in its dynamic response due to these factors. If the effects of traffic and the environment are not properly accounted for, it could lead to false positive alerts. This thesis proposes and tests a methodology for detecting and localizing damage in bridges subjected to traffic loads and environmental variability. To apply this methodology, acceleration signals from accelerometers placed on the deck of a cable-stayed bridge in China were analyzed as part of a real monitoring effort. This data bank enabled the implementation of the algorithm on real signals. Due to the non-stationary nature of the recorded data, the Fast Fourier Transform (FFT) is not applicable. Instead, a more modern approach, Variational Mode Decomposition (VMD), is used to decompose the signal into Intrinsic Mode Functions (IMF). The Hilbert Transform is then employed to extract instantaneous frequencies, which represent damage-sensitive features in this context. Furthermore, environmental effects are removed from the damage-sensitive features using Principal Component Analysis (PCA), a method for reducing the dimensionality of datasets while preserving interpretability and avoiding data loss. Finally, damage detection and localization are achieved using a clustering technique (K-means Machine Learning (ML) algorithm). By using symbolic objects to reduce data quantity and applying a moving time window technique to the damage-sensitive features, the proposed approach efficiently and accurately identifies and locates damage under transient vibrational loads in varying temperature conditions. The results demonstrate the method's effectiveness, suggesting its potential application in structural health monitoring of more complex and real structures. In addition, this thesis explores the integration of blockchain technology and smart contracts to further enhance the structural health monitoring process. Blockchain provides a secure and immutable ledger for recording damage data, ensuring data integrity and transparency. Smart contracts automate the response process by triggering predefined repair actions upon certification of damage. This integration not only improves the reliability and security of damage detection but also ensures timely and efficient interventions, thereby enhancing the overall safety and longevity of bridges.