Analyses of a structural health monitoring system on bridges through AI approaches

The structural health of bridges is a critical factor in ensuring the safety and longevity of infrastructure. Traditional methods for assessing damage, such as visual inspections, are labour-intensive and prone to human error, leading to delays in maintenance and repair. This research focuses on developing an advanced computer-based system for automatic detection of damage in bridges, leveraging machine learning approaches for acceleration and displacement analysis. A pre-trained deep learning model trained on acceleration measured during vibration tests carried out in laboratory on prestressed concrete beams characterized by different defects or damages, was adapted to detect normal and damage condition on a real prestressed concrete deck. Superstructure decks are made of posttensioned PC beams or composite steel-concrete beams/caissons. The analyses of the accelerations have been performed also when heavy trucks passed on the bridges. This allowed a correlation between the weight of the trucks and the acceleration. The outcomes of this research are preliminary, but they expected to contribute to the development of automated monitoring platforms that provide real-time assessment and continuous structural health monitoring, ultimately reducing the need for manual inspections and improving infrastructure management efficiency.