Artificial Intelligence-Aided Life-Cycle Assessment Strategies

The assessment of the fatigue behavior of bridges holds great importance in ensuring the reliability of infrastructure networks, formulating comprehensive maintenance plans, and upholding stringent safety standards. Traditionally, this necessitates the monitoring of the strain or stress response of bridges under daily traffic loads. Nevertheless, the installation and maintenance of strain sensors incur considerably higher costs and demand labor-intensive efforts compared to the utilization of acceleration sensors. In an endeavor to surmount these challenges, this study introduces an innovative indirect sensing methodology harnessing the potential of deep learning architectures. This novel approach aims to substitute real strain sensors with virtual ones, leveraging measured acceleration responses to estimate strain signals. The efficacy of these architectures is shown through the presentation of two numerical case studies and the application of this methodology in a real-world scenario. These demonstrations serve to underscore the suitability and effectiveness of the proposed approach in bridge fatigue assessment.