Population-based Structural health monitoring: Advancing Foundations, mEthods, and real-world validations

Civil engineering structures such as wind turbines and bridges are critical to economic development, public safety, and environmental sustainability. Structural Health Monitoring (SHM) safeguards these assets by providing essential information on their condition. However, current SHM practices are limited to individual structures, restricting their scalability and their capacity to capture diverse operational and damage-state conditions. To overcome this, Population-Based SHM (PBSHM) has emerged as a promising paradigm that exploits knowledge shared across populations of similar structures, enhancing SHM's accuracy, robustness, and coverage. Yet, PBSHM remains largely unvalidated on complex, real-world applications. This project will advance PBSHM in three steps. First, it will quantify the effects of population size, operational variations, and structural dissimilarity on PBSHM performance, establishing practical thresholds and guidelines for PBSHM applications. Second, it will develop adaptable, interpretable, and data-efficient learning methods tailored for PBSHM by integrating Graph Neural Networks (GNNs), Physics-Enhanced Machine Learning (PEML) and transfer learning. Third, to ensure practical relevance, the developed thresholds, guidelines, and methods will be validated using available real-world datasets, including data from over 30 wind turbines and more than 400 short- and medium-span bridges. The outcomes will deliver accurate, robust, and scalable PBSHM methodologies that enable robust anomaly detection and support proactive maintenance for industry and government partners. Scientifically, the project will extend SHM beyond the conventional individual-based paradigm. Societally and economically, it will promote safer, more resilient, and sustainable infrastructure systems.