Advancing PEAR: development of a Bridge Benchmark Datasets for PBSHM research

Population-Based Structural Health Monitoring (PBSHM) is an emerging field in Structural Health Monitoring that leverages data from multiple structures to enhance the assessment of individual structures. Unlike traditional SHM, which generally relies on data from a single structure, PBSHM utilises collective knowledge from a population to facilitate increasing the knowledge on an individual structure. Transfer learning enables the inference from a source structure to a target structure within the population. One of the limitations of this method is that a lot of transfer-learning methods require data models that are trained using substantial amounts of high-quality data which can be difficult to obtain. To support PBSHM research, the concept of the Population-based SHM Engineered Asset Resource (PEAR) has been introduced. PEAR is conceptualised as a benchmark dataset containing semi-realistic structures and associated data intended to drive the development and validation of PBSHM methodologies. This work advances the PEAR prototype by developing complete populations for two types of bridges, along with their associated data. The pipelines for generating these populations are presented, detailing how they produce structural data and PBSHM-specific models. Additionally, a simple analysis of the generated populations is conducted, demonstrating their utility in PBSHM research and showcasing the potential of PEAR as a resource for current and future PBSHM research.