Digital twin technology for bridge structural health monitoring based on agent model

The rapid development of China’s transportation infrastructure has posed new challenges for the structural health monitoring (SHM) of bridges. While the digital twin technology offers fresh perspectives for SHM, a mature method to generate a suitable digital twin model for bridges is still lacking. This study proposes a method for establishing a digital twin model for SHM of bridges, using the five-dimensional model of digital twins and agent model framework. First, a probabilistic benchmark model is introduced to create an agent model for bridges under operational conditions. This model maps the relationship between the physical bridge and its digital twin by applying a probability reference finite element model (FEM) to adjust the distribution of physical parameters. The probabilistic FEM updates theoretical responses by comparing measurement data with FEM results, then establishes a probability distribution of the virtual bridge’s structural parameters most aligned with the actual response. Second, a vehicle-mounted monitoring system and various sensors are used to collect dynamic response and traffic load data. A probability model is developed to obtain the traffic load distribution on the bridge. Safety thresholds are defined, and the corresponding bridge response under different safety levels is analyzed. Finally, a mapping model between the physical and digital spaces is proposed to generate an accurate digital twin model, reflecting the bridge's structural response under real vehicle loads. This approach enhances the precision of SHM in bridge infrastructure.