Sensing and information extraction towards a population based structural health monitoring (PBSHM) framework for bridges

Abstract

Existing bridge management systems are reliant on visual inspections; however, visual inspections are subjective. Therefore, there much of the state-of-the-art approaches to bridge SHM tend to be data based. Increasingly available datasets have led to the development of machine learning tools; however, bridges are large and complex structures, and collecting data from them can be challenging. Most bridges have insufficient historical data available for machine learning bridge SHM approaches to be applied. Population based structural health monitoring (PBSHM) is a new approach to SHM that aims to facilitate information sharing between similar structures. Applied to bridges, the PBSHM approach could potentially facilitate information sharing between bridges, meaning bridges may no longer be considered in isolation, and datasets could be leveraged between sets of similar bridges. PBSHM has been demonstrated on simple structures with few elements to date. Heterogeneous populations of structures (e.g. bridges) have not been explored yet. Therefore, the work in this thesis investigates the potential for implementing a PBSHM framework to bridges . The work first checks that Irreducible Element Models can be used to describe bridges, and Attributed Graphs can be generated and compared for similarity. For a pool of six bridges, each described with IE models and compared as AG's, mode tests are carried out to check that similarity metrics between pairs of bridges are reflective of the similarity between their dynamic responses. A novel approach to calculating bridge displacement is then proposed that is robust across bridge and load types. The approach is implemented to a set of similar beam and slab bridges, where it is observed that the bridges with high similarity metrics exhibit similar displacement responses from similar loading. Largely, the work of this thesis shows that a PBSHM approach to bridge SHM could potentially be implemented to facilitate knowledge transfer between bridges.

Thesis is embargoed until 31 July 2025.

Date of Award	Jul 2024
Original language	English
Awarding Institution	Queen's University Belfast
Sponsors	Engineering and Physical Sciences Research Council (EPSRC)
Supervisor	David Hester (Supervisor), Su Taylor (Supervisor) & Roger Woods (Supervisor)