Temperature-based measurement interpretation of critical civil infrastructure
File(s)
Author(s)
Glashier, Theo
Type
Thesis
Abstract
Critical civil infrastructure enables modern society, however it is ageing and under increasing pressure from population growth and more frequent, extreme weather. There is a clear need to utilise limited maintenance budgets more effectively through smarter workflows, informed by the current condition of structures. This thesis tackles civil structural health monitoring challenges that relate to the low quality of data measured on operational structures and the inherent effect of environmental and operational variability on structural datasets. A novel data preparation workflow is introduced to remove data acquisition and sensor anomalies from structural datasets, enhancing their quality and interpretability. The signal processing combines: i) dataset completeness verification; ii) time synchronisation error correction; iii) a new statistical outlier analysis; iv) operational condition identification through a novel traffic load labelling algorithm; and v) measurement noise removal. An iterative regression-based thermal response prediction (IRBTRP) methodology is developed to optimise the regression models required to characterise the thermal response of civil structures. The process of determining the optimal model hyperparameters is automated to enable the IRBTRP methodology to be easily applied across a wide range of applications. Accurate thermal response predictions are provided for a 2-week dataset of a Laboratory Truss, a 2-month commissioning and a 2-year operational dataset of the MX3D Bridge, including periods measured over one year after the training data. The application of damage detection techniques to the ‘as-measured’ and ‘IRBTRP methodology-corrected’ MX3D Bridge datasets shows that controlled and naturally-occurring structural changes in the commissioning and operational bridge measurements, respectively, are detected earlier and with greater certainty by using the IRBTRP methodology. The removal of thermal effects enables the structural changes to be detected on sensors that otherwise do not show a variation in their response. These results validate a temperature-based measurement interpretation approach to enable smarter maintenance interventions for critical infrastructure.
Version
Open Access
Date Issued
2025-05-23
Date Awarded
01/12/2025
Advisor
Buchanan, Craig
Publisher Department
Department of Civil and Environmental Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)