A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Machine Learning-Based Deterioration Modeling of Highway Bridges Considering Climatic Conditions
Highway bridges are critical infrastructure systems that facilitate efficient and optimal vehicle flow along the transportation network. However, these assets are rapidly deteriorating because of their ubiquitous nature and insufficient maintenance programs. Climate change is expected to exacerbate this issue. Precise deterioration modelling is essential to maintain sustainable function of highway bridges. Research on condition assessment of highway bridges is increasingly being reported. Nonetheless, few models incorporated climate-based factors. This study aims to create a comprehensive, data-driven deterioration model for highway bridges that considers operational and climatic factors. The study utilized datasets from two main sources: the National Bridge Inventory (NBI) database and the Long-Term Pavement Performance (LTPP) dataset, both managed by the Federal Highway Administration (FHWA) in the USA. Data mining algorithms, feature engineering, and hyperparameter tuning techniques are exploited to calibrate the condition assessment model. The considered models include a k-nearest neighbor, decision tree, random forest, gradient-boosted trees, deep learning, and support vector machine. Evolutionary optimization is utilized to facilitate automated hyperparameter tuning. Several Performance metrics tests are employed to validate the practicality and accuracy of the developed models utilizing a new set of data that was never employed in the calibration process. The gradient-boosted trees model yielded the most promising results, with a mean relative error of 3.6%. In addition, the predictive importance of some climatic factors, especially the freezing index and mean temperature average was signified through the analysis. The developed model can assist transportation agencies in establishing optimal rehabilitation programs for highway bridges.
Machine Learning-Based Deterioration Modeling of Highway Bridges Considering Climatic Conditions
Highway bridges are critical infrastructure systems that facilitate efficient and optimal vehicle flow along the transportation network. However, these assets are rapidly deteriorating because of their ubiquitous nature and insufficient maintenance programs. Climate change is expected to exacerbate this issue. Precise deterioration modelling is essential to maintain sustainable function of highway bridges. Research on condition assessment of highway bridges is increasingly being reported. Nonetheless, few models incorporated climate-based factors. This study aims to create a comprehensive, data-driven deterioration model for highway bridges that considers operational and climatic factors. The study utilized datasets from two main sources: the National Bridge Inventory (NBI) database and the Long-Term Pavement Performance (LTPP) dataset, both managed by the Federal Highway Administration (FHWA) in the USA. Data mining algorithms, feature engineering, and hyperparameter tuning techniques are exploited to calibrate the condition assessment model. The considered models include a k-nearest neighbor, decision tree, random forest, gradient-boosted trees, deep learning, and support vector machine. Evolutionary optimization is utilized to facilitate automated hyperparameter tuning. Several Performance metrics tests are employed to validate the practicality and accuracy of the developed models utilizing a new set of data that was never employed in the calibration process. The gradient-boosted trees model yielded the most promising results, with a mean relative error of 3.6%. In addition, the predictive importance of some climatic factors, especially the freezing index and mean temperature average was signified through the analysis. The developed model can assist transportation agencies in establishing optimal rehabilitation programs for highway bridges.
Machine Learning-Based Deterioration Modeling of Highway Bridges Considering Climatic Conditions
RILEM Bookseries
Banthia, Nemkumar (editor) / Soleimani-Dashtaki, Salman (editor) / Mindess, Sidney (editor) / Assad, Ahmed (author) / Bouferguene, Ahmed (author)
Interdisciplinary Symposium on Smart & Sustainable Infrastructures ; 2023 ; Vancouver, BC, Canada
Smart & Sustainable Infrastructure: Building a Greener Tomorrow ; Chapter: 92 ; 1039-1051
RILEM Bookseries ; 48
2024-02-20
13 pages
Article/Chapter (Book)
Electronic Resource
English
Deterioration modeling for highway bridges
| British Library Conference Proceedings | 1996
Facilities Modeling: Deterioration Models for Highway Bridges
| British Library Conference Proceedings | 1993
Facilities Modeling: Deterioration Models for Highway Bridges
| British Library Conference Proceedings | 1993
Quantitative Deterioration Modelling for Highway Bridges
| British Library Conference Proceedings | 1994