A Data-driven Framework for Damage Diagnosis of Coastal Bridges
Coastal bridges are exposed to multiple hazardous conditions including the corrosive environment, strong winds, storm surges and waves, abutment scour, possible vessel collisions and so forth. All these factors deteriorate the bridges, reduce their load-carrying capacities and even cause the bridges to collapse. A well-known catastrophic example is the I-35 Bridge that collapsed in Minneapolis during the summer of 2007. Therefore, accurate integrity evaluation and damage diagnosis of bridges will significantly enhance public safety and the nation's economic development. This research project proposes a novel data-driven framework to implement damage diagnosis (damage localization and quantification) for coastal bridges. Pattern recognition through supervised machine learning methods is conducted to identify the damage. Different machine learning methods have been tried and evaluated. The Artificial Neural Networks (ANNs) approach has been demonstrated accurate and efficient and thus adopted in the framework. To reduce the computational cost, a two-step diagnosis strategy is used in the framework where damage localization is conducted in the first step through classification and damage quantification is carried out in the second step through classification\regression. Damage sensitive features including the normalized modal frequencies variation, mode shapes variation and modal curvatures are extracted to train the model. The established framework was examined on a Finite Element (FE) model of a 3-span reinforced concrete bridge. It is found that single/multiple damage location presence in the bridge girder and the pier can be detected and quantified with high accuracy. In addition, bridge scour can also be quantified well.
- Record URL:
- Record URL:
-
Corporate Authors:
Louisiana State University, Baton Rouge
Department of Civil and Environmental Engineering
Baton Rouge, LA United States 70803Louisiana Transportation Research Center
Baton Rouge, LA United StatesLouisiana Department of Transportation and Development
Baton Rouge, LA United StatesFederal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC United States 20590 -
Authors:
- Sun, Chao
- Zhang, Zhiming
- Publication Date: 2017-6
Language
- English
Media Info
- Media Type: Digital/other
- Edition: Final Report
- Features: Appendices; Figures; Photos; References; Tables;
- Pagination: 74p
Subject/Index Terms
- TRT Terms: Bridges; Coasts; Finite element method; Machine learning; Neural networks; Structural health monitoring
- Subject Areas: Bridges and other structures; Data and Information Technology; Highways; Maintenance and Preservation;
Filing Info
- Accession Number: 01835755
- Record Type: Publication
- Report/Paper Numbers: LTRC Project Number: 17-4TIRE, State Project Number: DOTLT1000138
- Files: NTL, TRIS, ATRI, USDOT, STATEDOT
- Created Date: Feb 10 2022 9:07AM