United States Department of Transportation
i
Earthquake-Induced Damage Classification of Bridges Using Artificial Neural Network
-
2024-09-01
-
[PDF-12.09 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report January 2021 – June 2024
-
Corporate Publisher:
-
Abstract:Ensuring the resilience of bridges against earthquakes is crucial for the rapid recovery of communities. Speedy and accurate damage assessments are essential for pinpointing the most affected areas and facilitating prompt decision-making. Current methods (i.e., fragility curves and estimates) are often slow and inaccurate. To address these issues, we developed a novel damage evaluation model for highway bridges using two three-layer Artificial Neural Network (ANN) models: A and B. These models predict earthquake-induced bridge damage levels (none, slight, moderate, extensive, and collapse) of a particular type of bridges based on intensity measures from historical ground motion records. Key highway bridges were selected for modeling, and their damage states were assessed using a damage index through nonlinear time history analysis (NLTHA). This generated a comprehensive dataset for training, validating, and testing the ANN models. Our approach significantly outperforms traditional fragility estimations by offering rapid and accurate damage predictions. Model A and Model B achieved 90.74% and 95.43% accuracies, respectively. This improved accuracy enhances immediate response efforts and supports proactive decision-making, thereby bolstering earthquake preparedness and resilience.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
