Physics-informed neural network for damage identification in railroad bridges

Veluthedath Shajihan, Shaik Althaf

Permalink

https://hdl.handle.net/2142/124660 Copy

Description

Title

Physics-informed neural network for damage identification in railroad bridges

Author(s)

Veluthedath Shajihan, Shaik Althaf

Issue Date

2024-04-25

Director of Research (if dissertation) or Advisor (if thesis)

Chowdhary, Girish

Department of Study

Computer Science

Discipline

Civil Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Physics-informed Neural Network
• Damage Identification
• Structural Health Monitoring
• Railroad Bridges

Language

eng

Abstract

Railroad bridges are a crucial component of the U.S. freight rail system which accounts for moving more than 40 percent of freight in the country, playing a critical role in the U.S. economy. The aging infrastructure coupled with the increasing train traffic poses a safety hazard and risks disruption of services. While identifying damages and performing holistic assessments of railroad bridges remain challenging tasks. This research proposes a physics-informed neural network (PINN) based approach for damage identification and model updating of truss railroad bridges. The proposed framework adopts an unsupervised learning method, leveraging train wheel loads and measured responses at bridge nodes as inputs. The PINN model explicitly incorporates the governing differential equations of system dynamics using a recurrent neural network (RNN) based architecture with a custom Runge-Kutta 4th order (RK-4) integrator cell. This approach enables the identification of damage ratios and localization of damaged members in the bridge. To validate the performance of the proposed approach, a case study is conducted on the Calumet bridge in Chicago, utilizing a simplified 2D model with simulated damage scenarios. The results demonstrate the model’s ability to accurately identify and quantify damage under various conditions while maintaining low false-positive rates. Furthermore, the proposed updating pipeline is designed to seamlessly incorporate prior knowledge gathered from site inspections and drone surveys, enabling a context-aware updating and assessment of bridge’s condition.

Graduation Semester

2024-05

Type of Resource

Text

Handle URL

https://hdl.handle.net/2142/124660

Copyright and License Information

Copyright 2024 Shaik Althaf Veluthedath Shajihan

Owning Collections