Federated Learning for Railway Safety Analysis and Prediction

Railway networks constitute intricate cyber-physical systems, engaging with various transportation entities and additional cyber-physical systems. For instance, a railway crossing necessitates interactions between the railway network and road traffic management systems. This complex web of interactions between diverse entities presents intriguing challenges for research in effectively modeling railway operations through data analytics and enhancing safety measures at railway crossings using data-driven approaches. In their previous work with partner institution, the University of California Riverside (UCR), the research team adopted the spectral clustering technique to understand emerging accident patterns from historical data and identify underlying similarities in such patterns. The team then adopted kernel ridge regression to predict the number of accidents based on the selected factors from identified patterns, making the first attempt to reduce the number of accidents. However, due to the large volume of data and high computation complexity, the amount of data involved in learning was only 10% of that available. Moreover, the work assumes that all data are centrally available. In practice, data from different transportation entities may not be shared or collected to a center for processing, due to privacy issues. To address these issues, this project proposes to develop a federated learning framework to enable (a) parallel computation when data volume is too large, and (b) local data processing when privacy is a concern.

• Record URL:
  • https://www.utrgv.edu/railwaysafety/_files/documents/research/operations/exhibit-d_utcrs_utrgv_federated_learning_for_railway_safety_prediction_xu_2024.pdf
• Supplemental Notes:
  • The University of California Riverside is a partner for this project.

Language

• English

Project

• Status: Active
• Funding: $72169
• Contract Numbers:

  69A3552348340

• Sponsor Organizations:

  Office of the Assistant Secretary for Research and Technology

  University Transportation Centers Program
  Department of Transportation
  Washington, DC  United States  20590

  University of Texas Rio Grande Valley

  1201 W. University Dr
  Edinburg, TX  United States  78539
• Managing Organizations:

  University of Texas Rio Grande Valley

  1201 W. University Dr
  Edinburg, TX  United States  78539
• Project Managers:

  Stearns, Amy

• Performing Organizations:

  University of Texas Rio Grande Valley

  1201 W. University Dr
  Edinburg, TX  United States  78539
• Principal Investigators:

  Xu, Ping

• Start Date: 20240601
• Expected Completion Date: 20250831
• Actual Completion Date: 0
• USDOT Program: University Transportation Centers Program

Subject/Index Terms

• TRT Terms: Artificial intelligence; Data analysis; Information processing; Machine learning; Railroad grade crossings; Railroad safety; Safety analysis
• Subject Areas: Data and Information Technology; Railroads; Safety and Human Factors;

Filing Info

• Accession Number: 01924856
• Record Type: Research project
• Source Agency: University Transportation Center for Railway Safety
• Contract Numbers: 69A3552348340
• Files: UTC, RIP
• Created Date: Jul 22 2024 8:24AM