Laser Triangulation and Deep Neural Networks for Rail Safety Inspections — Phase 2: Framework for Isolating and Reporting Track Changes

Details

• Creators:
  de Oliveira Lima, Arthur ; Edwards, J. Riley ; Barbosa, Ian Germoglio ; Zadeh Arjomandi, Sadaf Shafie ; Palma, Paula ; Dersch, Marcus S.
• Corporate Creators:
  University of Illinois at Urbana-Champaign
• Corporate Contributors:
  United States. Department of Transportation. Federal Railroad Administration. Office of Research, Development and Technology
• Subject/TRT Terms:
  Algorithms Artificial Intelligence Inspection Lateral Resistance Neural Networks Railroad Safety Railroad Vehicle Operations Three Dimensional Flow Track Buckling Track Lateral Strength
• Publication/ Report Number:
  DOT/FRA/ORD-26/03
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States
• Edition:
  October 2020 – December 2023
• Contracting Officer:
  Stuart, Cameron
• Corporate Publisher:
  United States. Department of Transportation. Federal Railroad Administration
• Abstract:
  This research builds upon previous FRA-funded studies using the LRAIL 3D laser triangulation scanning system and Deep Convolutional Neural Networks (DCNNs) to identify track health attributes. The focus lies on developing technology-agnostic methodologies for processing LRAIL data, catering to diverse end-users in the railway sector. At the component level, a track component health index (TCHI) is introduced, offering a numerical evaluation of track health based on component conditions. Specific indexes for ballast, crossties, and fasteners are combined into a global TCHI, providing a holistic understanding of rail superstructure condition. From a system level, the research also introduces the track strength index (TSI) to numerically assess track buckling resistance. Machine vision-based inspections prove valuable for infrastructure owners, aiding in condition detection and tracking. This approach offers a holistic and quantifiable understanding of track health and strength, enhancing maintenance strategies and overall railway safety. Visualization of data supports decision-makers in prioritizing maintenance, potentially reducing the risk of track-caused derailments.
• Format:
  PDF
• Funding:
  693JJ6-20-C-000035
• Collection(s):
  Federal Railroad Administration
• Main Document Checksum:
  urn:sha-512:b74a4f387b2dabe90fe9dbc97ec6b26eddfb4ed98e7a6722e2bbbaa53498a7994e501629587dc4eb97d3140322460e5b1a8176e1372a215acff01ebc7a835cc7
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/89889/dot_89889_DS1.pdf
• File Type:
  [PDF - 2.30 MB ]