LATAM-SHM-2026

Study of railway vehicle monitoring by infrared thermal cameras: from consideration of sensor technology to detection and tracking algorithms

  • Dumoulin, Jean (Université Gustave Eiffel)
  • Merainani, Boualem (Université Gustave Eiffel)
  • Toullier, Thibaud (Université Gustave Eiffel)

Please login to view abstract download link

Rail transport represents a crucial sector of the economy. In 2021, the freight transport activities in the European Union-27 are estimated to amount of 409.6 billion tonne-kilometre and the passenger transports to 265.2 billion passenger-kilometre. As the demand for rail services continues to grow, safety remains a primary concern. As a critical vehicle component, axle bearing is subjected to intense mechanical and thermal stresses during operation. This component has to be monitored along the railway network. The consequences of undetected hot box can be dramatic, posing imminent threats to the safety of the entire rail system, including the potential for derailment and fires. Generally, hot box detectors incorporate two categories of sensing systems: inboard sensors and those integrated into the track also called wayside detectors. While these systems are indeed effective, the high installation and maintenance costs can be a significant obstacle to their widespread deployment across the rail network. This problem cannot be understated given the fact that a bearing failure can occur within few minutes of passing a hot-box detector without triggering an alarm. Additionally, the reliance on external facilities, like triggers, introduces a dependency that, if compromised, could impact the overall efficacy of the temperature monitoring system. In this study, we first introduce the problematic of way side monitoring of railway vehicle by infrared thermal camera. A discussion is proposed on sensor technologies used and tested versus final application. Preliminary experiments carried out on a railway open to traffic are presented. Results obtained by different image processing approaches are presented and discussed. Conclusion and perspectives are proposed. Acknowledgments Authors would like to acknowledge the BRIGHTER HORIZON project. BRIGHTER has received funding from the Chips Joint Undertaking (JU) under grant agreement No 101096985. The JU receives support from the European Union’s Horizon Europe research and innovation program and France, Belgium, Portugal, Spain, Turkey.