About
Large-scale accessible datasets are vital for further progress and algorithm development in the railway field. Unfortunately available datasets are rare and the vast majority of academic work is based on the simulation of vehicle-track interaction. While this provides a solid base for deep understanding of transient events it is not as optimal for practical maintenance, as it still requires a human operator to interpret the collected data, the amount of which can easily prove overwhelming. This could be overcome by using real, non-ideal datasets as a basis for simulations, algorithm development and method verification, eventually leading to the creation of reliable algorithms that could be applied directly to real-time measurement systems and enabling autonomous analysis of the state of the track. This paper introduces a dataset built for this purpose. Data were collected over a year from a track-side inertial measurement system that captured several vehicle-track interactions a day on a double crossover turnout. We encourage researchers to process the data and provide feedback regarding potential improvements as the system can be adapted to suit their needs.
Publications
- Sramota, Jan; Skavhaug, Amund. (2018) RailCheck: A WSN-Based System for Condition Monitoring of Railway Infrastructure. 2018 21st Euromicro Conference on Digital System Design (DSD).
- Sramota, Jan; Lundteigen, Mary Ann; Petersen, Stig; Skavhaug, Amund. (2019) RailCheck: Functional Safety for Wireless Condition Monitoring of Railway Turnouts and Level Crossings. 2019 IEEE Intelligent Transportation Systems Conference (ITSC).
- Journal Paper - Work in Progress
Dataset
This website presents an ongoing long-term test carried out on railway track in Norway and the resulting dataset. The RailCheck dataset is published under the 'All Rights Reserved' licence and is available for non-commercial academic and research use. Institutions will be granted access based on their request, and the licence permits publication of the outcomes of their work. A small proportion of the data will also be made available under the same licence without registration for demonstration purposes - dataset.