Enhancing Predictive Maintenance Through Detection of Unrecorded Track Work

Predictive maintenance can help infrastructure managers to reduce costs and improve railway availability while ensuring safety. However, its accuracy depends on reliable data from various sources, especially track measurement data. When analysing track data over time, historical maintenance actions must be considered, as otherwise the interpretation of the data would be misleading. This research aims to address inconsistencies in recorded maintenance data by detecting unrecorded track works through track geometry evaluations. The main goal is to provide the foundations for accurate descriptions of track behaviour, supporting the implementation of effective predictive maintenance regimes. As part of the research, three different approaches are analysed and evaluated, whereby two of them are based on cross-sectional analyses and the third one detects track works in longitudinal track dimension. The results show that the CRAB algorithm produces the most statistically significant results. Conversely, the cumulative track geometry-based algorithm provides a homogeneous representation of past maintenance work and a result that is statistically only marginally inferior. Consequently, these two methods are best suited to build the foundation for making accurate cross-sectional conclusions about track geometry behaviour. This allows for the verification and enhancement of existing maintenance databases.

Dieses Werk wurde unter der Creative-Commons-Lizenz Namensnennung 4.0 International (CC-BY 4.0) veröffentlicht und darf unter den Lizenzbedinungen vervielfältigt, verbreitet, öffentlich zugänglich gemacht, sowie abgewandelt und bearbeitet werden. Dabei muss der Urheber bzw. Rechteinhaber genannt und die Lizenzbedingungen eingehalten werden.