Investigating Root Causes of Railway Track Geometry Deterioration – A Data Mining Approach
Tutkimustuotos › › vertaisarvioitu
|Julkaisu||Frontiers in Built Environment|
|DOI - pysyväislinkit|
|Tila||Julkaistu - 3 elokuuta 2020|
Railway track geometry deterioration indicates degradation in the underlying track structures. Monitoring and predicting this behavior are important as is investigating the root causes contributing to the deterioration. Without knowing the causes, assigned remediation might not result in a long-lasting correction. However, there is little research regarding the pragmatic aspects of investigating the root causes of track geometry deterioration utilizing real-world data sources. For this purpose, a new method was explored. After reviewing methodologies, the chosen approach was an association rule data mining method: General Unary Hypotheses Automaton (GUHA). The initial data used in data mining comprise data from asset management and multiple measurement systems, including a track geometry measurement vehicle, a track stiffness measurement device, ground penetrating radar, and lidar. The results of the GUHA data mining are hypotheses based on the initial data and can be used to indicate the most common and uncommon types of structures regarding their track geometry deterioration behavior and the attributes governing the behavior of a certain structure type. Therefore, the GUHA method was found to be a suitable method for investigating the root causes of track geometry deterioration from comprehensive railway track structure data.