Railroads have long used track inspection vehicles that capture data on a defined interval as the vehicle travels longitudinally down the track. These data can be captured on a constant interval, near-constant interval, or temporally. The data are normally evaluated in real time against exception thresholds, which define locations of safety or maintenance concern. In addition, railways evaluate multiple inspection runs of the same track over time to determine the rates of degradation. Since most inspection vehicles rely on the operator input of location events (mile posts, signals, switches, etc.) and tachometer-based distance measurement, the resulting inspection data are often longitudinally misaligned, on occasion by as much as several hundred feet. This paper presents an application of classical and advanced time-series methods for aligning data longitudinally. It also presents a novel methodology for aligning data with a nonconstant sampling interval, utilizing a combination of classical techniques.
Methods for aligning near-continuous railway track inspection data
2020-08-01
13 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch