A short-range prediction model for track quality index

Abstract

In the process of train operation, the interaction between train wheels and railway track underneath leads to changes in track geometry. Changes in track geometry result in track irregularity. For managing the mean value of track irregularity, Railway Maintenance Department in China uses track quality index (TQI) to quantify track irregularity of a unit track section. TQI is used for planning and scheduling railway track maintenance and guiding track maintenance work in China. Based on the characteristics of changes in TQI and in accordance with differential and integral principles, a novel track irregularity prediction technique has been developed in this article. This technique utilizes track waveform data generated by a track geometry car to build different track irregularity prediction models for each unit section and use different models to make predictions for TQIs of different unit sections along the railway track on each day in a future short-range period of time. In this article, the technique was applied to 25 sets of track waveforms generated for Beijing–Shanghai railway up-going track (Jing–Hu up-going track) administrated by Jinan Railway Bureau and errors in predicted TQI were analysed in both spatial and temporal dimensions. The error analysis results show that some maintenance works, such as temping, levelling, and aligning have significant influence on TQI predictions. The methods for weakening the influence of such track maintenance and the usage of TQI predictions were also discussed.