Abstract
Usually, railway accidents are caused by train derailment, the mechanical failure of tracks, such as broken rails often caused by lack of railway condition monitoring. Such monitoring could identify track surface faults, such as squats, that act as a catalyst for the track to crack and ultimately break. The research presented in this paper enables real-time identification of railway track faults using image processing techniques such as Canny edge detection and 2D discrete wavelet transformation. The Canny edge detection outperforms traditional track damage detection techniques including Axle Based Acceleration using Inertial Measurement Units and is as reliable as Fiber Bragg Grating. The Canny edge detection employed can identify squats in real-time owing to its specific threshold amplitude using a camera module mounted on a specially designed handheld Track Recording Vehicle (TRV). The 2D discrete wavelet transformation validates the insinuation of the Canny edge detector regarding track damage and furthermore determines damage severity, by applying high sub band frequency filter. The entire algorithm works on a Raspberry Pi 3 B+ utilizing an OpenCV API. When tested using an actual rail track, the algorithm proved reliable at determining track surface damage in real-time. Although wavelet transformation performs better than Canny edge detection in terms of determining the severity of track surface damage, it has processing overheads that become a bottleneck in real-time. To overcome this deficiency a very effective two-stage process has been developed.
Original language | English |
---|---|
Pages (from-to) | 53-60 |
Number of pages | 8 |
Journal | Annals of Emerging Technologies in Computing |
Volume | 4 |
Issue number | 2 |
DOIs | |
Publication status | Published - 1 Apr 2020 |
Keywords
- Railway condition monitoring
- real-time
- Canny edge detection
- Wavelet transformation
- squats