Bane NOR has the responsibility for the Norwegian rail network infrastructure consisting of over 4000 km track. Maintaining such a large infrastructure while maintaining a high level of punctuality represents a significant challenge. Today the current standard practices for detecting damage relies on resource intensive manual inspections. Other significant challenges include detecting rolling-stock damage and detecting objects in the rail line that present a potential collision risk.
Bane NOR are upgrading their signal network to use the state-of-the-art European Signalling system - ERTMS. As an essential part of the ERTMS upgrade, trackside fibre optic cable based communication links are being installed nationally. Rail-DAS aims to exploit the same infrastructure to create an acoustic sensing based solution for rail infrastructure condition and collision risk monitoring. The solution will provide early warning of issues, allowing problem resolution before they become critical or cause accidents, and will therefore help to avoid downtime, reduce maintenance and renewal costs, and improve safety on the rail network. The solution will also help to reduce the time spent on the rail lines by inspection and maintenance engineers. The Rail DAS project will involve the collection of fibre optic acquired acoustic data over long time periods, and will develop data processing algorithms that use the data reveal the condition and collision risk monitoring situation over extended sections of rail.
The Rail-DAS project aims to exploit the free capacity of fibre optic communication cables currently being installed in the ground alongside the Bane NOR network, to listen for characteristic acoustic signatures of condition risk (generated as a train passes) and collision risk (generated by objects moving nearby the rails). At the heart of the concept lies Distributed Acoustic Sensing (DAS). DAS consists of a passive optical fibre, that can be 10s of km in length, and an “interrogator” box at one end. The interrogator sends laser pulses along the fibre, and a small amount of the light is scattered from each location along the fibre back to the interrogator. The optical path experienced by the backscatter is modulated by acoustic fields, allowing for disturbances to be spatially resolved to within a few meters along the entire length of the fibre. A profound advantage of distributed fibre optic technology is that the low-cost passive fibre itself functions as the sensor. Furthermore, in contrast to a large numbers of individual point sensors requiring individual power sources, a distributed fibre optic sensor requires power only at the interrogator end of the fibre. Unlike their electrical counterparts, fibre optic sensors are immune to electromagnetic interference, a valuable property for electrically powered railroads like in Norway. By the development and application of machine learning algorithms alongside DAS, Rail-DAS will allow for near-continuous condition and collision risk monitoring over the entire rail network, and allow for preventative maintenance and other interventions to be carried out before problems become critical. The main elements of Rail-DAS are:
1) Real-time, network wide detection and location of rail infrastructure damage using DAS
2) Real-time, network wide detection and location of rolling stock with wheel damage using DAS
3) Real-time, network wide detection and location of objects on the rail line using DAS