Research short
AI tool predicts delays to keep trains moving
New artificial intelligence tool PREPAIR generates predictions on network delays, enabling operators and rail staff to quickly detect congested areas and reroute.
PREPAIR updates the information available to control staff every 15 minutes.
The National Rail Communication Centre (NRCC) is already using this transformative tool, which is trained on historical Network Rail delay data. Installed on a large screen in the NRCC’s control room, PREPAIR uses machine learning to analyse data from moving trains and create a picture of how existing delays could ripple across the network.
The tool was born from a feasibility study funded through RSSB’s Data Sandbox+ competition. Developed by Frazer-Nash Consultancy with support from Lampada Digital Solutions, it provides an aggregated view of data from trains moving through various timing points across the network. Using the information it captures in this way, PREPAIR can create visualisations of predicted rail traffic. It also updates the information available to control staff every 15 minutes.
In the longer term, having access to this sort of predictive data could:
help operators better understand when and where significant delays are likely
determine how to manage the effects of those delays
provide more comprehensive information to the travelling public.
Frazer-Nash Consultancy is now working with Thales Ground Transportation Systems to refine PREPAIR and ready it for broader use. The installation of this system at the NRCC will be a key stepping stone for rolling it out in control centres around Britain.
Read more on research project ‘Rapid Evaluation and Planning Analysis Infrastructure for Railways (REPAIR) (COF-DSP-08)’, which led to the development of the PREPAIR tool.
View page
