LNER‘s machine learning experts have developed technology that can help to predict train delays.

The technology has successfully completed a trial at Peterborough and Newark Northgate stations. During the trial, dwell times of LNER trains at Peterborough and Newark Northgate stations improved.

The data that was collected during the trial showed that over 450 potential delays had been avoided.

Following this, LNER is now introducing the technology across its station network, and hopes that it will contribute to keeping trains running on time.

The technology collects data from previous train performance. It then adds factors such as the number of people travelling and the weather conditions. It then uses this information to highlight services which may be delayed at an LNER station.

All LNER staff can access the tool via their work mobile phone. They will use it to identify services which have the potential for delay, as well as the expected cause of this delay. This enables station staff to plan accordingly, and to adapt the support they provide to more closely meet the needs of passengers.

The operator, which experienced major disruption to its services last month and the previous month, hopes that this process will allow train services and passengers’ journeys to run more smoothly.

“We saw an opportunity to provide a solution for our teams, which would help them to not only deliver fantastic service to customers, but also enable them to get ahead of any potential issues. This solution really is the embodiment of the saying ‘to be forewarned is to be forearmed.’ It highlights pinch points, enabling teams to be proactive in reducing the reason for delay.

“The data captured from the trial has been positive, so we’re now rolling out the capability across our whole station network. It’s not just the time saved either, we’re also considering the carbon savings of trains being on the move more and dwelling at platforms less. So, this solution may have even more benefits than initially estimated.”

Steven Lloyd, Machine Learning Product Lead for LNER