As part of the Rail Safety Standards Board (RSSB) Data Sandbox+ project, our technical director Daniel Chick, recently took part in a virtual ministerial briefing to showcase how data analytics and machine learning (ML) can better predict and mitigate delays across the rail network.
Daniel joined Rail Minister, Chris Heaton-Harris, along with RSSB and Risk Solutions to demonstrate how Zipabout, in conjunction with Birmingham University, have developed a new advanced ML model to support existing operational systems, such as Darwin.
For years Darwin has been used by the UK rail industry as its default information platform, responsible for providing arrival and departure predictions, platform numbers, delay estimates, schedule changes and cancellations in real time. However, as the system is reliant on purely operational inputs, and doesn’t account for other data sources that may influence delays and disruption, inaccurate predictions often lead to a level of distrust between passengers and the operators providing them with service information.
To combat this, Zipabout is working with its partners at Birmingham University to build on the industry’s capability to deliver accurate disruption information to passengers. Together we have created a ML prediction model that uses a vast array of real-time and historic data sources, which have been combined to create a series of 200+ ‘features’ that may influence delays and disruption. The model is then deployed to our unique real-time data processing platform (provided by our technology partner KX).
Using large quantities of rail data in real-time (including train movement and operational data, weather data, passenger demand and cascading disruption), we are then able to generate more accurate delay predictions on a huge scale.
This new approach, which is an extension of existing academic research undertaken by the Centre of Excellence in Digital Systems (CEDS) team at Birmingham University, enables us to create delay predictions that account for many of the internal and external factors that impact upon delays. The ML model automatically accounts for the variations in the influence of these factors on how delays are propagated through the network, both temporally over the course of a working day, or spatially across the different stations on the rail network.
As part of the project, Zipabout has also explored how improved real-time delay and disruption messaging can influence passenger behaviour and improve the customer experience. Through our unique Passenger Connect information service that delivers personalised disruption messaging to passengers, we are exploring how the timing and wording of delay and disruption communications can help set rail passenger’s expectations, and even improve operational efficiency across the network.
We are currently undertaking a real-world customer trial with LNER and will be collating direct customer feedback at scale to help refine and develop the solution further.
In response to the briefing, Rail Minister Chris Heaton-Harris said: “While the current guidance means people should not be using our railways wherever possible, work must continue to build a more reliable and punctual service for the future.
“Harnessing data and new technology is crucial to modernise and improve our transport network, and we are determined to drive innovation through competitions like this. These projects will help
the industry tackle bottlenecks, delays and improve accessibility, and I look forward to seeing the crucial role they can play in improving journeys for passengers.”
Find more information about the Data Sandbox+ projects here