The rail industry has a national target to deliver a Public Performance Measure (PPM) of 92.5% by March 2019. In areas where lines are most congested, punctuality has been declining in recent years. Costs associated with performance related delay minutes is estimated to be worth £695 million/year. Therefore, with increasing demands on services and a need to improve the reliability and availability of the railway system network, capacity must be utilised in the most efficient way.
Five university-led teams, including one led by UKRRIN Centre of Excellence in Infrastructure lead – University of Southampton, will collaborate with Train Operating Companies (Greater Anglia, South Western Railway, Southeastern Railway, Great Western Railway, Mersey Rail) and Network Rail to explore novel approaches to using and analysing data to reduce the impact of reactionary delays and improve dwell times. The aim is to help the industry improve network performance by developing tools to support and optimise performance management in order to ultimately improve the customer experience.
Three of the five funded projects are focusing on predicting and avoiding reactionary delays using novel techniques such as AI, machine learning, graph theory and neural network models, and interactive visualisation techniques. The outputs will include intelligent solutions such as decision support systems and software able to predict delays and provide real-time strategies. The other two projects are applying similar techniques to predict variations in dwell time and create an interface for communication with front-line staff.
Data analysis techniques applied in the different studies have the potential to provide more robust information and therefore bring operational savings for Network Rail, the operators and improve the overall experience of rail customers.
For more information, visit the ‘Data Sandbox’ hub or contact Giulia Lorenzini at RSSB.