Predictive performance management is at the heart of the SESAR’s Airport operations centre (APOC).* It relies on having access to real time data from various data sources provided by different APOC stakeholders; it then finds the best techniques for making accurate predictions.
As part of its contribution to SESAR, EUROCONTROL launched a study early in 2016 to identify key APOC processes that could be enhanced by data-driven predictions and by machine learning algorithms (DDP & ML). The study’s aim was to develop a case study to illustrate how shared data and advanced analytics could be used successfully to support APOC’s development.
A consortium was selected: it was made up of the University College London, the University of Virginia and Heathrow Airport. Together, they were able to provide the optimum balance between analytical and operational expertise while ensuring access to the best possible available data.
More than a quarter of all passengers at Heathrow Airport connect to other flights. So, it is critical to ensure that all processes involved in their connection journey are optimised. The study team focused on predicting the passengers’ transfer journey using various databases and applying a Classical and Regression Trees (CART) technique to build the predictive model.
An eight-hour live trial took place on 19 July, with two-hour window predictions being made every five minutes. The predictions were checked visually with a camera looking at a specific area in Terminal 5 and the videos filmed were analysed further against actual data.
The live trial clearly demonstrated that such techniques can provide accurate forecasts (together with prediction intervals) which can help flow managers better understand the key factors that influence passengers’ connection time – and they can help improve passenger services in real time.
In addition, better predictions of passengers’ transfer activities can also improve the accuracy and stability of the Target-Off-Block-Time (TOBT) which is critical for optimised Air Traffic Flow Management in Europe.
The live trial also highlighted how all stakeholders benefit from having a joint conversation around the availability, use, and sharing of data. The output from the machine learning technique gave insights beyond anything that a single stakeholder group could learn on its own.
The final report will become available at the end of September.
Tom Garside, Head of Integrated Planning & Performance at Heathrow, said: “This study has demonstrated how the latest analytical techniques, using real time data, can be used to improve the experience of connecting passengers, and to support aircraft punctuality. We are now looking at how we deploy this approach into live operations and to apply similar techniques to other airport processes.”
* SESAR Project 06.03.01 / OFA05.01.01.