Inject probabilistic traffic predictions into DCB prototype tools (objective 3.a)
Measure the improvements in term of traffic prediction accuracy (objective 3.b).
|Description of work
P04.01 – Apply Probabilistic Traffic Prediction to ATC Planning. The last proposed part of the project would be to use the model obtained in P03.01 and P03.02 and to implement it into DCB prototype tools. Prototype tools DCB are under development (at the time of writing) in the context of ongoing SESAR activities such as VP720. This work will probably continue in the context of SESAR 2020 PJ9 and related ongoing DCB research. It should be possible with minimal effort to adapt some of these so as to present the data sets generated by WP3. (If adaptation proves too costly then the main HMI features should be reproduced in a dedicated tool). These prototype HMIs can be then used to illustrate to experts in the domain how the probabilistic aspects of the traffic forecast can be represented and applied. The ultimate aim would be to be able to produce and display simulated probabilistic traffic forecasts alongside non-probabilistic traffic forecast (as used currently) for specific situations. If time allows it may be possible to also consider closed loop effects where the mechanisms of demand capacity balancing influence the flights, perhaps by modelling how DCB impacts flights using queues.
P04.02 – Measure the improvements in term of traffic prediction accuracy. We would then measure the improvement obtained by our model. These improvements would be taken into account in the deliverables of WP05 and WP06. Current traffic forecasts do not consider uncertainty numerically. One result of this as they are subject to seemingly random change as time passes. By quantifying the uncertainty, probabilistic traffic forecasts should be less prone to the unexpected variation observed today. This can be tested by, for example comparing successive traffic forecasts for the same place and time as the underlying flight data evolves.