A Data-Driven comparison of Trajectory Prediction methodologies using ADS-B data
V.A. Pereboom (TU Delft - Aerospace Engineering)
Joost Ellerbroek – Graduation committee member (TU Delft - Control & Simulation)
Jacco Hoekstra – Mentor (TU Delft - Control & Operations)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Better predictability of future aircraft locations will result in more efficient air-traffic management operations and therefore an increase in airspace capacity. However, flight paths tend to have a stochastic character causing their predictability to decrease with the look-ahead time. In this research a quantitative analysis is made of the statistical properties of flight trajectory deviations and their dependency on the look-ahead time based on historical ADS-B flight segments and corresponding flight plan data. An empirical boundary on the prediction horizon for future aircraft locations in en-route flights is determined based on a deterministic state propagation method. Using these results, an analysis is made of how these uncertainty distributions affect conflict detection performance. A comparison is therefore made between deterministic, probabilistic and intent based conflict detection methods when applied to medium-term conflict scenarios. Results show that for lower look-ahead time windows up to 10 minutes, a deterministic approach proves to be the most reliable. Only beyond this time window, the use of flight plan information proves to have beneficial effects on the detection performance. The outcome of this research can be used as a benchmark for the development of novel conflict detection approaches.