| 研究実績の概要 |
This research develops a novel synergistic ground holding algorithm based on real-time air traffic pattern classification and off-line buffer optimization. When the expected airborne holding time will exceed a certain constant buffer value, this excess waiting is set as ground holding, i.e. aircraft are kept on the ground before departure. In our research, we consider various real-world uncertainties to determine the optimal buffer applied by the ground holding program. We then build a simulated database and develop a machine-learning-based traffic pattern classifier which, based on traffic features, predicts the optimal ground holding control parameters and potential savings within mean absolute percentage error of 17.96% of the potential optimal ones.
|
