A Novel Predictability Performance Metric and Its Forecast Using Machine Learning Techniques

Trajectory predictability is a paramount cornerstone of trajectory-based operations. The uncertainty in the four-dimensional position of aircraft affects the number of flights that the Air Traffic Control service is able to manage. Consequently, airspace capacity is directly impacted by a poor predictability performance. This paper presents a methodology that forecasts predictability performance in pre-tactical phase for traffic flows. The methodology will allow the Network Manager to establish preventive measures to avoid undesired impact on the flow of traffic in the event of predictability degradation. Moreover, if a lack of predictability is recurrently detected for a volume of airspace, strategic measures could be taken to optimise airspace design.