Methodology of air traffic flow clustering and 3-D prediction of air traffic density in ATC sectors based on machine learning models

The increase in the demand for aircraft operations has caused the ATM system to become overloaded as it no longer has sufficient capacity to respond to this increase in demand. For this reason, many projects have emerged with the aim of increasing the capacity of the ATM system through the development of new technologies. This paper proposes a solution that would allow predicting and evaluating the traffic density in a three-dimensional basis in one or several ATC sectors. The final goal of this methodology is to analyse the complexity of these ATC sectors. This paper proposes, first, the two-dimensional structuring of traffic density in a set of air traffic flows identified from historical operational data in the sector of analysis. In the vertical dimension, the traffic will be still structured in flight levels. As a subsequent step, a prediction of this structured traffic density is attempted by means of machine learning models. This proposed methodology will try to facilitate the work of the ATC service by allowing them to have a picture of the air traffic organisation of the ATC sector before the real operation occurs. The application of this methodology will allow the adjustment of the ATC service resources. In addition, it will allow the complexity of the sectors to be assessed, as this complexity will strongly depend on how the traffic is structured.