View Video Presentation: https://doi.org/10.2514/6.2021-2956.vid

This paper describes efforts to autonomously classify flight activities at small general aviation airports. The research team has Automatic Dependent Surveillance – Broadcast equipment installed at several small airports in the continental U.S. Data from these devices are consolidated in an Amazon Web Services cloud-based database and computing platform. A number of steps are taken to filter and clean the original data. Individual location records are stitched together to form flights. Supervised learning and clustering techniques are then applied to these flights to classify them according to their operations types. Because small airports are often connected to flight training schools, a particular focus of the project is classifying activities that are more typical of training procedures, and which would be less prevalent at airports where active training activities were not taking place. At the end of the classification process all detected flights receive a label indicating their operation type. This information can later be used to produce counts per operation per airport.