Machine Learning Approach for Finding Similar Weather-Impacted Situations in En Route Airspace

摘要

In this paper, we present methods of identifying situations of similar weather impacts in en route airspace in the National Airspace System (NAS). A cluster analysis technique is described that exploits the use of path, trajectory, and flow similarity metrics. A total of 13 metrics are defined, and a detailed comparison example is performed for 5 metrics. The research supports a machine learning approach for finding similar weather-impacted situations in en route airspace in the NAS, and using those similar situations as a basis for selecting Traffic Management Initiatives (TMIs) for the current day air traffic situation.