Identification of Safety Critical Scenarios for Airlines using Machine Learning in Filter Trees​

During flight, civil aircraft record data that are analyzed within Flight Data Monitoring (FDM) activities of an airline. Based on the characteristics of the operation and network, the flights can be filtered according to aircraft types, departure and arrival airports, and runways. The considered criteria that describe a set of flights are called a filter. When several filters are considered, they often can be arranged according to their hierarchical structure in a filter tree. Subsequently, every filter in this tree can be assigned certain values that describe the performance, e.g. in terms of safety, of the underlying flights. For example, the mean and the standard deviation of the landing masses can be assigned to any available flights arriving at Munich airport on runway 08L. Finally, outstanding filters can be identified using machine learning algorithms for a (potentially very big) filter tree.     «

During flight, civil aircraft record data that are analyzed within Flight Data Monitoring (FDM) activities of an airline. Based on the characteristics of the operation and network, the flights can be filtered according to aircraft types, departure and arrival airports, and runways. The considered criteria that describe a set of flights are called a filter. When several filters are considered, they often can be arranged according to their hierarchical structure in a filter tree. Subse...     »