A method for determining the severity of impending losses of separation (LOSs) is proposed. It is based on the FAA (Federal Aviation Administration) separation conformance category for classification of operational errors. A recently proposed short-term conflict detection algorithm for terminal airspace is enhanced with this severity concept. The alerts from the resulting algorithm are compared with the Conflict Alert (CA) currently in the field, which is a legacy system for automated short-term conflict detection. Three complementary sets of aircraft track data are employed. The first set is real-world data with documented LOSs due to operational errors. It allows determination of average alert lead time while providing regression testing of the algorithm. The second set is realistic data from human-in-the-loop experiments with no visual approaches allowed and with known intervention from controllers or pilots available. As a result more objective determination of false alert rate is possible. The third set is real-world data with unknown mixed operations of Instrument Landing System (ILS) and visual approaches but with CA alerting data available from the FAA. The comparison with CA indicates that the algorithm produces a similar total number of alerts but with a much larger safety buffer and a much lower false alert rate. The study also suggests that a high-severity conflict prediction option may be used for aircraft performing visual approaches to satisfy the controller's moral responsibility for those aircraft.
展开▼
