A predictive aircraft landing speed model using neural network

Expected increases in air traffic demand have stimulated the development of automated tools intended to assist the air traffic controller in accurately and precisely spacing aircraft landing at congested airports. Such tools require an accurate landing-speed prediction to increase throughput while decreasing the need for controller' interventions for avoiding separation violations. There are many practical challenges to developing an accurate landing-speed model that has acceptable prediction errors. This paper focuses on a near-term implementation, using readily available information, to model the final approach speed profile from the top-of-descent phase of flight to the landing runway. The developed models accurately predicted the landing speed, for the MD-80 aircraft type operations at the Dallas/Fort Worth airport, 95% of times with error margins of 12.6% for the low-and-no gust and 12% for high gust conditions, respectively. Also, the models reduced the uncertainties of the landing speed predictions by at least 9.5% for both gust conditions from the current state-of-the-art predictions.