US 12,462,693 B2
System and method for flight arrival time prediction
Morten Tobias Jorgensen, Killaloe (IE); and Ayse Gul Ozkan, Hampshire (GB)
Assigned to SITA Information Networking Computing UK Limited, Middlesex (GB)
Appl. No. 17/613,980
Filed by SITA Information Networking Computing UK Limited, Middlesex (GB)
PCT Filed May 27, 2020, PCT No. PCT/GB2020/051279
§ 371(c)(1), (2) Date Nov. 24, 2021,
PCT Pub. No. WO2020/240177, PCT Pub. Date Dec. 3, 2020.
Claims priority of application No. 1907502 (GB), filed on May 28, 2019.
Prior Publication US 2022/0215760 A1, Jul. 7, 2022
Int. Cl. G08G 5/30 (2025.01); G08G 5/52 (2025.01); G08G 5/55 (2025.01); G08G 5/76 (2025.01)
CPC G08G 5/30 (2025.01) [G08G 5/52 (2025.01); G08G 5/55 (2025.01); G08G 5/76 (2025.01)] 19 Claims
OG exemplary drawing
 
1. A computer implemented method for predicting a flight arrival time of a given aircraft flight, between an origin airport and a destination airport, of a given aircraft based on a set of features, the method comprising:
receiving, at an input interface, the set of features and a scheduled departure time of the given aircraft flight from the origin airport;
receiving, at the input interface, a flight plan route comprising a plurality of waypoints indicating a flight path for the given aircraft flight;
determining, by a processor, an estimated standard terminal arrival route for the given aircraft from a plurality of standard terminal arrival routes of the destination airport as the standard terminal arrival route with an entry point that is closest in distance to a final waypoint of the flight plan route;
receiving, at the input interface, an indication of a wake vortex size and an estimated landing order for each of a plurality of aircraft estimated to be flying between the entry point of the estimated standard terminal arrival route and a runway of the destination airport, wherein the estimated landing order represents an order that each aircraft of the plurality of aircraft is estimated to land on the runway of the destination airport;
determining, by a first predictive unit, a predicted time delay of a flight departure time of the given aircraft flight from the origin airport by applying a first trained machine learning predictive model to a first plurality of features of the set of features,
wherein the first trained machine learning predictive model is a first regression model comprised of a first artificial neural network,
wherein the first plurality of features comprises a plurality of samples of airport weather data indicating weather conditions at the origin airport at respective points in time and an actual or estimated arrival delay of the given aircraft at the origin airport from a previous flight of the given aircraft as a percentage of a time duration between a scheduled arrival time of the previous flight of the given aircraft and the scheduled departure time of the given aircraft flight, and
wherein values for features relating to the airport weather data are weighted based on an intensity of the weather conditions;
determining, by a second predictive unit, a predicted time duration of the given aircraft flight from the origin airport to the entry point of a standard terminal arrival route by applying a second trained machine learning predictive model to a second plurality of features of the set of features,
wherein the second trained machine learning predictive model is a second regression model comprised of a second artificial neural network, and
wherein the second plurality of features comprises the estimated standard terminal arrival route, an estimated flight time between the origin airport and the entry point of the estimated standard terminal arrival route, an actual or predicted take-off time for the given aircraft flight, and a distance from the origin airport to the destination airport;
determining, by the processor, a minimum spacing between each pair of adjacent aircraft based on the estimated landing order and the indication of the wake vortex size for each of the plurality of aircraft;
determining, by the processor, an aggregate wake vortex spacing distance for a standard terminal approach route by summing the minimum spacing between each pair of adjacent aircraft;
determining, by a third predictive unit, a predicted time duration of the given aircraft flight from the entry point of the standard terminal arrival route for the destination airport to landing on the runway of the destination airport by applying a third trained machine learning predictive model to a third plurality of features of the set of features, the third plurality of features comprising the aggregate wake vortex spacing distance for the standard terminal arrival route,
wherein the third trained machine learning predictive model is a third regression model comprised of a third artificial neural network, and
wherein the predicted time duration of the given aircraft flight from the entry point of the standard terminal arrival route for the destination airport to landing on the runway of the destination airport is based, at least in part, on a runway configuration of the destination airport during an associated time window;
determining, by the processor, the predicted flight arrival time of the given aircraft flight at the destination airport by adding the predicted time delay of a flight departure time of the given aircraft flight from the origin airport, the predicted time duration of the given aircraft flight from the origin airport to the entry point of the standard terminal arrival route for the destination airport and the predicted time duration of the given aircraft flight from the entry point of the standard terminal arrival route for the destination airport to landing on the runway of the destination airport and a received scheduled departure time of the given aircraft flight from the origin airport; and
outputting, by an output interface, the predicted flight arrival time of the given aircraft flight at the destination airport.
 
18. A system for predicting a flight arrival time of a given aircraft flight, between an origin airport and a destination airport, of a given aircraft based on a set of features, the system comprising:
receiving, by an input interface, the set of features, a scheduled departure time of the given aircraft flight from the origin airport, a flight plan route comprising a plurality of waypoints indicating a flight path for the given aircraft flight, and an indication of a wake vortex size and estimated landing order for each of a plurality of aircraft estimated to be flying between an entry point of an estimated standard terminal arrival route and a runway of the destination airport, wherein the estimated landing order represents an order that each aircraft of the plurality of aircraft is estimated to land on the runway of the destination airport;
determining the estimated standard terminal arrival route for the given aircraft from a plurality of standard terminal arrival routes of the destination airport as the standard terminal arrival route, wherein the entry point associated with the standard terminal arrival route is closest in distance to a final waypoint of the flight plan route;
determining, by a first predictive unit, a predicted time delay of a flight departure time of the given aircraft flight from the origin airport by applying a first trained machine learning predictive model to a first plurality of features of the set of features,
wherein the first trained machine learning predictive model is a first regression model comprised of a first artificial neural network,
wherein the first plurality of features comprises a plurality of samples of airport weather data indicating weather conditions at the origin airport at respective points in time and an actual or estimated arrival delay of the given aircraft at the origin airport from a previous flight of the given aircraft as a percentage of a time duration between a scheduled arrival time of the previous flight of the given aircraft and the scheduled departure time of the given aircraft flight, and
wherein values for features relating to the airport weather data are weighted based on an intensity of the weather conditions;
determining, by a second predictive unit, a predicted time duration of the given aircraft flight from the origin airport to the entry point of the estimated standard terminal arrival route by applying a second trained machine learning predictive model to a second plurality of features of the set of features,
wherein the second trained machine learning predictive model is a second regression model comprised of a second artificial neural network, and
wherein the second plurality of features comprises the estimated standard terminal arrival route, an estimated flight time between the origin airport and the entry point of the estimated standard terminal arrival route, an actual or predicted take-off time for the given aircraft flight, and a distance from the origin airport to the destination airport;
determining a minimum spacing between each pair of adjacent aircraft based on the estimated landing order and the indication of the wake vortex size for each of the plurality of aircraft;
determining an aggregate wake vortex spacing distance for a standard terminal approach route by summing the minimum spacing between each pair of adjacent aircraft;
determining, by a third predictive unit, a predicted time duration of the given aircraft flight from the entry point of the standard terminal arrival route for the destination airport to landing on the runway of the destination airport by applying a third trained machine learning predictive model to a third plurality of features of the set of features, the third plurality of features comprising the determined aggregate wake vortex spacing distance for the standard terminal arrival route,
wherein the third trained machine learning predictive model is a third regression model comprised of a third artificial neural network, and
wherein the predicted time duration of the given aircraft flight from the entry point of the standard terminal arrival route for the destination airport to landing on the runway of the destination airport is based, at least in part, on a runway configuration of the destination airport during an associated time window;
determining the predicted flight arrival time of the given aircraft flight at the destination airport by adding the predicted time delay of a flight departure time of the given aircraft flight from the origin airport, the predicted time duration of the given aircraft flight from the origin airport to the entry point of the standard terminal arrival route for the destination airport and the predicted time duration of the given aircraft flight from the entry point of the standard terminal arrival route for the destination airport to landing on the runway of the destination airport and a received scheduled departure time of the given aircraft flight from the origin airport; and
outputting, by an output interface, the predicted flight arrival time of the given aircraft flight at the destination airport.