US 12,136,348 B2
Systems and methods for flight performance parameter computation
Antonio Gracia Berna, Madrid (ES); Javier Lopez Leones, Madrid (ES); Ruben Vega Astorga, Madrid (ES); Maria del Pozo Dominguez, Madrid (ES); and Manuel Polaina Morales, Munich (DE)
Assigned to THE BOEING COMPANY, Arlington, VA (US)
Filed by THE BOEING COMPANY, Chicago, IL (US)
Filed on Feb. 12, 2021, as Appl. No. 17/175,255.
Claims priority of application No. 20382469 (EP), filed on Jun. 2, 2020.
Prior Publication US 2021/0375141 A1, Dec. 2, 2021
Int. Cl. G08G 5/00 (2006.01); G05D 1/00 (2024.01); G06N 3/044 (2023.01)
CPC G08G 5/003 (2013.01) [G05D 1/101 (2013.01); G08G 5/0017 (2013.01); G08G 5/0052 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A device for flight performance parameter computation, the device comprising:
a memory configured to store an aircraft performance model, the aircraft performance model based on historical flight data of one or more aircraft, wherein the aircraft performance model includes a recurrent neural network layer;
a network interface configured to receive real-time time-series flight data from a data bus of a first aircraft; and
a processor configured to:
receive, via the network interface, the real-time time-series flight data;
generate, based on the real-time time-series flight data and the aircraft performance model, predicted aircraft performance parameters;
generate a recommended aircraft setting based on the predicted aircraft performance parameters;
provide the predicted aircraft performance parameters and the recommended aircraft setting to a display device;
automatically update a flight control setting based on the recommended aircraft setting; and
update the aircraft performance model based on the real-time time-series flight data and observed aircraft performance parameters, wherein said update the aircraft performance model causes the processor to:
generate a plurality of candidate aircraft performance models based on the aircraft performance model, a first subset of the real-time time-series flight data, and a second subset of the observed aircraft performance parameters, wherein each of the plurality of candidate aircraft performance models is generated by updating at least one hyperparameter of the aircraft performance model, wherein the at least one hyperparameter is updated based on a comparison of an output of the aircraft performance model and a parameter of the second subset of the observed aircraft performance parameters, and wherein the output is generated based on data from the first subset of the real-time time-series flight data;
provide a third subset of the real-time time-series flight data, and a fourth subset of the observed aircraft performance parameters to the plurality of candidate aircraft performance models to generate a plurality of prediction error metrics; and
select a particular candidate aircraft performance model from the plurality of candidate aircraft performance models based on the plurality of prediction error metrics.