Aircraft control systems
AssigneeAirbus Operations GmbH
InventorTing Yu Au
Life of this patent
Estimating the term…
Abstract
A test system for modifying a graph-based trained classifier, configured to output control data for controlling the aircraft system according to a graph model representing the aircraft system. The test system is configured to obtain scenario data, control data, and validation data. The test system generates a custom loss score based on differences between the validation data and the control data and modifies the graph-based trained classifier based on the custom loss score, the scenario data, and the control data. A computer-implemented method for modifying the graph-based trained classifier and a storage medium comprising instructions to perform the method are also provided.
Claims23
A test system for modifying a graph-based trained classifier that is configured to control an aircraft system according to a graph model representing the aircraft system,
- wherein the graph-based trained classifier is configured to output control data for controlling the aircraft system in response to a set of operating inputs representing an operating state of an aircraft, the control data representing a path for the graph model, and
- wherein the test system comprises at least one memory and at least one processor, the test system comprising program code stored in the at least one memory and that, when executed by the at least one processor, causes the test system to:
- obtain scenario data comprising a set of operating inputs representing an operating state of an aircraft;
- obtain control data representing a path for the graph model that corresponds to the scenario data;
- obtain validation data representing a set of candidate paths for the graph model;
- generate a custom loss score based on differences between the validation data and the control data, wherein an absolute value of the custom loss score reflects a degree to which the control data deviates from the validation data;
- and
- modify the graph-based trained classifier based on the custom loss score, the scenario data, and the control data.
A computer-implemented method for modifying a graph-based trained classifier configured to control an aircraft system according to a graph model representing the aircraft system,
- wherein the graph-based trained classifier is configured to output control data for controlling the aircraft system in response to a set of operating inputs representing an operating state of an aircraft, the control data being representing a path for the graph model, and
- wherein the method comprises:
- obtaining scenario data comprising a set of operating inputs representing an operating state of an aircraft;
- obtaining control data representing a path for the graph model that corresponds to the scenario data;
- obtaining validation data representing a set of candidate paths for the graph model;
- generating a custom loss score based on differences between the validation data and the control data, wherein an absolute value of the custom loss score reflects a degree to which the control data deviates from the validation data;
- and
- modifying the graph-based trained classifier based on the custom loss score, the scenario data, and the control data.
A computer-implemented method for modifying a graph-based trained classifier for operating a control system according to a graph model representing the control system,
- wherein the graph-based trained classifier is configured to output control data in response to a set of operating inputs representing an operating scenario for the control system, the control data representing a path for the graph model, and
- wherein the method comprises:
- obtaining scenario data comprising a set of operating inputs representing an operational state of the control system;
- obtaining control data representing a respective path for the graph model that corresponds to the scenario data;
- obtaining validation data representing a set of candidate paths for the graph model;
- generating a custom loss score based on differences between the validation data and the control data, wherein an absolute value of the custom loss score reflects a degree to which the control data deviates from the validation data;
- and
- modifying the graph-based trained classifier based on the custom loss score, the scenario data, and the control data.
Dependencies are inferred from the claim text.