Machine learning based airflow sensing for aircraft

Patent metadata

Field	Value
Publication number	US-11119504-B2
Application number	US-201816184554-A
Country	US
Kind code	B2
Filing date	Nov 8, 2018
Priority date	Nov 8, 2018
Publication date	Sep 14, 2021
Grant date	Sep 14, 2021

Abstract

Using a set of airflow sensors disposed on an airfoil of an aircraft, first airflow data including an amount of airflow experienced at each airflow sensor at a first time is measured. Using a trained neural network model, the first airflow data is analyzed to determine an airflow state of the aircraft. In response to determining that the aircraft is in the abnormal airflow state, a control surface and a power unit of the aircraft are adjusted. Responsive to the adjusting, the aircraft is returned to the normal airflow state.

First claim

What is claimed is:

1. A method comprising:
measuring, using a set of airflow sensors disposed on an airfoil of an aircraft, first airflow data comprising an amount of airflow experienced at each airflow sensor at a first time;
analyzing, using a trained neural network model trained to classify airflow data from a first airflow sensor of the set of airflow sensors into either a first state or a second state, the first airflow data to determine an airflow state of the aircraft;
adjusting, in response to determining that the aircraft is in an abnormal airflow state, at least one member from a set comprising (i) a control surface and (ii) a power unit of the aircraft; and
returning, responsive to the adjusting, the aircraft to a normal airflow state.

2. The method of claim 1 , wherein the abnormal airflow state comprises a stalled state.

3. The method of claim 1 , wherein the aircraft comprises a rotary aircraft and the abnormal airflow state comprises a vortex ring state.

4. The method of claim 1 , wherein the aircraft comprises a rotary aircraft and the abnormal airflow state comprises a blade stall state.

5. The method of claim 1 , wherein the abnormal airflow state comprises a disrupted airflow state.

6. The method of claim 1 , further comprising:
measuring, using the set of airflow sensors, second airflow data comprising an amount of airflow experienced at each airflow sensor at a second time, the second time being earlier than the first time; and
training, using training data associating the second airflow data of each airflow sensor with one of (i) the normal airflow state and (ii) the abnormal airflow state, a neural network model.

7. The method of claim 6 , further comprising:
measuring second control input data of the aircraft at the second time, the second control input data comprising a position of the control surface of the aircraft;
measuring second energy consumption data of the aircraft at the second time;
training, by correlating the second airflow data, the second control input data, and the second energy consumption data, a second neural network model;
measuring first control input data of the aircraft at the first time; and
predicting, using the trained second neural network model, the first airflow data and the first control input data, an energy consumption rate of the aircraft.

8. The method of claim 7 , further comprising:
training, by correlating the second airflow data, the second control input data, and the second energy consumption data, a third model;
measuring, at the first time, an attitude of the aircraft;
analyzing, using the trained third model, the first airflow data, the first control input data, the predicted energy consumption rate, and the attitude to determine an optimal energy consumption rate;
adjusting, in response to determining that the predicted energy consumption rate is greater than the optimal energy consumption rate, the control surface of the aircraft, the adjusting causing a control surface setting matching a control surface setting associated with the optimal energy consumption rate;
adjusting, in response to determining that the predicted energy consumption rate is greater than the optimal energy consumption rate, the power unit of the aircraft, the adjusting causing a power unit setting matching a power unit setting associated with the optimal energy consumption rate.

9. A computer usable program product comprising one or more computer-readable storage media, and program instructions stored on at least one of the one or more storage media, the stored program instructions when executed by a processor causing operations comprising:
measuring, using a set of airflow sensors disposed on an airfoil of an aircraft, first airflow data comprising an amount of airflow experienced at each airflow sensor at a first time;
analyzing, using a trained neural network model trained to classify airflow data from a first airflow sensor of the set of airflow sensors into either a first state or a second state, the first airflow data to determine an airflow state of the aircraft;
adjusting, in response to determining that the aircraft is in an abnormal airflow state, at least one member from a set comprising (i) a control surface and (ii) a power unit of the aircraft; and
returning, responsive to the adjusting, the aircraft to a normal airflow state.

10. The computer usable program product of claim 9 , wherein the abnormal airflow state comprises a stalled state.

11. The computer usable program product of claim 9 , wherein the aircraft comprises a rotary aircraft and the abnormal airflow state comprises a vortex ring state.

12. The computer usable program product of claim 9 , wherein the aircraft comprises a rotary aircraft and the abnormal airflow state comprises a blade stall state.

13. The computer usable program product of claim 9 , wherein the abnormal airflow state comprises a disrupted airflow state.

14. The computer usable program product of claim 9 , further comprising:
measuring, using the set of airflow sensors, second airflow data comprising an amount of airflow experienced at each airflow sensor at a second time, the second time being earlier than the first time; and
training, using training data associating the second airflow data of each airflow sensor with one of (i) the normal airflow state and (ii) the abnormal airflow state, a neural network model.

15. The computer usable program product of claim 14 , further comprising:
measuring second control input data of the aircraft at the second time, the second control input data comprising a position of the control surface of the aircraft;
measuring second energy consumption data of the aircraft at the second time;
training, by correlating the second airflow data, the second control input data, and the second energy consumption data, a second neural network model;
measuring first control input data of the aircraft at the first time; and
predicting, using the trained second neural network model, using the trained second neural network model, the first airflow data and the first control input data, an energy consumption rate of the aircraft.

16. The computer usable program product of claim 15 , further comprising:
training, by correlating the second airflow data, the second control input data, and the second energy consumption data, a third model;
measuring, at the first time, an attitude of the aircraft;
analyzing, using the trained third model, the first airflow data, the first control input data, the predicted energy consumption rate, and the attitude to determine an optimal energy consumption rate;
adjusting, in response to determining that the predicted energy consumption rate is greater than the optimal energy consumption rate, the control surface of the aircraft, the adjusting causing a control surface setting matching a control surface setting associated with the optimal energy consumption rate; and
adjusting, in response to determining that the predicted energy consumption rate is greater than the optimal energy consumption rate, the power unit of the aircraft, the adjusting causing a power unit setting matching a power unit setting associated with the optimal energy consumption rate.

17. The computer usable program product of claim 9 , wherein the stored program instructions are stored in the one or more computer-readable storage media in a data processing system, and wherein the stored program instructions are transferred over a network from a remote data processing system.

18. The computer usable program product of

Assignees

• IBM

Inventors

• Zhou Wang
• Hu Huan
• Tan Wei
• Chen Qianwen

Classifications

• G06N3/08 Primary

  Learning methods · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G05D1/0816 Primary

  to ensure stability · CPC title