Aircraft maintenance message prediction

Patent metadata

Field	Value
Publication number	US-10787278-B2
Application number	US-201715721494-A
Country	US
Kind code	B2
Filing date	Sep 29, 2017
Priority date	Sep 29, 2017
Publication date	Sep 29, 2020
Grant date	Sep 29, 2020

Abstract

A method and apparatus for maintaining a vehicle, such as an aircraft. A plurality of maintenance messages generated during operation of the vehicle are stored to form a plurality of stored maintenance messages. The stored maintenance messages are filtered to remove from the stored maintenance messages those maintenance messages that are correlated to minimum equipment list actions to form filtered stored maintenance messages. A predicted maintenance message is generated from the filtered stored maintenance messages by applying a machine learning algorithm to the filtered stored maintenance messages. The predicted maintenance message may be used to perform a maintenance operation on the vehicle.

First claim

What is claimed is:

1. A method of maintaining an aircraft, the method comprising:
one or more computing devices:
storing a plurality of maintenance messages generated during a plurality of legs of flights of the aircraft to form a plurality of stored maintenance messages;
filtering the plurality of stored maintenance messages to remove minimum-equipment-list-induced signals;
using machine learning to generate a predicted maintenance message based on the plurality of stored maintenance messages, wherein:
maintenance message data of the plurality of stored maintenance messages and flight deck effect data are employed from a number of previous flight legs to a current leg to produce the predicted maintenance message;
the predicted maintenance message comprises a prediction of a future occurrence of a maintenance message for the aircraft; and
the machine learning is applied after removal of the minimum-equipment-list-induced signals from the plurality of stored maintenance messages; and

providing the predicted maintenance message to a user device, whereby a maintenance operation is performed on the aircraft, the maintenance operation comprising proactive or preventative maintenance relating to the predicted maintenance message.

2. The method of claim 1 , wherein storing the plurality of maintenance messages comprises storing the plurality of maintenance messages generated during a plurality of legs of flights of a plurality of aircraft including the aircraft.

3. The method of claim 1 , further comprising generating a feature set comprising a number of stored maintenance messages of a plurality of aircraft.

4. The method of claim 3 , further comprising extracting feature vectors from the feature set, wherein:
filtering the stored maintenance messages further comprises removing from the feature vectors those of the feature vectors that are correlated to minimum equipment list actions, whereby filtered feature vectors are formed; and
generating the predicted maintenance message comprises generating the predicted maintenance message from the filtered feature vectors by applying a machine learning algorithm to the filtered feature vectors, such that the machine learning algorithm does not train on feature vectors correlated to minimum equipment list actions.

5. The method of claim 4 , wherein extraction of feature vectors comprises determining maintenance messages that are uncorrelated to feature real time events.

6. The method of claim 5 , further comprising using a rule differentiation procedure to distinguish maintenance messages that are correlated to minimum equipment list actions and maintenance messages that are not correlated to minimum equipment list actions.

7. The method of claim 1 , wherein generating the predicted maintenance message from the stored plurality of maintenance messages comprises applying a machine learning algorithm to the stored plurality of maintenance messages to train prediction of future events.

8. A method of maintaining a vehicle, the method comprising:
one or more hardware processors implementing a maintenance predictor by:
storing a plurality of maintenance messages generated during operation of the vehicle to form a plurality of stored maintenance messages;
filtering the stored maintenance messages to remove from the stored maintenance messages those maintenance messages that are correlated to minimum equipment list actions, the filtering forming filtered stored maintenance messages;
generating a predicted maintenance message based on the filtered stored maintenance messages by applying a machine learning algorithm to the filtered stored maintenance messages to train prediction of future maintenance message events for the vehicle, wherein:
maintenance message data of the filtered stored maintenance messages and flight deck effect data are employed from a number of previous flight legs to a current leg to produce the predicted maintenance message;
patterns of errors in the filtered stored maintenance messages and flight deck effects are evaluated to predict future real time events; and
the machine learning algorithm is applied after removal of the maintenance messages that are correlated to minimum equipment list actions, such that the machine learning algorithm does not train on stored maintenance messages correlated to minimum equipment list actions; and

facilitating a presentation of the predicted maintenance message for performing a maintenance operation on the vehicle, the maintenance operation comprising proactive or preventative maintenance relating to the predicted maintenance message.

9. The method of claim 8 , wherein storing the plurality of maintenance messages comprises storing the plurality of maintenance messages generated by a plurality of vehicles including the vehicle.

10. The method of claim 8 further comprising:
generating a feature set from the plurality of stored maintenance messages; and
extracting feature vectors from the feature set, wherein:
filtering the stored maintenance messages further comprises removing from the feature vectors those of the feature vectors that are correlated to minimum equipment list actions, whereby filtered feature vectors are formed; and
generating the predicted maintenance message comprises generating the predicted maintenance message from the filtered feature vectors by applying a machine learning algorithm to the filtered feature vectors, such that the machine learning algorithm does not train on feature vectors correlated to minimum equipment list actions.

11. The method of claim 8 , further comprising using a rule differentiation procedure to distinguish maintenance messages that are correlated to minimum equipment list actions from maintenance messages that are not correlated to minimum equipment list actions.

12. The method of claim 8 , wherein the vehicle is an aircraft and storing the plurality of maintenance messages comprises storing the plurality of maintenance messages generated during a plurality of legs of flights of the aircraft to form the plurality of stored maintenance messages.

13. An apparatus, comprising:
one or more processors implementing:
a real-time event storage storing a plurality of maintenance messages generated by a vehicle during operation of the vehicle to form a plurality of stored maintenance messages;
a maintenance predictor generating a predicted maintenance message based on the plurality of stored maintenance messages, wherein:
patterns of errors in the plurality of the stored maintenance messages and flight deck effects are evaluated to predict future real time events;
the predicted maintenance message comprises a prediction of a future occurrence of a maintenance message for the vehicle of the plurality of stored maintenance messages generated by the vehicle; and
a machine learning algorithm is applied to the plurality of stored maintenance messages after minimum-equipment-list-induced signals have been removed from the plurality of stored maintenance messages, such that the machine learning algorithm is not trained with data correlated to minimum equipment list actions; and

a predicted maintenance message deliverer sending the predicted maintenance message to a device associated with a maintenance personnel to facilitate performance of a maintenance operation on the vehicle.

14. The apparatus of claim 13 , wherein the real-time event storage stores the plurality of maintenance messages comprising a plurality of maintenance messages generated during operation of a plurality of vehicles including the vehicle.

15. The apparatus of claim 13 , further comprising a

Assignees

• Boeing Co

Inventors

• Huber David J
• Stepp Nigel D
• Lu Tsai-Ching

Classifications

• G05B23/0221 Primary

  Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods · CPC title

• G07C5/006

  Indicating maintenance · CPC title

• G07C5/0808

  Diagnosing performance data (testing of vehicles G01M17/00; testing of electrical installation on vehicles G01R31/005) · CPC title

• B64F5/40 Primary

  Maintaining or repairing aircraft · CPC title

• G07C5/085

  using electronic data carriers · CPC title