Search

CN-114626401-B - Maintenance computing system and method for aircraft using predictive classifier

CN114626401BCN 114626401 BCN114626401 BCN 114626401BCN-114626401-B

Abstract

A computing system (10) includes a processor (12) and a non-volatile memory (20) storing executable instructions that, in response to execution by the processor (12), cause the processor (12) to execute an inspection classifier (22) including at least a first artificial intelligence model (22 a), the inspection classifier (22) configured to receive runtime event input data (28A-C) from a plurality of data sources associated with an aircraft, the data sources including structural health monitoring sensors equipped on the aircraft, extract features (22 f) of the runtime event input data (28A-C), determine a predicted inspection classification (54A) based on the extracted features (22 f), the predicted inspection classification (54A) being one of a plurality of candidate inspection classifications (30 Aa-C), and output the predicted inspection classification (54A).

Inventors

  • G. J. wells
  • G.E. Jogson

Assignees

  • 波音公司

Dates

Publication Date
20260508
Application Date
20211214
Priority Date
20201214

Claims (17)

  1. 1.A maintenance computing system (10), comprising: a processor (12) and a non-volatile memory (20) storing executable instructions that, in response to execution by the processor (12), cause the processor (12) to: -running an inspection classifier (22) comprising at least a first artificial intelligence model (22 a), the inspection classifier (22) being configured to: Receiving runtime event input data (28A-C) from a plurality of data sources associated with a vehicle, the data sources including structural health monitoring sensors equipped on the vehicle; extracting features (22 f) of the runtime event input data (28A-C); Determining a predicted inspection classification (54A) based on the extracted features (22 f), the predicted inspection classification (54A) being one of a plurality of candidate inspection classifications (30 Aa-c), and Outputting the predicted inspection classification (54A); Wherein the processor (12) is configured to: receiving user input (38) of an employed check classification (30A) for the runtime event input data (28A-C), and Performing feedback training of the first artificial intelligence model (22A) using the runtime event input data (28A-C) and the employed inspection classification (30A) as a feedback training data pair (31A), and Running a repair classifier (24) comprising at least a second artificial intelligence model (24 a), the repair classifier (24) being configured to: receiving run-time inspection input data (30A, 48A-C) including inspection related input data (48A-C) and said employed inspection classification (30A); extracting inspection features (24 f) of the runtime inspection input data (30 a,48 a-C); determining a predicted repair class (54B) based on the extracted inspection features (24 f), the predicted repair class (54B) being one of a plurality of candidate repair classes (30 Ba-d); outputting the predicted repair classification (54B); receiving user input (38) of a employed repair class (30B) for said run-time inspection input data (30A, 48A-C), and Feedback training of the second artificial intelligence model (24 a) is performed using the inspection related input data (48A-C) and the employed repair classification (30B) as a feedback training data pair (31B).
  2. 2. The maintenance computing system (10) of claim 1, wherein the inspection classifier (22) has been trained on inspection classifier training data (27) comprising inspection training input data (29A) and associated inspection ground truth labels (29B), the inspection training input data (29A) comprising structural health data from one or more structural health monitoring sensors equipped on the vehicle, and the inspection ground truth labels (29B) are inspection classifications (130A) of user inputs (38) associated with the inspection training input data (29A), the inspection classifications (130A) of the user inputs (38) selected from the plurality of candidate inspection classifications (30 Aa-c).
  3. 3. The maintenance computing system (10) of claim 2: Wherein the inspection classifier training data (27) further includes at least one of camera images, audio data, or dimensional measurements, and Wherein the runtime event input data (28A-C) further includes at least one of camera images, audio data, or dimensional measurements.
  4. 4. The maintenance computing system (10) of any of claims 1-3, wherein the one or more structural health monitoring sensors are selected from the group consisting of inertial accelerometers, inertial gyroscopes, strain gauges, displacement transducers, air velocity sensors, and temperature sensors.
  5. 5. The maintenance computing system (10) of claim 1, wherein the repair classifier (24) has been trained on repair classifier training data (47), the repair classifier training data (47) comprising repair training input data (49A) and associated ground truth labels (49B), the repair training input data (49A) comprising imaging learning and electrical measurements, and the ground truth labels (49B) being user-entered repair classifications (30B) associated with the repair training input data (49A), the user-entered repair classifications (30B) selected from the plurality of candidate repair classifications (30 Ba-d).
  6. 6. The maintenance computing system (10) of claim 1, wherein the processor (12) further executes a monitoring classifier (26) including at least a third artificial intelligence model (26 a), the monitoring classifier (26) configured to: Receiving runtime service input data (30B, 58A-C) comprising service related input data (58A-C) and a service classification (30B) employed; extracting a service feature (26 f) of the runtime service input data (30 b,58 a-C); Determining a predicted monitoring classification (54C) based on the extracted repair features (26 f), the predicted monitoring classification (54C) being one of a plurality of candidate monitoring classifications (30 Ca-C); Outputting the predicted supervisory classification (54C); receiving user input (38) of a employed monitoring class (30C) for the runtime maintenance input data (30B, 58A-C), and Feedback training of the third artificial intelligence model (26 a) is performed using the runtime maintenance input data (30 b,58 a-C) and the employed monitoring classification (30C) as a feedback training data pair (31C).
  7. 7. The maintenance computing system (10) of claim 6, wherein the service-related input data (58A-C) includes at least one of a service material or a service type.
  8. 8. A maintenance computing method (300), comprising: -running an inspection classifier (22) using a processor (12) and associated memory (20), the inspection classifier (22) comprising at least a first artificial intelligence model (22 a), running the inspection classifier (22) comprising: receiving runtime event input data (28A-C) from a plurality of data sources associated with a vehicle, the data sources including structural health monitoring sensors equipped on the vehicle; extracting features (22 f) of the runtime event input data (28A-C); deciding a predicted inspection classification (54A) based on the extracted features (22 f), the predicted inspection classification (54A) being one of a plurality of candidate inspection classifications (30 Aa-c); outputting the predicted inspection classification (54A); receiving user input (38) of an employed check classification (30A) for the runtime event input data (28A-C); performing feedback training of the first artificial intelligence model (22 a) using the runtime event input data (28A-C) and the employed inspection classification (30A) as a feedback training data pair (31A), and Running a repair classifier comprising at least a second artificial intelligence model, running the repair classifier comprising: receiving run-time inspection input data, the run-time inspection input data including inspection-related input data and an employed inspection classification; extracting inspection features of the runtime inspection input data; determining a predicted repair class based on the extracted inspection features, the predicted repair class being one of a plurality of candidate repair classes; Outputting the predicted repair classification; receiving user input of an employed maintenance classification for the runtime inspection input data, and Feedback training of the second artificial intelligence model is performed using the runtime inspection input data and the employed repair classification as a feedback training data pair.
  9. 9. The maintenance computing method of claim 8, further comprising: Before running the inspection classifier, training the inspection classifier on the inspection classifier training data, the inspection classifier training data comprising training input data comprising structural health data from the structural health monitoring sensors equipped on the vehicle and an associated ground truth tag, and the ground truth tag being a user-entered inspection classification associated with the training input data, the user-entered inspection classification selected from the plurality of candidate inspection classifications.
  10. 10. The maintenance computing method of claim 9, Wherein the inspection classifier training data further includes at least one of camera images, audio data, or dimensional measurements, and Wherein the runtime event input data further comprises at least one of camera images, audio data, or dimensional measurements.
  11. 11. The maintenance computing method of claim 8, wherein the structural health monitoring sensor is selected from the group consisting of inertial accelerometers, inertial gyroscopes, strain gauges, displacement transducers, air velocity sensors, temperature sensors.
  12. 12. The maintenance computing method of claim 8, further comprising: The repair classifier is trained on repair classifier training data prior to running the repair classifier, the repair classifier training data including repair classifier training input data including imaging learning and electrical measurements and associated ground truth labels, and the ground truth labels being user-entered repair classifications associated with the repair classifier training input data, the user-entered repair classifications being selected from the plurality of candidate repair classifications.
  13. 13. The maintenance computing method of claim 8, further comprising: running a monitoring classifier including at least a third artificial intelligence model, the running the monitoring classifier including: receiving runtime service input data, the runtime service input data comprising service related input data and a service classification employed; extracting maintenance characteristics of the runtime maintenance input data; Determining a predicted monitoring class based on the extracted repair features, the predicted monitoring class being one of a plurality of candidate monitoring classes; Outputting the predicted monitoring classification; receiving user input of a employed monitoring classification for the runtime maintenance input data, and And performing feedback training of the third artificial intelligence model using the runtime maintenance input data and the employed monitoring classification as a feedback training data pair.
  14. 14. The maintenance computing method of claim 13, wherein the service related input data includes at least one of a service material or a service type.
  15. 15. The maintenance computing method of claim 13, further comprising: before running the monitor classifier, training the monitor classifier on monitor training data, the monitor training data comprising monitor training input data and associated ground truth labels, the monitor training input data comprising imaging learning and electrical measurements, and the ground truth labels being user-entered repair classifications associated with repair training input data, the user-entered repair classifications being selected from the plurality of candidate repair classifications.
  16. 16. A maintenance computing system, comprising: A processor and a non-volatile memory storing executable instructions that, in response to execution by the processor, cause the processor to: a run-time inspection classifier configured to determine a predicted inspection classification based on run-time event input data from a structural health monitoring sensor equipped on a vehicle; outputting the predicted inspection classification; receiving user input of an employed check classification for the runtime event input data; performing feedback training of the inspection classifier using the runtime event input data and the employed inspection classification as a feedback training data pair; Running a repair classifier to determine a predicted repair classification based on run-time inspection input data, the run-time inspection input data including inspection-related input data and an employed inspection classification; Outputting the predicted repair classification; receiving user input of an employed maintenance classification for the runtime inspection input data, and Feedback training of the repair classifier is performed using the runtime inspection input data and the employed repair classification as a feedback training data pair.
  17. 17. The maintenance computing system of claim 16, wherein the processor is further configured to: Running a monitoring classifier to determine a predicted monitoring class based on run-time service input data, the run-time service input data including service related input data and a service class employed; Outputting the predicted monitoring classification; receiving user input of a employed monitoring classification for the runtime maintenance input data, and And performing feedback training of the monitoring classifier by using the runtime maintenance input data and the adopted monitoring classification as a feedback training data pair.

Description

Maintenance computing system and method for aircraft using predictive classifier Technical Field The present invention relates generally to machine learning and artificial intelligence processes and systems, and in particular to deciding on appropriate classifications for maintenance events, repairs, and repair tracking. For example, the techniques described herein may be applied to data related to an aircraft. Background Aircraft maintenance, including inspection after a flight event, initial repair, and subsequent inspections and repairs, is important to keeping the aircraft in flight and is also a significant cost to the vehicle operator. Such maintenance typically involves non-destructive evaluation of a portion of the aircraft followed by initial repair. The use of non-destructive evaluation techniques can be labor intensive and subject to manual judgment regarding the proper inspection technique, maintenance and tracking plan being practiced. Disclosure of Invention This summary is not an extensive overview of the specification. It is intended to neither identify key or critical elements of the specification nor delineate any scope of the particular embodiments of the specification or any scope of the claims. Its sole purpose is to present some concepts of the specification in a simplified form as a prelude to the more detailed description that is presented in the present disclosure. A computing system is disclosed that includes a processor and a non-volatile memory storing executable instructions that, in response to execution by the processor, cause the processor to execute an inspection classifier that includes at least a first Artificial Intelligence (AI) model. The inspection classifier is configured to receive runtime event input data from a plurality of data sources associated with the aircraft. The data source includes structural health monitoring sensors equipped on board the aircraft. The inspection classifier extracts features of the runtime event input data, decides a predicted inspection classification based on the extracted features, and outputs the predicted inspection classification. The predicted inspection classification is one of a plurality of candidate inspection classifications. The features, functions, and advantages that have been discussed can be achieved independently in various embodiments or may be combined in yet other embodiments, further details of which can be seen with reference to the following description and drawings. Drawings FIG. 1 shows a diagram depicting a maintenance system according to an example embodiment of the subject disclosure. Fig. 2 shows a diagram depicting an inspection classifier, a maintenance classifier and a monitoring classifier of the maintenance system according to fig. 1. Fig. 3A shows a diagram depicting a training phase of the inspection classifier according to fig. 1, including an initial training phase and a feedback training phase. Fig. 3B shows a diagram depicting a training phase of the maintenance classifier according to fig. 1, including an initial training phase and a feedback training phase. Fig. 3C shows a diagram depicting a training phase of the supervised classifier according to fig. 1, including an initial training phase and a feedback training phase. FIG. 4A shows a diagram depicting possible inspection, repair, and monitoring classifications that may be output by the inspection, repair, and monitoring classifiers, respectively, in accordance with an example of the subject disclosure. FIG. 4B shows a diagram depicting possible inspection, repair, and monitoring classifications that may be output by the inspection, repair, and monitoring classifiers, respectively, according to another example of the subject disclosure. Fig. 5 shows a diagram depicting a multi-dimensional feature space of the inspection classifier according to fig. 1. FIG. 6 shows a diagram of an exemplary graphical user interface of the inspection classifier of FIG. 4A in accordance with the example embodiment of the subject disclosure. FIG. 7 illustrates a process flow diagram of the operation of the maintenance system of FIG. 1 for feedback training at runtime, in accordance with one specific example of the subject disclosure. FIG. 8A illustrates a flow chart of a training phase of a maintenance calculation method for use in connection with aircraft maintenance, according to one example of the subject disclosure. FIG. 8B is a continuation of the flow chart of FIG. 8A, showing the runtime phase of the method, illustrating the processing steps involved in using an inspection classifier that includes feedback training. FIG. 8C is a continuation of the flow chart of FIG. 8B, showing the runtime phase of the method, illustrating the processing steps involved in using a maintenance classifier that includes feedback training. FIG. 8D is a continuation of the flow chart of FIG. 8C showing the runtime phase of the method illustrating the processing steps involved in