CN114897193A - A decision-making method and decision-making system for aircraft structure maintenance based on human-in-the-loop - Google Patents

Abstract

The invention discloses a human-in-the-loop-based aircraft structure maintenance decision-making method and decision-making system, comprising the following steps: obtaining damage inspection data of the aircraft structure of a target case; Case features and analyze the case feature attribute weights; according to the case feature attribute weights, obtain a source case similar to the target case, and output the optimal maintenance decision of the target case; according to the optimal maintenance decision, through human-computer interaction, output decision-making in conclusion. According to the damage inspection data of the aircraft structure of the target case, the invention can expressly and accurately generate the maintenance decision result.

Description

A decision-making method and decision-making system for aircraft structure maintenance based on human-in-the-loop

technical field

The invention relates to a human-in-the-loop-based aircraft structure maintenance decision-making method and decision-making system, belonging to the technical field of aircraft structure maintenance.

Background technique

Timely and accurate damage measurement and assessment, and efficient repair plan formulation are the core contents of solving civil aircraft structural repair problems. Foreign civil aircraft represented by Boeing and Airbus have established a support platform covering "damage monitoring, assessment, reporting and repair program management" for the efficient formulation of repair plans. Domestic research on structural repair evaluation and verification has been started, but most of the research results are at the level of theoretical analysis and simulation verification. Manual Structural Repair Program Development Techniques. In the case study of civil aircraft structure, it is necessary to select some attributes as comprehensively as possible for different types of civil aircraft according to their structural and functional characteristics, manufacturers and models to fully reflect the characteristics of civil aircraft cases. For highly complex and specialized aircraft, there will inevitably be problems such as redundancy and correlation of attributes. Therefore, it is necessary to choose an appropriate method to extract effective features from numerous attributes.

Traditional maintenance decision-making is directly related to human ability and experience. The selected maintenance attributes and decision-making algorithms are relatively fixed in practice, which restricts the rapid improvement of aircraft structure maintenance decision-making ability. At the same time, the maintenance decision of aircraft structure needs to comprehensively consider the overall reasonable maintenance timing, take into account the maintenance method of resources and capabilities, weigh the damage situation and the rapid recovery measures of maintenance capabilities, and establish the optimal maintenance resource guarantee and technical state recovery under complex damage conditions. Therefore, it is urgent to develop a set of intelligent decision-making methods suitable for aircraft structure maintenance, which can quickly and accurately target the damage inspection data of the aircraft structure, and give the maintenance decision of the damaged structure, so as to realize the integrity of the actual operation of the aircraft, the reliability of the mission and the reliability of the aircraft. Effectively improve the usability of the whole machine.

The information disclosed in this Background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person of ordinary skill in the art.

SUMMARY OF THE INVENTION

The purpose of the present invention is to overcome the deficiencies in the prior art, and to provide a human-in-the-loop-based aircraft structure maintenance decision-making method and decision-making system, which can expressly and accurately generate maintenance decision-making results according to the damage inspection data of the aircraft structure of the target case.

To achieve the above object, the present invention adopts the following technical solutions to realize:

In one aspect, the present invention discloses a human-in-the-loop-based aircraft structure maintenance decision-making method, comprising the following steps:

Obtain the damage inspection data of the aircraft structure of the target case;

According to the damage inspection data of the aircraft structure of the target case, based on the maintenance decision model algorithm library and the maintenance case library, extract the case features of the target case and analyze the case feature attribute weights;

According to the weight of the feature attribute of the case, a source case similar to the target case is obtained;

According to the source case, output the optimal maintenance decision for the target case;

According to the optimal maintenance decision, a decision conclusion is output through human-computer interaction; wherein, if the optimal maintenance decision is correct, the decision conclusion is directly output; if the optimal maintenance decision is wrong, the target case is manually verified The decision conclusion and update the optimization maintenance decision model algorithm library and maintenance case library.

Further, the case characteristics of the target case include, but are not limited to, the ATA part number, the structure type of the damage part, the cause of the damage, the damage type, the damage size, the inspection method, and the repair type.

Further, the case features are selected by a dimensionality reduction algorithm such as AHP or PCA.

Further, based on the maintenance decision model algorithm library and the maintenance case library, the weight of the case feature attribute is determined by the entropy weight method.

Further, according to the weight of the feature attribute of the case, the weighted K-nearest neighbor retrieval algorithm based on Euclidean distance is used to calculate the similarity of the case, and the source case similar to the target case is obtained.

Further, according to the source case, based on the transfer learning of similar cases, the optimal maintenance decision of the target case is output.

Further, in response to an error in the optimal maintenance decision output by the human-computer interaction module,

If the target case is a new case, manually input the decision conclusion of the target case, form a new maintenance decision, and enter the maintenance case database as a new maintenance case; at the same time, trigger the case self-learning algorithm to perform case self-learning and update the case Feature and case feature attribute weight, optimize maintenance decision model algorithm library;

If the target case is not a new case, manually revise the decision conclusion of the target case, and update the maintenance case database; at the same time, trigger the case self-learning algorithm to perform case self-learning, and optimize the maintenance decision model algorithm library.

On the other hand, the present invention discloses a decision-making system based on a human-in-the-loop aircraft structure maintenance decision-making method, including a data acquisition module, an analysis data module, a maintenance decision model algorithm library, a maintenance case library, a source case module, and an optimal maintenance decision. models and human-computer interaction modules,

Wherein, the data acquisition module is used to acquire the damage inspection data of the aircraft structure of the target case;

The data analysis module is configured to extract the case features of the target case and analyze the case feature attribute weights based on the maintenance decision model algorithm library and the maintenance case library according to the damage inspection data of the aircraft structure of the target case;

the maintenance case database for storing maintenance cases of the aircraft structure;

The maintenance decision model algorithm library is used to store the maintenance decision model algorithm of the aircraft structure;

The source case module is used to obtain a source case similar to the target case according to the weight of the feature attribute of the case;

The optimal maintenance decision model is used for outputting the optimal maintenance decision of the target case according to the source case;

The human-computer interaction module is used to output a decision conclusion according to the optimal maintenance decision; if the optimal maintenance decision is correct, directly output the decision conclusion; if the optimal maintenance decision is wrong, manually verify The decision conclusion of the target case is updated to optimize the maintenance decision model algorithm library and maintenance case library.

Compared with the prior art, the beneficial effects achieved by the present invention:

Based on the maintenance case library and the maintenance decision model algorithm library, the invention extracts the case features of the target case, and deeply digs the multi-dimensional case information through the common features between the cases, determines the source case similar to the target case, and obtains the target case. optimal maintenance decision.

The invention also adopts a human-computer interaction module, and feedbacks the optimal maintenance decision based on the human-in-the-loop method, so as to obtain the maintenance decision conclusion more accurately; meanwhile, the maintenance case database and the maintenance decision model algorithm database are updated and optimized, so that the input After a new target case, the maintenance decision conclusion can be obtained more quickly.

Description of drawings

Figure 1 is a flow chart of a man-in-the-loop based aircraft structure maintenance decision-making method.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings. The following examples are only used to illustrate the technical solutions of the present invention more clearly, and cannot be used to limit the protection scope of the present invention.

Example 1

This embodiment 1 provides a human-in-the-loop-based aircraft structure maintenance decision-making method, including the following steps:

Step 1: Obtain the damage inspection data of the aircraft structure of the target case.

Step 2: According to the damage inspection data of the aircraft structure of the target case, based on the maintenance decision model algorithm library and the maintenance case library, extract the case features of the target case and analyze the weight of the case feature attributes.

Specifically, the type of case characteristics of the target case that affects the maintenance decision can be selected from historical aircraft structural repair cases according to engineering experience, such as ATA part number, damage part structure type, damage cause, damage type, damage size, inspection On the other hand, dimensionality reduction algorithms such as AHP and PCA can also be used for the features in the repair cases to select the main case features that affect the decision-making of aircraft structural maintenance.

Based on the maintenance case database, after extracting the case features of the target case, the attribute weights of the case features are determined. In order to avoid the influence of human subjectivity on the weight of case feature attributes, the entropy weight method is used to determine the weight of case feature attributes.

First, normalize the case feature attribute matrix, the specific expression is as follows:

Among them, y _ij represents the standardized case feature attribute matrix, x _ij represents the case feature attribute matrix, both y _ij and x _ij are matrices with m rows and n columns, m represents the row number of the case feature attribute matrix, and n represents the case feature attributes The number of columns of the matrix, i represents the row coordinate of the case feature attribute matrix, j represents the column coordinate of the case feature attribute matrix, and x _i represents the ith feature attribute value of the source case x.

Secondly, the information entropy value E _j of each case feature attribute is obtained, and the specific expression is as follows:

Among them, E _j represents the information entropy value of the feature attribute of the jth case, k is an intermediate parameter, and p _ij represents the matrix after normalizing y _ij .

Finally, determine the weight of each case feature attribute:

_0≤ωj ≤1

Among them, ω _j represents the weight of the feature attribute of the jth case.

Step 3: According to the weight of the feature attributes of the case, the weighted K-nearest neighbor retrieval algorithm based on Euclidean distance is used to calculate the similarity of the case, and the source case similar to the target case is obtained.

details as follows:

The weighted Euclidean distance D(X,Y) is

Among them, x _i represents the ith feature attribute value of the source case x, i=1,2,3,...,n; y _i represents the ith feature attribute value of the target case y; n represents the case feature attribute value number; ω _j represents the weight of the jth case feature attribute; d(x _i , y _i ) ² represents the Euclidean distance between the target case and the source case, s _i represents the ith feature on the n-dimensional case feature space .

The similarity S(X,Y) between the target case and the source case is:

S(X,Y)=1/(1+D(X,Y))

Step 4: According to the source case, output the optimal maintenance decision of the target case.

Specifically, the source cases can be divided into similar model cases and similar model cases according to different types. Through the information transfer learning of the source cases, considering the similarity and matching degree, the optimal maintenance decision of the target case is obtained. .

Step 5: According to the optimal maintenance decision, through human-computer interaction, output the decision conclusion.

Specifically, the optimal maintenance decision obtained through transfer learning takes the maintenance decision information feedback of the human-in-the-loop as the triggering condition, and uses manual input to verify the maintenance decision conclusion. If the optimal maintenance decision is correct, the maintenance decision conclusion of the damaged structure is directly output.

Step 6: In response to the error of the optimal maintenance decision output by the human-computer interaction module:

If the target case is a new case, manually input the decision conclusion of the target case to form a new maintenance decision, and enter the maintenance case database as a new maintenance case; at the same time, trigger the case self-learning algorithm to conduct case self-learning, and update the case characteristics and case characteristics Attribute weight, optimization maintenance decision model algorithm library;

If the target case is not a new case, manually revise the decision conclusion of the target case and update the maintenance case database; at the same time, trigger the case self-learning algorithm to conduct case self-learning and optimize the maintenance decision model algorithm library.

Step 7: After the update and optimization is completed, enter the target case and repeat the above steps.

Example 2

The second embodiment discloses a decision-making system based on the aircraft structure maintenance decision-making method of the first embodiment, including a data acquisition module, an analysis data module, a maintenance decision-making model algorithm library, a maintenance case library, a source case module, an optimal maintenance decision-making model and Human-computer interaction module,

Among them, the data acquisition module is used to acquire the damage inspection data of the aircraft structure of the target case;

The data analysis module is used to extract the case features of the target case and analyze the weight of the case feature attributes based on the maintenance decision model algorithm library and maintenance case library based on the damage inspection data of the aircraft structure of the target case;

Maintenance case library for storing maintenance cases for aircraft structures;

Maintenance decision model algorithm library, which is used to store the maintenance decision model algorithm of aircraft structure;

The source case module is used to obtain the source case similar to the target case according to the weight of the feature attribute of the case;

The optimal maintenance decision model is used to output the optimal maintenance decision of the target case according to the source case;

The human-computer interaction module is used to output the decision conclusion according to the optimal maintenance decision; if the optimal maintenance decision is correct, it will directly output the decision conclusion; if the optimal maintenance decision is wrong, manually verify the decision conclusion of the target case and update the optimization Maintenance decision model algorithm library and maintenance case library.

The technical idea of the present invention is to quickly generate an optimal maintenance decision based on the input damage inspection data of the aircraft structure, use a human-computer interaction module to verify the optimal maintenance decision, and feed back, update and optimize the entire decision-making system, and use iteratively to make the system more efficient. Make decisions quickly and accurately.

The above are only the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, without departing from the technical principle of the present invention, several improvements and modifications can also be made. These improvements and modifications It should also be regarded as the protection scope of the present invention.

Claims (8)

1. A method for deciding the maintenance of airplane structure based on man-in-the-loop includes such steps as providing a decision-making unit,

acquiring damage inspection data of an airplane structure of a target case;

extracting case characteristics of the target case and analyzing case characteristic attribute weights based on a maintenance decision model algorithm library and a maintenance case library according to damage inspection data of the airplane structure of the target case;

obtaining a source case similar to the target case according to the case characteristic attribute weight;

outputting an optimal maintenance decision of the target case according to the source case;

outputting a decision conclusion through man-machine interaction according to the optimal maintenance decision; if the optimal maintenance decision is correct, a decision conclusion is directly output; and if the optimal maintenance decision is wrong, manually verifying the decision conclusion of the target case and updating an optimal maintenance decision model algorithm library and a maintenance case library.

2. The human-in-loop based aircraft structure repair decision method as claimed in claim 1, wherein the case characteristics of the target case include, but are not limited to, ATA site number, damage site structure type, damage cause, damage type, damage size, inspection mode, repair type.

3. The human-in-loop-based aircraft structure maintenance decision method as claimed in claim 1, wherein the case characteristics are selected using a dimensionality reduction algorithm such as an analytic hierarchy process or a principal component analysis process.

4. The human-in-loop-based aircraft structure maintenance decision method as claimed in claim 1, wherein the case characteristic attribute weight is determined by an entropy weight method based on a maintenance decision model algorithm library and a maintenance case library.

5. The human-in-loop-based aircraft structure maintenance decision method as claimed in claim 1, wherein case similarity is calculated by using a weighted K-nearest neighbor search algorithm based on Euclidean distance according to the case characteristic attribute weight to obtain a source case similar to a target case.

6. The human-in-loop-based aircraft structure maintenance decision method as claimed in claim 1, wherein the optimal maintenance decision of the target case is output based on similar case transfer learning according to the source case.

7. The human-in-loop based aircraft structure maintenance decision method as claimed in claim 1, wherein in response to an error in the optimal maintenance decision output by the human-machine interaction module,

if the target case is a new case, manually inputting a decision conclusion of the target case to form a new maintenance decision, and entering a maintenance case library as a newly added maintenance case; simultaneously triggering a case self-learning algorithm to perform case self-learning, updating case characteristics and case characteristic attribute weights, and optimizing a maintenance decision model algorithm library;

if the target case is not a new case, manually revising the decision conclusion of the target case and updating a maintenance case base; and simultaneously triggering a case self-learning algorithm to perform case self-learning and optimizing a maintenance decision model algorithm library.

8. The decision-making system for the human-in-loop-based aircraft structure maintenance decision-making method according to any one of claims 1 to 7, comprising a data acquisition module, an analysis data module, a maintenance decision model algorithm library, a maintenance case library, a source case module, an optimal maintenance decision model and a human-computer interaction module,

the data acquisition module is used for acquiring damage inspection data of the airplane structure of the target case;

the data analysis module is used for extracting case characteristics of the target case and analyzing case characteristic attribute weights according to damage inspection data of the airplane structure of the target case and based on a maintenance decision model algorithm library and a maintenance case library;

the maintenance case library is used for storing maintenance cases of the aircraft structure;

the maintenance decision model algorithm library is used for storing maintenance decision model algorithms of the aircraft structure;

the source case module is used for obtaining a source case similar to the target case according to the case characteristic attribute weight;

the optimal maintenance decision model is used for outputting an optimal maintenance decision of a target case according to the source case;

the human-computer interaction module is used for outputting a decision conclusion according to the optimal maintenance decision; if the optimal maintenance decision is correct, directly outputting a decision conclusion; and if the optimal maintenance decision is wrong, manually verifying the decision conclusion of the target case and updating the optimal maintenance decision model algorithm library and the maintenance case library.

Images