Avionics fault knowledge fusion method oriented to distributed collaboration

Abstract

The present invention relates to the field of avionics equipment. Provided is an avionics fault knowledge fusion method oriented to distributed collaboration. The method comprises the following steps: S1, acquiring multi-source fault knowledge data of avionics equipment, and performing analysis processing and screening on the acquired multi-source fault knowledge data of the avionics equipment; S2, for the problem of knowledge conflicts across different timelines, comprehensively calculating machining fault knowledge on the basis of attribute similarity, in order to complete fault knowledge fusion; S3, for the problem of fault knowledge conflicts from different sources, completing fault knowledge fusion by means of fault element annotation and identification; and S4, constructing a fault knowledge graph, and storing fused fault knowledge. The present invention can mitigate knowledge conflicts during both the initial construction of a fault knowledge model and the expansion of existing fault knowledge models, thereby improving the accuracy, reliability and integrity of knowledge model construction for avionics information equipment, and thus providing better support for fault localization and maintenance of the avionics equipment.

Description

A distributed collaborative-oriented avionics fault knowledge fusion method TECHNICAL FIELD

[0001] The present application relates to the field of avionics equipment, in particular to a distributed collaborative-oriented avionics fault knowledge fusion method. BACKGROUND

[0002] On the one hand, the fault knowledge of avionics information equipment is of various types and sources, including FMEA, fault cases, fault trees and other types of fault knowledge, and various types of fault knowledge are constructed by different personnel, and the fault knowledge from different sources may differ in specific text description and fault entity relationship construction, and the knowledge may conflict; on the other hand, the fault knowledge on the same source but different timelines also has fusion problems. The fault knowledge base is updated continuously over time, and the new and old fault knowledge may also conflict. In order to fuse the conflicting knowledge and improve the accuracy, reliability and integrity of the knowledge model construction of avionics information equipment, a distributed collaborative-oriented avionics fault knowledge fusion method is proposed.

[0003] SUMMARY

[0004] The present application aims to provide a distributed collaborative-oriented avionics fault knowledge fusion method to solve the knowledge conflict problem in the initial construction of the fault knowledge model and the expansion of the existing fault knowledge model, and improve the accuracy, reliability and integrity of the knowledge model construction of avionics information equipment.

[0005] The present application provides a distributed collaborative-oriented avionics fault knowledge fusion method, which comprises the following steps:

[0006] S1, acquiring multi-source fault knowledge data of avionics equipment, and analyzing, processing and screening the acquired multi-source fault knowledge data of avionics equipment;

[0007] S2, for the knowledge conflict problem on different timelines, processing fault knowledge based on attribute similarity comprehensive calculation, and completing fault knowledge fusion;

[0008] S3, for the fault knowledge conflict problem of different sources, completing fault knowledge fusion by means of fault element labeling and identification;

[0009] S4, constructing a fault knowledge graph and storing the fused fault knowledge.

[0010] Further, in step S1, FMEA, FTA and fault cases of avionics equipment are selected as multi-source fault knowledge data of avionics equipment.

[0011] Further, step S1 comprises:

[0012] Analyzing the meaning of fault knowledge data, screening the table headers that can be used to build fault knowledge model as entity types, and building entity relationships to support fault knowledge information reading;

[0013] When selecting fault knowledge data attributes, the meanings of fault knowledge data attributes and the correlation between attributes are comprehensively considered according to expert knowledge, and the attributes that can effectively represent fault knowledge are selected as entity types;

[0014] Each FMEA, FTA and fault case is read into a group of entities, and then the group of entities is converted into a plurality of entity pairs according to the correlation between the corresponding entity types of the group of entities, and the fault knowledge triple is constructed by combining the relationship between the entities in the entity pairs.

[0015] Further, the entity types include: fault mode, model, system, complete equipment, whole machine, component, case name, fault model, fault whole machine or unit, fault phenomenon, diagnosis process, maintenance measure, fault reason, fault detection method, this level influence, high level influence, final influence, improvement measure, use compensation measure, fault tree name, test item name and monitoring point.

[0016] Further, in step S2, the cosine similarity is used to calculate the similarity of each attribute of each FMEA, FTA and fault case; then the weighted average method is used to calculate the overall similarity between the two data, and compared with the determined co-reference resolution threshold to determine whether the two data are the same entity, and then the data are merged and stored, that is, the fault knowledge fusion is completed.

[0017] Further, step S3 includes:

[0018] The fault mode and fault phenomenon in FMEA, FTA and fault case are further extracted, and the construction of fault element and its alias library is completed by intelligent labeling software;

[0019] Based on the sentence segmentation method in the field of natural language, an automatic extraction algorithm of fault element alias is constructed to extract the text description of fault element alias from the fault phenomenon text;

[0020] Based on the method of text similarity matching, the fuzzy matching of fault element alias description in fault description and existing fault element alias is carried out;

[0021] Based on the mapping relationship between fault element alias-fault element-fault mode established in the construction process of fault element, different sources of fault knowledge are mapped to the same fault mode.

[0022] Further, the construction steps of fault element and its alias library include:

[0023] A fault element representation system of an aviation integrated electronic information equipment is established; the fault element refers to a key word of condensed representation of a fault which is sorted and combed from the knowledge of fault phenomena and fault modes according to the experience of experts; the categories of the fault element include a component unit, a performance representation and a fault state. A fault element library is formed through the sorting of expert knowledge, the relationship between the fault element and the alias of the fault element is manually established by using a fault element intelligent labeling software, and then the construction of the basic library of the fault element and the alias of the fault element is completed by combining intelligent identification and manual labeling.

[0024] Further, the automatic extraction step of the alias of the fault element includes:

[0025] According to the open source natural language statistical dictionary of the third party library, a prefix dictionary of words in the statistical dictionary is constructed;

[0026] Then, the input sentence is subjected to directed acyclic graph construction according to the prefix dictionary; then a probability maximum path is found on the directed acyclic graph by using a dynamic programming method, and the path is subjected to word segmentation;

[0027] Finally, for the uncollected words, the most possible hidden state sequence is found out by using the Viterbi algorithm based on the hidden Markov model, and the word sequence which is possibly the alias of the fault element is found out.

[0028] Further, the step of fuzzy matching of the fault element includes:

[0029] Based on the longest common subsequence method, the similarity of all possible aliases of the fault element extracted and all aliases of the fault element in the alias library is calculated;

[0030] The similarity measurement threshold is set, and the extracted alias of the fault element and the existing alias of the fault element with the similarity greater than the similarity measurement threshold are taken as a matching pair, so as to realize the fuzzy matching of the alias of the fault element.

[0031] Further, the step S4 includes:

[0032] For the extracted and processed knowledge, the entity entries with conflicts are cleaned or merged and stored according to the threshold comparison result, and are converted into RDF format data, so as to realize the unified expression of knowledge; the aviation integrated electronic information equipment fault knowledge graph is constructed based on Neo4j, the entities and the relationship between the entities are constructed first, then the attributes are taken as the description of the internal characteristics of the entities, and finally the multi-modal aviation fault knowledge fusion is completed.

[0033] As described above, since the above technical solutions are adopted, the present application has the following beneficial effects:

[0034] 1、The present application fuses the fault knowledge of avionics equipment from different sources or different timelines in depth, eliminates the conflicts between the knowledge, guarantees the accuracy, reliability and integrity of the knowledge expression in the process of knowledge base expansion, and then achieves the efficiency of equipment fault positioning and maintenance and repair, which has important significance and value for the operation and maintenance process of avionics equipment.

[0035] 2、The above-mentioned distributed cooperative-oriented avionics fault knowledge fusion method of the present application can be applied to multiple fields. ACCURACY

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings in the embodiments will be briefly introduced as follows, and it should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0037] Fig. 1 is a flow chart of the distributed cooperative-oriented avionics fault knowledge fusion method in the embodiment of the present application.

[0038] Fig. 2a is a cosine similarity matrix calculation result diagram between fault phenomena in the embodiment of the present application.

[0039] Fig. 2b is a cosine similarity matrix calculation result diagram between whole machine fault elements in the embodiment of the present application.

[0040] Fig. 2c is a cosine similarity matrix calculation result diagram between module fault elements in the embodiment of the present application.

[0041] Fig. 2d is a cosine similarity matrix calculation result diagram between fault reasons in the embodiment of the present application. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application, and obviously, the described embodiments are part of the embodiments of the present application, but not all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0043] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. All other embodiments obtained by those skilled in the art on the basis of the embodiments in the present application without creative labor belong to the scope of protection of the present application.

[0044] EMBODIMENT

[0045] As shown in Figure 1, the embodiment proposes a distributed collaborative-oriented avionics fault knowledge fusion method, including the following steps:

[0046] S1, obtaining avionics equipment multi-source fault knowledge data, analyzing and processing and screening the obtained avionics equipment multi-source fault knowledge data;

[0047] Select the FMEA (Failure Mode and Effects Analysis Table), FTA (Fault Tree) and fault cases of avionics equipment as the avionics equipment multi-source fault knowledge data, and analyze and process and screen the avionics equipment multi-source fault knowledge data. Specifically:

[0048] Analyze the meaning of fault knowledge data, select the table header that can be used to construct the fault knowledge model as the entity type, and construct the entity relationship to support fault knowledge information reading; when selecting fault knowledge data attributes, consider the meaning of each data attribute and the association relationship between the attributes according to expert knowledge, select the attributes that can effectively represent the fault knowledge as the entity type; read each FMEA, FTA and fault case into a group of entities, and then convert the group of entities into a plurality of entity pairs according to the association relationship between the entity types corresponding to the group of entities, and construct fault knowledge triples by combining the relationship between the entities in the entity pairs. The form of the triple is <head entity, relationship, tail entity>.

[0049] Specifically, the entity types extracted from the avionics equipment fault knowledge data include: fault mode, model, system, complete equipment, whole piece, component, case name, fault model, fault whole machine or unit, fault phenomenon, diagnosis process, maintenance measure, fault reason, fault detection method, this level influence, high level influence, final influence, improvement measure, use compensation measure, fault tree name, test item name and monitoring point. The relationship type of the aviation integrated electronic information equipment ontology is defined as shown in Table 1, and the attribute type of the aviation integrated electronic information equipment ontology is defined as shown in Table 2.

[0050] Table 1, definition table of relationship type of aviation integrated electronic information equipment ontology:

[0051] Table 2, definition table of attribute type of aviation integrated electronic information equipment ontology:

[0052] S2, for the knowledge conflict problem in different timelines, process the fault knowledge based on attribute similarity comprehensive calculation, and complete the fault knowledge fusion. Specifically:

[0053] Taking FMEA data with different timelines as an example, the cosine similarity is used to calculate the similarity between each pair of attributes, that is, the similarity between two vectors in the TF matrix is directly calculated, as shown in FIGS. 2a, 2b, 2c and 2d, which is actually to solve the cosine value of the angle between two vectors, that is, the dot product divided by the length of the two vectors:

[0054] wherein, similarity represents the similarity between vector A and vector B, A i and B i represent the components of vector A and vector B, respectively.

[0055] The similarity of each attribute of each FMEA data is calculated, and then the overall similarity between the two data is calculated using the weighted average method. The overall similarity is compared with the determined co-reference resolution threshold to determine whether the two data are the same entity. After the data are merged and stored, the fault knowledge fusion is completed.

[0056] S3, for the conflict problem of fault knowledge from different sources, the fault element marking and recognition are used to complete the fault knowledge fusion;

[0057] wherein, the fault mode and the fault phenomenon in the FMEA, the FTA and the fault case are further extracted, and the fault element and the alias library thereof are constructed by using the intelligent marking software. Then, the fault element alias automatic extraction algorithm is constructed based on the sentence segmentation method in the natural language field, and the possible fault element alias description is extracted from the fault phenomenon text. Then, the fuzzy matching between the fault element alias description in the fault description and the existing fault element alias is carried out based on the text similarity matching method. Finally, based on the mapping relationship between the fault element alias-fault element-fault mode established in the construction process of the fault element, the fault knowledge from different sources is mapped to the same fault mode.

[0058] wherein, the construction steps of the fault element and the alias library thereof include:

[0059] An aviation integrated electronic information equipment fault element representation system is established. The fault element refers to the key words for representing the condensed fault, which are extracted from the fault phenomenon, the fault mode and other knowledge according to the expert experience. The categories of the fault element include the component unit, the performance representation and the fault state. The fault element library is formed by the expert knowledge, the relationship between the fault element and the fault element alias is established by using the intelligent marking software, and the construction of the fault element-fault element alias basic library is completed by combining the intelligent recognition and the manual marking.

[0060] wherein, the automatic extraction steps of the fault element alias include:

[0061] Firstly, according to the open source natural language statistical dictionary of Python third-party library jieba, a prefix dictionary of words in the statistical dictionary is constructed; then, the input sentence is constructed into a directed acyclic graph (DAG) according to the prefix dictionary; then, a probability maximum path is found on the DAG by using a dynamic programming method, and the sentence is segmented according to the path; finally, for the uncollected words, based on a Hidden Markov Model (HMM model), a Viterbi algorithm is used to find the most possible hidden state sequence, and all possible word sequences of the fault alias are found.

[0062] The step of fuzzy matching of the fault element includes:

[0063] Firstly, the similarity of the extracted possible fault element alias and all fault element aliases in the fault element alias library is calculated based on a Longest Common Subsequence (LCS) method; then, a similarity measure threshold is set, and the extracted fault element alias-existing fault element alias pair with a similarity greater than the similarity measure threshold is taken as a matching pair, so as to realize fuzzy matching of the fault element alias.

[0064] S4, constructing a fault knowledge graph and storing the fused fault knowledge.

[0065] For the extracted and processed knowledge, the entity entries with conflicts are cleaned or merged and stored according to the threshold comparison result, and are converted into RDF format data to realize unified expression of the knowledge; based on Neo4j (a commonly used graph database), an aviation integrated electronic information equipment fault knowledge graph is constructed, the entities and the relationship between the entities are constructed first, then the attributes are taken as the description of the internal characteristics of the entities, and finally the multi-modal aviation fault knowledge fusion is completed.

[0066] The above-mentioned aviation electronic fault knowledge fusion method oriented to distributed cooperation can be applied to multiple fields. First, after extracting entity, relationship and attribute information from the multi-source health state knowledge data of the analysis object; then, in view of the fact that different entities in different timeline records of fault knowledge texts actually refer to the same thing, co-reference resolution is realized by calculating attribute similarity; then, a fault element and its alias library are constructed, and fault knowledge from different sources but pointing to the same fault mode is combined through the phenomenon of fault phenomenon-fault element alias-fault element-fault mode; finally, the extracted and processed knowledge is converted into RDF format data, knowledge is stored in the database, and an aviation integrated electronic information equipment fault knowledge graph is constructed through Neo4j, thereby completing the fusion and unified expression of the aviation fault knowledge model. Through deep fusion of aviation electronic equipment fault knowledge from different sources or the same source but at different timelines, the conflicts between the knowledge are eliminated, the accuracy, reliability and integrity of the knowledge expression in the knowledge base expansion process are ensured, and then the efficiency of equipment fault positioning and maintenance is ensured, which has important significance and value for the aviation electronic equipment operation process.

[0067] The above merely illustrates the preferred embodiments of the present application, but should not be used to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A distributed-collaboration-oriented avionics fault knowledge fusion method, characterized in that, The method comprises the following steps: S1, acquiring multi-source fault knowledge data of avionics equipment, and analyzing and screening the acquired multi-source fault knowledge data of avionics equipment; S2, for knowledge conflict problems on different timelines, processing fault knowledge based on attribute similarity comprehensive calculation, and completing fault knowledge fusion; S3, for fault knowledge conflict problems of different sources, completing fault knowledge fusion through fault element labeling and identification; S4, constructing a fault knowledge graph and storing the fused fault knowledge.

2. The distributed-collaboration-oriented avionics failure knowledge fusion method according to claim 1, characterized in that, In step S1, FMEA, FTA and fault cases of avionics equipment are selected as multi-source fault knowledge data of avionics equipment.

3. The distributed-collaboration-oriented avionics fault knowledge fusion method according to claim 2, characterized in that, Step S1 comprises: Analyzing the meaning of fault knowledge data, selecting the table header that can be used to construct the fault knowledge model as the entity type, and constructing the entity relationship to provide support for fault knowledge information reading; When selecting fault knowledge data attributes, the meanings of the attributes of the fault knowledge and the correlation between the attributes are considered, and the attributes that can effectively represent the fault knowledge are selected as the entity type; Each FMEA, FTA and fault case is read into a group of entities, and then the group of entities is converted into a plurality of entity pairs according to the correlation between the entity types corresponding to the group of entities, and a fault knowledge triple is constructed by combining the relationship between the entities in the entity pairs.

4. The distributed-collaboration-oriented avionics fault knowledge fusion method according to claim 3, characterized in that, The entity types include: fault mode, model, system, complete equipment, whole machine, component, case name, fault model, fault whole machine or unit, fault phenomenon, diagnosis process, maintenance measure, fault reason, fault detection method, this level influence, high level influence, final influence, improvement measure, use compensation measure, fault tree name, test item name and monitoring point.

5. The distributed-collaboration-oriented avionics failure knowledge fusion method according to claim 4, characterized in that, In step S2, the cosine similarity is used to calculate the attribute similarity of each FMEA, FTA and fault case; then the weighted average method is used to calculate the overall similarity between two data, and the overall similarity is compared with the determined co-reference resolution threshold to determine whether the two data are the same entity, and then the data are merged and stored, that is, the fault knowledge fusion is completed.

6. The distributed-collaboration-oriented avionics fault knowledge fusion method according to claim 4, characterized in that, Step S3 comprises: The fault mode and fault phenomenon in FMEA, FTA and fault cases are further extracted, and the construction of the fault element and its alias library is completed through intelligent labeling software; The fault element alias automatic extraction algorithm is constructed based on the sentence segmentation method in the natural language field, and the text description of the fault element alias is extracted from the fault phenomenon text; The fuzzy matching of the fault element alias description in the fault description and the existing fault element alias is carried out based on the text similarity matching method; Based on the mapping relationship between the fault element alias-fault element-fault mode established in the fault element construction process, different sources of fault knowledge are mapped to the same fault mode.

7. The distributed-collaboration-oriented avionics failure knowledge fusion method according to claim 6, characterized in that, The construction steps of the fault element and its alias library comprise: A fault element representation system of the comprehensive electronic information equipment of aviation is established; the fault element refers to the key words of condensed representation of faults which are sorted and combed from the knowledge of fault phenomenon and fault mode according to the expert experience; the categories of the fault element include component unit, performance representation and fault state; the fault element library is formed through the sorting of expert knowledge, the relationship between the fault element and the alias of the fault element is manually established by using the intelligent labeling software of the fault element, and then the construction of the basic library of the fault element and the alias of the fault element is completed by combining the intelligent recognition and manual labeling.

8. The distributed-collaboration-oriented avionics fault knowledge fusion method according to claim 6, characterized in that, The automatic extraction steps of the alias of the fault element include: According to the open source natural language statistical dictionary of the third party library, a prefix dictionary of words in the statistical dictionary is constructed; Then, the input sentence is subjected to directed acyclic graph construction according to the prefix dictionary; then a probability maximum path is found on the directed acyclic graph by using the dynamic programming method, and the path is subjected to word segmentation; Finally, for the uncollected words, the most possible hidden state sequence is found out by using the Viterbi algorithm based on the hidden Markov model, and all the word sequences which are the aliases of the fault element are found out.

9. The distributed-collaboration-oriented avionics fault knowledge fusion method according to claim 6, characterized in that, The steps of the fuzzy matching of the fault element include: Based on the longest common subsequence method, the similarity of all possible fault element aliases extracted and all fault element aliases in the fault element alias library is calculated; The similarity measurement threshold is set, the extracted fault element alias and the existing fault element alias pair with the similarity greater than the similarity measurement threshold are taken as the matching pair, and the fuzzy matching of the alias of the fault element is realized.

10. The distributed-collaboration-oriented avionics fault knowledge fusion method according to claim 1, characterized in that, Step S4 includes: For the extracted and processed knowledge, the conflicting entity entries are cleaned or merged and stored according to the threshold comparison result, and are converted into RDF format data to realize the unified expression of knowledge; the fault knowledge graph of the comprehensive electronic information equipment of aviation is constructed based on Neo4j, the entities and the relationship between the entities are constructed first, then the attributes are taken as the description of the internal characteristics of the entities, and finally the multi-modal aviation fault knowledge fusion is completed.

Images