CN-121637263-B - Unmanned aerial vehicle fault diagnosis method, device, equipment and medium based on knowledge graph

CN121637263BCN 121637263 BCN121637263 BCN 121637263BCN-121637263-B

Abstract

The application discloses an unmanned aerial vehicle fault diagnosis method, device, equipment and medium based on a knowledge graph, and relates to the technical field of unmanned aerial vehicle fault diagnosis, comprising the steps of integrating symptom vectors based on data characteristics obtained by interpreting unmanned aerial vehicle parameters; the symptom vector comprises reference symptoms, the reference symptoms comprise reference symptom parameters and occurrence probability of the reference symptom parameters, a target knowledge graph model is built according to data characteristics, symptom nodes, which are mapped in the target knowledge graph model, of target symptoms with the occurrence probability larger than an activation threshold are activated to obtain an activated model, path searching is performed according to the activated model to obtain a plurality of target fault propagation paths of each target fault, fault probability of the target fault is calculated based on path information and symptom information of the plurality of target fault propagation paths, the symptom information comprises occurrence probability and number of the symptom nodes, and the target fault and the fault probability are integrated to obtain a target diagnosis report. The accuracy and efficiency of unmanned aerial vehicle fault diagnosis can be improved.

Inventors

GONG YUZHAO
YANG XIAO
CHEN XIAOPING
XIE YUYAO
GUAN TING
FENG AITING
MA HONGYU
YI SHUANG
YANG DA
LI TING

Assignees

中航(成都)无人机系统股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260205

Claims (9)

1. The unmanned aerial vehicle fault diagnosis method based on the knowledge graph is characterized by comprising the following steps of: Interpreting unmanned aerial vehicle parameters to obtain data features, and integrating symptom vectors based on the data features, wherein the symptom vectors comprise a plurality of reference symptoms, and the reference symptoms comprise reference symptom parameters and occurrence probability of each reference symptom parameter; Constructing a target knowledge graph model according to the data characteristics and reference data, wherein the reference data comprises an troubleshooting manual for unmanned aerial vehicle fault analysis and diagnosis, expert experience and a system component architecture of the unmanned aerial vehicle; extracting target symptoms with occurrence probability larger than an activation threshold value from the symptom vector, and activating symptom nodes of the target symptoms mapped in a target knowledge-graph model to obtain an activated knowledge-graph model; Performing path search according to the activated knowledge graph model to obtain a plurality of target fault propagation paths of each target fault, and calculating fault probability of the target fault based on path information and symptom information of the plurality of target fault propagation paths; Integrating target faults and fault probability according to a preset diagnosis report template to obtain a target diagnosis report; Wherein the calculating the fault probability of the target fault based on the path information and symptom information of the plurality of target fault propagation paths includes: Estimating the importance degree of the target path of the target fault propagation path based on the path weight, the path length, the occurrence probability of the relevant symptom nodes and the number of the relevant symptom nodes in the target fault propagation path according to the importance degree calculation formula; calculating the fault probability of the target fault according to the number of corresponding relevant symptom nodes of a plurality of target fault propagation paths and the importance degree of the target paths; Wherein, importance degree calculation formula: ; Wherein, the The target path importance degree of the target fault propagation path is represented; representing a kth path weight in the target fault propagation path; representing the occurrence probability of the ith relevant symptom node in the target fault propagation path; representing the number of relevant symptom nodes in the target fault propagation path.
2. The unmanned aerial vehicle fault diagnosis method based on the knowledge-graph according to claim 1, wherein the calculating the fault probability of the target fault according to the number of corresponding relevant symptom nodes of the plurality of target fault propagation paths and the importance degree of the target paths comprises: determining the maximum path importance degrees corresponding to a plurality of target fault propagation paths corresponding to the target faults; when the maximum path importance degree is smaller than the preset importance degree, calculating symptom quantity factors corresponding to the target faults by utilizing the quantity of relevant symptom nodes corresponding to the target faults determined based on the plurality of target fault propagation paths; And calculating the fault probability of the target fault based on the maximum path importance degree and the symptom quantity factor.
3. The unmanned aerial vehicle fault diagnosis method based on the knowledge-graph according to claim 1, wherein the calculating the fault probability of the target fault according to the number of corresponding relevant symptom nodes of the plurality of target fault propagation paths and the importance degree of the target paths comprises: determining the maximum path importance degrees corresponding to a plurality of target fault propagation paths corresponding to the target faults; When the maximum path importance degree is not smaller than the preset importance degree, carrying out logarithmic scaling processing on the maximum path importance degree to obtain a processing result, calculating symptom quantity factors corresponding to the target faults by utilizing the quantity of relevant symptom nodes corresponding to the target faults determined based on the target fault propagation paths, and determining average path importance degrees corresponding to the target fault propagation paths corresponding to the target faults; and calculating the fault probability of the target fault based on the average path importance degree, the symptom quantity factor and the processing result.
4. The unmanned aerial vehicle fault diagnosis method based on the knowledge-graph according to claim 1, wherein the performing a path search according to the activated knowledge-graph model to obtain a plurality of target fault propagation paths for each target fault comprises: Performing path search according to the activated knowledge graph model to obtain a plurality of temporary fault propagation paths; and counting a plurality of target fault propagation paths corresponding to each target fault in the temporary fault propagation paths according to the temporary fault propagation paths.
5. The unmanned aerial vehicle fault diagnosis method based on the knowledge-graph according to claim 1, wherein the interpreting unmanned aerial vehicle parameters to obtain data features comprises: Performing type interpretation on unmanned aerial vehicle parameters to identify parameter types of the unmanned aerial vehicle parameters, performing system interpretation on the unmanned aerial vehicle parameters to identify parameter sources of the unmanned aerial vehicle parameters, and performing state interpretation on unmanned aerial vehicle parameters of discrete types to determine triggering conditions; determining data characteristics based on the parameter types, the parameter sources and the triggering conditions; Accordingly, the data-feature-based syndrome vector comprises: Determining the triggered unmanned aerial vehicle parameters of discrete types as reference symptom parameters of discrete types, and determining the occurrence probability of the reference symptom parameters according to the triggering conditions; Determining overrun parameters as continuous type reference symptom parameters, and determining occurrence probability of the reference symptom parameters according to overrun ranges, wherein the overrun parameters are obtained by overrun analysis on continuous type unmanned aerial vehicle parameters so as to extract overrun parameters; the symptom vector is integrated based on the discrete type of reference symptom parameter and the continuous type of reference symptom parameter, and the occurrence probability of the reference symptom parameter.
6. The unmanned aerial vehicle fault diagnosis method based on the knowledge-graph according to any one of claims 1 to 5, wherein the interpreting unmanned aerial vehicle parameters to obtain data features comprises: Analyzing the unmanned aerial vehicle parameters to obtain parameter original values, and calculating parameter engineering values according to the product of the scaling factors and parameter target values, wherein the parameter target values are the difference values of the parameter original values and the offset values; performing continuity check on the time stamp corresponding to the parameter engineering value and performing time alignment to obtain a complete data set under a unified time axis; the complete data set is interpreted to obtain data features.
7. Unmanned aerial vehicle fault diagnosis device based on knowledge graph, characterized by comprising: The system comprises a vector integration module, a symptom vector, a data analysis module and a data analysis module, wherein the vector integration module is used for interpreting unmanned aerial vehicle parameters to obtain data characteristics and integrating symptom vectors based on the data characteristics; The model construction module is used for constructing a target knowledge graph model according to the data characteristics and reference data, wherein the reference data comprises an troubleshooting manual for unmanned aerial vehicle fault analysis and diagnosis, expert experience and a system component architecture of the unmanned aerial vehicle; The node activation module is used for extracting target symptoms with occurrence probability larger than an activation threshold value from the symptom vector, and activating symptom nodes of the target symptoms mapped in the target knowledge-graph model to obtain an activated knowledge-graph model; The path searching module is used for performing path searching according to the activated knowledge graph model to obtain a plurality of target fault propagation paths of each target fault, and calculating the fault probability of the target fault based on the path information and the symptom information of the plurality of target fault propagation paths; The diagnosis report integrating module is used for integrating the target faults and the fault probability according to a preset diagnosis report template to obtain a target diagnosis report; The path search module further includes: A path importance degree evaluation unit for evaluating the target path importance degree of the target fault propagation path based on the path weight, the path length, the occurrence probability of the relevant symptom nodes and the number of the relevant symptom nodes in the target fault propagation path according to the importance degree calculation formula; the fault probability calculation unit is used for calculating the fault probability of the target fault according to the number of corresponding relevant symptom nodes of the plurality of target fault propagation paths and the importance degree of the target paths; Wherein, importance degree calculation formula: ; Wherein, the The target path importance degree of the target fault propagation path is represented; representing a kth path weight in the target fault propagation path; representing the occurrence probability of the ith relevant symptom node in the target fault propagation path; representing the number of relevant symptom nodes in the target fault propagation path.
8. An electronic device, comprising: A memory for storing a computer program; A processor for executing the computer program to implement the knowledge-graph-based unmanned aerial vehicle fault diagnosis method as claimed in any one of claims 1 to 6.
9. A computer readable storage medium for storing a computer program, wherein the computer program when executed by a processor implements the knowledge-graph-based unmanned aerial vehicle fault diagnosis method of any one of claims 1 to 6.

Description

Unmanned aerial vehicle fault diagnosis method, device, equipment and medium based on knowledge graph Technical Field The invention relates to the technical field of unmanned aerial vehicle fault diagnosis, in particular to an unmanned aerial vehicle fault diagnosis method, device, equipment and medium based on a knowledge graph. Background Currently, the fault diagnosis mode of an unmanned aerial vehicle is generally that an on-board flight parameter recorder is carried out to record flight parameters of each system in the flight process, flight parameter data in the aircraft recorder is downloaded after the aircraft lands, then the flight parameter data is manually read and analyzed through special flight parameter interpretation software, and fault diagnosis is carried out according to self experience, a comparison troubleshooting manual or a knowledge base. In fault diagnosis of an actual aircraft system, diagnosis knowledge has the characteristics of nested relation, complex logic and multidirectional connection, so that the traditional method is large in time consumption and low in efficiency when analyzing fault relation, and too depends on an expert knowledge base. The existing unmanned aerial vehicle ground fault diagnosis technology mostly adopts a frame-based production rule and relational database or relies on a statistical model, and aims to construct a structured and layered automatic diagnosis model to support fault reasoning. However, when the technology is used for dealing with the complex-relation and obscure-knowledge scenes of unmanned aerial vehicle systems, the obvious defects of incomplete knowledge system, poor diagnosis automation effect, difficult representation of complex knowledge logic, complex model construction process, high sample data requirement, low reasoning efficiency and difficulty in meeting the high-efficiency diagnosis requirement are exposed. In summary, how to improve the accuracy and efficiency of unmanned aerial vehicle fault diagnosis is a current problem to be solved. Disclosure of Invention In view of the above, the present invention aims to provide a method, a device, equipment and a medium for unmanned aerial vehicle fault diagnosis based on a knowledge graph, which can improve the accuracy and efficiency of unmanned aerial vehicle fault diagnosis, and the specific scheme is as follows: In a first aspect, the application discloses an unmanned aerial vehicle fault diagnosis method based on a knowledge graph, which comprises the following steps: Interpreting unmanned aerial vehicle parameters to obtain data features, and integrating symptom vectors based on the data features, wherein the symptom vectors comprise a plurality of reference symptoms, and the reference symptoms comprise reference symptom parameters and occurrence probability of each reference symptom parameter; Constructing a target knowledge graph model according to the data characteristics and reference data, wherein the reference data comprises an troubleshooting manual for unmanned aerial vehicle fault analysis and diagnosis, expert experience and a system component architecture of the unmanned aerial vehicle; extracting target symptoms with occurrence probability larger than an activation threshold value from the symptom vector, and activating symptom nodes of the target symptoms mapped in a target knowledge-graph model to obtain an activated knowledge-graph model; Performing path search according to the activated knowledge graph model to obtain a plurality of target fault propagation paths of each target fault, and calculating fault probability of the target fault based on path information and symptom information of the plurality of target fault propagation paths; and integrating the target faults and the fault probability according to a preset diagnosis report template to obtain a target diagnosis report. Optionally, the calculating the fault probability of the target fault based on the path information and the symptom information of the plurality of target fault propagation paths includes: Estimating the importance degree of the target path of the target fault propagation path based on the path weight, the path length, the occurrence probability of the relevant symptom nodes and the number of the relevant symptom nodes in the target fault propagation path according to the importance degree calculation formula; calculating the fault probability of the target fault according to the number of corresponding relevant symptom nodes of a plurality of target fault propagation paths and the importance degree of the target paths; Wherein, importance degree calculation formula: ; Wherein, the The target path importance degree of the target fault propagation path is represented; representing a kth path weight in the target fault propagation path; representing the occurrence probability of the ith relevant symptom node in the target fault propagation path; representing the number of relevant symptom