Spacecraft FMEA application method based on SysML model and knowledge graph

Abstract

The invention relates to a spacecraft FMEA application method based on a SysML model and a knowledge graph. The spacecraft FMEA application method comprises the steps that logic analysis is conducted on functional faults of a spacecraft, and a functional fault logic model and a fault mode basic element semantic model are obtained; according to the function fault logic model and the fault mode basic element semantic model, utilizing an SysML model tool to construct an FMEA digital model of the spacecraft; an XML metadata conversion method is adopted, and the FMEA digital model is converted into FMEA structured data; mapping concepts in the fault mode basic element semantic model and subclasses and relationships thereof into concepts, subclasses and relationships of an FMEA knowledge graph ontology model to obtain the FMEA knowledge graph ontology model; fMEA structural data and historical fault information are obtained through FMEA digital model conversion, an FMEA knowledge graph ontology model is instantiated, and an FMEA knowledge graph is established; based on the FMEA knowledge graph, spacecraft FMEA analysis is carried out, automatic statistical calculation of fault mode occurrence frequency and fault mode reduction measure recommendation are achieved, and the aim of reducing spacecraft design risks and on-orbit operation and maintenance risks is achieved.

Description

Technical Field

[0001] The present invention relates to a method for applying spacecraft FMEA based on SysML model and knowledge graph, belonging to the technical field of spacecraft risk management. Background Art

[0002] With the deep integration of the new generation of artificial intelligence technology and advanced manufacturing technology, the digital, networked and intelligent manufacturing management mode will reshape the entire life cycle of spacecraft product design, manufacturing, operation and maintenance, etc. Failure mode and effects analysis is an important task for spacecraft risk management.

[0003] There are some deficiencies in the construction of traditional FMEA analysis: 1) It is manually created and prone to human errors. High labor costs usually result in it being only executed once at the final stage of product design. The improvement requirements for safety, reliability, availability and maintainability design identified through FMEA are also discovered only after system integration testing, and the cost of implementing design changes is high; 2) The analysis and evaluation process rely on the experience and knowledge of model designers and historical failure data. It is difficult to learn from each other the FMEA analysis experience of the same system and single-unit products, and there are still blind spots in failure analysis risks. The above reasons lead to relatively high potential risks of spacecraft products after the implementation of FMEA.

[0004] In order to overcome the above shortcomings of traditional FMEA analysis, currently, the method of digital modeling of FMEA is mostly used for failure mode and effects analysis. The FMEA digital model analysis method mainly includes two steps: model definition and model conversion. In terms of the definition of the FMEA digital model, the existing modeling representation methods of FMEA in the SysML model are divided into state definition method, meta-model representation method and architecture modeling method. The state definition method regards that when the system is in a certain state and cannot execute a certain activity in the SysML activity diagram and state diagram as a failure mode; the meta-model method supports the establishment of a risk model library based on the extension mechanism of SysML - concepts such as package elements. Its core is to first define the generalization relationship of SysML to inherit the attributes in the basic model, and then concretize the abstract attributes in the model through redefinition. The failure propagation path is represented by a proxy interface in the meta-model; the architecture modeling method starts from the key data items of FMEA and a series of sequential activities of FMEA analysis. Each failure mode corresponds to an FMEA system structure model table. The above methods are mainly developed based on the AltiRica traversal engine. The depth of analysis of the state diagram determines the quality of the FMEA generated by the algorithm. The method is highly complex and depends on the nested level of the state diagram. From the perspective of FMEA data management, the FMEA information stored in documents is not conducive to communication, sharing and reuse among multiple teams, resulting in problems such as poor consistency of reliability and safety analysis results caused by unsmooth FMEA data and low utilization rate. Summary of the Invention

[0005] The technical problem solved by the present invention is to overcome the deficiencies of the prior art, and provide a method for applying spacecraft FMEA based on SysML model and knowledge graph. By using the forward design method of model-based systems engineering (MBSE) and managing the failure data of the entire life cycle of spacecraft products with knowledge graph, more comprehensive risk analysis can be completed in various stages of spacecraft design verification, research, manufacturing, and on-orbit operation and maintenance.

[0006] The technical solution of the present invention is: a method for applying spacecraft FMEA based on SysML model and knowledge graph, the method comprising the following steps:

[0007] S1. In the design verification stage, conduct a logical analysis of the functional failures of the spacecraft to obtain a functional failure logical model and a semantic model of the basic elements of the failure mode; according to the functional failure logical model and the semantic model of the basic elements of the failure mode, use the SysML modeling tool to construct a digital FMEA model of the spacecraft;

[0008] S2. Adopt the XML metadata conversion method to convert the digital FMEA model into FMEA structured data;

[0009] S3. Map the concepts, their subclasses and relationships in the semantic model of the basic elements of the failure mode to the concepts, subclasses and relationships of the FMEA knowledge graph ontology model to obtain the FMEA knowledge graph ontology model;

[0010] S4. Extract the unstructured failure mode data in the research, manufacturing and on-orbit operation and maintenance stages to obtain structured failure information;

[0011] S5. Convert the digital FMEA model in the design stage into FMEA structured data and the structured failure information in the research, manufacturing and on-orbit operation and maintenance stages, and establish an FMEA knowledge graph according to the FMEA knowledge graph ontology model;

[0012] S6. Based on the FMEA knowledge graph, realize the automatic statistical calculation of the occurrence frequency of the failure mode and the recommendation of failure mode reduction measures.

[0013] Preferably, the unstructured failure mode data includes product documents in the research and manufacturing stage, quality zeroing reports and analysis monthly reports in the on-orbit operation and maintenance stage.

[0014] Preferably, the specific steps of step S1 are as follows:

[0015] Sort out the system states where the functional requirements are not met, use the system states where the functional requirements are not met as the failure modes, and establish a functional failure logical model; the system states where the functional requirements are not met include functional overflow, functional non-compliance, functional non-realization, functional degradation, and functional violation;

[0016] For the fault modes in the functional fault logic model, construct a semantic model of the basic elements of the fault mode to describe the fault modes corresponding to the abnormal functional states;

[0017] According to the functional fault logic model and the corresponding semantic model of the basic elements of the fault mode, obtain the FMEA digital model.

[0018] Preferably, the functional fault logic model includes a unit functional fault model, a system functional fault model, an interface fault model, and a fault propagation model. The construction process is as follows:

[0019] Decompose the functional requirements of the spacecraft according to the top-down method, sort out the system states where the functional requirements are not met, and use the system states where the functional requirements are not met as fault modes to obtain the system functional fault model and the single-unit functional fault model; the system functional fault model represents the system fault mode and its impact on the overall mission or system function; the single-unit functional fault model represents the single-unit fault mode and its impact on the single-unit mission or single-unit function;

[0020] Analyze the fault propagation between spacecraft single-units, between single-units and systems, and between systems according to the bottom-up method to obtain the interface fault model and the fault propagation model. The interface fault model represents the interface fault mode and its impact on the overall mission or system function; the fault propagation model represents the propagation chain of faults layer by layer and level by level.

[0021] Preferably, the semantic model of the basic elements of the fault mode defines four general class concepts and their subclasses and relationships and one proprietary class concept and its subclasses and relationships. The general class concepts include functional state, failure mode, mitigation measures, and risk priority number. The general class concepts describe the fault modes in the single-unit functional fault model and the system functional fault model; the proprietary class concept includes compatibility, error propagation, and transmission parameter coordination failure; the proprietary class concept is used to describe the fault modes of the interface fault model and the fault propagation model.

[0022] Preferably, the construction steps of the FMEA digital model are as follows:

[0023] S2.1. According to the mission requirements, construct a SysML activity diagram, and the constructed SysML activity diagram represents the activities required to complete the mission; the SysML activity diagram includes the abnormal state and the normal state of the activities to complete the mission;

[0024] S2.2. Connect the abnormal state of the activities to complete the mission in the SysML activity diagram with its normal state through a violation relationship; the violation relationship is represented by the SysML metamodel definition;

[0025] S2.3. Instantiate the semantic model of the basic elements of the failure mode in the SysML state diagram for the failure modes in the SysML activity diagram obtained after performing step 2.2;

[0026] S2.4. Construct a SysML metamodel containing failure mode attribute information on the SysML metamodel diagram according to the semantic model of the basic elements of the failure mode;

[0027] S2.5. According to the SysML metamodel containing failure mode attribute information, instantiate the concept types in the SysML metamodel in the SysML state diagram to obtain the FMEA digital model.

[0028] Preferably, the methods for instantiating the general class concepts in the semantic model of the basic elements of the failure mode include:

[0029] Use the SysML state diagram and metamodel extension mechanism to represent functional states, failure modes, and mitigation measures;

[0030] Use the SysML parameter diagram model to represent the risk priority number class;

[0031] The methods for instantiating the proprietary class concepts in the semantic model of the basic elements of the failure mode include:

[0032] Use the metamodel extension mechanism to represent compatibility, error propagation, and transmission parameter coordination failures.

[0033] Preferably, the steps for constructing the FMEA knowledge graph are as follows:

[0034] Determine the ontology schema layer: Expand and semantically map the concepts, their subclasses, and relationships in the semantic model of the basic elements of the failure mode into the concepts, subclasses, and relationships of the FMEA knowledge graph ontology model to obtain the FMEA knowledge graph ontology model;

[0035] Knowledge acquisition: Use the XML metadata conversion method to convert the FMEA digital model into structured data, and obtain the structured data converted from the FMEA digital model and the historical failure information in the research, manufacturing, and on-orbit operation and maintenance stages; the historical failure information includes unstructured and structured data;

[0036] Knowledge extraction: According to the concepts, subclasses, and relationships in the FMEA knowledge graph ontology model, use a general information extraction model to extract the entities and relationships related to the failure mode from the unstructured FMEA document data in the historical failure information to obtain FMEA structured data;

[0037] Knowledge conversion: According to the top-down construction method of the knowledge graph, organize data based on the schema layer of the FMEA knowledge graph ontology model, and convert the structured FMEA data obtained from the digital model conversion of the spacecraft's FMEA and the structured fault information obtained from knowledge extraction into RDF triples for knowledge representation;

[0038] Knowledge fusion: For synonym terms and co-referential terms in the RDF triple data, establish or utilize an existing domain-specific thesaurus for entity alignment and co-reference resolution to achieve knowledge fusion and knowledge connection across units and departments;

[0039] Knowledge storage: Store the RDF triple data in an RDF database or a property graph database to form an FMEA knowledge graph.

[0040] Preferably, the calculation method of the occurrence frequency of the failure mode is as follows:

[0041] Input the failure mode node in the FMEA knowledge graph, and query the number N of quality problem cases related to the failure mode that have occurred in the FMEA knowledge graph within a period of time T:

[0042] Occurrence frequency of failure mode = Number of quality problem cases N / Time T.

[0043] Preferably, the recommendation method for the mitigation measures of the failure mode to be mitigated is as follows:

[0044] Define the meta-path between the entity type nodes and other entity type nodes on the FMEA knowledge graph ontology model:

[0045] For the FMEA knowledge graph ontology schema layer G = (N, R), the meta-path P between the entity type node N1 and the entity type node N l is represented as:

[0046]

[0047] where, N i is the i-th entity type node, the relationship R l is the l-th relationship type, and the path length is l - 1

[0048] If there is a path composed of certain nodes and relationships in the FMEA knowledge graph data layer that satisfies the meta-path pattern on the ontology model, then this path is called an instance of the meta-path. The round-trip meta-path is usually symmetric. Define the symmetric meta-path p of the entity's own node on the FMEA knowledge graph data layer G = (V, E):

[0049]

[0050] where, n iis an instance node on the data layer of the FMEA knowledge graph, r l is an instance relationship node on the FMEA knowledge graph;

[0051] According to the PathSim algorithm for measuring the similarity of nodes of the same type in a heterogeneous network, the similarity between the node of the failure mode to be reduced and other nodes of the same type of failure mode is determined on the data layer of the FMEA knowledge graph:

[0052]

[0053] where, |{p x→y : p x→y ∈M}| is the number of meta-paths from node x to node y; M is the set of all meta-paths on the data layer of the FMEA knowledge graph;

[0054] |{p x→x : p x→x ∈M}| is the number of symmetric meta-paths of node x;

[0055] |{p y→y : p y→y ∈M}| is the number of symmetric meta-paths of node y;

[0056] Determine the failure mode node with the highest similarity to the node of the failure mode to be reduced, denoted as the reference failure mode node, and recommend the reduction measures corresponding to the reference failure mode node to the failure mode to be reduced.

[0057] The beneficial effects of the present invention compared with the prior art are:

[0058] (1). Based on the advantages of the forward design of model-based systems engineering technology and system-level requirements modeling, the present invention uses SysML to carry out the definition of the FMEA digital model. This FMEA digital model integrates risk assessment activities and design activities, decomposes and verifies system requirements in advance, takes the closed-loop reduction and control of failures as the core, and utilizes the forward design advantages of model-based systems engineering (MBSE) to bring FMEA forward to the early design stage, solving the problem of high implementation cost of design stage changes.

[0059] (2). The present invention utilizes XML-based data exchange technology and knowledge graph multi-source data integration technology to construct an FMEA knowledge graph, which runs through the design, manufacturing, operation and maintenance stages, and achieves the goal of reducing the quality risk of spacecraft products caused by changes, errors or inconsistencies by strengthening the linkage between the design and operation and maintenance links. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is the overall architecture diagram of the embodiment of the present invention;

[0061] Figure 2It is the V diagram of the functional fault logic model in the embodiment of the present invention;

[0062] Figure 3 It is the state analysis and basic semantic model of the P diagram of the functional fault logic model in the embodiment of the present invention;

[0063] Figure 4 It is the activity diagram in the FMEA digital model in the embodiment of the present invention;

[0064] Figure 5 It is the meta-model diagram in the FMEA digital model in the embodiment of the present invention;

[0065] Figure 6 It is the construction and management flow chart of the FMEA knowledge graph in the embodiment of the present invention;

[0066] Figure 7 It is the ontology layer of the FMEA knowledge graph in the embodiment of the present invention;

[0067] Figure 8 It is the SysML state diagram model of the abnormal current fault mode of the satellite two-degree-of-freedom gyro motor in the embodiment of the present invention;

[0068] Figure 9 It is the visualization diagram of the satellite fault mode knowledge graph in the embodiment of the present invention;

[0069] Figure 10 It is the abnormal current fault chain of the two-degree-of-freedom gyro motor in the satellite fault mode knowledge graph in the embodiment of the present invention;

[0070] Figure 11 It is the query result of the occurrence frequency of the two-degree-of-freedom gyro fault mode in the embodiment of the present invention. Detailed implementation manners

[0071] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0072] As Figure 1 shown, the present invention provides a method for applying spacecraft FMEA based on the SysML model and the knowledge graph, and the method includes the following steps:

[0073] S1. In the design verification stage, conduct a logical analysis of the functional faults of the spacecraft to obtain the functional fault logic model and the semantic model of the basic elements of the fault mode; according to the functional fault logic model and the semantic model of the basic elements of the fault mode, use the SysM modeling tool to construct the FMEA digital model of the spacecraft;

[0074] S2. Adopt the XML metadata conversion method to convert the FMEA digital model into FMEA structured data;

[0075] S3. Map the concepts, their subclasses, and relationships in the semantic model of the basic elements of the failure mode into the concepts, subclasses, and relationships of the FMEA knowledge graph ontology model to obtain the FMEA knowledge graph ontology model;

[0076] S4. Extract the unstructured failure mode data in the development, manufacturing, and on-orbit operation and maintenance phases to obtain structured failure information;

[0077] S5. Convert the FMEA digital model in the design verification phase to obtain FMEA structured data and the structured failure information in the development, manufacturing, and on-orbit operation and maintenance phases, and establish an FMEA knowledge graph according to the FMEA knowledge graph ontology model;

[0078] S6. Based on the FMEA knowledge graph, realize the automatic statistical calculation of the occurrence frequency of the failure mode and the recommendation of failure mode reduction measures.

[0079] As can be seen from Figure 1 the FMEA application method based on the SysML model and the knowledge graph provided by the present invention includes three parts: FMEA digital model management, FMEA data management based on the knowledge graph, and FMEA knowledge graph application.

[0080] 1. FMEA digital model management

[0081] FMEA digital model management includes two parts of work: FMEA digital model definition and FMEA model conversion. Through FMEA digital model definition and conversion, with the failure mode as the core, the various works in the design and development stage are organically linked, and a unified technical logic is formed with the SysML digital model of the failure mode. The requirements of six indicators, namely reliability, maintainability, safety, testability, supportability, and environmental adaptability (referred to as six reliability properties), are implemented through FMEA modeling and conversion methods in the three stages of design requirement confirmation, requirement implementation, and requirement verification.

[0082] In the actual engineering situation, the integrated evaluation technology of six reliability properties closely focuses on "function architecture - failure logic". Therefore, the functional failure logic model is the basis of FMEA analysis. Based on the functional failure logic model, relevant semantic definitions of the failure mode are carried out, and then a digital FMEA model is formed based on the SysML modeling language. A bridge is established between the FMEA model and data management through XML data conversion technology, forming an integrated technology system for FMEA digital management.

[0083] FMEA digital model management takes the closed-loop reduction and control of faults as the core, integrates the design and analysis of six aspects such as reliability into the main line of the design and development of the model's functions and performance, and from the perspective of systems engineering, represents the fault mode with the system state where the functional requirements are not met, highlighting the impact of the uncompleted function on the overall task or system, and establishes a functional fault logic model. The functional fault logic model includes a unit functional fault model, a system functional fault model, an interface fault model, and a fault propagation model. The construction process is as follows:

[0084] As Figure 2 shown, decompose the functional requirements of the spacecraft according to the top-down method, sort out the system states where the functional requirements are not met, and use the system states where the functional requirements are not met as the fault mode to obtain the system functional fault model and the single-unit functional fault model; the system functional fault model represents the system fault mode and its impact on the overall task or system function; the single-unit functional fault model represents the single-unit fault mode and its impact on the single-unit task or single-unit function;

[0085] Analyze the fault propagation (including interface faults and propagation faults) between spacecraft single-units, between single-units and systems, and between systems according to the bottom-up method to obtain the interface fault model and the fault propagation model. The interface fault model represents the interface fault mode and its impact on the overall task or system function; the fault propagation model represents the propagation chain of faults passing layer by layer and step by step.

[0086] As Figure 3 shown, the semantic model of the basic elements of the fault mode defines four general class concepts and one proprietary class concept. Each class concept contains corresponding subclasses and relationships. The general class concepts include functional state, failure mode, reduction measures, and risk priority number. The general class concepts describe the fault modes in the single-unit functional fault model and the system functional fault model; the proprietary class concept includes compatibility, error propagation, and transmission parameter coordination faults; the proprietary class concept is used to describe the fault modes of the interface fault model and the fault propagation model.

[0087] The fault states of the functional fault model include deviations from the expected functions, outputs of unexpected systems. The states of deviation from the expected functions include function degradation (partial functions are implemented), function violation (functions cannot be implemented), and function overflow (beyond the function scope). The outputs of unexpected systems mainly refer to "noise" factors, unexpected interfaces or conditions and interactions that can lead to function failures. The sources of noise mainly include variations between components, internal and external environments, fatigue, and user usage conditions. The failure mode categories include failure causes, failure effects (whether single points, local effects, high-level effects, and final effects are formed), and compensation measures; the reduction measure categories include corrective measures and fault detection methods. The risk priority number describes the quantitative indicators of the failure mode, including severity, hazard, and detectability. The special categories of interface connection faults are mainly divided into several types of faults such as compatibility faults, error propagation faults, and transmission parameter coordination faults.

[0088] In a specific embodiment of the present invention, typical single machines related to key subsystems and key links of a spacecraft are selected, and the semantic model of the basic elements of their failure modes is defined. The described attributes include 10 attribute names: failure mode category, failure mode, failure cause, failure effect, compensation measure, corrective measure, risk priority number (severity, hazard, and detectability), and fault detection method.

[0089] As Figure 3 shown, specifically, based on function diagrams, function flowcharts, boundary diagrams, and interface diagrams, the system-level qualitative and quantitative requirements, subsystem / component scopes, and system interaction degrees are analyzed. The P-diagram is used to analyze the expected inputs (signals) and outputs (functions), and the inputs and outputs are determined as specific functional states, and the fault states of function anomalies are also determined accordingly.

[0090] After determining the functional fault logic model, as Figure 4 and Figure 5 shown, the SysML activity diagram, state diagram, and meta-model extension mechanism are used to represent the semantic model of the basic elements of the failure mode, and an FMEA digital model is constructed.

[0091] The construction process of the FMEA digital model is specifically as follows:

[0092] (1). According to the task requirements, a SysML activity diagram is constructed. The constructed SysML activity diagram represents the activities required to complete the task; the SysML activity diagram includes the abnormal state and normal state of the activities to complete the task;

[0093] (2). The abnormal state of the activities to complete the task in the SysML activity diagram is connected to its normal state through a violation relationship; the violation relationship is represented by the definition of the SysML meta-model.

[0094] (3) For the failure modes in the SysML activity diagram obtained after performing step (2), instantiate the semantic model of the basic elements of the failure mode in the SysML state diagram;

[0095] (4) Construct the semantic model of the basic elements of the failure mode in the functional failure logic model on the SysML metamodel diagram to obtain the SysML metamodel containing the attribute information of the failure mode;

[0096] (5) According to the SysML metamodel containing the attribute information of the failure mode, instantiate the concept types in the SysML metamodel in the SysML state diagram to obtain the FMEA digital model.

[0097] The general class concept method for instantiating the semantic model of the basic elements of the failure mode includes:

[0098] (1) Use the SysML state diagram and metamodel extension mechanism to represent the functional state, failure mode, and mitigation measures;

[0099] (2) Use the SysML parameter diagram model to represent the risk priority number class;

[0100] The proprietary class concept method for instantiating the semantic model of the basic elements of the failure mode includes:

[0101] Use the metamodel extension mechanism to represent compatibility, error propagation, and transmission parameter coordination failures.

[0102] 2. FMEA Data Management Based on Knowledge Graph

[0103] FMEA data management based on knowledge graph includes:

[0104] 2.1. Adopt the XML metadata conversion method to convert the FMEA digital model into FMEA structured data;

[0105] Convert the FMEA digital model defined by SysML into structured data according to the XML metadata conversion method. The steps are as follows:

[0106] S2.1.1. Traverse the XML document corresponding to the FMEA digital model to extract the FMEA knowledge elements, and the FMEA knowledge elements are represented by package elements in the XML document;

[0107] S2.1.2. The method for extracting the FMEA knowledge elements by using the relationship extraction method is: trace the relationships between the nodes in the XML document, systematically understand the composition of the XML file, use the unique identity of the package element (packagedElement) to represent the own attribute tags of each node, and obtain the structure, relationships, and attributes of the nodes from them to generate the FMEA structured data.

[0108] 2.2 Map the concepts, their subclasses, and relationships in the semantic model of the basic elements of failure modes to the concepts, subclasses, and relationships of the FMEA knowledge graph ontology model to obtain the FMEA knowledge graph ontology model;

[0109] Specifically, conduct FMEA knowledge graph ontology modeling based on the basic semantics of the functional failure logic model, and represent the general class concepts of failure modes and the specific class concepts of connection failures using Protege based on the SysML metamodel to obtain the FMEA knowledge graph ontology model.

[0110] The definition of the ontology model complies with the requirements of QJ 3050A Guidelines for Failure Mode, Effects and Criticality Analysis of Aerospace Products. Based on the functional failure logic model, it is agreed that the hierarchy of system function classes in general class concepts is the tasks that equipment, systems, subsystems, equipment, and components need to complete, generally represented by a mission profile, that is, the sequence of events and environments experienced by the product within the specified mission time.

[0111] Based on the semantic model of the basic elements of failure modes, in the FMEA knowledge graph ontology model, the general class concepts of failure modes include qualitative class concepts such as failure mode classes and mitigation measure classes, as well as quantitative class descriptions such as the risk probability of failure modes and the risk priority number. The specific class concepts of connection failures include types such as interface failure models and failure propagation models, and are specifically divided into specific types such as compatibility, error propagation, and transmission parameter coordination failures.

[0112] 2.3 Use the FMEA digital model to convert to FMEA structured data, instantiate the FMEA knowledge graph ontology model, and establish the FMEA knowledge graph;

[0113] In this step, use knowledge graph technology to integrate historical failure information in the design, manufacturing, and operation and maintenance stages to establish the FMEA knowledge graph as the authoritative source of truth. As Figure 6 shown, the construction of the FMEA knowledge graph includes main steps such as knowledge acquisition, knowledge extraction, knowledge transformation, knowledge fusion, and knowledge storage.

[0114] The steps to construct the FMEA knowledge graph are as follows:

[0115] (1) Determine the ontology schema layer: Map the concepts, their subclasses, and relationships in the semantic model of the basic elements of failure modes to the concepts, subclasses, and relationships of the FMEA knowledge graph ontology model to obtain the FMEA knowledge graph ontology model, as Figure 7 shown;

[0116] (2) Knowledge acquisition: Obtain the structured data converted from the FMEA digital model through the XML data conversion method in the design verification stage and the historical failure information of research, manufacturing, and on-orbit operation and maintenance;

[0117] In terms of knowledge acquisition, the data source is the FMEA structure data of the full life cycle of typical spacecraft products obtained after processing; the knowledge graph data source includes the structured data converted from the FMEA digital model through the XML data conversion method in the design verification stage, and the historical fault information in the research, manufacturing, and on-orbit operation and maintenance stages. The historical fault information includes unstructured and structured data.

[0118] (3) Knowledge extraction: According to the concepts, subclasses, and relationships in the FMEA knowledge graph ontology model, a general information extraction model is used to extract the entities and relationships related to the failure mode from the unstructured FMEA document data in the historical fault information to obtain FMEA structured data.

[0119] The present invention processes the unstructured FMEA document data by applying information extraction technology, and uses a general information extraction model (Universal Information Extraction, UIE) to extract historical fault information such as product documents in the research and manufacturing stage, quality zeroing reports and analysis monthly reports in the on-orbit operation and maintenance stage. The UIE model has the general information processing ability of a large-scale text structure pre-trained model, and uses a prompt learning mechanism based on the ontology schema layer to process small sample data related to faults, which can improve the accuracy and recall rate of data related to aerospace quality problems.

[0120] (4) Knowledge conversion: According to the top-down construction method of the knowledge graph, organize the data according to the schema layer of the FMEA knowledge graph ontology model, and convert the structured data obtained from the conversion of the FMEA digital model of the spacecraft and the structured fault data obtained from knowledge extraction into RDF triples for knowledge representation.

[0121] (5) Knowledge fusion: For the synonym terms and co-referential terms of the RDF triple data, establish or utilize the existing domain-specific thesaurus for entity alignment and co-referential resolution to achieve knowledge fusion and knowledge connection across units and departments.

[0122] RDF triples include a subject, a predicate, and an object.

[0123] (6) Knowledge storage: Store the RDF triple data in an RDF database or an attribute graph database to form an FMEA knowledge graph, and perform unified networked storage management based on knowledge graph technology, serving as an authoritative truth source covering the failure mode information of the entire life cycle of the spacecraft.

[0124] Store the triple data in an RDF database or a property graph database. The advantage of the RDF database lies in the unique resource identifier of knowledge, which is convenient for knowledge sharing and view establishment. The property graph database, with Neo4j graph database as a typical representative, features a user-friendly interface, a declarative graph query language, ACID transactions, and high performance. It supports operations such as depth-first or breadth-first traversal of the graph model, entity node degree statistics, multi-hop path query, and critical path finding, and is suitable for multi-dimensional statistical analysis of failure modes and the establishment of correlation analysis.

[0125] 3. Application of FMEA Knowledge Graph

[0126] Utilize the FMEA knowledge graph to achieve knowledge sharing and critical failure mode analysis, and realize intelligent analysis and decision support for FMEA.

[0127] The retrieval technology of the FMEA knowledge graph is based on the Cyber query language of the Neo4j graph database, supporting operations such as data addition, deletion, modification, and query. Among them, the query retrieval operation supports statistical analysis tasks such as node query, relationship query, multi-hop path query, shortest path finding, and node in-degree and out-degree statistics.

[0128] By retrieving and analyzing the FMEA knowledge graph, the frequency of failure modes within the entire life cycle of a key single machine can be automatically given. For high-frequency failure modes, through the calculation of the shortest path within multiple paths and the analysis of related problem causes, the most likely cause of the failure mode can be given.

[0129] The calculation method of the occurrence frequency of the failure mode is as follows:

[0130] Input the failure mode node in the FMEA knowledge graph, and query the number N of quality problem cases related to the failure mode that have occurred in the FMEA knowledge graph within a period of time T.

[0131] Occurrence frequency of failure mode = Number of quality problem cases N / Time T

[0132] By defining the meta-path of the FMEA knowledge graph to perform chain modeling on the failure mode, similarity analysis of occurred failure cases can be carried out, and specific reduction measures for related failure modes can be inferred, providing a decision basis for the failure closed-loop disposal process based on the severity of the failure consequence impact.

[0133] The recommended method for the reduction measures is as follows:

[0134] (1). Define the meta-path between entity type nodes and other entity type nodes on the ontology model of the FMEA knowledge graph;

[0135] In fault chain modeling, meta-paths are used to represent the characteristics of fault mode nodes in the entire knowledge graph. Different meta-paths represent different semantics. For the ontology schema layer G=(N, R) of the FMEA knowledge graph, the meta-path P between entity type node N1 and entity type node N l is represented as:

[0136]

[0137] where N i is the i-th entity type node, and the relationship R l is the l-th relationship type, and the path length is l-1;

[0138] (2) If there is a path composed of certain nodes and relationships in the data layer of the FMEA knowledge graph that satisfies the meta-path pattern, then this path is called an instance of the meta-path. The round-trip meta-path is usually symmetric. Define the symmetric meta-path p of the entity's own node on the data layer G=(V, E) of the FMEA knowledge graph:

[0139]

[0140] where n i is an instance node on the FMEA knowledge graph, and r l is an instance relationship node on the FMEA knowledge graph.

[0141] (3) According to the similarity measurement algorithm for nodes of the same type in the heterogeneous network, determine the similarity between the fault mode node to be reduced and other fault mode nodes of the same type in the ontology schema layer of the FMEA knowledge graph;

[0142]

[0143] where, |{p x→y : p x→y ∈M}| is the number of meta-paths from instance node x to instance node y on the data layer of the FMEA knowledge graph; M is the set of all meta-paths on the data layer of the FMEA knowledge graph.

[0144] |{p x→x : p x→x ∈M}| is the number of symmetric meta-paths of instance node x on the data layer of the FMEA knowledge graph;

[0145] |{p y→y : p y→y ∈M}| is the number of symmetric meta-paths of instance node y on the FMEA data layer;

[0146] (4) Determine the failure mode node with the highest similarity to the failure mode node to be eliminated, denoted as the reference failure mode node, and recommend the elimination measures corresponding to the reference failure mode node to the failure mode to be eliminated.

[0147] If the number of meta-paths between two instance nodes is large, it indicates a high similarity between the two failure mode nodes, implies a high correlation between the two failure modes, and the elimination measures of the two are similar. When one failure mode occurs, the measures of the other failure mode are more likely to be recommended to a position with a higher priority in the elimination decision.

[0148] In summary, a method for applying spacecraft FMEA based on SysML model and knowledge graph provided by the present invention includes three parts: FMEA digital model management, FMEA data management based on knowledge graph, and FMEA knowledge graph application. First, by analyzing the functional failure logic model and the semantic model of the basic elements of the corresponding failure mode, the FMEA digital model is represented using the SysML metamodel and state diagram of the system modeling language. Second, the FMEA digital model in the design verification stage is converted into FMEA structured data through the XML metadata conversion method, and the FMEA structured data obtained by processing the product document data related to FMEA in the research, development, and manufacturing stage and the quality zeroing report and monthly analysis report data in the on-orbit operation and maintenance stage of the spacecraft using the general information extraction model. The above structured data is used with knowledge graph technology to achieve unified networked storage management of FMEA knowledge, and a database covering the failure mode information of the entire life cycle of the spacecraft model is established as the authoritative truth source. Third, through on-graph reasoning and calculation, functions such as automatic calculation of the occurrence degree of failure modes and matching recommendation of elimination measures for failure modes are realized, so as to achieve automatic elimination decision-making for spacecraft failures and early warning of high-incidence risks, and ultimately achieve the goal of reducing the design risk and on-orbit operation risk of the spacecraft.

[0149] Embodiment:

[0150] The present invention will be further described below in conjunction with the satellite failure management embodiment.

[0151] The main function of the satellite control subsystem is to complete the attitude control and orbit control of the satellite in each mission stage from satellite-rocket separation to on-orbit operation until the end of its life. Timely detection of faults, handling of faults, analysis of historical fault experience, and timely improvement of design weaknesses are important cornerstones for maintaining the normal operation of the satellite in orbit. Taking the failure of a typical single machine of a two-degree-of-freedom gyro as an example, an FMEA digital model is defined, FMEA data is managed based on the knowledge graph, and FMEA knowledge graph application is carried out.

[0152] The two-degree-of-freedom gyro plays a crucial role in satellite attitude control. It is used to accurately measure the angular velocity of the satellite around three orthogonal axes (or two mutually perpendicular axes), and is the basis for maintaining the stable pointing of the satellite and performing attitude adjustment. The prerequisites for its normal function include the normal operation of the bearings, the normal output of the registers, and the normal angular velocity. If the above conditions are not met, it will cause functional failures of the two-degree-of-freedom gyro. The main failure modes and failure severity levels of the two-degree-of-freedom gyro are as follows, and the main causes include jamming, fatigue, and single-event effects in the space environment.

[0153] Table 1 On-orbit failure modes of the two-degree-of-freedom gyro

[0154]

[0155] According to the semantic model of the basic elements of the failure mode, the digital model of the FMEA of the two-degree-of-freedom gyro defined using the SysML state diagram and the meta-model extension mechanism is as Figure 8 shown.

[0156] In addition, the failure information of the two-degree-of-freedom gyro in the development, manufacturing, and on-orbit operation and maintenance stages is obtained, and the unstructured data is extracted through a general information extraction model to obtain structured data. The digital model of the FMEA of the two-degree-of-freedom gyro in the design verification stage is converted into FMEA structured data through XML metadata, and the above FMEA structured data is stored in the Neo4j database under the guidance of the FMEA ontology model to construct an FMEA knowledge graph, as Figure 9 、 Figure 10 shown.

[0157] 1) Count the occurrence frequency of quality problem cases related to the failure modes of a single unit. After removing duplicates through failure cause inspection, the occurrence frequency of the failure modes within the statistical time period can be determined, as Figure 11 shown:

[0158] MATCH(d:standaloneName)-[m:failureMode]->(f:standaloneFailureMode),(f:standaloneFailureMode)-[r:relevantTo]->(n:qualityProblemCase)WHERE d.standaloneName='two-degree-of-freedom gyro'RETURNf.standaloneFailureModeName,count(r)

[0159] Assume that each occurrence of the failure mode is caused by a different reason. Through automatic calculation, it can be seen that the occurrence frequency of the two types of failure modes of the two-degree-of-freedom gyro based on the number of case occurrences is as follows: FailureMode("Abnormal motor current of the two-degree-of-freedom gyro")[occurrence] = 9. Determine the statistical time period T, and the occurrence degree of this failure is 9 / T

[0160] 2) By executing the similarity calculation algorithm for nodes of the same type in the heterogeneous network of the fault mode knowledge graph PathSim, similar single-machine fault modes with high correlation and large influence range are discovered and recommended to the front positions in the satellite fault mode reduction decision-making.

[0161] An example definition of the symmetric meta-path of the fault mode is as follows:

[0162]

[0163] Mode

[0164] Among them, P represents the meta-path of the fault mode in the ontology layer, and the arrow represents the relationship, that is, the path direction.

[0165] N1, N2, N3, and N4 respectively correspond to the entity node types of single-machine fault mode, quality problem case, fault cause, and fault impact; R1, R2, R3, and R4 are the corresponding relationship types of relevantTo, cause, effect, and effect respectively.

[0166] An example definition of the symmetric meta-path of the fault cause is as follows:

[0167]

[0168] Among them, P2 represents the meta-path of the fault cause in the ontology layer, and the arrow represents the relationship, that is, the path direction. N1, N2, N3, and N4 respectively correspond to the entity node types of fault cause, fault impact, single-machine fault mode, and quality problem case; R1, R2, R3, and R4 are the corresponding relationship types of effect, effect, relevantTo, and cause respectively.

[0169] Calculate the similarity between fault modes through the similarity measurement algorithm for nodes of the same type in the heterogeneous network:

[0170]

[0171] Among them, the numerator represents the number of path instances from the fault mode node to the fault cause node; the denominator represents the number of path instances from the fault mode node to the fault mode node and the number of path instances from the fault cause node to the fault cause node.

[0172] Based on the data layer of the FMEA knowledge graph, an adjacency matrix from the fault mode to the fault cause of the two-degree-of-freedom gyro is given, and the value on the matrix is the number of times the corresponding fault mode occurs due to this cause:

[0173]

[0174]

[0175] By calculation, it is found that the similarity between the angular velocity anomaly in failure mode 2 and the data output anomaly in failure mode 3 is the highest, the failure causes are more similar, and the mitigation measures for the two failure modes are more similar and substitutable. According to the calculation results, the disposal measures for the angular velocity anomaly failure mode can be recommended to the data output anomaly failure. After expert manual verification, the result is correct.

[0176] Although the present invention has been disclosed above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solution of the present invention without departing from the spirit and scope of the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

Claims

1. A spacecraft FMEA application method based on SysML model and knowledge graph, characterized in that The steps include: S1. In the design verification phase, the functional failures of the spacecraft are logically analyzed to obtain the functional failure logic model and the semantic model of the basic elements of the failure mode; based on the functional failure logic model and the semantic model of the basic elements of the failure mode, the SysML modeling tool is used to build the FMEA digital model of the spacecraft; S2. Use XML metadata conversion method to convert FMEA digital model into FMEA structured data; S3, mapping the concepts and their subclasses and relationships in the semantic model of the basic elements of the failure mode to the concepts, subclasses, and relationships of the FMEA knowledge graph ontology model to obtain the FMEA knowledge graph ontology model; S4. Extract unstructured fault mode data during the development and manufacturing and on-orbit operation and maintenance phases to obtain structured fault information; S5. Convert the FMEA digital model in the design verification phase to obtain FMEA structured data and structured fault information in the development and manufacturing and on-orbit operation and maintenance phases, and establish an FMEA knowledge graph according to the FMEA knowledge graph ontology model; S6. Based on the FMEA knowledge graph, automatic statistical calculation of the frequency of failure mode occurrence and recommendation of failure mode reduction measures are realized.

2. The spacecraft FMEA application method based on SysML model and knowledge graph according to claim 1 is characterized in that: Unstructured failure mode data include product documents in the development and manufacturing phase, quality zeroing reports in the on-orbit operation and maintenance phase, and monthly analysis reports.

3. The spacecraft FMEA application method based on SysML model and knowledge graph according to claim 1 is characterized in that: The specific steps of step S1 are as follows: Sort out the system states where the functional requirements are not met, use the system states where the functional requirements are not met as the failure mode, and establish a functional failure logic model; the system states where the functional requirements are not met include function overflow, function failure, function failure, function degradation, and function violation; According to the failure mode in the functional failure logic model, a semantic model of the basic elements of the failure mode is constructed to describe the failure mode corresponding to the functional abnormal state; According to the functional failure logic model and the corresponding semantic model of the basic elements of the failure mode, the FMEA digital model is obtained.

4. The spacecraft FMEA application method based on SysML model and knowledge graph according to claim 3 is characterized in that: The functional fault logic model includes a unit functional fault model, a system functional fault model, an interface fault model, and a fault propagation model. The construction process is as follows: The functional requirements of the spacecraft are decomposed in a top-down manner, and the system states that do not meet the functional requirements are sorted out. The system states that do not meet the functional requirements are used as failure modes to obtain the system functional failure model and the single-machine functional failure model; the system functional failure model represents the system failure mode and its impact on the overall mission or system function; the single-machine functional failure model represents the single-machine failure mode and its impact on the single-machine mission or single-machine function; According to the bottom-up method, the fault propagation between single-machine and single-machine, single-machine and system, and system and system is analyzed to obtain the interface fault model and fault propagation model. The interface fault model represents the interface fault mode and its influence on the overall mission or system function. The fault propagation model represents the propagation chain in which faults are transmitted layer by layer and level by level.

5. The spacecraft FMEA application method based on SysML model and knowledge graph according to claim 4 is characterized in that: The failure mode basic element semantic model defines four general class concepts and their subclasses and relationships and one proprietary class concept and its subclasses and relationships. The general class concepts include functional status, failure mode, mitigation measures, and risk sequence number. The general class concepts describe the failure modes in single-machine function failure models and system function failure models; the proprietary class concepts include compatibility, error propagation, and transmission parameter coordination failure; the proprietary class concepts are used to describe the failure modes of interface failure models and fault propagation models.

6. The spacecraft FMEA application method based on SysML model and knowledge graph according to claim 1 is characterized in that: The steps to build the FMEA digital model are as follows: S2.

1. Construct a SysML activity diagram according to the task requirements, wherein the SysML activity diagram represents the activities required to complete the task; the SysML activity diagram includes abnormal states and normal states of the task completion activities; S2.2, connect the abnormal state of the task completion activity in the SysML activity diagram with its normal state through a violation relationship; wherein the violation relationship is represented by the SysML metamodel definition; S2.3, for the failure mode in the SysML activity diagram obtained after executing step 2.2, instantiate the failure mode basic element semantic model in the SysML state diagram; S2.4, constructing a SysML metamodel containing fault mode attribute information on the SysML metamodel diagram according to the semantic model of the basic elements of the fault mode; S2.

5. Based on the SysML metamodel containing the failure mode attribute information, the concept types in the SysML metamodel are instantiated in the SysML state diagram to obtain the FMEA digital model.

7. The spacecraft FMEA application method based on SysML model and knowledge graph according to claim 5 is characterized in that: The generic class concept methods in the instantiated failure mode basic element semantic model include: Use SysML state diagrams and metamodel extension mechanisms to represent functional states, failure modes, and mitigation measures; The risk sequence number class is represented using the SysML parameter diagram model; The methods for instantiating the proprietary class concepts in the semantic model of the failure mode basic elements include: The metamodel extension mechanism is used to represent compatibility, error propagation, and transmission parameter coordination failures.

8. The spacecraft FMEA application method based on SysML model and knowledge graph according to claim 1 is characterized in that: The steps to construct the FMEA knowledge graph are as follows: Determine the ontology model layer: expand the semantics of concepts and their subclasses and relationships in the semantic model of the basic elements of the fault mode into concepts, subclasses, and relationships of the FMEA knowledge graph ontology model to obtain the FMEA knowledge graph ontology model; Knowledge acquisition: Use XML metadata conversion method to convert the FMEA digital model into structured data, obtain the structured data converted from the FMEA digital model and historical fault information during the development and manufacturing and on-orbit operation and maintenance stages; the historical fault information includes unstructured and structured data; Knowledge extraction: According to the concepts, subclasses, and relationships in the FMEA knowledge graph ontology model, a general information extraction model is used to extract entities and relationships related to failure modes from the unstructured FMEA document data in the historical failure information to obtain FMEA structured data; Knowledge conversion: According to the top-down construction method of the knowledge graph, the data is organized according to the FMEA knowledge graph ontology model pattern layer, and the FMEA structured data obtained by converting the FMEA digital model of the spacecraft and the structured fault information obtained by knowledge extraction are converted into RDF triples for knowledge representation; Knowledge fusion: For synonyms and coreference terms of RDF triple data, establish or use existing domain-specific lexicons to perform entity alignment and coreference resolution, and achieve knowledge fusion and knowledge connectivity across units and departments; Knowledge storage: Store RDF triple data in the RDF database or property graph database to form a FMEA knowledge graph.

9. The spacecraft FMEA application method based on SysML model and knowledge graph according to claim 1, characterized in that: The method for calculating the frequency of occurrence of the failure mode is as follows: Enter the failure mode node in the FMEA knowledge graph and query the number of quality problem cases N related to the failure mode that have occurred in the FMEA knowledge graph within a period of time T: Frequency of failure mode occurrence = number of quality problem cases N / time T.

10. The spacecraft FMEA application method based on SysML model and knowledge graph according to claim 1, characterized in that: The recommended mitigation measures for the failure mode to be mitigated are as follows: Define the meta-path between entity type nodes and other entity type nodes on the FMEA knowledge graph ontology model: For the FMEA knowledge graph ontology model layer G = (N, R), entity type node N1 and entity type node N l The meta-path P between is expressed as: Among them, N i is the i-th entity type node, relationship R l is the lth relationship type, and the path length is l-1; If there are some paths composed of nodes and relationships in the FMEA knowledge graph data layer that satisfy the meta-path pattern on the ontology model, then the path is called an instance of the meta-path. The round-trip meta-path is usually symmetric. The symmetric meta-path p of the entity's own node on the FMEA knowledge graph data layer G=(V,E) is defined as: Among them, n i is an instance node on the FMEA knowledge graph data layer, r l It is the instance relationship node on the FMEA knowledge graph; According to the similarity measurement algorithm of nodes of the same type in heterogeneous networks, the similarity between the failure mode node to be eliminated and other failure mode nodes of the same type is determined on the FMEA knowledge graph data layer: Among them, |{p x→y :p x→y ∈M}| is the number of meta-paths from node x to node y; M is the set of all meta-paths on the FMEA knowledge graph data layer; |{p x→x :p x→x ∈M}| is the number of symmetric meta-paths of node x; |{p y→y :p y→y ∈M}| is the number of symmetric meta-paths of node y; The fault mode node with the greatest similarity to the fault mode node to be mitigated is determined and recorded as the reference fault mode node, and the mitigation measures corresponding to the reference fault mode node are recommended to the fault mode to be mitigated.

Images