Fault mode effect analysis method and device based on mbse-llm

Abstract

This application relates to the interdisciplinary field of artificial intelligence and reliability systems engineering, and provides a failure mode effect analysis (FMEA) method and apparatus based on MBSE-LLM. The method first constructs a logical model and severity category set of the analysis object based on MBSE, then enhances the logical model. Next, it generates enhanced component descriptions for the components in the enhanced logical model. Using LLM, it constructs cue words based on these enhanced component descriptions and preset failure occurrence clues, embedding the cue words into each component to obtain a list of potential failure modes, thereby constructing a structured failure record. Finally, it performs FMEA. This method leverages the structured knowledge representation capabilities of the model in systems engineering and the semantic understanding and reasoning capabilities of the large language model to achieve automated and intelligent identification of potential failure modes in complex systems, effectively improving the accuracy and efficiency of FMEA.

Description

Technical Field

[0001] This application relates to the interdisciplinary field of artificial intelligence and reliability systems engineering, and in particular to a failure mode effect analysis method and apparatus based on MBSE-LLM. Background Technology

[0002] Failure Mode and Effect Analysis (FMEA) is a key method in reliability design analysis. Traditional analysis mainly relies on manual work by domain experts, which suffers from inefficiency, strong subjectivity, and the tendency to miss potential failure modes, making it difficult to adapt to the rapid iteration requirements of complex systems. With the development of artificial intelligence, Large Language Models (LLMs) have demonstrated powerful capabilities in knowledge reasoning and text generation, making intelligent FMEA possible. However, directly using LLMs for failure mode analysis can easily lead to "illusion" phenomena, generating erroneous results that do not conform to engineering realities, and lacking a precise understanding of domain knowledge.

[0003] To alleviate this problem, Retrieval-augmented Generation (RAG) technology has been introduced to improve the accuracy of responses through external knowledge retrieval. However, the performance of this method heavily depends on the text slicing strategy, the quality of the embedding model, and the size of the vector database, and it generally suffers from limitations such as incomplete recall information, weak semantic connections, and insufficient cross-document logical understanding.

[0004] Therefore, there is an urgent need for an intelligent analysis method that can deeply integrate systems engineering knowledge and improve semantic understanding and reasoning capabilities. Summary of the Invention

[0005] In view of this, embodiments of this application provide a method and apparatus for failure mode effect analysis based on MBSE-LLM, in order to solve the problems of low efficiency and insufficient accuracy of failure mode analysis in the prior art.

[0006] A first aspect of this application provides a failure mode effect analysis method based on MBSE-LLM, comprising:

[0007] A logical model of the analysis object is constructed using the Model-Based Systems Engineering (MBSE) approach, defining a set of severity categories; the logical model includes... Logical levels Each level It includes a set of components that make up this layer. , It is a positive integer greater than 1. greater than or equal to 0 and less than or equal to 0 equal positive integers, It is a positive integer greater than 0;

[0008] By injecting retrieval-enhanced knowledge into the logical model, an enhanced logical model is obtained.

[0009] For each component in the enhanced logical model LLM is used to generate enhanced component descriptions; the enhanced component descriptions cover functional details, connectivity relationships, and redundancy mechanism elements. greater than or equal to 1 and less than or equal to 1 Positive integers;

[0010] Using the Large Language Model (LLM), guided prompts are constructed based on the enhanced component descriptions and preset fault occurrence clues of each component. These prompts are then embedded into each component of the enhanced logic model to obtain a list of potential fault modes.

[0011] Construct a structured fault record, which includes the fault mode, the corresponding fault occurrence clues, and the task stage;

[0012] Failure Mode and Effects Analysis (FMEA) is performed based on the enhanced logical model, enhanced component description, structured fault records, and severity category set.

[0013] A second aspect of this application provides a failure mode effect analysis apparatus based on MBSE-LLM, comprising:

[0014] The building module is configured to construct a logical model of the analysis object using the Model-Based Systems Engineering (MBSE) approach, defining a set of severity categories; the logical model includes... Logical levels Each level It includes a set of components that make up this layer. , It is a positive integer greater than 1. greater than or equal to 0 and less than or equal to positive integers, It is a positive integer greater than 0;

[0015] The enhancement module is configured to inject enhanced retrieval knowledge into the logical model, resulting in an enhanced logical model.

[0016] The generation module is configured to generate each component in the enhanced logical model. LLM is used to generate enhanced component descriptions; the enhanced component descriptions cover functional details, connectivity relationships, and redundancy mechanism elements. greater than or equal to 1 and less than or equal to 1 Positive integers;

[0017] The embedding module is configured to use the Large Language Model (LLM) to construct guiding prompts based on the enhanced component descriptions and preset fault occurrence clues of each component, and embed the prompts into each component in the enhanced logic model to obtain a list of potential fault modes.

[0018] The build module is also configured to build a fault structure record, which includes the fault mode, the corresponding fault occurrence clues, and the task stage.

[0019] The analysis module is configured to perform failure mode effect analysis based on the enhanced logical model, enhanced component descriptions, fault structured records, and severity category sets.

[0020] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0021] The beneficial effects of the embodiments in this application compared with the prior art are:

[0022] This application first constructs a logical model and severity category set of the analysis object, enhances the logical model, and then generates enhanced component descriptions for the components in the enhanced logical model. Using LLM (Limited Language Modeling), it constructs cue words based on these enhanced component descriptions and preset fault occurrence clues, embedding these cue words into each component to obtain a list of potential fault modes. This leads to the construction of a structured fault record including fault modes, corresponding fault occurrence clues, and task stages. Finally, it uses the enhanced logical model, enhanced component descriptions, structured fault records, and severity category set to perform fault mode impact analysis. This method, based on the model's structured knowledge representation capabilities in systems engineering and the semantic understanding and reasoning capabilities of a large language model, achieves automated and intelligent identification of potential fault modes in complex systems, effectively improving the accuracy and efficiency of fault mode impact analysis. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1This is a flowchart illustrating a failure mode effect analysis method based on MBSE-LLM provided in an embodiment of this application.

[0025] Figure 2 This is a schematic diagram of the hierarchical structure of a certain ground mobile platform.

[0026] Figure 3 This is a schematic diagram of the second-level logical model of a certain ground mobile platform.

[0027] Figure 4 This is a text illustration converted into Markdown format, using the first page of the ground mobile platform design document as an example.

[0028] Figure 5 This is a flowchart illustrating another failure mode effect analysis method based on MBSE-LLM provided in this application embodiment.

[0029] Figure 6 This is a diagram of the MBSE-guided LLM fault analysis interaction mechanism provided in the embodiments of this application.

[0030] Figure 7 This is a schematic diagram of a fault propagation analysis logic provided in an embodiment of this application.

[0031] Figure 8 This is another schematic diagram of fault propagation analysis logic provided in the embodiments of this application.

[0032] Figure 9 This is a schematic diagram of a fault mode effect analysis device based on MBSE-LLM provided in an embodiment of this application.

[0033] Figure 10 This is a schematic diagram of the electronic device provided in the embodiments of this application. Detailed Implementation

[0034] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0035] The following will describe in detail, with reference to the accompanying drawings, a method and apparatus for failure mode effect analysis based on MBSE-LLM according to embodiments of this application.

[0036] As mentioned above, directly using LLM for fault mode analysis can easily lead to "illusion" phenomena, generating erroneous results that do not conform to engineering realities, and lacking a precise understanding of domain knowledge. While analysis methods incorporating RAG technology can improve the accuracy of responses through external knowledge retrieval, their performance heavily relies on text slicing strategies, the quality of the embedding model, and the size of the vector database, and generally suffers from limitations such as incomplete recall information, weak semantic connections, and insufficient cross-document logical understanding.

[0037] In view of this, this application provides a failure mode effect analysis method based on MBSE-LLM, which aims to combine the advantages of structured knowledge representation in model-based systems engineering with the reasoning ability of large language models to achieve more efficient, accurate and comprehensive failure mode identification.

[0038] Figure 1 This is a flowchart illustrating a failure mode and effect analysis method based on MBSE-LLM provided in an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0039] In step S101, the logical model of the analysis object is constructed using the MBSE method, and a set of severity categories is defined.

[0040] The logical model includes Logical levels Each level It includes a set of components that make up this layer. , It is a positive integer greater than 1. greater than or equal to 0 and less than or equal to positive integers, It is a positive integer greater than 0.

[0041] In step S102, retrieval enhancement knowledge is injected into the logical model to obtain the enhanced logical model.

[0042] In step S103, for each component in the enhanced logic model, an enhanced component description is generated using LLM.

[0043] The enhanced component description covers functional details, connectivity relationships, and redundancy mechanism elements; greater than or equal to 1 and less than or equal to 1 Positive integers.

[0044] In step S104, guiding prompts are constructed using LLM based on the enhanced component descriptions and preset fault occurrence clues of each component, and the prompts are embedded into each component in the enhanced logic model to obtain a list of potential fault modes.

[0045] In step S105, a fault structure record is constructed.

[0046] The structured fault record includes the fault mode, the corresponding fault occurrence clues, and the task stage.

[0047] In step S106, failure mode impact analysis is performed based on the enhanced logical model, enhanced component description, fault structured record, and severity category set.

[0048] In some embodiments of this application, the method may be executed by a server or by a terminal device with certain processing capabilities.

[0049] In some embodiments of this application, a logical model of the analysis object can first be constructed using the MBSE method, defining a set of severity categories. Then, retrieval-enhanced knowledge is injected into the logical model to obtain an enhanced logical model.

[0050] In some implementations, each component in the enhanced logic model can be... The process involves generating enhanced component descriptions. Then, using a Large Language Model (LLM), guided prompts are constructed based on the enhanced component descriptions and preset fault occurrence clues for each component. These prompts are then embedded into each component in the enhanced logic model to obtain a list of potential fault modes.

[0051] Furthermore, a structured record of faults, including fault modes, corresponding fault occurrence clues, and task stages, can be constructed. Then, based on the enhanced logical model, enhanced component descriptions, structured fault records, and severity category sets, fault mode impact analysis can be performed.

[0052] According to the technical solution provided in the embodiments of this application, firstly, a logical model and a set of severity categories for the analysis object are constructed. The logical model is then enhanced. Next, enhanced component descriptions are generated for the components in the enhanced logical model. Based on these enhanced component descriptions and preset fault occurrence clues, LLM is used to construct cue words. These cue words are embedded into each component to obtain a list of potential fault modes. Furthermore, a structured record of faults, including fault modes, corresponding fault occurrence clues, and task stages, is constructed. Finally, the enhanced logical model, enhanced component descriptions, structured fault records, and set of severity categories are used to perform fault mode impact analysis. This method, based on the structured knowledge representation capability of the model in systems engineering and the semantic understanding and reasoning capability of the large language model, achieves automated and intelligent identification of potential fault modes in complex systems, effectively improving the accuracy and efficiency of fault mode impact analysis.

[0053] In some embodiments of this application, constructing the logical model of the analysis object using the MBSE method can be done by assuming the overall analysis object system is as follows: The MBSE method is used to divide the analyzed object into multiple logical levels based on its actual hardware architecture, denoted as... , of which each layer It includes a set of components that make up this layer. .in this way, It can represent the first The first in the layer Each component, also known as a physical unit, For the first The total number of components in a layer.

[0054] Within each defined hierarchy, identify and create all entity component nodes that constitute that hierarchy, and then perform component... Define its attribute tuple ;in, A unique identifier for the component. For component name, This describes the functionality of the components. All node components can be stored in the structured data format JSON.

[0055] After defining the component nodes at each level, you can further establish connections between components. Connections can be formed by port definitions and connection lines: Let the port set be... Each port Must be with one and only one component Associated, with a unique port identifier Port type and port direction Let the set of connection relations be... Each connection A directed edge from one output port to one input port is defined as: Among them, the connection types include three categories: matter, information, and energy, namely... Text attributes This provides a supplementary explanation of the connection relationship, highlighting that the connection lines have directionality.

[0056] in, Indicates Boolean type, Indicates integer. Indicates floating-point type, Represents a character type; Indicates the input port type. Indicates the output port type; It is a positive integer; express A unique identifier, express input port, express The output port, express Connection type; To represent information, Indicates energy. It refers to matter.

[0057] The logical model constructed through the above steps is represented as a set of nodes. and edge set The directed graph formed The data is stored in the structured data format JSON to obtain the logical model of the object being analyzed.

[0058] In some embodiments of this application, defining the severity category set can be a combination of The consequences of the failure, defining a set of severity categories. .

[0059] Taking a certain ground mobile platform as an example, this platform serves as the mobile vehicle for a multi-purpose robot. By loading different functional modules, it can perform various tasks such as patrolling, transportation, and rescue, and is applied in fields such as industrial inspection, smart logistics, and disaster relief. Therefore, the platform is required to possess strong mobility, high speed, and the ability to climb slopes, overcome obstacles, and wade through water; strong environmental adaptability, allowing it to move freely on terrains such as grasslands, sandy areas, mountains, roads, and indoor spaces; the ability to withstand high vibration and impact loads, a temperature range of -25 to +50℃, and harsh electromagnetic environments; and primarily remote control operation, capable of local path planning. The physical structure of the ground mobile platform includes a control box, remote control box, power box, power supply, vehicle body, and suspension. Dividing these components hierarchically according to the steps described above yields the following... Figure 2 The hierarchical structure shown. From Figure 2 As can be seen, the ground mobile platform can be divided into 3 layers.

[0060] Taking the control box in the second layer of the ground mobile platform as an example, set the control box ID to P-004, name it "Control Box", and describe its function as "Positioning distance (under normal working conditions) ≥ 3m; Control and sensing: direction, speed, attitude". Save it in JSON format as follows:

[0061] {

[0062] "name": "Control Box",

[0063] "id": "P-004",

[0064] "parent_id": "P-000",

[0065] "parent_names": "Ground Mobile Platform",

[0066] "function": "Location distance (under normal working conditions) ≥3m\nControl and sensing: direction, speed, attitude",

[0067] "level": "subsystem",

[0068] "top_agreement_level": "Ground mobile platform",

[0069] "agreement_level": "Ground Mobile Platform",

[0070] }

[0071] Furthermore, taking the second level of the ground mobile platform as an example, the constructed logical model is as follows: Figure 3 As shown.

[0072] In some embodiments of this application, injecting retrieval enhancement knowledge into the logical model may include: parsing the enhanced knowledge of the analysis object to obtain a preset format string; using a dynamic semantic aggregation slicing strategy and a sliding window aggregation method to convert the preset format string into a semantically coherent set of knowledge slices; wherein, the dynamic semantic aggregation slicing is at least based on a pre-trained embedding model; each knowledge slice includes a semantically complete text segment, and each knowledge slice corresponds to an embedding vector; importing the set of knowledge slices and the embedding vectors corresponding to each knowledge slice into a vector database; extracting any The original query suggestions are used to construct query vectors for each component using a pre-trained embedding model. Then, similarity calculations and re-ranking are performed on the query vectors of each component in a vector database to obtain... The recall results; The recall results and original query suggestions are concatenated into complete suggestion terms, guiding the LLM to filter and summarize the recall results, resulting in enhanced suggestions. Repeat the above steps until all instances have been traversed. .

[0073] In other words, when injecting retrieval enhancement knowledge into the logical model, the non-handwritten electronic design and operation and maintenance data of the analysis object (such as PDF, Word, WPS, scanned drawings, XML format, etc.) can be converted into a structured text intermediate format using an optical character recognition model. Then, the document content is parsed page by page, and a Markdown format string containing semantic tags for headings, paragraphs, and lists is output.

[0074] Taking the first page of the ground mobile platform design document as an example, the text converted to Markdown format is as follows: Figure 4 As shown.

[0075] Next, a dynamic semantic aggregation slicing strategy is adopted. First, the Markdown text is initially segmented according to natural sentences to obtain a sentence sequence. Then, a pre-trained embedding model is used to map each sentence into a dense vector. Finally, sliding window aggregation is performed starting from the first sentence. The pre-trained embedding model can be, for example, Yinka.

[0076] The sliding window aggregation process can be as follows: initialize the current slice vector and slice content; for subsequent statements, calculate their cosine similarity with the current slice vector; if the cosine similarity is greater than or equal to a preset similarity threshold, then merge the subsequent statements into the current slice and update the slice vector to the vector transformed from the new content; otherwise, end the current slice, treat it as a complete knowledge fragment, and start a new slice with the subsequent statement as the starting point; this process continues until all statements have been processed, and finally a semantically coherent set of knowledge slices is obtained.

[0077] Taking the first paragraph of the first page of the ground mobile platform design document as an example, after being segmented into 4 sentences according to the original text's period, semantic similarity is calculated using the Yinka embedding model. At a preset similarity threshold of 0.65, the first 3 sentences, due to their high semantic coherence (similarities of 0.72 and 0.68 respectively), are aggregated into a single knowledge fragment. The 4th sentence, with a similarity of 0.63 to the current segment, is below the preset similarity threshold and is therefore separated into a separate fragment. Ultimately, the original text is segmented into two semantically complete knowledge fragments. The first paragraph states: "The ground mobile general-purpose platform is a mobile driving carrier for multi-purpose robots, capable of..." By loading different functional modules, it can perform various tasks such as patrolling, transportation, and rescue, and is applied in fields such as industrial inspection, smart logistics, and disaster relief. Therefore, the platform is required to have strong mobility, high speed, and the ability to climb slopes, overcome obstacles, and ford water; strong environmental adaptability, allowing it to move freely on grasslands, sandy areas, mountains, roads, and indoor environments. The second paragraph states: "It can withstand high vibration and impact loads, a temperature range of -25 to +50℃, and harsh electromagnetic environments; in terms of operation, it is primarily remote-controlled and capable of local path planning."

[0078] The generated knowledge slice set and its embedding vectors can then be imported into a vector database, and the components can be analyzed one by one. Specific analysis steps include:

[0079] 1) For any component Analyze the data and extract its attributes to generate original query suggestions. , will Input the above embedding model to obtain the query vector. ;

[0080] 2) Perform similarity calculation in the vector database and retrieve the M most similar slices. M is a positive integer;

[0081] 3) Improve recall results using a re-ranking model The similarity is reordered, and the K slices with the highest similarity after reordering are selected as the final recall results. ;

[0082] 4) The final recall results Compared with the original query suggestion The words are concatenated into complete prompts to guide the large model in filtering and summarizing the recall results.

[0083] 5) Repeat the above steps until all components in the logical model have been analyzed.

[0084] Taking the "suspension" of a ground mobile platform as an example, the system generates a query prompt: "Typical failure modes, failure mechanisms, and historical cases related to the 'suspension' component." After being encoded by the Yinka embedding model, this query retrieves the knowledge slices generated in step 2 from the FAISS vector database, recalling several semantically similar fragments. After being re-ranked by bce-reranker, the returned Top-K results mainly include general descriptions of the platform's environmental adaptability and vibration loads, such as: "Able to withstand high vibration and impact loads" and "Can move freely in terrains such as grassland, sand, and mountains." After concatenating these fragments with the original query to form prompts, the large model summarizes the answer: "Based on the provided data, it can be inferred that the suspension system needs to withstand high vibration and impact loads and maintain stable operation in complex terrains (such as sandy areas, mountains, and grasslands). Therefore, potential failure modes may include structural fatigue fracture, shock absorber failure, loosening or wear of connecting parts, etc., and the failure mechanism may be related to long-term alternating loads, impact overloads, or material aging. However, the provided corpus does not contain specific historical failure cases or measured failure data related to suspension components. The above analysis is based on reasonable inferences about the system environment and performance requirements. It is recommended to further verify this analysis by combining it with specific tests or operation and maintenance records."

[0085] In some embodiments of this application, for each component in the enhanced logic model Generate enhanced component descriptions, including: extracting The original description, which includes at least descriptive information, basic functions, hierarchical affiliation, and interface definitions; in a vector database, using Using component identifiers, functional keywords, and port types as search criteria, relevant technical document snippets are retrieved. Design prompts guide the LLM to integrate the search results with the original descriptions, generating a comprehensive summary of functional details, connectivity relationships, and redundancy mechanisms. Enhanced component description.

[0086] Among them, the preset fault occurrence clues include at least one of the following: loss of specified function, discontinuous execution of function, incomplete execution of function, performance deviation, premature or delayed execution of function or execution of undesired function; the task stage is the task stage in which each fault mode occurs, as determined by the LLM based on the task stage list of preset analysis objects.

[0087] In other words, any component in the logical model can be extracted. The description, basic functions, hierarchical affiliation, and interface definitions of the component are used to call the aforementioned vector database. Relevant technical document fragments are automatically retrieved using component identifiers, functional keywords, and port types as search criteria. At the same time, prompt words are designed to guide the large model to integrate the search results with the original description, generating an enhanced component description that covers functional details, connection relationships, and redundancy mechanism elements.

[0088] Taking the ground mobile platform control box as an example, the initial functional description manually added to the model is: "The control box is used for positioning at a distance (under normal working conditions) ≥3m, and has the functions of controlling and sensing direction, speed, and attitude." After calling the vector database and summarizing by the large model, the enhanced description is generated: "The control box is a key subsystem of the ground mobile general platform, responsible for the platform's command parsing, power management, and system start-up and shutdown control functions. Its core functions include connecting or disconnecting the power supply of the whole machine, receiving and processing drive and operation commands from the remote control terminal, outputting control signals to actuators such as the drive motor, and monitoring the power supply status of the lithium battery pack in real time. Under normal working conditions, the positioning distance of the control box can reach ≥3 meters, and it has the functions of controlling and sensing direction, speed, and attitude. The control box is connected to the lithium battery pack through the power supply line, and interacts with components such as the drive motor, sensors, and communication modules through the signal interface. Its internal components include key components such as the main switch, fuse, electrical connector, backup power supply, and interface module."

[0089] Then, based on the enhanced component description, guiding prompts are constructed according to the fault occurrence clues of specified function loss, discontinuous function execution, incomplete function execution, performance deviation, premature or delayed function execution, and execution of unexpected functions. These prompts are then embedded into the final recall results obtained when retrieving enhanced knowledge in the logical model. Based on this, the LLM generates a list of potential failure modes that conforms to engineering practice.

[0090] Additionally, by combining the task phase list of the analysis object, the LLM can be guided to determine the task phase in which each failure mode occurs, and the failure mode, corresponding failure occurrence clues, and task phase can be bound together as a structured record as input for subsequent impact analysis.

[0091] Taking the control circuit board in the mobile platform control box as an example, we analyze its potential failure modes and the mission stages in which they occur. The mission stages are selected from "start-up-recovery-wading-reconnaissance-full-stage". The partial analysis results are shown in Table 1.

[0092] Table 1. Analysis Results of Potential Failure Modes and Occurrence Stages of Control Circuit Boards

[0093]

[0094] In some embodiments of this application, failure mode effect analysis is performed based on the enhanced logical model, enhanced component description, and fault structured record, including: Each fault mode Based on the enhanced logical model, the affected objects of the target fault are identified, and a fault impact propagation path is established. This path includes local impact, higher-level impact, and final impact. Locally affected objects include directly related components of the target fault; higher-level affected objects include the sub-levels of directly related components within the enhanced logical model; and final impact objects include the top-level components within the enhanced logical model. Using the fault impact propagation path, the fault mode of the target fault, and the description of the enhanced components as prompts, the LLM is guided to generate natural language descriptions of the local impacts. Higher-level influences on natural language description and the natural language description of the ultimate impact ;Will Semantic matching with the severity category set yields... Severity category; combination The severity category yields a structured output; each category is iterated over. All and iterate through The failure mode effect analysis results were obtained.

[0095] In other words, when performing failure mode and effect analysis, components can be used. Each fault mode First, the recipient is identified based on the logical model information. The affected objects include components that have direct relationships in the logical model. The subsystem to which it belongs, i.e., the higher-level affected object. and top-level objects These are denoted as local impact, higher-level impact, and final impact, respectively, thus establishing the propagation path of the fault's impact.

[0096] Then, the identified fault propagation paths, fault modes, and enhanced component descriptions are used as prompt words to guide the LLM to generate natural language descriptions of local effects, higher-level effects, and final effects, respectively denoted as... , , .

[0097] Finally With respect to the set of severity categories defined above Perform semantic matching to determine the severity category, and then match the determined severity category with... Together they constitute the structured output.

[0098] Taking the fault modes of "signal relay interruption", "failure to power on", and "discontinuous or abrupt control command output" of the ground mobile platform control box as examples, the analysis results are shown in Table 2:

[0099] Table 2. Failure Mode Analysis Results of Control Box

[0100]

[0101] In some embodiments of this application, failure mode effect analysis may further include failure cause analysis. This failure cause analysis may include the following steps: [The text abruptly ends here, so the translation stops as well.] The design prompts guide LLM to analyze failure causes from three dimensions: causative factors, environmental factors, and human factors. The causative factors dimension includes failures caused by related components at the same level or by components within the component itself. The environmental factor dimension includes the effects of physical, chemical, and biological environmental factors. The human factors dimension includes failures caused by human error. Based on the failure causes identified in each dimension, the design prompts guide LLM to propose design improvement measures and usage compensation measures. All failure causes, design improvement measures, and usage compensation measures are then structured and linked to failure modes. Forming a failure mode Failure Mode and Effects Analysis (FMEA) records.

[0102] Taking the "failure to transmit control signals" fault mode of the control circuit board as an example, the analysis results are shown in Table 3:

[0103] Table 3 Control Circuit Board Fault Mode FMEA Record

[0104]

[0105] In some embodiments of this application, after completing the Failure Mode and Effects Analysis (FMEA), the components in the logical model can be individually checked using FMEA records based on the semantic synthesis analysis of LLM to establish a causal chain for the propagation of failure effects.

[0106] For any component Each fault mode The design prompts guide the LLM to check this item sequentially. Locally affected objects ;in, .

[0107] If LLM is used to determine The failure modes contain elements related to this. of For fault modes with the same semantics, their descriptions can be compared with the current one. of Semantic merging is performed to obtain an updated natural language description of the local effects. Otherwise, Added as a new failure mode In the failure modes.

[0108] If LLM is used to determine The existence of this For faults with the same semantic meaning, then we can utilize Replace the fault cause description in this document with the one in this document. The updated fault mode is obtained by describing the fault cause with the same semantic meaning. Otherwise, Added as a new cause of failure Among the causes of the malfunction.

[0109] Furthermore, regarding Each fault mode It can also design prompts to guide the LLM to check this item sequentially. Higher-level influence targets .

[0110] If LLM is used to determine The failure modes contain elements related to this. of For fault modes with the same semantics, their descriptions can be compared with the current one. of Semantic merging is performed to obtain an updated natural language description with higher-level influence. Otherwise, Added as a new failure mode In the failure modes.

[0111] If LLM is used to determine The existence of this For faults with the same semantic meaning, then we can utilize Replace the fault cause description in this document with the one in this document. The updated fault mode is obtained by describing the fault cause with the same semantic meaning. Otherwise, Added as a new cause of failure Among the causes of the malfunction.

[0112] Furthermore, regarding Each fault mode Furthermore, prompts can be designed to guide the LLM to check this item sequentially. of .

[0113] If all of the analysis objects are determined The existence of this Descriptions with the same semantics can be compared with this description. Perform semantic merging to obtain the updated natural language description of the final impact. Otherwise, retain this. .

[0114] Finally, the results of the above inspections are combined to construct a cross-level fault propagation causal chain based on the relationship between the related objects in the logical model, with a unified fault mode, fault cause, and fault impact.

[0115] Taking the failure mode "Unable to transmit control signals" of the ground mobile platform control circuit board and its local effect "Interruption of control signal input to the push rod drive circuit" as an example, LLM, guided by prompts, performs semantic checks on the analyzed failure modes of the "push rod drive circuit." The analyzed failure modes include: "Failed to transmit speed / direction signals as expected," "Incorrect signal transmission timing," "Inferior transmission signal integrity," and "Abnormal multi-channel signal processing." After comparison, it is determined that "Failed to transmit speed / direction signals as expected" is semantically consistent with "Unable to transmit control signals," thus generating the unified expression "Failed to transmit control signals as expected."

[0116] Upon further examination of the cause of this fault mode, LLM found that the original fault cause, "abnormal front-end input signal (fuse 1, fuse 2, fuse 3 or control board output signal error) causing the drive circuit to fail to respond correctly," already contained the semantics of the fault cause to be added, "control board cannot transmit control signal." Therefore, the original fault cause was kept unchanged.

[0117] Subsequently, LLM performed semantic analysis on the two fault modes involved, namely "control circuit board" and "push rod drive circuit 1"—"control circuit board cannot transmit control signal" and "push rod drive circuit does not transmit control signal as expected"—and determined that their impact was consistent at the system level. They were then merged into a unified system-level statement: "control box does not output control signal as expected".

[0118] Furthermore, the higher-level impact of the ground mobile platform control circuit board's fault mode "Unable to transmit control signals"—"Control box does not output control signals as expected"—was further examined. Guided by prompts, LLM performed a semantic check on the existing fault modes of the higher-level object "Control box" of "Control circuit board." An existing fault mode, "Control box does not output control signals as expected," was found and directly merged. Further examination of the fault causes for this fault mode revealed existing causes such as: "Program errors or hardware damage to the control logic chip or microprocessor, leading to abnormal control signal generation," "Abnormal front-end input signals (such as loss or errors in the remote control box control signal), leading to misjudgment of the control logic and thus failure to output control signals as expected," and "Poor soldering of internal circuits or poor connector contact, causing interruption of the control signal path or timing disorder." No semantically identical fault causes were found, so "Control circuit board cannot transmit control signals" was added as a new fault cause to this fault mode.

[0119] Next, the final impact of the ground mobile platform control circuit board failure mode "Unable to transmit control signals" was checked as "Ground mobile platform control signal interruption". Guided by the prompt words, LLM compared "Ground mobile platform control signal interruption" with all the final impact statements that had been analyzed. A similar semantic statement "Ground mobile platform control command lost" was found. The semantics were matched and all other identical final impact statements were unified as "Ground mobile platform control command lost".

[0120] Finally, by correlating all the information obtained from the above checks, we can obtain the same-level fault propagation causal chain: "Control circuit board cannot transmit control signals - push rod drive circuit does not transmit control signals as expected", and the cross-level fault propagation causal chain: "Control circuit board cannot transmit control signals - control box does not output control signals as expected - ground mobile platform control commands are lost" and "Push rod drive circuit does not transmit control signals as expected - control box does not output control signals as expected - ground mobile platform control commands are lost".

[0121] Figure 5 This is a flowchart illustrating another failure mode effect analysis method based on MBSE-LLM provided in this application embodiment. Figure 5 As shown, the method includes the following steps:

[0122] Step 1: Construct a logical model of the object being analyzed and define severity categories.

[0123] Step 2: Retrieval Enhancement Knowledge Injection. This retrieval enhancement knowledge injection can be achieved based on modeling information and outputs RAG (Research-Augmented Query) answers.

[0124] Step 3: LLM-based Failure Mode and Effects Analysis. This analysis can be performed based on the modeling information and severity category definitions obtained in Step 1, as well as the RAG responses output in Step 2, and yields preliminary FMEA results.

[0125] Step 4: Semantic Synthesis Analysis and Fault Propagation Causal Chain Construction Based on LLM. This analysis and fault propagation causal chain construction can be achieved based on the preliminary FMEA results and the RAG response content output in Step 2.

[0126] Figure 6 This is a diagram of the MBSE-guided LLM fault analysis interaction mechanism provided in an embodiment of this application. For example... Figure 6 As shown, the component name, top-level object, and parent-level object in the logical model can be input into the LLM. The LLM is used to enhance the initial component description to obtain the enhanced component description. The enhanced component description is then input into the LLM. Furthermore, the retrieval enhancement method can be injected into the LLM to obtain the component name and analysis target from the LLM and to provide the RAG response content back to the LLM.

[0127] LLM receives the analysis objective, analysis requirements, and output requirements, automatically generates prompts, and automatically retrieves the analysis results based on these prompts. These results can include failure mode analysis, local effects, higher-level effects and ultimate effects analysis, severity categories, failure cause analysis, and analysis of design improvements and compensatory measures.

[0128] Figure 7 and Figure 8 This is a schematic diagram of the fault propagation analysis logic provided in an embodiment of this application. For example... Figure 7 As mentioned above, faults can propagate within the same level or across levels. For example, a known fault in component A can be propagated to component B at the same level, or a known fault in component A can be propagated to component B at a higher level.

[0129] like Figure 8 As shown, the propagation correlation analysis of faults between components can be determined based on fault modes, fault effects, and fault causes. Furthermore, by semantically merging the associated fault modes, fault effects, and fault causes, the updated final effects and severity can be obtained.

[0130] All of the above-mentioned optional technical solutions can be combined in any way to form the optional embodiments of this application, and will not be described in detail here.

[0131] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0132] Figure 9This is a schematic diagram of a failure mode effect analysis device based on MBSE-LLM provided in an embodiment of this application. Figure 9 As shown, the device includes:

[0133] Module 901 is configured to construct a logical model of the analysis object using the Model-Based Systems Engineering (MBSE) approach, defining a set of severity categories; wherein, the logical model includes Logical levels Each level It includes a set of components that make up this layer. , It is a positive integer greater than 1. greater than or equal to 0 and less than or equal to positive integers, It is a positive integer greater than 0.

[0134] Enhancement module 902 is configured to inject retrieval enhancement knowledge into the logical model to obtain the enhanced logical model.

[0135] Module 903 is configured to generate each component in the enhanced logical model. LLM is used to generate enhanced component descriptions; the enhanced component descriptions cover functional details, connectivity relationships, and redundancy mechanism elements. greater than or equal to 1 and less than or equal to 1 Positive integers.

[0136] Embedding module 904 is configured to use the Large Language Model (LLM) to construct guiding prompts based on the enhanced component descriptions and preset fault occurrence clues of each component, and embed the prompts into each component in the enhanced logic model to obtain a list of potential fault modes.

[0137] Module 901 is also configured to build a fault structure record, which includes fault modes, corresponding fault occurrence clues, and task phases.

[0138] Analysis module 905 is configured to perform failure mode effect analysis based on the enhanced logical model, enhanced component description, fault structured record, and severity category set.

[0139] According to the technical solution provided in the embodiments of this application, firstly, a logical model and a set of severity categories for the analysis object are constructed. The logical model is then enhanced. Next, enhanced component descriptions are generated for the components in the enhanced logical model. Based on these enhanced component descriptions and preset fault occurrence clues, LLM is used to construct cue words. These cue words are embedded into each component to obtain a list of potential fault modes. Furthermore, a structured record of faults, including fault modes, corresponding fault occurrence clues, and task stages, is constructed. Finally, the enhanced logical model, enhanced component descriptions, structured fault records, and set of severity categories are used to perform fault mode impact analysis. This method, based on the structured knowledge representation capability of the model in systems engineering and the semantic understanding and reasoning capability of the large language model, achieves automated and intelligent identification of potential fault modes in complex systems, effectively improving the accuracy and efficiency of fault mode impact analysis.

[0140] In some implementations, the preset fault occurrence clues include at least one of the following: loss of specified function, discontinuous execution of function, incomplete execution of function, performance deviation, premature or delayed execution of function, or execution of undesired function; the task phase is the task phase in which each fault mode occurs, as determined by the LLM based on the task phase list of preset analysis objects.

[0141] In some implementations, failure mode effect analysis is performed based on the enhanced logical model, enhanced component descriptions, and structured fault records, including: Each fault mode Based on the enhanced logical model, the affected objects of the target fault are identified, and a fault impact propagation path is established. This path includes local impact, higher-level impact, and final impact. Locally affected objects include directly related components of the target fault; higher-level affected objects include the sub-levels of directly related components within the enhanced logical model; and final impact objects include the top-level components within the enhanced logical model. Using the fault impact propagation path, the fault mode of the target fault, and the description of the enhanced components as prompts, the LLM is guided to generate natural language descriptions of the local impacts. Higher-level influences on natural language description and the natural language description of the ultimate impact ;Will Semantic matching with the severity category set yields... Severity category; combination The severity category yields a structured output; each category is iterated over. All and iterate through The failure mode effect analysis results were obtained.

[0142] In some implementations, failure mode effect analysis also includes failure cause analysis; failure cause analysis includes: for failure modes The design prompts guide LLM to analyze failure causes from three dimensions: causative factors, environmental factors, and human factors. The causative factors dimension includes failures caused by related components at the same level or by components within the component itself. The environmental factor dimension includes the effects of physical, chemical, and biological environmental factors. The human factors dimension includes failures caused by human error. Based on the failure causes identified in each dimension, the design prompts guide LLM to propose design improvement measures and usage compensation measures. All failure causes, design improvement measures, and usage compensation measures are then structured and linked to failure modes. Forming a failure mode Failure Mode and Effects Analysis (FMEA) records.

[0143] In some implementations, after completing the failure mode and effect analysis, the method further includes: Each fault mode The design prompts guide the LLM to check this item sequentially. Locally affected objects ;in, ; Response to using LLM to determine The failure modes contain elements related to this. of For fault modes with the same semantics, the description of the fault mode with the same semantics will be compared with this. of Semantic merging is performed to obtain an updated natural language description of the local effects. Otherwise, Added as a new failure mode In the failure modes; in response to using LLM to determine The existence of this Using semantically identical fault causes Replace the fault cause description in this document with the one in this document. The updated fault mode is obtained by describing the fault cause with the same semantic meaning. Otherwise, Added as a new cause of failure Among the causes of the malfunction.

[0144] In some implementations, after completing the failure mode and effect analysis, the method further includes: Each fault mode The design prompts guide the LLM to check this item sequentially. Higher-level influence targets ; Response to using LLM to determine The failure modes contain elements related to this. of For fault modes with the same semantics, the description of the fault mode with the same semantics will be compared with this. of Semantic merging is performed to obtain an updated natural language description with higher-level influence. Otherwise, Added as a new failure mode In the failure modes; in response to using LLM to determine The existence of this Using semantically identical fault causes Replace the fault cause description in this document with the one in this document. The updated fault mode is obtained by describing the fault cause with the same semantic meaning. Otherwise, Added as a new cause of failure Among the causes of the malfunction.

[0145] In some implementations, after completing the failure mode and effect analysis, the method further includes: Each fault mode The design prompts guide the LLM to check this item sequentially. of ; In response to determining all objects corresponding to the analysis object The existence of this Descriptions with the same semantics will be compared with this description. Perform semantic merging to obtain the updated natural language description of the final impact. Otherwise, retain this. .

[0146] In some implementations, fault cause analysis may also include: constructing a cross-level fault propagation causal chain for the analysis object based on the updated fault modes and fault causes.

[0147] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0148] Figure 10 This is a schematic diagram of the electronic device provided in an embodiment of this application. For example... Figure 10 As shown, the electronic device 10 of this embodiment includes: a processor 1001, a memory 1002, and a computer program 1003 stored in the memory 1002 and executable on the processor 1001. When the processor 1001 executes the computer program 1003, it implements the steps in the various method embodiments described above. Alternatively, when the processor 1001 executes the computer program 1003, it implements the functions of each module / unit in the various device embodiments described above.

[0149] Electronic device 10 may be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 10 may include, but is not limited to, a processor 1001 and a memory 1002. Those skilled in the art will understand that... Figure 10 This is merely an example of electronic device 10 and does not constitute a limitation on electronic device 10. It may include more or fewer components than shown, or different components.

[0150] The processor 1001 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0151] The memory 1002 can be an internal storage unit of the electronic device 10, such as a hard disk or RAM of the electronic device 10. The memory 1002 can also be an external storage device of the electronic device 10, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, FlashCard, etc., equipped on the electronic device 10. The memory 1002 can also include both internal and external storage units of the electronic device 10. The memory 1002 is used to store computer programs and other programs and data required by the electronic device.

[0152] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0153] If an integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable medium may include: any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0154] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for failure mode effect analysis based on MBSE-LLM, characterized in that, include: A logical model of the analysis object is constructed using the Model-Based Systems Engineering (MBSE) method, defining a set of severity categories; wherein, the logical model includes Logical levels Each level It includes a set of components that make up this layer. , It is a positive integer greater than 1. greater than or equal to 0 and less than or equal to positive integers, It is a positive integer greater than 0; The logical model is then injected with enhanced retrieval knowledge to obtain the enhanced logical model. for each component in the enhanced logical model , generating an enhanced component description using a large language model (LLM); the enhanced component description encompasses functional details, connection relationships, and redundancy mechanism elements; is a positive integer greater than or equal to 1 and less than or equal to . Using LLM to construct guiding prompts based on enhanced component descriptions and preset fault occurrence clues for each component, and embedding the prompts into each component in the enhanced logic model, a list of potential fault modes is obtained; Construct a structured fault record, which includes fault mode, corresponding fault occurrence clues, and task stage; Fault mode impact analysis is performed based on the enhanced logical model, the enhanced component description, the fault structure record, and the severity category set. The failure mode effect analysis (FMEA) based on the enhanced logical model, the enhanced component description, and the structured failure record includes: right Each fault mode Based on the enhanced logical model, the affected objects of the target fault are determined, and a fault impact propagation path is established; wherein, the fault impact propagation path includes local impact, higher-level impact, and final impact, the locally affected objects include the directly related components of the target fault, the higher-level affected objects include the sub-levels to which the directly related components belong in the enhanced logical model, and the final affected objects include the top level in the enhanced logical model; propagating path of the fault impact, the fault mode of the target fault, and the enhanced component description as prompt words to guide the LLM to generate natural language descriptions of the local impact, the high-level impact, and the final impact, respectively ;​​ The semantically matching with the set of severity categories, resulting in severity category of the Combining the above The severity category yields a structured output; Traverse each All and iterate through The failure mode effect analysis results were obtained.

2. The method according to claim 1, characterized in that, The preset fault occurrence clues include at least one of the following: loss of specified function, discontinuous execution of function, incomplete execution of function, performance deviation, premature or delayed execution of function, or execution of unexpected function; The task phase refers to the task phase in which each failure mode occurs, as determined by the LLM based on a list of task phases for a preset analysis object.

3. The method according to claim 1, characterized in that, The failure mode impact analysis also includes failure cause analysis; The fault cause analysis includes: The fault mode The design prompts guide the LLM to analyze the causes of failures from the perspectives of causation, environment, and human factors. The causation dimension includes failures caused by related components at the same level or by components within the LLM itself. The environment dimension includes the effects of physical, chemical, and biological environmental factors. The human factors dimension includes failures caused by human error. Based on the fault causes identified in each dimension, design prompts guide the LLM to propose design improvement measures and use compensation measures; All failure causes, the aforementioned design improvements, and the use of compensatory measures are linked to the failure mode in a structured manner. The aforementioned failure mode is formed. Failure Mode and Effects Analysis (FMEA) records.

4. The method according to claim 3, characterized in that, After completing the failure mode and effect analysis, the method further includes: right Each fault mode The design prompts guide the LLM to check this item sequentially. Locally affected objects ;in, ; In response to determining using the LLM The failure modes contain elements related to this. of For fault modes with the same semantics, the description of the fault mode with the same semantics is compared with this. of Semantic merging is performed to obtain an updated natural language description of the local effects. Otherwise, the aforementioned Added as a new failure mode In the failure modes; In response to determining the LLM The existence of this The fault causes with the same semantics, using the above Replace the fault cause description in this document with the one in this document. The updated fault mode is obtained by describing the fault cause with the same semantic meaning. Otherwise, the aforementioned Added as a new cause of failure Among the causes of the malfunction.

5. The method according to claim 4, characterized in that, After completing the failure mode and effect analysis, the method further includes: right Each fault mode The design prompts guide the LLM to check this item sequentially. Higher-level influence targets ; In response to determining using the LLM The failure modes contain elements related to this. of For fault modes with the same semantics, the description of the fault mode with the same semantics is compared with this. of Semantic merging is performed to obtain an updated natural language description with higher-level influence. Otherwise, the aforementioned Added as a new failure mode In the failure modes; In response to determining the LLM The existence of this The fault causes with the same semantics, using the above Replace the fault cause description in this document with the one in this document. The updated fault mode is obtained by describing the fault cause with the same semantic meaning. Otherwise, the aforementioned Added as a new cause of failure Among the causes of the malfunction.

6. The method according to claim 5, characterized in that, After completing the failure mode and effect analysis, the method further includes: right Each fault mode The design prompts guide the LLM to check this item sequentially. of ; In response to determining all of the analysis objects The existence of this Descriptions with the same semantics, and the descriptions with the same semantics are compared with this. Perform semantic merging to obtain the updated natural language description of the final impact. Otherwise, retain this. .

7. The method according to claim 6, characterized in that, The method further includes: Construct a cross-level fault propagation causal chain for the analyzed object based on the updated fault modes and fault causes.

8. A failure mode effect analysis device based on MBSE-LLM, characterized in that, include: The building module is configured to construct a logical model of the analysis object using the Model-Based Systems Engineering (MBSE) approach, defining a set of severity categories; wherein, the logical model includes Logical levels Each level It includes a set of components that make up this layer. , It is a positive integer greater than 1. greater than or equal to 0 and less than or equal to positive integers, It is a positive integer greater than 0; The enhancement module is configured to inject enhanced retrieval knowledge into the logical model to obtain an enhanced logical model; The generation module is configured to generate each component in the enhanced logical model. The enhanced component description is generated using a large language model (LLM); the enhanced component description covers functional details, connectivity relationships, and redundancy mechanism elements. greater than or equal to 1 and less than or equal to 1 Positive integers; The embedding module is configured to use the Large Language Model (LLM) to construct guiding prompts based on the enhanced component descriptions and preset fault occurrence clues of each component, and embed the prompts into each component in the enhanced logic model to obtain a list of potential fault modes. The building module is also configured to build a fault structure record, which includes fault modes, corresponding fault occurrence clues, and task stages. The analysis module is configured to perform fault mode impact analysis based on the enhanced logical model, the enhanced component description, the fault structure record, and the severity category set. The failure mode effect analysis (FMEA) based on the enhanced logical model, the enhanced component description, and the structured failure record includes: right Each fault mode Based on the enhanced logical model, the affected objects of the target fault are determined, and a fault impact propagation path is established; wherein, the fault impact propagation path includes local impact, higher-level impact and final impact, the locally affected objects include the directly related components of the target fault, the higher-level affected objects include the sub-levels to which the directly related components belong in the enhanced logical model, and the final affected objects include the top level in the enhanced logical model; Using the fault impact propagation path, the fault mode of the target fault, and the description of the enhanced component as prompts, the LLM is guided to generate natural language descriptions of local effects. Higher-level influences on natural language description and the natural language description of the ultimate impact ; The Semantic matching is performed with the set of severity categories to obtain Severity category; Combining the above The severity category yields a structured output; Traverse each All and iterate through The failure mode effect analysis results were obtained.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

Images