An Automated Construction Method for FMECA Based on Multi-Signal Flow Graphs

Abstract

This invention discloses an automated FMECA construction method based on multi-signal flow graphs. The method constructs an ontology knowledge structure based on multi-signal flow graphs to automate the generation of FMECA analysis tables from complex signal flow graphs. The method comprises three main steps: basic fault mode screening based on propagation relationships, structured knowledge extraction from multi-signal flow graphs based on ontology constraints, and automatic FMECA generation based on structured knowledge. The method analyzes the signal interaction relationships of multi-signal flow graphs using fault propagation logic, filters valid fault modes, and generates standardized FMECA entries by combining structured ontology knowledge, thereby achieving efficient automated construction of FMECA files in complex systems. This method can dynamically update FMECA files and supports collaborative improvement throughout the entire lifecycle, significantly improving construction efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to an automated method for constructing FMECA files, and more particularly to an automated method for constructing FMECA files based on multi-signal flow graphs. Background Technology

[0002] FMECA (Failure Mode, Effects, and Criticality Analysis) is an analytical method widely used in the design and evaluation of complex systems. It aims to identify potential failure modes, analyze their impact on the system, and assess their hazard, and is of great significance for ensuring the reliability of complex products.

[0003] However, with the increasing scale of complex systems and the growing sophistication of functional structures in fields such as aviation and aerospace, the difficulty of constructing FMECA documents has also increased. Expert-experience-driven FMECA construction methods often have significant blind spots in analyzing failure modes involving multi-component interconnections, failing to fully cover reliability requirements. Furthermore, traditional FMECA document construction is often limited to the product design phase, failing to incorporate the vast amount of fault reasoning knowledge accumulated during operation. This results in a lack of comprehensive reliability knowledge, incrementally supplemented by experience, during subsequent product design and maintenance. Due to these shortcomings, traditional FMECA construction methods are insufficient to meet the high-quality reliability assurance requirements of today's complex equipment, necessitating a new FMECA construction method with automatic improvement and incremental capabilities.

[0004] Meanwhile, the trend of model management based on multi-signal flow graphs is becoming increasingly prominent throughout the modern reliability lifecycle. Multi-signal flow... Figure 1 On the one hand, it allows for incremental modifications, thus effectively improving as product fault diagnosis knowledge accumulates; on the other hand, it can integrate fragmented fault propagation knowledge through the model's graph structure, thereby effectively inferring richer potential fault propagation knowledge. Automated construction of FMECA based on multi-signal flow graphs can achieve collaborative improvement and incremental functionality of FMECA throughout the product lifecycle at the lowest cost within a modern reliability assurance framework, which has significant engineering implications.

[0005] Therefore, the FMECA construction method based on multi-signal flow graphs needs further optimization to achieve automated construction of FMECA analysis tables from multi-signal flow graphs. On one hand, the method needs to have the ability to analyze complex signal flow graphs to ensure accurate identification of the interaction logic between signals and fault propagation paths. On the other hand, it needs to lower the barrier to entry through framework tools, enabling developers and users to quickly generate FMECA analysis tables and intuitively display the analysis results. This method can not only improve the efficiency of FMECA analysis of complex systems but also provide important technical support for intelligent fault analysis.

[0006] In summary, given the increasing inadequacy of existing FMECA methods for meeting the demands of complex multi-signal systems, this invention proposes a generalized FMECA generation framework based on multi-signal flow graphs, providing an automated method for constructing FMECA tables from multi-signal flow graphs. This method not only satisfies the multi-signal analysis requirements of complex systems but also automates and generalizes the analysis process, significantly improving the efficiency and accuracy of FMECA construction. Summary of the Invention

[0007] This invention addresses the problems of high construction difficulty, large knowledge blind spots, and difficulty in dynamic evolution of existing FMECA construction methods in complex systems. It provides an automatic FMECA construction method based on multi-signal flow graphs. This method fully utilizes multi-signal flow graphs that evolve synchronously with product lifecycle knowledge. Through ontology knowledge extraction, filtering, and transformation, it achieves automatic FMECA construction based on product reliability knowledge, thereby meeting the requirements of modern complex system reliability for the automatic improvement and evolution of FMECA throughout the product lifecycle.

[0008] The automated FMECA construction method based on multi-signal flow graphs comprises three main steps: basic fault mode screening based on propagation relationships, structured knowledge extraction from multi-signal flow graphs based on ontology constraints, and automatic FMECA generation based on structured knowledge. The basic fault mode screening step, based on propagation relationships, intelligently filters invalid or redundant fault modes using fault propagation logic and relationships, extracting a simplified set of basic fault modes. The structured knowledge extraction step, based on ontology constraints, analyzes the signal propagation paths and interaction logic in the multi-signal flow graphs, constructing three ontology knowledge structures to provide an efficient and consistent knowledge foundation for subsequent FMECA generation. The automatic FMECA generation step, based on structured knowledge, automatically generates FMECA entries using standardized ontology knowledge structures and the filtered set of basic fault modes, achieving efficient automation of the entire process from multi-signal flow graph to FMECA analysis.

[0009] The features of this invention are:

[0010] (1) By using knowledge entities and associated ontology constraints, a correspondence method between multi-signal flow graph structure information and FMECA entries was constructed, thereby overcoming the blind spots in FMECA files caused by manual fault analysis based on the structure information of multi-signal flow graphs.

[0011] (2) By performing ontology analysis of the fault propagation structure within the multi-signal flow graph, the fault entities in the graph structure are screened for necessity, thus fulfilling the requirement for simplification of FMECA file entries in industrial scenarios.

[0012] (3) By using an automated structured knowledge filling method, the function of automatically building FMECA files based on multi-signal flow graphs was realized, thereby leveraging the dynamic evolution characteristics of multi-signal flow graphs to realize the dynamic evolution of FMECA files throughout the product lifecycle. Attached Figure Description

[0013] Figure 1 This is a flowchart of an automated FMECA construction method based on multi-signal flow graphs.

[0014] Figure 2 The flowchart shows a structured knowledge extraction method for multi-signal flow graphs based on ontology constraints.

[0015] Figure 3 This is a flowchart of a basic failure mode screening method based on propagation relationships.

[0016] Figure 4 This is a flowchart of the FMECA automatic generation method based on structured knowledge. Detailed Implementation

[0017] The following description, in conjunction with the accompanying drawings, details an automated FMECA construction method based on multi-signal flow graphs provided by this invention.

[0018] This invention provides an automated FMECA construction method based on multi-signal flow graphs. Overall, this method achieves the transformation process from multi-signal flow graphs to FMECAs through three main stages: structured knowledge extraction from multi-signal flow graphs based on ontology constraints, basic fault mode filtering based on propagation relationships, and automatic FMECA generation based on structured knowledge. The method forms three ontology knowledge structures based on the propagation relationships in the multi-signal flow graphs: fault mode-fault mode, fault mode-signal, and fault mode-component. Irrelevant fault modes are filtered out during the filtering stage, and finally, complete FMECA entries are generated based on the filtered set of fault modes. The overall flowchart of the automated FMECA construction method based on multi-signal flow graphs is shown below. Figure 1 As shown.

[0019] 1. Structured Knowledge Extraction of Multi-Signal Flow Graphs Based on Ontology Constraints

[0020] The aforementioned ontology-constrained multi-signal flow graph structured knowledge extraction step extracts structured knowledge from the multi-signal flow graph. Structured knowledge extraction is the process of parsing a complex multi-signal flow graph into an easily operable ontology knowledge structure, mainly including multi-signal flow... Figure 3 The method involves five steps: preliminary extraction of multi-signal flow graph knowledge based on core ontology knowledge structure; generation of fault mode-fault mode ontology knowledge structure; generation of fault mode-signal ontology knowledge structure; generation of fault mode-component ontology knowledge structure; and comprehensive extraction of structured knowledge. The flowchart of the ontology-constrained multi-signal flow graph structured knowledge extraction method is shown below. Figure 2 As shown.

[0021] 1.1. Preliminary Extraction of Multi-Signal Flow Graph Knowledge

[0022] The preliminary extraction step of the multi-signal flow graph knowledge involves initial entity extraction from the multi-signal flow graph using the ontology structure of fault propagation. The ontology structure of fault propagation mainly consists of entities and entity associations, corresponding to nodes and edges in the multi-signal flow graph, respectively. The entities primarily include two types: fault modes and signals, corresponding to the concepts of "fault" and "measurement point" in the multi-signal flow graph concept, respectively. During extraction, entities retain only the text and type information of the multi-signal flow graph nodes, serving as the entity's name and type. The entity relationships mainly include two types of fault propagation relationships: those between fault modes and those between fault modes and signals, consistent with the edge relationships defined between corresponding entities in the multi-signal flow graph. During extraction, entity associations retain only the source entity text and target entity text information of the multi-signal flow graph nodes.

[0023] 1.2. Generation of Fault Mode-Fault Mode Ontology Knowledge Structure

[0024] The generation steps of the fault mode-fault mode ontology knowledge structure involve parsing the signal propagation path and recording the causal relationships between each fault mode. The method recursively checks the connection paths in the edge list, determines whether a node is a fault mode based on its type, traverses all propagation paths in the signal flow graph, backtracks from the signal output to the input, and marks the corresponding upstream and downstream fault modes one by one. Using a graph traversal algorithm, each pair of fault modes with a direct propagation relationship is recorded, generating an ontology knowledge structure in the form of upstream fault mode: [downstream fault mode 1, downstream fault mode 2...]. 1.3. Generation of Fault Mode-Signal Ontology Knowledge Structure

[0025] The generation steps of the fault mode-signal ontology knowledge structure involve extracting the measurement point numbers related to the current fault mode by calling the measurement point number extraction function in the auxiliary function during signal flow graph parsing, obtaining the text field representing the signal information of the measurement point, and forming the association between the fault mode and the signal. The signal range affected by each fault mode is analyzed one by one, and each signal is marked as being controlled by which fault modes. The generated ontology knowledge structure is a fault mode: [signal 1, signal 2...], providing a basis for subsequent determination of whether a signal is related to a specific fault mode.

[0026] 1.4. Generation of Failure Mode-Component Ontology Knowledge Structure

[0027] The generation steps of the fault mode-component ontology knowledge structure further identify the component to which each fault mode belongs by extracting module information and propagation relationships between modules from the signal flow graph. Using the mapping between nodes and system names, this step binds each fault mode to its associated system (component). If a fault mode node is not mapped to a system, it means it is not associated with a component. The generated ontology knowledge structure is in the form of fault mode:component name, which facilitates the exclusion of fault modes belonging to non-target components during the filtering phase.

[0028] 1.5. Comprehensive Extraction of Structured Knowledge

[0029] Through the above steps, the multi-signal flow graph structured knowledge extraction step based on ontology constraints transforms the multi-signal flow graph into three ontology knowledge structures through recursion and filtering rules. Specifically, fault mode-fault mode connections are directly extracted using an edge list; fault mode-signal connections are made using measurement point information; and fault mode-component connections are made based on system mapping.

[0030] 2. Basic Failure Mode Screening Based on Propagation Relationships

[0031] The basic fault mode screening steps based on propagation relationships simplify the fault mode set through screening rules based on propagation relationships. These steps mainly include five steps: initialization of screening conditions, screening based on the fault mode-component table, screening based on the fault mode-signal table, screening based on the fault mode-fault mode table, and comprehensive screening results. The flowchart of the basic fault mode screening method based on propagation relationships is shown below. Figure 3 As shown.

[0032] 2.1. Initialization of Filtering Criteria

[0033] The initialization steps of the filtering conditions read the structured knowledge, namely, fault mode-fault mode (a table of propagation relationships between fault modes), fault mode-signal (a table of association between fault modes and signals), and fault mode-component (a table of association between fault modes and components).

[0034] 2.2. Filtering based on the failure mode-component list

[0035] In the filtering step based on the fault mode-component table, if a fault mode does not belong to the target component, it is deleted from the candidate set. In practice, this requires traversing the candidate fault mode set and searching for the corresponding component information for each fault mode. If the component for a fault mode is not within the target component range, the fault mode is deleted.

[0036] 2.3. Filtering based on fault mode - signal table

[0037] In the filtering step based on the fault mode-signal table, if a fault mode does not affect any signal, it is considered a redundant fault mode and should be removed from the candidate set. Specifically, for each candidate fault mode, its corresponding signal influence list (fault mode-signal) is searched. If the signal list is empty, it means that the fault mode does not affect any signal, and the fault mode is deleted.

[0038] 2.4. Filtering based on the Fault Mode-Fault Mode Table

[0039] In the filtering step based on the fault mode-fault mode table, if a fault mode has no downstream propagation relationship (i.e., does not affect other fault modes) and does not directly or indirectly affect the propagation path of the target fault mode, it is considered a redundant fault mode. If a fault mode has multiple downstream fault modes but is not bound to a signal, and the fault mode does not affect the target fault mode, it can be deleted. In specific implementation, the candidate fault mode set needs to be traversed, and the downstream fault mode information in the fault mode-fault mode table needs to be checked. If a fault mode has no downstream fault modes and is not on any influencing path, the fault mode is removed; if a fault mode has multiple downstream fault modes that are not bound to a signal, and its downstream modes have no effect on the target mode, the mode is deleted.

[0040] 2.5. Overall Screening Results

[0041] In the comprehensive screening step, the set of screened fault modes will be returned as the "basic fault mode set" after screening. These fault modes are all directly related to the target component, signal, and propagation path. This set will serve as the basis for the next step of FMECA generation.

[0042] 3. Automatic generation of FMECA based on structured knowledge

[0043] The structured knowledge-based automatic FMECA generation process, based on a filtered set of fault modes, combines three ontology knowledge structures to generate standardized FMECA entries. It mainly includes three steps: preprocessing of the ontology knowledge structure, automatic filling of FMECA entries, and merging and generating FMECA files. The flowchart of the structured knowledge-based automatic FMECA generation method is shown below. Figure 4 As shown.

[0044] 3.1. Preprocessing of Ontology Knowledge Structure

[0045] The preprocessing steps of the ontology knowledge structure first traverse each system configuration, creating a mapping table from node number to system name for each node, so that the system name to which it belongs can be quickly located by node number when generating the FMECA table; then, based on the node type, valid nodes and test point nodes are filtered out, and valid node diagrams and valid test point diagrams are generated. The valid node diagram contains all valid node numbers (fault nodes and test nodes), and the valid test point diagram contains only test point node numbers (test nodes and switch nodes).

[0046] 3.2. Automatic filling of FMECA entries

[0047] The automatic filling steps for the FMECA entries first traverse each node. If the node type is a fault node, the function to obtain the measurement point number is used to recursively extract the numbers of all measurement point nodes from the edges associated with the current node, and obtain their text information. The detection method is represented by the text of the measurement point node. Then, the upstream connecting edges of the current node are traversed, and the text of the upstream nodes (also fault nodes) is extracted as the cause of the fault. Finally, the downstream nodes of the current fault node are traversed, and multi-level influences (local influences, higher-level influences, and final influences) are extracted recursively. The hierarchical relationship is recursively processed, recording a maximum of three levels of influence. If a certain level has no downstream node, it is filled with an empty string.

[0048] 3.3. Generating FMECA files by merging

[0049] The FMECA file merging and generation step saves all analysis results and calls the merge function to generate a formatted Excel spreadsheet. During output, cells with duplicate content are merged and centered. The FMECA spreadsheet follows the filling specifications shown in Table 1 to fill in the structured entry information formed in the previous step. The formatted FMECA Excel spreadsheet then combines the filled entry information according to the specifications in Table 2 to generate the Excel spreadsheet.

[0050] Table 1. FMECA Table Column Definitions and Contents

[0051]

[0052] Table 2 FMECA Examples

[0053]

[0054] Once the export is complete, the automated construction method for FMECA based on multi-signal flow graphs is finished.

Claims

1. An automated FMECA construction method based on multi-signal flow graphs, characterized in that: the automated FMECA construction method achieves efficient automatic generation and dynamic evolution of FMECA entries through ontology knowledge extraction and automated fault mode screening of multi-signal flow graphs; the automated FMECA construction method includes a multi-signal flow graph structured knowledge extraction step based on ontology constraints, a basic fault mode screening step based on propagation relationships, and an FMECA automatic generation step based on structured knowledge; the multi-signal flow graph structured knowledge extraction step based on ontology constraints is as follows: 1.

1. Preliminary Extraction of Multi-Signal Flow Graph Knowledge The preliminary extraction step of the multi-signal flow graph knowledge involves initial entity extraction from the multi-signal flow graph using the ontology structure of fault propagation. This ontology structure is divided into entities and entity associations, corresponding to nodes and edges in the multi-signal flow graph, respectively. Entities include two types: fault modes and signals, corresponding to the concepts of "fault" and "measurement point" in the multi-signal flow graph. During extraction, entities retain only the text and type information of the multi-signal flow graph nodes, serving as the entity's name and type. Entity associations include two types of fault propagation relationships: those between fault modes and those between fault modes and signals, consistent with the edge relationship definition between corresponding entities in the multi-signal flow graph. During extraction, entity associations retain only the source entity text and target entity text information of the multi-signal flow graph nodes. 1.

2. Generation of Fault Mode-Fault Mode Ontology Knowledge Structure The generation steps of the fault mode-fault mode ontology knowledge structure are as follows: First, the signal propagation path is parsed to record the causal relationship between each fault mode; second, the connection path in the edge list is recursively checked to determine whether it is a fault mode by the node type; third, all propagation paths in the multi-signal flow graph are traversed, tracing back from the output end to the input end of the signal, and the corresponding upstream fault mode and downstream fault mode are marked one by one; fourth, the graph traversal algorithm is used to record each pair of fault modes with a direct propagation relationship, generating an ontology knowledge structure in the form of upstream fault mode: [downstream fault mode 1, downstream fault mode 2...]. 1.

3. Generation of Fault Mode-Signal Ontology Knowledge Structure The steps for generating the fault mode-signal ontology knowledge structure are as follows: when parsing a multi-signal flow graph, the measurement point number extraction function in the auxiliary function is called to extract the measurement point number related to the current fault mode, obtain the text field representing the signal information of the measurement point, and form the association between the fault mode and the signal; the signal range affected by each fault mode is analyzed one by one, and each signal is marked as being controlled by which fault modes; the generated fault mode-signal ontology knowledge structure is fault mode: [signal 1, signal 2...]; 1.

4. Generation of Failure Mode-Component Ontology Knowledge Structure The generation steps of the fault mode-component ontology knowledge structure are as follows: by extracting module information and propagation relationships between modules from the multi-signal flow graph, the component to which each fault mode belongs is identified; and by using the mapping between nodes and component names, each fault mode is bound to its component. If a fault mode node is not mapped to a component, it means that it is not associated with a component; the generated fault mode-component ontology knowledge structure is in the form of fault mode:component name; 1.

5. Comprehensive Extraction of Structured Knowledge Through steps 1.1-1.4 above, the multi-signal flow graph structured knowledge extraction step based on ontology constraints transforms the multi-signal flow graph into three ontology knowledge structures through recursion and filtering rules. The basic fault mode screening steps based on propagation relationships simplify the fault mode set through screening rules based on propagation relationships. The steps include five steps: initialization of screening conditions, screening based on fault mode-component table, screening based on fault mode-signal table, screening based on fault mode-fault mode table, and comprehensive screening results. The structured knowledge-based automatic FMECA generation steps are based on the filtered fault mode set and combine three ontology knowledge structures to generate standardized FMECA entries. The steps include three steps: preprocessing of ontology knowledge structures, automatic filling of FMECA entries, and merging and generating FMECA files.

2. The FMECA automated construction method according to claim 1, characterized in that: The multi-signal flow graph structured knowledge extraction step in the FMECA automated construction method constructs a correspondence method between multi-signal flow graph structure information and FMECA entries through knowledge entities and associated ontology constraints, thereby overcoming the blind spots in FMECA files caused by manual fault analysis based on the structure information of multi-signal flow graphs.

3. The FMECA automated construction method according to claim 1, characterized in that: The basic fault mode screening step in the FMECA automated construction method uses ontology analysis of the fault propagation structure within the multi-signal flow graph to perform necessary screening of fault entities in the graph structure, thus fulfilling the requirement for simplification of FMECA file entries in industrial scenarios.

4. The FMECA automated construction method according to claim 1, characterized in that: The FMECA automatic generation step in the described FMECA automated construction method realizes the automatic construction function of FMECA files based on multi-signal flow graphs through an automated structured knowledge filling method. This enables the dynamic evolution of FMECA files throughout the entire product lifecycle by leveraging the dynamic evolution characteristics of multi-signal flow graphs.

Images