Multi-signal flow graph incremental construction method based on case-level FTA

Abstract

The invention discloses a multi-signal flow graph incremental construction method based on case-level FTA (Fourier transform algorithm). According to the multi-signal flow graph incremental construction method, on the basis of case-level FTA, a node type-template picture binding relation is defined to achieve FTA analysis protocol self-definition based on ontology constraint, OR gate nodes are deleted in a depth-first mode, and attributes of edges are correspondingly adjusted to achieve automatic updating of a multi-signal flow graph. And performing text similarity judgment on node names, merging similar nodes and correspondingly adjusting attributes of edges to realize dynamic incremental updating of the multi-signal flow graph. The multi-signal flow graph incremental construction method comprises the following three steps: firstly, realizing FTA analysis protocol self-definition based on ontology constraint by defining a node type-template picture binding relationship; then deleting'or gate 'nodes through depth-first and correspondingly adjusting attributes of edges to realize automatic updating of the multi-signal flow graph; and finally, performing text similarity judgment on node names, merging similar nodes and correspondingly adjusting attributes of edges to realize dynamic incremental updating of the multi-signal flow graph.

Description

Technical Field

[0001] The present invention relates to a multi-signal flow graph incremental construction method, in particular to a multi-signal flow graph incremental construction method based on case-level FTA. Background Art

[0002] For complex systems, there are many components that are coupled and interact with each other. When a fault occurs, the situation is complicated and difficult to judge, which makes it difficult to diagnose and maintain the fault. Therefore, it is necessary to model the fault state to study the diagnosis and maintenance strategies under fault conditions. Traditional physical modeling methods have the disadvantages of high modeling difficulty and long modeling time when facing complex systems. In contrast, knowledge-based modeling is based on the interaction and interaction relationship between the various components of the system. It can well model complex systems and express fault states, and is a good modeling method. There are many knowledge-based models, among which the multi-signal flow graph model has received widespread attention and application due to its simplicity, intuitiveness, efficiency and ease of use.

[0003] When constructing a multi-signal flow graph using traditional methods, first, it is necessary to obtain and clarify the principles and structural block diagram of the system; second, it is necessary to divide the entire system into subsystems, submodules, etc. as required; then define the signals in the system and associate the system components with the signals; finally, add tests to the system and associate measurement points according to the test plan and requirements to obtain the multi-signal flow graph of the system. However, this method only considers the construction of the multi-signal flow graph, and does not consider the dynamic update of the multi-signal flow graph based on the increase of experience. Especially in the face of large-scale and multi-changing systems, the update based on the fact of local fault zeroing is cumbersome. Therefore, how to achieve automatic update of the multi-signal flow graph based on the increase of experience is still a technical problem that needs to be solved urgently.

[0004] Fault Tree Analysis (FTA) is a method that uses logical reasoning to describe the relationship between dangerous factors and fault events in a system. FTA files are mainly used to formally summarize the cause analysis of specific fault cases. In terms of data structure, they are manifested as a tree structure that reflects the fault propagation relationship and can be directly converted into a multi-signal flow graph. Therefore, based on the case-level FTA, the automatic update of the multi-signal flow graph can be achieved. At the same time, in view of the large-scale and multi-variable characteristics of the system, each specific fault case is subjected to FTA analysis to obtain a case-level FTA file. When the multi-case FTA is converted into a multi-signal flow graph, there is a possibility of merging. Therefore, based on the case-level FTA, the dynamic incremental update of the multi-signal flow graph can be achieved. Therefore, the present invention proposes a multi-signal flow graph incremental construction method based on case-level FTA to realize the automatic dynamic incremental update of the multi-signal flow graph, so as to better dynamically model the fault state of large-scale and multi-variable systems, and then perform fault diagnosis and maintenance. Summary of the invention

[0005] The present invention provides a method for incrementally constructing a multi-signal flow graph to solve the problem that it is difficult to automatically and dynamically update a multi-signal flow graph of a large-scale, multi-change system. The method can realize automatic and dynamic incremental updates of multi-signal flow graphs, thereby better dynamically modeling the fault state of a large-scale, multi-change system, and then performing fault diagnosis and maintenance.

[0006] The multi-signal flow graph incremental construction method mainly realizes the automatic dynamic incremental update of the multi-signal flow graph based on the case-level FTA. Specifically, the multi-signal flow graph incremental construction method includes three core steps: node definition, knowledge structure transformation, and multi-case FTA autonomous merging.

[0007] The node definition, by defining the binding relationship between the node type and the template image in the FTA, opens the template image bound to each node type to the user, solves the problem of inconsistent drawing numbers of fault tree data from multiple sources in industrial scenarios, and realizes the customized function of the FTA parsing protocol based on ontology constraints. The knowledge structure transformation, by depth-first deleting the "OR gate" node in the FTA, and adjusting the node connection relationship and modifying the edge attributes accordingly, automatically eliminates the difference in local definition between FTA and the multi-signal flow graph, and realizes the automatic update function of the multi-signal flow graph based on FTA. The multi-case FTA autonomous merging, by performing text similarity judgment based on Jaccard similarity on the node names of different case FTAs, merges similar nodes, and adjusts the edge attributes accordingly according to the corresponding relationship before and after the name merger, eliminates the problem of style differences among multiple people when writing FTA, thereby realizing the dynamic incremental update function of the multi-signal flow graph based on multi-case FTA.

[0008] The characteristics of the present invention are:

[0009] (1) The present invention defines the binding relationship between the node type and the template image in FTA, opens the template image bound to each node type to the user, solves the problem of inconsistent drawing numbers of fault tree data from multiple sources in industrial scenarios, and realizes the customization function of FTA parsing protocol based on ontology constraints;

[0010] (2) The present invention deletes the "OR gate" nodes in FTA in a depth-first manner, adjusts the node connection relationship and modifies the edge attributes accordingly, thereby automatically eliminating the differences in local definitions between FTA and the multi-signal flow graph, and realizing the automatic update function of the multi-signal flow graph based on FTA;

[0011] (3) The present invention performs text similarity judgment based on Jaccard similarity on the node names of different case FTAs, merges similar nodes, and adjusts the attributes of the edges accordingly according to the corresponding relationship before and after the name merger, thereby eliminating the problem of style differences among multiple people when writing FTAs, thereby realizing the dynamic incremental update function of the multi-signal flow graph based on multi-case FTA. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 Method steps for incremental construction of multi-signal flow graphs based on case-level FTA

[0013] Figure 2 Define step principles for nodes

[0014] Figure 3 Principles of knowledge structure transformation steps

[0015] Figure 4 Principles of autonomous merger steps for multi-case FTA DETAILED DESCRIPTION

[0016] The multi-signal flow graph incremental construction method provided by the present invention is described in detail below with reference to the accompanying drawings.

[0017] The present invention provides a method for incrementally constructing a multi-signal flow graph based on case-level FTA. Figure 1 As shown, the incremental construction method of multi-signal flow graph based on case-level FTA includes three core steps: node definition, knowledge structure transformation, and autonomous merging of multi-case FTA.

[0018] 1. Node definition steps

[0019] The node definition step, by defining the binding relationship between the node type and the template image in FTA, opens the template image bound to each node type to the user, solves the problem of inconsistent drawing numbers of multi-source fault tree data in industrial scenarios, and realizes the FTA parsing protocol customization function based on ontology constraints, as shown in the attached Figure 2 As shown in Table 1, the common node types and their meanings in FTA are shown in Table 1. By defining the binding relationship between the node types and template images in FTA, the template images bound to each node type are opened to users. Users can customize the FTA of the fault case based on the FTA graphical basic components formed by binding template images of each node type. By crawling the node graph information in FTA, the attribute information of all nodes and edges in FTA is obtained, as shown in Table 2.

[0020] Table 1 Common node types and their meanings of FTA

[0021] Node Type meaning Basic Events It is not necessary to find out the cause of the event. The event with a clear fault mechanism and known probability of occurrence is at the bottom of the fault tree. Result Event Includes top events and intermediate events caused by other events or combinations of events OR Gate The OR gate means that only one of the input events needs to occur for the output event to occur. AND Gate The AND gate indicates that the output event occurs when all input events occur at the same time.

[0022] Table 2 FTA composition and property information

[0023]

[0024] 2. Knowledge structure transformation steps

[0025] The knowledge structure transformation step described above automatically eliminates the differences in local definitions between FTA and the multi-signal flow graph by depth-first deleting the "OR gate" nodes in FTA, and correspondingly adjusting the node connection relationship and modifying the edge attributes, thereby realizing the automatic update function of the multi-signal flow graph based on FTA, as shown in the attached figure. Figure 3 As shown. The basic events (root nodes) of FTA generally have a certain degree of observability, so they can be used as measurement points of multi-signal flow graphs; while the result events of FTA generally correspond to the fault events to be inferred, so they are used as fault events of multi-signal flow graphs; the "AND gate" nodes of FTA can be directly converted into "AND gate" nodes of multi-signal flow graphs. The difference in local definition between FTA and multi-signal flow graphs is reflected in: there is an "OR gate" node in FTA, which means that the output event occurs only if at least one of the input events occurs; however, there is no similar "OR gate" node in multi-signal flow graphs, and multiple inputs are directly connected to one output to indicate that the output occurs only if at least one of the inputs occurs; therefore, when converting FTA into a multi-signal flow graph, it is necessary to process the "OR gate" node first to eliminate the difference in local definition between FTA and multi-signal flow graphs.

[0026] The main steps of the knowledge structure transformation step are as follows:

[0027] 1) Filter “OR gate” nodes;

[0028] 2) Filter non-OR gate nodes;

[0029] 3) Establishing a connection relationship table of the “OR gate” node, including the “OR gate” node and the target node it points to;

[0030] 4) Traverse the “OR gate nodes” in sequence, expand the edges of the “OR gate” nodes, and directly connect the input events of the “OR gate” nodes with the output events;

[0031] 5) Determine whether there are multiple "OR" nodes directly connected. If so, recursively expand according to the depth-first principle. The depth-first principle means that if the target node connected to the "OR" node is also an "OR" node,

[0032] If the target node connected to the OR node is not an OR node, then the edge of the current OR node is expanded directly. Otherwise, go to step (6);

[0033] 6) Transform the basic events, result events and “AND gate” nodes of FTA;

[0034] 7) Obtain the processed multi-signal flow graph nodes and edges.

[0035] 3. Steps for autonomous merger of multiple FTA cases

[0036] The multi-case FTA autonomous merging step performs text similarity judgment based on Jaccard similarity on the node names of different case FTAs, merges similar nodes, and adjusts the attributes of the edges accordingly according to the corresponding relationship before and after the name merger, thereby eliminating the problem of style differences among multiple people when writing FTAs, thereby realizing the dynamic incremental update function of the multi-signal flow graph based on multi-case FTA. Jaccard similarity refers to the number of words in the intersection of two texts divided by the number of words in the union. Assume that the set composed of each character of text A is A, the set composed of each character of text B is B, the intersection of characters in text A and text B is A∩B, the union of characters in text A and text B is A∪B, and the number of words contained in the character set is represented by |.|. The Jaccard similarity calculation formula between text A and text B is as follows:

[0037]

[0038] The main steps of the multi-case FTA autonomous merger process are as follows:

[0039] 1) Calculate the Jaccard similarity of the node names of different case FTAs;

[0040] 2) Determine whether the Jaccard similarity is greater than a threshold. If so, proceed to step (3). Otherwise, proceed to step (5);

[0041] 3) Merge similar nodes whose Jaccard similarity is greater than the threshold;

[0042] 4) According to the corresponding relationship before and after the name merger, the attributes of the edge are adjusted accordingly, including the starting point ID, starting point name, end point ID, and end point name of the edge;

[0043] 5) Merge the remaining non-similar nodes and related edges of FTA;

[0044] 6) Obtain FTA based on autonomous merger of multiple cases.

Claims

1. A method for incrementally constructing a multi-signal flow graph based on case-level FTA, characterized by: The incremental construction method of the multi-signal flow graph includes a node definition step of realizing a customized FTA parsing protocol based on ontology constraints by defining a node type-template image binding relationship, a knowledge structure transformation step of realizing automatic updating of the multi-signal flow graph by depth-first deleting "OR gate" nodes and adjusting the attributes of the edges accordingly, and a multi-case FTA autonomous merging step of realizing dynamic incremental updating of the multi-signal flow graph by merging similar nodes based on text similarity judgment of node names and adjusting the attributes of the edges accordingly.

2. The method for incrementally constructing a multi-signal flow graph according to claim 1, characterized in that: The node definition step in the incremental construction method of the multi-signal flow graph defines the binding relationship between the node type and the template image in the FTA, and opens the template image bound to each node type to the user, thereby solving the problem of inconsistent drawing numbers of fault tree data from multiple sources in industrial scenarios and realizing the customization function of the FTA parsing protocol based on ontology constraints.

3. The method for incrementally constructing a multi-signal flow graph according to claim 1, characterized in that: The knowledge structure transformation step in the incremental construction method of the multi-signal flow graph deletes the "OR gate" nodes in the FTA in a depth-first manner, and adjusts the node connection relationship and modifies the attributes of the edges accordingly, thereby automatically eliminating the differences in local definitions between the FTA and the multi-signal flow graph, and realizing the automatic update function of the multi-signal flow graph based on the FTA.

4. The method for incrementally constructing a multi-signal flow graph according to claim 1, characterized in that: The multi-case FTA autonomous merging step in the multi-signal flow graph incremental construction method performs text similarity judgment on the node names of different case FTAs ​​based on Jaccard similarity, merges similar nodes, and adjusts the attributes of the edges accordingly according to the corresponding relationship before and after the name merger, thereby eliminating the problem of style differences among multiple people when writing FTAs, thereby realizing the dynamic incremental update function of the multi-signal flow graph based on multi-case FTA.

Images