Equipment damage identification knowledge graph construction, identification methods and systems
By constructing a knowledge graph for equipment damage identification and combining it with relevant equipment standards and test reports, we have achieved rapid and accurate identification and analysis of equipment damage patterns and influencing factors. This solves the problem of insufficient semantic information acquisition capabilities in traditional systems and provides an efficient knowledge management and damage identification method.
Patent Information
- Application Number
- CN202311726007.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-15
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-12-15
AI Technical Summary
Traditional equipment damage identification systems lack the ability to acquire semantic information about damage under complex factors, making it difficult to quickly and effectively identify equipment damage patterns.
A knowledge graph for equipment damage identification is constructed by extracting domain knowledge and scenario knowledge from equipment-related standard systems and inspection and testing reports, dividing them into multiple knowledge ontologies and event ontologies, establishing a semantic network, and forming a knowledge graph for equipment damage identification to achieve bidirectional collaborative analysis of damage patterns and influencing factors.
It enables rapid and accurate identification and analysis of equipment damage, provides an efficient knowledge management system, can efficiently and accurately locate damage pathways, and improves the efficiency and accuracy of equipment damage identification.
Smart Images

Figure CN117972103B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of data analysis, equipment damage identification, and equipment damage identification knowledge management, and in particular to a knowledge graph construction, identification method, and system for equipment damage identification. Background Technology
[0002] Equipment refers to equipment used in industrial production for storage, transportation, reaction, separation and other processes that need to withstand certain pressures. It often contains media with characteristics such as high temperature, high pressure, flammability, explosiveness and high toxicity. Equipment damage identification has always been a research hotspot in the field of safety assurance technology.
[0003] In current traditional equipment damage identification systems, the accumulated information is in the form of structured and unstructured data such as file systems and traditional databases. When damage identification involves complex factors, this limits the ability to quickly obtain effective information at the semantic level of damage identification.
[0004] Knowledge graphs can map knowledge domains, use visualization techniques to describe knowledge resources and their carriers, and mine, analyze, construct, draw, and display knowledge and the relationships between them, thereby improving the ability to efficiently acquire semantic information.
[0005] Therefore, how to design a knowledge graph model for equipment damage identification to quickly process large amounts of semantic information has become an urgent technical problem to be solved. Summary of the Invention
[0006] To address the challenge of rapid damage pattern identification under the influence of complex factors in equipment, this invention proposes a knowledge graph construction method for equipment damage identification. This method leverages the abundant semantic information in inspection and testing reports regarding the complex influencing factors of equipment damage to achieve bidirectional collaborative analysis of damage patterns and influencing factors, thereby facilitating rapid identification and analysis of equipment damage.
[0007] The present invention proposes a method for constructing a knowledge graph for equipment damage identification, comprising the following steps:
[0008] S1. Extract domain knowledge from the selected equipment-related standard system and scenario knowledge from equipment inspection and testing archives;
[0009] S2. Divide the domain knowledge into knowledge ontology in multiple defined directions, establish semantic hierarchical relationships for each knowledge ontology, and form a semantic network as the standard system knowledge concept hierarchy; the standard system knowledge concept hierarchy represents the hierarchical relationship of the concept of damage identification in equipment-related standards;
[0010] The scenario knowledge is divided into multiple categories of event ontology. For different event ontology, semantic hierarchical relationships are established to form a semantic network as the hierarchical structure of event entity relationship system. The hierarchical structure of event entity relationship system represents the various event entities that caused the damage as determined by the inspection and testing report and their interrelationships, including the ranking of influencing factors and the relationships between each factor.
[0011] S3. Merge nodes with the same name in the knowledge concept level and the event entity relationship level of the standard system to form a knowledge graph for equipment damage identification.
[0012] Preferably, for pressure-bearing equipment, S2 divides the domain knowledge into four directions: damage mode, main influencing factors, affected materials, and easily corroded equipment. The knowledge ontology of each direction is extracted and referred to as the damage mode knowledge ontology, the main influencing factors knowledge ontology, the affected materials knowledge ontology, and the easily corroded equipment knowledge ontology, respectively.
[0013] Preferably, the construction of the knowledge concept hierarchy of the standard system in S2 includes the following steps:
[0014] SA21, Divide domain knowledge into knowledge ontology in multiple defined directions;
[0015] SA22. Establish semantic layers for each knowledge ontology, extract concept nodes from the semantic layers, establish relationships between the layers to form a semantic network that corresponds one-to-one with the knowledge ontology, and let the sum of the semantic networks corresponding to all knowledge ontologs be used as the standard system knowledge concept layers.
[0016] Preferably, in SA22, the relationship between nodes in the semantic hierarchy of the knowledge ontology is described as a relational model.
[0017] Preferably, in S2, the scenario knowledge is divided into four categories: device file, process parameters, media composition, and inspection and testing report, which are respectively referred to as device file event, process parameter event, media composition event, and inspection and testing report event.
[0018] Preferably, the event entity relationship system is represented as a tree structure. Entities are extracted from the tree structure as nodes, and the root node, child nodes, and leaf nodes are divided according to the semantic relationship between the nodes.
[0019] Preferably, for device archive events, the extracted nodes include: materials, manufacturing processes, and welding processes;
[0020] For media composition events, the extracted entities include: major organic components, acid value, salt content, and chloride ion content;
[0021] For process parameter events, the extracted entities include: flow rate, temperature, concentration, pressure, and material ratio;
[0022] For incidents involving inspection and testing reports, the extracted entities include: the test results, the tested equipment, the damaged area, the morphology of the damage, and the cause of the damage.
[0023] Preferably, in S2, the knowledge concept hierarchy map of the standard system is represented as SG =<SN,SR> SN represents a concept node in the knowledge ontology; SR represents node relationships, indicated by semantic relationship edges between two nodes; the event entity relationship hierarchy is represented by EG =<EN,ER> EN represents the entity node in the event; ER represents the node relationship, which is represented by the semantic relationship edge between two nodes;
[0024] In S3, the knowledge graph representation of equipment damage identification, KG, is:
[0025] KG = { <n> , <r>}
[0026] <n>∈(SN∪EN)
[0027] <r>∈(SR∪ER)
[0028] R={T,D,G}
[0029] G=(node_A,relation,node_B)
[0030] node_A∈ <n>
[0031] node_B∈ <n>
[0032] in, <n>For a set of nodes, <r>Let R be the set of edges. <r>For any edge in the equation, T represents the relation attribute, D represents the relation direction, G represents the triple, node_A and node_B represent nodes, and relation represents the semantic relation and its relation attribute.
[0033] This invention proposes an equipment damage identification method that employs the aforementioned equipment damage identification knowledge graph construction method. This method can quickly and accurately achieve equipment damage factor mining and context analysis. First, it obtains the equipment damage identification knowledge graph and the equipment inspection and testing report. Then, it extracts the event ontology from the inspection and testing report, finds the corresponding nodes in the equipment damage pattern knowledge graph, performs matching analysis based on the damage pattern to which the node belongs, and determines the damage pattern. Finally, it filters the node relationships to establish an association network, forming a context for mining potential influencing factors.
[0034] The present invention proposes a knowledge graph construction system for equipment damage identification, comprising a memory and a processor. The memory stores a computer program, and the processor is connected to the memory. The processor is used to execute the computer program to implement the equipment damage identification method of the knowledge graph construction method for equipment damage identification.
[0035] The advantages of this invention are:
[0036] (1) This invention proposes a method for constructing a knowledge graph for equipment damage identification. By constructing a knowledge ontology for equipment damage identification, based on concept nodes and event entity nodes, and using the hierarchical relationship of knowledge concepts and the relationship of event entities between nodes as node relationships, a knowledge graph is formed, providing a fast and efficient knowledge management system for rapid identification and analysis of equipment damage.
[0037] (2) In this invention, domain knowledge based on equipment-related standards is divided into knowledge ontology in multiple directions, and a standard system knowledge concept hierarchy is constructed in each direction; scenario knowledge based on practical experience is divided into multiple categories, and an event entity relationship hierarchy is constructed in each category. Both the standard system knowledge concept hierarchy and the event entity relationship hierarchy are represented in the form of semantic networks, with a clear structure, which is conducive to the extraction of concept nodes in the standard system knowledge concept hierarchy and entity nodes in the event entity relationship hierarchy, as well as the representation of the relationships between nodes.
[0038] (3) In this invention, after representing the knowledge ontology and scene knowledge in the form of nodes and node relationships, the merging of multiple networks is achieved by merging nodes, thereby forming a knowledge graph for equipment damage identification, realizing a clear and efficient knowledge graph construction method and accurate graph representation.
[0039] (4) The present invention also proposes a method for identifying equipment damage. Based on a knowledge graph, it can use node search and filtering to efficiently and accurately find the damage path and achieve efficient and accurate damage identification. Attached Figure Description
[0040] Figure 1 for;
[0041] Figure 2(a) shows the semantic hierarchy relationship map of the knowledge ontology of hydrochloric acid corrosion damage morphology.
[0042] Figure 2(b) shows the semantic hierarchy relationship map of the knowledge ontology of ammonium chloride corrosion damage morphology.
[0043] Figure 2(c) shows the semantic hierarchy relationship map of the knowledge ontology of sulfuric acid corrosion damage morphology.
[0044] Figure 3(a) shows the semantic hierarchy relationship map of the knowledge ontology of the main influencing factors of hydrochloric acid corrosion.
[0045] Figure 3(b) shows the semantic hierarchy relationship map of the knowledge ontology of the main influencing factors of ammonium chloride corrosion.
[0046] Figure 3(c) shows the semantic hierarchy relationship map of the knowledge ontology of the main influencing factors of sulfuric acid corrosion.
[0047] Figure 4(a) shows the semantic hierarchy relationship map of the knowledge ontology of materials affected by hydrochloric acid corrosion.
[0048] Figure 4(b) shows the semantic hierarchy relationship map of the knowledge ontology of materials affected by ammonium chloride corrosion.
[0049] Figure 4(c) shows the semantic hierarchy relationship graph of the knowledge ontology of materials affected by sulfuric acid corrosion.
[0050] Figure 5(a) shows the semantic hierarchy relationship graph of the knowledge ontology of hydrochloric acid-corroded equipment.
[0051] Figure 5(b) shows the semantic hierarchy relationship graph of the knowledge ontology of equipment that is easily corroded by ammonium chloride.
[0052] Figure 5(c) shows the semantic hierarchy relationship graph of the knowledge ontology of equipment that is easily corroded by sulfuric acid.
[0053] Figure 6 Transforming device archive events into a semantic network graph;
[0054] Figure 7 Transforming process parameter events into semantic network graphs;
[0055] Figure 8 The event is transformed into a semantic network graph.
[0056] Figure 9 To examine the transformation of test report events into a semantic network graph;
[0057] Figure 10 A knowledge graph for identifying damage to pressure equipment. Detailed Implementation
[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0059] This implementation proposes a knowledge graph for equipment damage identification, which takes into account both the knowledge concepts of the standard system based on industry standard knowledge and the practical experience based on testing and inspection.
[0060] Reference Figure 1 The method for constructing the device damage recognition knowledge graph includes the following steps.
[0061] S1. Extract domain knowledge from the selected equipment-related standard system and scenario knowledge from equipment inspection and testing archives.
[0062] For pressure equipment, domain knowledge includes damage modes, key influencing factors, affected materials, and easily corroded equipment.
[0063] Scenario knowledge includes equipment files, process parameters, media composition, and inspection and testing reports; equipment files include materials, manufacturing processes, and welding processes; media composition includes main organic components, acid value, salt content, and chloride ion content; process parameters include flow rate, temperature, concentration, pressure, and material ratio; inspection and testing reports include: test conclusions, the tested equipment, the damaged location, the damage morphology, and the cause of the damage.
[0064] S2. Divide the domain knowledge into knowledge ontology in multiple defined directions, establish semantic hierarchical relationships for each knowledge ontology, and form a semantic network as the standard system knowledge concept hierarchy; the standard system knowledge concept hierarchy represents the hierarchical relationship of the concept of damage identification in equipment-related standards;
[0065] The scenario knowledge is divided into multiple categories of event ontology. For different event ontology, semantic hierarchical relationships are established to form a semantic network as the hierarchical structure of event entity relationships. The hierarchical structure of event entity relationships represents the various event entities that caused the damage as determined by the inspection and testing report and their interrelationships, including the ranking of influencing factors and the relationships between each factor.
[0066] Specifically, in this implementation, the knowledge concept level of the standard system adopts a relational network, also known as a relational model; the event entity relational system level adopts a tree structure network, or simply a tree structure.
[0067] The knowledge concept hierarchy map of the standard system is represented by SG =<SN,SR> SN represents a concept node in the knowledge ontology; SR represents node relationships, which are represented by semantic relationship edges between two nodes.
[0068] When constructing the knowledge concept hierarchy of the standard system, the domain knowledge is first divided into knowledge ontology in multiple directions, such as damage morphology knowledge ontology, main influencing factor knowledge ontology, affected material knowledge ontology, and easily corroded equipment knowledge ontology. Then, a semantic hierarchy is established for each knowledge ontology, the concept nodes in the semantic hierarchy are extracted, and the hierarchy is connected by relationships to form a semantic network that corresponds one-to-one with the knowledge ontology. The sum of the semantic networks corresponding to all knowledge ontology is taken as the knowledge concept hierarchy of the standard system.
[0069] In this embodiment, when constructing the semantic network of the knowledge concept level of the standard system, the relationship between each node in the semantic level of the knowledge ontology is described as a relational model, that is, the knowledge concept level of the standard system adopts a relational network model.
[0070] The hierarchical diagram of the event entity relationship system is represented as EG =<EN,ER> EN represents the entity node in the event; ER represents the node relationship, which is represented by the semantic relationship edge between two nodes.
[0071] When constructing the event entity relationship hierarchy, the scenario knowledge is first divided into multiple defined events, such as equipment file events, process parameter events, media composition events, and inspection and testing report events. Then, a tree structure network is established for each event as the event entity relationship hierarchy. Entities are extracted as nodes from the tree structure, and root nodes, child nodes, and leaf nodes are defined based on the semantic relationships between nodes.
[0072] S3. Merge nodes with the same name in the knowledge concept level and the event entity relationship level of the standard system to form a knowledge graph for equipment damage identification.
[0073] Equipment Damage Identification Knowledge Graph KG = { <n> , <r>},in, <n>For a set of nodes, <r>Let R be the set of edges. <r>Any edge in the middle;
[0074] <n>∈(SN∪EN)
[0075] <r>∈(SR∪ER)
[0076] R = {T, D, G}
[0077] T represents relation attribute, D represents relation direction, and G represents triple;
[0078] G = (node_A, relation, node_B)
[0079] node_A∈ <n>
[0080] node_B∈ <n>
[0081] node_A and node_B represent nodes, and relation represents semantic relations and their attributes.
[0082] In specific implementation, in step S3, the event entity relationship hierarchy can be mapped to the standard system knowledge concept hierarchy to form a knowledge graph for equipment damage identification.
[0083] The following section uses the loss identification of atmospheric and vacuum distillation units in pressure equipment as an example to illustrate the above-mentioned method for constructing a knowledge graph for equipment damage identification.
[0084] First, a knowledge graph for damage identification of pressure equipment is constructed. The domain knowledge is selected from GBT30579-2022 "Damage Pattern Recognition of Pressure Equipment", and the scenario knowledge is selected from the inspection and testing report of an atmospheric and vacuum distillation unit of a petrochemical enterprise.
[0085] Then, the knowledge ontology of damage morphology, main influencing factors, affected materials, and easily corroded equipment is extracted from the domain knowledge.
[0086] Damage morphology knowledge ontology refers to the semantic data describing typical damage morphologies of hydrochloric acid corrosion and ammonium chloride corrosion in standards. For example, the description of hydrochloric acid corrosion damage morphology in "Damage Pattern Recognition of Pressure Equipment" states: Hydrochloric acid corrosion of carbon steel and low-alloy steel can manifest as uniform corrosion; when the medium is locally concentrated or at the dew point, it manifests as localized corrosion or under-deposit corrosion; 300 series and 400 series stainless steel can exhibit pitting corrosion, forming pits with a diameter of millimeters, and may even develop into penetrating pits. Regarding ammonium chloride corrosion damage morphology: White, green, or brown salt-like deposits are often present at the corrosion sites. If water washing or purging is performed during shutdown, these deposits will be removed, and they may no longer be noticeable upon visual inspection; under-deposit corrosion is usually localized corrosion, such as pitting corrosion.
[0087] The knowledge ontology of main influencing factors refers to the semantic data analyzing the main causes of hydrochloric acid corrosion in the standard. For example, the description of the main influencing factors of hydrochloric acid corrosion in "Damage Pattern Recognition of Pressure Equipment" is as follows: 1) Hydrochloric acid concentration: The corrosion rate increases with increasing hydrochloric acid concentration; Ammonium chloride or amine hydrochloride deposits in heat exchangers and pipelines easily absorb moisture from process materials or injected washing water, and can hydrolyze under the deposits to form an acidic solution. When the pH value of the aqueous solution is lower than 4.5, it is more corrosive to carbon steel and low alloy steel; 2) Temperature: The corrosion rate increases with increasing temperature.
[0088] Description of the main influencing factors of ammonium chloride: 1) Scale (crystallization) degree: When high-temperature materials cool, ammonium chloride salts will precipitate and form scale (crystallization). The degree depends on the concentration of ammonium chloride and the temperature. Even if the temperature is much higher than the dew point temperature of water, such as 205°C, it will still corrode pipes and equipment; 2) Moisture: Ammonium chloride salts are hygroscopic and deliquescent. A small amount of water can cause serious corrosion, and the maximum corrosion rate can exceed 2.5 m / year. When ammonium chloride precipitates at a temperature higher than the dew point temperature of water, it may be necessary to inject washing water to dissolve the ammonium chloride salts. Ammonium chloride salts also have strong water solubility and corrosiveness. When mixed with water, they form an acidic solution. Some neutralized amines react with chlorides to form hydrochloric acid amine salts with similar effects.
[0089] The knowledge ontology of affected materials refers to the semantic data of metallic materials commonly used in pressure equipment and susceptible to hydrochloric acid corrosion, as listed in the standard. For example, the description of materials affected by hydrochloric acid corrosion in "Damage Pattern Recognition of Pressure Equipment" includes: carbon steel, low alloy steel, 300 series stainless steel, and 400 series stainless steel.
[0090] Materials affected by ammonium chloride corrosion are described in order of increasing corrosion resistance: carbon steel, low alloy steel, 300 series stainless steel, Monel 400, duplex stainless steel, Inconel 800 and Inconel 825, Inconel 625 and Hastelloy C276, and titanium.
[0091] The knowledge ontology of easily corroded equipment refers to the semantic data of devices, equipment, and components listed in the standard that are susceptible to hydrochloric acid corrosion. For example, the "Damage Pattern Recognition of Pressure Equipment" specifies the following equipment susceptible to hydrochloric acid corrosion: 1) Atmospheric and vacuum distillation units: a) In the atmospheric distillation tower top system, the cooling of oil and gas at the top of the tower forms a condensate containing hydrochloric acid with a low pH value, which can cause rapid corrosion of pipes and heat exchangers (including shells, tube bundles, and tube boxes); b) The vacuum ejector and condensation equipment at the top of the vacuum distillation tower are also susceptible to hydrochloric acid corrosion; 2) Hydrogenation units: a) Organic chlorides in the feed hydrocarbons or circulating hydrogen enter the unit and react to produce hydrochloric acid; b) When ammonia and hydrogen chloride are present at the same time, ammonium chloride salts can be formed in many parts of the unit, such as the reaction product side of the feed / reaction product heat exchanger, and these ammonium chloride salts may also accumulate and concentrate in the aqueous phase of the reaction product system; c) Hydrogen chloride may also flow through the fractionation unit with the process flow, causing severe acid dew point corrosion at the water injection point and downstream.
[0092] Regarding equipment susceptible to ammonium chloride corrosion: 1) Atmospheric pressure tower top system: Scale may accumulate on the top of the atmospheric pressure tower, the upper trays inside the tower, the top pipes and heat exchangers, resulting in under-deposit corrosion: deposits may appear in low flow areas due to the condensation of ammonium chloride and / or ammonium hydrochloride salts from the gas phase; corrosion may occur when the circulating reflux material at the top of the tower contains ammonium chloride or ammonium hydrochloride salts; 2) Hydrogenation unit: Chloride scale may form in the reactor effluent, resulting in under-deposit corrosion; 3) Catalytic reforming unit: Chloride under-deposit corrosion may occur in the reactor effluent and the circulating hydrogen system.
[0093] For the knowledge ontology of damage morphology, a semantic hierarchical relationship is established, and concept nodes in the semantic hierarchy are extracted. The concept nodes are then connected using the semantic hierarchical relationship to form a semantic network, as shown in Table 1, Figure 2(a), Figure 2(b), and Figure 2(c).
[0094] Table 1. Semantic hierarchy, nodes, and relationships of the damage morphology knowledge ontology.
[0095]
[0096] For the knowledge ontology of the main influencing factors, a semantic hierarchical relationship is established, and concept nodes in the semantic hierarchy are extracted. The concept nodes are connected by the semantic hierarchical relationship to form a semantic network, as shown in Table 2, Figure 3(a), Figure 3(b) and Figure 3(c).
[0097] Table 2. Main Influencing Factors: Knowledge Ontology Semantic Level, Nodes, and Relationships
[0098]
[0099]
[0100] For the knowledge ontology of the affected materials, a semantic hierarchical relationship is established, and concept nodes in the semantic hierarchy are extracted. The concept nodes are connected by the semantic hierarchical relationship to form a semantic network, as shown in Table 3, Figure 4(a), Figure 4(b) and Figure 4(c).
[0101] Table 3. Semantic hierarchy, nodes, and relationships of the knowledge ontology of affected materials.
[0102]
[0103]
[0104] For the knowledge ontology of easily corroded equipment, a semantic hierarchical relationship is established, and concept nodes in the semantic hierarchy are extracted. The concept nodes are then connected using the semantic hierarchical relationship to form a semantic network, as shown in Table 4, Figure 5(a), Figure 5(b), and Figure 5(c).
[0105] Table 4. Semantic hierarchy, nodes, and relationships of the knowledge ontology for easily corroded equipment.
[0106]
[0107]
[0108]
[0109] The semantic data related to the inspection and testing archives of pressure equipment are divided into equipment archive events, process parameter events, media composition events, and inspection and testing report events.
[0110] Equipment file events refer to semantic data such as equipment design, material selection, manufacturing process, and geometric parameters; for example, a petrochemical equipment file of a certain enterprise indicates that the hydrogenation unit of the plant was produced in the same batch, and the main parameters are: heat exchanger heat transfer area 10m². 2 The design pressure is 1.0 MPa, the primary side water temperature is 85 / 60℃, and the primary side water pressure is 1.0 MPa. The material is carbon steel, and the welding method in the manufacturing process is manual electric arc welding. Hydrogen removal treatment is carried out after welding. The main geometric parameters include inlet and outlet pipe diameters, equipment external dimensions (length, width, height), etc.
[0111] Process parameter events refer to the process parameters of petrochemical units, such as the main process parameters of the atmospheric and vacuum distillation unit of this enterprise: atmospheric top absorption rate 10%~14%, atmospheric top pressure 0.1MPa, superheated steam pressure 0.34MPa, atmospheric top temperature 128±2℃, etc.
[0112] The media composition event refers to the main media components and impurities contained in the refined crude oil, such as the crude oil of this petrochemical enterprise containing 83% hydrocarbons, 0.12% nitrogen, 0.23% phosphorus, and 0.17% vanadium.
[0113] Inspection and testing report incidents refer to periodic inspection conclusion reports, such as the inspection report of a petrochemical enterprise's petrochemical pressure equipment, which pointed out that there were 17 corrosion pits caused by hydrochloric acid corrosion in the atmospheric and vacuum distillation unit, mainly concentrated in the pipeline of the tower top system, and the material was carbon steel.
[0114] For device archive events, entities are extracted, entity nodes are created, and the corresponding entity attributes become node attributes, forming a tree structure, as shown in Table 5 and... Figure 6 As shown;
[0115] Table 5 Semantic knowledge, nodes, and relationships of device archive events
[0116]
[0117]
[0118] For process parameter events, entities are extracted, entity nodes are created, and the corresponding entity attributes become node attributes, forming a tree structure, as shown in Table 6 and... Figure 7 As shown;
[0119] Table 6 Semantic Knowledge of Process Parameter Events, Nodes, and Relationships
[0120]
[0121] For events involving media composition, entities are extracted, entity nodes are created, and the corresponding entity attributes become node attributes, forming a tree structure, as shown in Table 7 and... Figure 7 As shown;
[0122] Table 7. Semantic knowledge of media components, events, nodes, and relationships.
[0123]
[0124] For the inspection and testing report event, nodes are extracted, and the corresponding entity attributes are node attributes, forming a tree structure, as shown in Table 8. Figure 8 As shown;
[0125] Table 8. Semantic Knowledge, Nodes, and Relationships of Inspection and Testing Reports
[0126]
[0127] Merge the nodes with the same name in the standard knowledge system hierarchy of the knowledge model for damage identification of pressure equipment and the event entity relationship hierarchy of periodic inspection and testing;
[0128] Establish a hierarchical relationship diagram of knowledge concepts in the standard system SG=<SN,SR> SN represents a node in the concept, and SR represents a node connection, which is represented by a semantic relationship edge between two nodes;
[0129] Create an entity-relationship diagram EG=<EN,ER> EN represents the entity node in the event, and ER represents the node connection, which is represented by the semantic relationship edge between two nodes;
[0130] Assume the knowledge graph KG for damage identification of pressure equipment has a graph structure consisting of a set of nodes and a set of edges, represented as KG = { <n> , <r>},
[0131] in, <n>For a set of nodes, <n>∈(SN∪EN),
[0132] <r>Let be the set of edges. <r>∈(SR∪ER), for any R={T,D,G}, T represents the relation attribute, D represents the relation direction, and G represents the triple;
[0133] Semantic relations are represented by triples (node_A, relation, node_B), where node_A and node_B represent entity nodes or concept nodes, and relation represents the semantic relation and its relation attributes.
[0134] In this embodiment, Figures 2(a)-5(c) They are assembled into a semantic network as a standard system of knowledge concept levels, and then... Figures 6-9 The tree structure is mapped onto the semantic network of the knowledge concept hierarchy in the standard system, merging nodes with the same name and attributes to form... Figure 10 The diagram shows a knowledge graph for identifying damage to pressure equipment.
[0135] Based on inspection and testing reports and archives of similar atmospheric pressure tower top systems, an event ontology was extracted. Comparison with relevant nodes in the knowledge graph revealed a higher similarity between the event ontology and the ammonium chloride corrosion data in the damage identification knowledge graph. This indicates that the equipment is prone to ammonium chloride corrosion, which may be accompanied by hydrochloric acid corrosion, but not sulfuric acid corrosion. Some results are shown in the table below:
[0136] Table 9. Comparison Results of Inspection and Testing Event Ontology and Damage Knowledge Graph for Atmospheric Pressure Tower Top System (Partial)
[0137]
[0138] After identifying the damage mode, further determine the possible damage forms such as pitting corrosion, cratering, or uniform corrosion.
[0139] Once it is determined that the atmospheric pressure tower top system is prone to ammonium chloride corrosion, further analysis can be performed by retrieving hydrochloric acid corrosion and factors affecting ammonium chloride from the knowledge graph. This allows for adjustments to be made to target the damage mode and avoid or reduce the impact of these factors.
[0140] Of course, those skilled in the art will recognize that the present invention is not limited to the details of the exemplary embodiments described above, but also includes the same or similar structures that can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.
[0141] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.
[0142] The technologies, shapes, and structures not described in detail in this invention are all known technologies.< / r> < / r> < / n> < / n> < / r> < / n> < / n> < / n> < / r> < / n> < / r> < / r> < / n> < / r> < / n> < / r> < / r> < / n> < / n> < / n> < / r> < / n> < / r> < / n>
Claims
1. A method for constructing a device damage identification knowledge graph, characterized in that, Includes the following steps: S1. Extract domain knowledge from the selected equipment-related standard system and scenario knowledge from equipment inspection and testing archives; S2. Divide the domain knowledge into knowledge ontology in multiple defined directions, establish semantic hierarchical relationships for each knowledge ontology, and form a semantic network as the standard system knowledge concept hierarchy; the standard system knowledge concept hierarchy represents the hierarchical relationship of the concept of damage identification in equipment-related standards; The scenario knowledge is divided into multiple categories of event ontology. For different event ontology, semantic hierarchical relationships are established to form a semantic network as the hierarchical structure of event entity relationship system. The hierarchical structure of event entity relationship system represents the various event entities that caused the damage as determined by the inspection and testing report and their interrelationships, including the ranking of influencing factors and the relationships between each factor. S3. Merge nodes with the same name in the knowledge concept level and event entity relationship level of the standard system to form a knowledge graph for equipment damage identification; For pressure equipment, S2 divides domain knowledge into four directions: damage mode, main influencing factors, affected materials, and easily corroded equipment. The knowledge ontology of each direction is extracted and referred to as the damage mode knowledge ontology, the main influencing factors knowledge ontology, the affected materials knowledge ontology, and the easily corroded equipment knowledge ontology, respectively. The construction of the knowledge concept hierarchy in the S2 standard system includes the following steps: SA21, Divide domain knowledge into knowledge ontology in multiple defined directions; SA22. Establish semantic layers for each knowledge ontology, extract concept nodes in the semantic layers, establish relationships between the layers to form a semantic network that corresponds one-to-one with the knowledge ontology, and let the sum of the semantic networks corresponding to all knowledge ontologs be the standard system knowledge concept layer. The event entity relationship system is represented as a tree structure. Entities are extracted from the tree structure as nodes, and the root node, child nodes, and leaf nodes are divided according to the semantic relationship between the nodes.
2. The method for constructing a knowledge graph for equipment damage identification as described in claim 1, characterized in that, In SA22, the semantic hierarchy of the knowledge ontology describes the relationships between nodes as a relational model.
3. The method for constructing a knowledge graph for equipment damage identification as described in claim 1, characterized in that, In S2, scenario knowledge is divided into four categories: equipment files, process parameters, media composition, and inspection and testing reports, which are respectively called equipment file events, process parameter events, media composition events, and inspection and testing report events.
4. The method for constructing a knowledge graph for equipment damage identification as described in claim 3, characterized in that, Regarding the device archive incident, the extracted nodes include: materials, manufacturing process, and welding process; For process parameter events, the extracted entities include: flow rate, temperature, concentration, pressure, and material ratio; For incidents involving inspection and testing reports, the extracted entities include: the test results, the tested equipment, the damaged area, the morphology of the damage, and the cause of the damage.
5. The method for constructing a knowledge graph for equipment damage identification as described in claim 1, characterized in that, In S2, let the hierarchical knowledge concept map of the standard system be represented as SG=<SN, SR> SN represents a concept node in the knowledge ontology; SR represents node relationships, indicated by semantic relationship edges between two nodes; the event entity relationship hierarchy is represented as EG=<EN, ER> EN represents the entity node in the event; ER represents the node relationship, which is represented by the semantic relationship edge between two nodes; In S3, the knowledge graph representation of equipment damage identification, KG, is: KG={ <n> , <r>}< / r> < / n> <n> ∈(SN∪EN)< / n> <r> ∈(SR∪ER)< / r> R={T, D, G} G=(node_A, relation, node_B) node_A∈ <n>< / n> node_B∈ <n>< / n> in, <n>For a set of nodes, <r>Let R be the set of edges. <r> For any edge in the equation, T represents the relation attribute, D represents the relation direction, G represents the triple, node_A and node_B represent nodes, and relation represents the semantic relation and its relation attribute.< / r> < / r> < / n> 6. A method for identifying equipment damage, characterized in that, First, the equipment damage identification knowledge graph and the equipment inspection and testing report are obtained. The equipment damage identification knowledge graph is obtained using the equipment damage identification knowledge graph construction method described in any one of claims 1-5. Then, the event ontology in the inspection and testing report is extracted, and the corresponding node in the equipment damage pattern knowledge graph is found. Matching analysis is performed based on the damage pattern to which the node belongs to determine the damage pattern. Then, the node relationships are filtered to establish an association network, forming a mining framework for potential influencing factors.
7. A knowledge graph construction system for equipment damage identification, characterized in that, It includes a memory and a processor. The memory stores a computer program, and the processor is connected to the memory. The processor is used to execute the computer program to implement the equipment damage identification method as described in any one of claims 1-5.
Citation Information
Patent Citations
Construction method of Guangcai porcelain knowledge base based on semantic ontology
CN112307772A
