Digital component process knowledge learning classification expression method based on knowledge graph

Abstract

The invention discloses a digital component process knowledge learning classification expression method based on a knowledge graph, which adopts a technical scheme of digital component process knowledge learning classification expression based on the knowledge graph to realize intelligent classification and digital expression of digital component process knowledge and form a uniform and standard digital component process knowledge expression form. According to the technical scheme, the technical problems of multiple types, multiple contents and complex storage formats of the process knowledge related to the digital component are solved, and the technical effects of improving the reusability of the process knowledge, ensuring the convenience and accuracy of process knowledge query and facilitating the carding and query of the digital component process knowledge are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of process system construction, and specifically relates to a knowledge graph and learning expression technology. Background Art

[0002] With the digital transformation of electronics manufacturing, the management and application of digital component process knowledge has become increasingly important. Digital component process knowledge refers to the digital representation of various process parameters, material properties, processing methods, and other knowledge involved in product design and manufacturing. It can improve product design efficiency, optimize manufacturing processes, and reduce production costs.

[0003] Existing methods for managing digital component process knowledge suffer from inconsistent knowledge representation, low knowledge retrieval efficiency, and untimely knowledge updates, hindering the effectiveness of digital component process knowledge applications. Patent CN2024108975766 discloses a digital representation of a three-dimensional process knowledge system based on a knowledge graph, and patent CN202310904023.4 discloses the systematic management and expression of process knowledge using an ontology. Existing technologies for the holistic digital representation of process knowledge lack research on the categorized representation, self-learning, and rapid construction of process knowledge, impacting the accuracy and reusability of process knowledge. Summary of the Invention

[0004] In order to solve the technical problems of digital components involving many types of process knowledge, rich contents and complex storage formats, a technical solution of digital component process knowledge learning and classification expression based on knowledge graph is adopted to realize the intelligent classification and digital expression of digital component process knowledge, and form a unified and standardized digital component process knowledge expression form, which produces the technical effects of improving the reusability of process knowledge, ensuring the convenience and accuracy of process knowledge query, and facilitating the sorting and query of digital component process knowledge.

[0005] Combined with the digital component assembly and welding process technology system, feature extraction is performed based on regular matching of keywords and a multimodal fusion model that integrates text, image information, and graph structure. Information is extracted from process design documents, instance process assembly flow cards, component manuals, equipment manuals, tooling materials, and process standard documents to form process knowledge.

[0006] Furthermore, the process design documents include: general processes involved in the digital component assembly and welding process, main manufacturing units, types of work, brief descriptions of process contents, brief descriptions of work step contents, equipment, tooling and tools, and precautions.

[0007] Furthermore, the example process assembly flow card includes: product name of the digital component, whole part drawing number, product process flow, process name, brief description of process content, work step name, brief description of work step content, equipment, tooling and tools, operating work, inspection tooling, part number, and part name.

[0008] Furthermore, the component manual includes: the component's material code, manufacturer, quality grade, process structure, packaging form, pin material, pin plating, shell material, and specific attribute information of the component.

[0009] Furthermore, the equipment manual includes: production line purpose, asset number, equipment name, equipment model, equipment manufacturer, current application production line, equipment typical indicator range, and equipment parameter information.

[0010] Furthermore, the tooling information includes: tooling purpose, tooling number, tooling name, tooling quantity, tooling structure diagram, and tooling parameter information.

[0011] Furthermore, process standard documents include: operating procedures and standard specifications used to guide actual production.

[0012] The digital component process knowledge is divided into process knowledge, component knowledge, process resource knowledge, and process decision rule knowledge. The application domain is determined according to the specific category, and an ontology-driven knowledge model is constructed. Based on the four-dimensional ontology architecture including entity set, relationship set, attribute set, and constraint set, the process decision rules are quantified.

[0013] Furthermore, using the process route as the main line, the digital component process knowledge is classified, and the process name, main manufacturing unit, operating type, brief description of the process content, operating procedures, step number, brief description of the step content, equipment, tooling and tools, cooperating unit, cooperating type, key process identification process name, step requirements, process parameter requirements, quality standard requirements are used to sort out the basic process data.

[0014] Furthermore, using components as the main line, the digital component process knowledge is classified, and the component data is sorted out using the basic information of the components, category codes, package attributes, pin attributes, structural attributes, body codes, and information on the process steps used.

[0015] Furthermore, using process resources as the main line, the digital component process knowledge is classified, and the process resource data is sorted out using equipment type, equipment name, equipment type, equipment capability range, equipment adjustment parameter range, tooling type, tooling name, tooling number, and tooling applicable characteristic parameters.

[0016] Furthermore, using process decision-making as the main line, the digital component process knowledge is classified, and the process decision rule data is sorted out using the established rule documents, operating procedures, and industry standards.

[0017] According to the digital component ontology knowledge model, the entity alignment algorithm, ontology matching, and relationship connection based on the graph attention alignment model are adopted to perform knowledge fusion on specific categories of digital component process knowledge, and construct a knowledge graph of specific categories from structured triple knowledge.

[0018] Furthermore, regular matching and OCR recognition technology are integrated to extract entity elements of specific categories, and the entity set is constructed using the type and unique code defined for each entity, which includes the type, unique code, and entity extraction content.

[0019] Furthermore, based on the relationship between the entity set and the ontology knowledge model, the entities are associated to form an "entity-relationship-entity" mapping, and a relationship set is constructed, which includes the head entity in the first three columns, the "entity-entity" relationship in the fourth column, and the tail entity in the last three columns.

[0020] Furthermore, the entity triples in the entity set and the triple structured data in the relationship set are sequentially embedded into the knowledge graph to instantiate the digital component process knowledge graph ontology model.

[0021] The knowledge graph is encapsulated as an external service, and components, process knowledge, process resources, and process decision documents are imported. Based on the digital component process knowledge fusion engine, a digital component process knowledge graph is constructed to realize the visualization of the knowledge graph.

[0022] Furthermore, the information of process design documents, example process assembly flow cards, component manuals, equipment manuals, tooling data, and process standard documents is used to import the files into the external service model.

[0023] Furthermore, the external service model classifies the digital component process knowledge according to the imported files, constructs an ontology model, and uses structured triple knowledge to generate a knowledge graph of specific categories.

[0024] Build and manage a data network of digital component process knowledge, classify and express components, process knowledge, process resources, and process decision learning, and provide external service models for knowledge acquisition, knowledge construction, and knowledge retrieval.

[0025] Furthermore, the categories, content, and relationships of the knowledge graph are reviewed. If approved, they are published; otherwise, they are edited or deleted.

[0026] Furthermore, a first-level query is performed on the nodes and relationships in the class knowledge graph from the visualization window to obtain the node, relationship attributes, and corresponding graph data structure, and multi-level query is realized through association query.

[0027] Furthermore, based on the temporal knowledge graph embedding model, incremental updates to the class knowledge graph are achieved. DETAILED DESCRIPTION

[0028] Step 1: In combination with the digital component assembly and welding process technology system, feature extraction is performed based on regular matching of keywords and a multimodal fusion model that integrates text, image information, and graph structure. Information is extracted from process design documents, instance process assembly flow cards, component manuals, equipment manuals, tooling materials, and process standard documents to form process knowledge.

[0029] In step 1, the process design documents include: general processes involved in the digital component assembly and welding process, main manufacturing units, types of work, brief descriptions of process contents, brief descriptions of work step contents, equipment, tooling and tools, and precautions.

[0030] The example process assembly flow card includes: product name of digital component, whole part drawing number, product process flow, process name, brief description of process content, work step name, brief description of work step content, equipment, tooling and tools, operation work, inspection tooling, part number, and part name.

[0031] The component manual includes: the component's material code, manufacturer, quality grade, process structure, packaging form, pin material, pin plating, shell material, and specific attribute information of the component.

[0032] The equipment manual includes: production line purpose, asset number, equipment name, equipment model, equipment manufacturer, current application production line, equipment typical indicator range, and equipment parameter information.

[0033] Tooling information includes: tooling purpose, tooling number, tooling name, tooling quantity, tooling structure diagram, and tooling parameter information.

[0034] Process standard documents include: operating procedures and standard specifications used to guide actual production.

[0035] Step 2: Divide the digital component process knowledge into process knowledge, component knowledge, process resource knowledge, and process decision rule knowledge. Determine the application domain based on the specific category, build an ontology-driven knowledge model, and quantify the process decision rules based on the four-dimensional ontology architecture containing entity set, relationship set, attribute set, and constraint set.

[0036] In step 2, step 21 is also included: using the process route as the main line, classify the digital component process knowledge, use process name, main manufacturing unit, operating type, brief description of process content, operating procedures, process step number, brief description of process step content, equipment, tooling and tools, cooperating unit, cooperating type, key process identification process name, process step requirements, process parameter requirements, quality standard requirements, and sort out the basic process data.

[0037] Step 22: Use components as the main line to classify digital component process knowledge, and use component basic information, category code, package attributes, pin attributes, structural attributes, body code, and information on process steps to sort out component data.

[0038] Step 23: Use process resources as the main line to classify digital component process knowledge, and use equipment type, equipment name, equipment type, equipment capability range, equipment adjustment parameter range, tooling type, tooling name, tooling number, and tooling applicable characteristic parameters to sort out process resource data.

[0039] Step 24: Using process decision-making as the main line, classify the digital component process knowledge, and use the established rule documents, operating procedures, and industry standards to sort out the process decision rule data.

[0040] Step 3: Based on the digital component ontology knowledge model, the entity alignment algorithm, ontology matching, and relationship connection based on the graph attention alignment model are used to fuse the specific categories of digital component process knowledge, and construct a knowledge graph of specific categories from the structured triple knowledge.

[0041] In step 3, step 31 is also included: integrating regular matching and OCR recognition technology to extract entity elements of specific categories, and using the type and unique code defined for each entity to construct an entity set, including the type, unique code, and entity extraction content.

[0042] Step 32: According to the relationship between the entity set and the ontology knowledge model, associate each entity to form an "entity-relationship-entity" mapping, and construct a relationship set, which includes the head entity in the first three columns, the "entity-entity" relationship in the fourth column, and the tail entity in the last three columns.

[0043] Step 33: Embed the entity triples in the entity set and the triple structured data in the relationship set into the knowledge graph in sequence to instantiate the digital component process knowledge graph ontology model.

[0044] Step 4: Encapsulate the knowledge graph as an external service, import components, process knowledge, process resources, and process decision documents, build a digital component process knowledge graph based on the digital component process knowledge fusion engine, and realize the visualization of the knowledge graph.

[0045] In step 4, step 41 is also included: using information of process design files, example process assembly flow cards, component manuals, equipment manuals, tooling data, and process standard files, the files are imported into the external service model.

[0046] Step 42: The external service model classifies the digital component process knowledge according to the imported files, constructs an ontology model, and uses the structured triple knowledge to generate a knowledge graph of a specific category.

[0047] Step 5: Build and manage a data network of digital component process knowledge, classify and express components, process knowledge, process resources, and process decision learning, and provide external service models for knowledge acquisition, knowledge construction, and knowledge retrieval.

[0048] In step 5, step 51 is also included: review the categories, content, and relationships of the knowledge graph. If approved, it will be published; otherwise, it will be edited or deleted.

[0049] Step 52: Perform a first-level query on the nodes and relationships in the class knowledge graph from the visualization window to obtain the node and relationship attributes and the corresponding graph data structure, and implement multi-level query through association query.

[0050] Step 53: Based on the temporal knowledge graph embedding model, incremental updates are implemented in the class knowledge graph.

[0051] The above are embodiments of the present invention and do not limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention are included in the protection scope of the present invention.

Claims

1. A method for learning and classifying digital component process knowledge based on knowledge graph, characterized in that: include: In combination with the digital component assembly and welding process technology system, feature extraction is performed based on regular keyword matching and a multimodal fusion model that integrates text, image information, and graph structure. Information is extracted from process design documents, instance process assembly flow cards, component manuals, equipment manuals, tooling materials, and process standard documents to form process knowledge. Digital component process knowledge is divided into process knowledge, component knowledge, process resource knowledge, and process decision rule knowledge. The application domain is determined according to the specific category, and an ontology-driven knowledge model is constructed. Based on the four-dimensional ontology architecture containing entity sets, relationship sets, attribute sets, and constraint sets, process decision rules are quantified. Based on the digital component ontology knowledge model, we use the entity alignment algorithm, ontology matching, and relationship connection based on the graph attention alignment model to fuse specific categories of digital component process knowledge, and construct a knowledge graph of specific categories from structured triple knowledge. Encapsulate the knowledge graph as an external service, import components, process knowledge, process resources, and process decision documents, build a digital component process knowledge graph based on the digital component process knowledge fusion engine, and realize the visualization of the knowledge graph; Build and manage a data network of digital component process knowledge, classify and express components, process knowledge, process resources, and process decision learning, and provide external service models for knowledge acquisition, knowledge construction, and knowledge retrieval.

2. The method for learning and classifying digital component process knowledge based on knowledge graph according to claim 1 is characterized in that: The process design document includes: general processes involved in the digital component assembly and welding process, main manufacturing unit, type of work, brief description of process content, brief description of work step content, equipment, tooling and tools, and precautions; the example process assembly flow card includes: product name of the digital component, whole part drawing number, process flow of the product, process name, brief description of process content, work step name, brief description of work step content, equipment, tooling and tools, operation work, inspection tooling, part number, and part name; the component manual includes: material code, manufacturer, quality grade, process structure, packaging form, pin material, pin plating, shell material, and specific attribute information of the component; the equipment manual includes: production line purpose, asset number, equipment name, equipment model, equipment manufacturer, current application production line, typical equipment indicator range, and equipment parameter information; the tooling data includes: tooling purpose, tooling number, tooling name, tooling quantity, tooling structure diagram, and tooling parameter information; the process standard documents include: operating procedures and standard specifications used to guide actual production.

3. The method for learning and classifying digital component process knowledge based on knowledge graph according to claim 1 is characterized in that: Also includes: Using the process route as the main line, the digital component process knowledge is classified, and the process name, main unit, operating type, brief description of the process content, operating procedures, step number, brief description of the process content, equipment, tooling and tools, cooperating unit, cooperating type, key process identification process name, step requirements, process parameter requirements, quality standard requirements are used to sort out the process basic data; using components as the main line, the digital component process knowledge is classified, and the component basic information, category code, package attributes, pin attributes, structural attributes, body code, and information on the used work steps are used to sort out the component data; using process resources as the main line, the digital component process knowledge is classified, and the process resource data is sorted out using equipment type, equipment name, equipment type, equipment capability scope, equipment adjustment parameter scope, tooling type, tooling name, tooling number, and characteristic parameters applicable to the tooling; using process decision as the main line, the digital component process knowledge is classified, and the process decision rule data is sorted out using the established rule documents, operating procedures, and industry standards.

4. The method for learning and classifying digital component process knowledge based on knowledge graph according to claim 1 is characterized in that: Also includes: By integrating regular matching and OCR recognition technology, specific categories of entity elements are extracted, and an entity set is constructed using the type and unique code defined for each entity, including the type, unique code, and entity extraction content; according to the relationship between the entity set and the ontology knowledge model, the entities are associated to form an "entity-relationship-entity" mapping, and a relationship set is constructed, including the head entity in the first three columns, the "entity-entity" relationship in the fourth column, and the tail entity in the last three columns; the entity three elements in the entity set and the triple structured data in the relationship set are sequentially embedded in the knowledge graph to instantiate the digital component process knowledge graph ontology model.

5. The method for learning and classifying digital component process knowledge based on knowledge graph according to claim 1 is characterized in that: Also includes: Use information from process design documents, instance process assembly flow cards, component manuals, equipment manuals, tooling materials, and process standard documents to import files into the external service model; the external service model classifies digital component process knowledge based on the imported files, constructs an ontology model, and uses structured triple knowledge to generate knowledge graphs for specific categories.

6. The method for learning and classifying digital component process knowledge based on knowledge graph according to claim 1 is characterized in that: Also includes: Review the categories, content, and relationships of the knowledge graph. If approved, publish it; otherwise, edit or delete it. Perform a first-level query on the nodes and relationships in the class knowledge graph from the visualization window to obtain the node and relationship attributes and the corresponding graph data structure, and implement multi-level query through associative query. Based on the temporal knowledge graph embedding model, implement incremental updates to the class knowledge graph.