Part process design scene knowledge pushing method, equipment and medium
By building part design scenarios and requirements models and combining process knowledge graphs to match knowledge, the problem of knowledge push in the existing technology does not meet the requirements, and the precise knowledge push in part process design is realized, and design efficiency and quality are improved.
Patent Information
- Application Number
- CN202411057043.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-02
- Publication Date
- 2024-11-08
AI Technical Summary
The existing knowledge push method is difficult to accurately match knowledge requirements when design scenarios change, resulting in the pushing knowledge inconsistent with the actual requirements and cannot meet the dynamic knowledge push requirements in the part process design process.
By building part design scenario model and design requirement model, obtain task requirement information, extract and analyze information, combine process knowledge graph to match scene entities, if there is no matching design case, build a knowledge matching sub-graph, and use similarity calculation method to filter and Push knowledge.
It realizes accurate matching and dynamic push of knowledge in part process design, improves design efficiency and quality, and meets the knowledge needs of specific design scenarios.
Smart Images

Figure CN118916527A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the technical field of process knowledge push, and specifically relates to a method, device and medium for pushing knowledge of part process design scenarios. Background Art
[0002] In the era of knowledge economy, knowledge plays a vital role in the scientific research and production of enterprises, and has become an important resource for production management and enterprise development. With the rapid growth of knowledge volume, knowledge spillover and knowledge loss are very common in knowledge-intensive enterprises. In knowledge-intensive work such as product design, effective knowledge push to designers can greatly improve the efficiency and quality of product design work. At present, knowledge push has been widely used in product design, but related research has difficulty in accurately pushing knowledge in response to changes in design scenarios and the accuracy of knowledge needs.
[0003] The current research on knowledge push methods can be mainly divided into knowledge push methods based on knowledge sequence pattern mining, collaborative filtering algorithm and ontology. The knowledge push method based on knowledge sequence pattern mining can complete the push of design knowledge through the similarity of knowledge sequence, but it needs to do a more complex processing process for related knowledge, and the knowledge processing is difficult. The knowledge push method based on collaborative filtering algorithm can use the similarity calculation method to mix multiple push algorithms to achieve the push of multiple knowledge, but its implementation process does not consider the specific needs of users for knowledge, and the knowledge pushed is less matched with the knowledge user. The knowledge push method based on ontology establishes a knowledge ontology and maps fuzzy attributes into the ontology, combines the ontology similarity measurement algorithm, and completes the knowledge push by matching the knowledge ontology. However, it selects knowledge through the similarity between ontology models, and needs to generate a large number of knowledge matching rules, which improves the accuracy of knowledge push results to a certain extent, but at the same time, the rules also bring certain limitations. The customization of its matching rules is complex and requires the introduction of professional knowledge for customization, which brings certain difficulties to knowledge push.
[0004] Therefore, existing knowledge push methods mostly focus on design tasks and knowledge needs, and ignore the accuracy requirements for different knowledge in design scenarios. As a result, most of the pushed knowledge is difficult to meet specific scenario requirements and cannot achieve accurate knowledge push. In addition, knowledge push for dynamic organization of knowledge usage scenarios in part process design is still in the research stage, and accurate and effective knowledge push has not yet met user needs. Summary of the invention
[0005] In view of the above analysis, the embodiments of the present invention aim to provide a method, device and medium for pushing knowledge of part process design scenarios, so as to combine typical scenarios of process design with knowledge requirements and realize accurate matching and dynamic pushing of process knowledge.
[0006] In a first aspect of the present application, a method for pushing knowledge of a part process design scenario is provided, comprising:
[0007] Obtain task requirement information for part process design;
[0008] Based on the constructed process knowledge matching model, the task requirement information is extracted and analyzed to obtain the scenario model information and the design requirement information;
[0009] Based on the scenario model information and the design requirement information, scenario entity matching is performed from the constructed process knowledge graph knowledge base to obtain matching design case information;
[0010] If there is no matching design case information, a knowledge matching subgraph is constructed;
[0011] Combined with the constructed process knowledge graph knowledge base, mapping matching is performed based on the constructed knowledge matching subgraph, and the matched knowledge set is pushed to the user as push information.
[0012] Optionally, the process of constructing the process knowledge matching model includes:
[0013] The process knowledge matching model includes: a part design scenario model and a design requirement model;
[0014] The construction of the part design scenario model includes:
[0015] Define the scheme design scenario, process route design scenario and process detailed design scenario;
[0016] Construct an ontology for each scenario, including identifying and defining scenario-related entities, attributes, and their relationships;
[0017] Define the relationship between different scenarios and build a part design scenario model;
[0018] The construction of the design requirement model includes:
[0019] Identify the required information during the design process;
[0020] Establishing a demand ontology, wherein the demand ontology includes basic information and design target information;
[0021] For each type of requirement information, define the relevant entities, attributes and their relationships, and build a design requirement model.
[0022] Optionally, based on the constructed process knowledge matching model, the task requirement information is extracted and parsed to obtain scenario model information and design requirement information including:
[0023] The design scenario is located through the predefined structure tree; based on the scenario model ontology corresponding to the design scenario, the relevant element information of the design scenario is extracted;
[0024] Extract design requirement information from user input or project documents;
[0025] The extracted information is input into the fine-tuned large model, and the information contained in the relevant ontology in the instantiated knowledge matching model is subjected to trunk extraction analysis to obtain the entities and relationships of the scenario model information and the design requirement information.
[0026] Optionally, based on the scenario model information and the design requirement information, performing scenario entity matching from the constructed process knowledge graph knowledge base to obtain matching design case information includes:
[0027] By parsing the entities and relationships of the scene model information and design requirement information, combined with the scene node entities and part entity information, and using the fine-tuned large model to generate retrieval statements for the graph database, typical cases are matched from the constructed process knowledge graph knowledge base to obtain matching design case information.
[0028] Optionally, constructing a knowledge matching subgraph includes:
[0029] The entities and relationships of the scene model information and design requirement information are converted into standard triples, represented in RDF form, and a knowledge matching subgraph containing scene entities is constructed based on the resource attributes and relationships represented by the triples.
[0030] Optionally, pushing the matched knowledge set to the user as push information includes:
[0031] A similarity calculation method is used to screen and sort the knowledge matching subgraphs; the similarity calculation includes entity semantic similarity, attribute similarity and comprehensive similarity;
[0032] The calculation expression of entity semantic similarity is:
[0033]
[0034] Sim(D1, D2) represents the semantic similarity between two entities, and Dis(D1, D2) represents the semantic distance between two entities. The expressions are as follows:
[0035]
[0036] d1i and d 2i Respectively represent the values of the i-th dimension in vectors D1 and D2, and n is the vector dimension;
[0037] Entity attributes consist of clear strings, and the attribute similarity calculation expression is:
[0038]
[0039] J(X,Y) represents the matching degree between set X and set Y, |X∩Y| represents the number of elements in the intersection of sets X and Y, and |X|, |Y| represent the number of elements in sets X and Y;
[0040] The expression of comprehensive similarity is:
[0041] S=w1Sim(D1,D2)+w2J(X,Y)
[0042] S is the comprehensive similarity, w1 is the weight of entity semantic similarity; w2 is the weight of entity attribute similarity, w1+w2=1.
[0043] Optionally, pushing the matched knowledge set to the user as push information includes:
[0044] Based on user demand information and combined with the big model, the matched knowledge sets are pushed in order of similarity from high to low.
[0045] Optionally, pushing the matched knowledge set to the user as push information includes:
[0046] Obtain prompt information and confirm the designer's demand form for knowledge collection;
[0047] The knowledge set is processed by using a large model to generate a knowledge set in the form of designer's needs;
[0048] The knowledge collection presented in the form of designer's needs is pushed to users as push information.
[0049] According to a second aspect of the present application, a device for pushing knowledge of part process design scenarios is provided, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, any one of the above-mentioned methods for pushing knowledge of part process design scenarios is implemented.
[0050] According to a third aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, any of the above-mentioned methods for pushing knowledge of part process design scenarios is implemented.
[0051] The method for pushing scene knowledge of part process design provided in the present application obtains task requirement information of part process design; based on the constructed process knowledge matching model, extracts and analyzes the task requirement information to obtain scene model information and design requirement information; based on the scene model information and design requirement information, performs scene entity matching from the constructed process knowledge graph knowledge base to obtain matching design case information; if there is no matching design case information, constructs a knowledge matching subgraph; combines the constructed process knowledge graph knowledge base, performs mapping matching based on the constructed knowledge matching subgraph, and pushes the matched knowledge set to the user as push information. In the typical process design link of parts, the present application can push appropriate and accurate process knowledge to process designers in combination with design scenarios and user design requirements, thereby improving the design efficiency and quality of part process design and laying the foundation for the intelligent process design process of parts. In addition, the present application also provides a scene knowledge push device for part process design and a computer-readable storage medium having the above-mentioned technical effects. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of this specification. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0053] Figure 1 A flowchart of a specific implementation method of the part process design scenario knowledge push method provided in this application;
[0054] Figure 2 A schematic diagram of another specific implementation process of the method for pushing knowledge of part process design scenarios provided in this embodiment;
[0055] Figure 3 A schematic diagram of the method for constructing a knowledge matching model;
[0056] Figure 4 Construct a flow chart for information parsing and matching subgraphs of the fused large model;
[0057] Figure 5 Matching flow charts for candidate knowledge subsets;
[0058] Figure 6 Design scenario knowledge push implementation diagram for typical process solutions;
[0059] Figure 7 Design scenario knowledge push implementation diagram for typical process routes;
[0060] Figure 8Design scenario knowledge push implementation diagrams in detail for typical processes;
[0061] Fig. 9 This is a structural block diagram of the equipment for pushing knowledge of the part process design scenario provided in this application. DETAILED DESCRIPTION
[0062] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. It should be noted that, in the absence of conflict, the embodiments in the present disclosure and the features in the embodiments can be combined, separated, interchanged and / or rearranged with each other. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.
[0063] The terms used here are for the purpose of describing specific embodiments, and are not intended to be restrictive. As used here, unless the context clearly indicates otherwise, the singular forms "one (kind, person)" and "said (the)" are also intended to include plural forms. In addition, when the terms "comprise" and / or "include" and their variations are used in this specification, it is explained that there are stated features, integral bodies, steps, operations, parts, assemblies and / or their groups, but it is not excluded that there are or add one or more other features, integral bodies, steps, operations, parts, assemblies and / or their groups. It should also be noted that, as used here, the terms "substantially", "approximately" and other similar terms are used as approximate terms and not as degree terms, so that they are used to explain the inherent deviations of the measured values, calculated values and / or the values provided that will be recognized by those of ordinary skill in the art.
[0064] A flowchart of a specific implementation method of the part process design scenario knowledge push method provided in this application is as follows Figure 1 As shown, the method specifically includes:
[0065] S101: Obtain task requirement information of part process design.
[0066] Specifically, specific design task requirements, such as part specifications, performance requirements, material types, etc., can be input through the interface, or task requirement information can be automatically extracted by reading design files (such as CAD files) or databases.
[0067] S102: Based on the constructed process knowledge matching model, extract and analyze the task requirement information to obtain scenario model information and design requirement information.
[0068] Among them, the process knowledge matching model KMM includes: part design scenario model PDS and design requirement model DEI. The expression of KMM is KMM={PDS,DEI}. PDS is a typical scenario such as PDS={scheme design, process route design, detailed process design}, and DEI is design requirement information such as DEI={basic information, design target information}. The PDS model and the DEI model are constructed using the ontology construction method. The construction of the PSD model mainly includes constructing a typical scenario ontology, describing the key element information of the scenario, and establishing the association relationship between typical scenarios. The construction of the DEI model mainly establishes the demand ontology, and the description information includes basic information and design target information.
[0069] Specifically, the construction of the part design scenario model includes:
[0070] S1021: Define the scheme design scenario, process route design scenario and process detailed design scenario.
[0071] S1022: Construct an ontology for each scenario, including identifying and defining scenario-related entities, attributes, and their relationships.
[0072] S1023: Define the association relationship between different scenarios and build a part design scenario model.
[0073] The construction of the design requirement model includes:
[0074] S1024: Identify the required information during the design process.
[0075] S1025: Establishing a demand ontology, wherein the demand ontology includes basic information and design target information.
[0076] S1026: For each type of requirement information, define the relevant entities, attributes and their relationships, and build a design requirement model.
[0077] Based on the constructed process knowledge matching model, the task requirement information is extracted and analyzed to obtain the scenario model information and design requirement information, including: locating the design scenario through the predefined structure tree; extracting the relevant element information of the design scenario based on the scenario model ontology corresponding to the design scenario. Extract the design requirement information from the user input or project file. Input the extracted information into the fine-tuned large model, perform trunk extraction analysis on the information contained in the relevant ontology in the instantiated knowledge matching model, and obtain the entities and relationships of the scenario model information and the design requirement information.
[0078] The specific steps are:
[0079] (1) Knowledge matching model information extraction instantiation
[0080] PDS information extraction: The process design nodes are predefined through the structure tree, and the scenario model can obtain relevant scenario model information by reading the structure tree. Take the forging process design process of ring forging parts as an example, where the process design is divided into several typical scenarios such as part scheme design, process route design, and detailed process design. The extraction of its scenario element information can be obtained by identifying the scenario model entity in the structure tree to obtain the relevant element information of the current typical design scenario;
[0081] DEI information extraction: basic information, including part material information, product delivery requirements, etc. It is the basic static process information inherent in the part, which will not change due to the design process. By identifying the unique identification information of the part (such as the ID information in the part library), the basic information of the current part is directly obtained from the part library. Design target information, through user command input and design task information, obtains the current design requirement information.
[0082] (2) Knowledge matching information analysis
[0083] Combining the advantages of large models in information processing, the embodiment of the present application uses a large model (GLM-3 large model) that has been fine-tuned by prompt words to perform trunk extraction analysis on the information contained in the relevant ontology in the instantiated knowledge matching model to obtain the entities and relationships of scene element information and demand information.
[0084] S103: Based on the scene model information and design requirement information, scene entity matching is performed from the constructed process knowledge graph knowledge base to obtain matching design case information.
[0085] Specifically, by parsing the entities and relationships of the scene model information and design requirement information, combined with the scene node entities and part entity information, and using the fine-tuned large model to generate retrieval statements for the graph database, typical cases are matched from the constructed process knowledge graph knowledge base to obtain matching design case information.
[0086] The search statement is a Cypher search statement. The knowledge base is a case knowledge base with a graph structure constructed in the form of a knowledge graph, including typical cases and example knowledge in various scenarios required to complete the process design process.
[0087] If typical case information of the design scenario is matched, the typical case information will be dynamically organized and pushed in combination with the large model. If no match is found, the next step is to build a matching sub-graph for the relevant entity information.
[0088] S104: If there is no matching design case information, construct a knowledge matching subgraph.
[0089] Constructing a knowledge matching subgraph includes: converting the entities and relationships of the scene model information and the design requirement information into standard triple forms, expressing them in RDF form, and constructing a knowledge matching subgraph containing scene entities based on the resource attributes and relationships represented by the triples.
[0090] S105: Combined with the constructed process knowledge graph knowledge base, mapping matching is performed based on the constructed knowledge matching subgraph, and the matched knowledge set is pushed to the user as push information.
[0091] In this step, a similarity calculation method is used to screen and sort the knowledge matching subgraphs; wherein the similarity calculation includes entity semantic similarity, attribute similarity and comprehensive similarity;
[0092] The calculation expression of entity semantic similarity is:
[0093]
[0094] Sim(D1, D2) represents the semantic similarity between two entities, and Dis(D1, D2) represents the semantic distance between two entities. The expressions are as follows:
[0095]
[0096] d 1i and d 2i Respectively represent the values of the i-th dimension in vectors D1 and D2, and n is the vector dimension;
[0097] Entity attributes consist of clear strings, and the attribute similarity calculation expression is:
[0098]
[0099] J(X,Y) represents the matching degree between set X and set Y, |X∩Y| represents the number of elements in the intersection of sets X and Y, and |X|, |Y| represent the number of elements in sets X and Y;
[0100] The expression of comprehensive similarity is:
[0101] S=w1Sim(D1,D2)+w2J(X,Y)
[0102] S is the comprehensive similarity, w1 is the weight of entity semantic similarity; w2 is the weight of entity attribute similarity, w1+w2=1.
[0103] Based on the process knowledge set obtained by the above comprehensive similarity calculation, based on user demand information and combined with the big model, the matched knowledge set is pushed in order of similarity from high to low.
[0104] Based on any of the above embodiments, pushing the matched knowledge set as push information to the user includes:
[0105] Obtain prompt information to confirm the designer's demand form for the knowledge set. The demand form may include any one or any combination of graph structure output, table structure output, and natural language output.
[0106] The knowledge set is processed using a large model to generate a knowledge set presented in the form of designer's needs.
[0107] The knowledge collection presented in the form of designer's needs is pushed to users as push information.
[0108] Compared with the traditional customized knowledge push process, the embodiment of the present application uses the existing capabilities of the fine-tuned large model to dynamically organize and push the optimal knowledge in the form of output by designing prompt information.
[0109] This application constructs scenario models and demand models for typical design scenarios, uses the information processing capabilities of large models to process the demand information of each scenario, matches the required knowledge by matching typical cases or building knowledge matching subgraphs, and obtains the optimal knowledge through similarity calculation. The generation capabilities of large models are used to dynamically organize and push the optimal knowledge, thereby improving the flexibility of knowledge application, providing designers with appropriate process knowledge, and improving the quality of process design.
[0110] In the process of part process design, it is necessary to determine the design tasks and delivery requirements, and then formulate a process plan based on the specific conditions of the parts, and carry out production and processing according to the formulated process plan to obtain part products that meet the requirements. In the process of formulating process plans, the quality of typical scenario design scenarios has a direct impact on the overall process planning. Accurately pushing typical design scenario knowledge to assist designers is an important way to improve the quality of process design.
[0111] Based on this, another specific implementation process diagram of the method for pushing knowledge of part process design scenarios provided in this embodiment is as follows: Figure 2 The process specifically includes the following steps:
[0112] Step 1: Construct a process knowledge matching model.
[0113] Figure 3 A schematic diagram of the knowledge matching model construction method is shown. Figure 3, the knowledge matching model includes a typical process design scenario model and a design requirement model; the design scenario model includes typical design nodes of process design, such as scheme design, process route design, and detailed process design; the design requirement model includes part information and design target information; the part information includes part name, material information, and delivery requirements, etc.; the design target information includes task information and user instruction information. Among them, the PDS model and the DEI model are constructed using the ontology construction method. The construction of the PSD model mainly includes constructing a typical scenario ontology, describing the key element information of the scenario, and establishing the association relationship between typical scenarios. For the construction of the DEI model, the demand ontology is mainly established, and the description information includes basic information and design target information.
[0114] Step 2: Process information processing through design process.
[0115] Step 2.1: Knowledge matching model information extraction instantiation
[0116] The process design information extraction mainly extracts information according to the knowledge matching model, including the acquisition of scene element information PDS and design requirement model information DEI.
[0117] Among them, the information extraction of process design scenario information PDS is based on the forging process design process of annular structure parts. Its typical design scenarios are scheme design, process route design, and detailed process design. The extraction of its scenario element information can obtain the relevant element information of the current typical design scenario description and its associated scenarios by identifying the scenario model ontology in the structure tree.
[0118] The acquisition of design requirement information DEI mainly includes the extraction of basic part information and design target information. Basic part information includes part name, material information, product delivery requirements, etc. It is the basic static process information inherent in the part and will not change due to the design process. By identifying the unique identification information of the part (such as the ID information in the parts library), the basic information of the current part is directly obtained from the parts library. Design target information, through user command input and design task information, obtains the current design requirement information.
[0119] Step 2.2: Requirement information analysis and processing
[0120] like Figure 4 As shown in the flowchart of information analysis and matching subgraph construction of the fusion big model, combined with the advantages of the big model in information processing, this application uses the big model (GLM-3 big model) fine-tuned with prompt words to perform trunk extraction and analysis on the information contained in the relevant ontology in the instantiated knowledge matching model to obtain the entities and relationships of scene element information and demand information.
[0121] First, the extracted information is input into the GLM-3 large model, and the main demand information is extracted from the input information by designing prompt words to extract seed entities and seed relationships. Taking the extracted input information "design scenario information (process route design), design requirement information (part information (noun: outer casing, material grade: GH4169, delivery requirements: solid solution + rough machining), instructions or user input information: What are the processing routes for this type of parts?)" as an example, the prompt words are designed as follows: "From the perspective of part forging process design, extract entities and relationships in the input information". After processing by the GLM-3 large model, relevant entities and relationships such as entity sets "outer casing, material grade, delivery requirements", etc., and relationship sets "outer casing-material grade, outer casing-process route", etc. are obtained.
[0122] Step 2.3 Typical case information matching
[0123] By parsing the relevant entities and relationship information, combined with the extracted seed entities (scenario node entities) and part entity information, the Cypher search statement of the graph database is generated by fine-tuning the large model to match typical case knowledge from the constructed knowledge base. Among them, the knowledge base is a case knowledge base with a graph structure constructed in the form of a knowledge graph, including typical cases and example knowledge in various scenarios required to complete the process design process. If the typical case information of the design scenario is matched, go to step 5, and dynamically organize and push the typical case information in combination with the large model. If there is no match, go to the next step to match the relevant entity information with the sub-graph construction.
[0124] Step 2.4: Matching subgraph construction
[0125] The fine-tuned large model has good semantic parsing performance. Combined with its excellent generation performance, it can automatically convert the extracted scene and demand information entities and relationships into standard triples, express them in RDF form, and directly use the resources it represents and their attributes and relationships to construct matching subgraphs, thereby achieving accurate construction of subgraphs.
[0126] Step 3: Subgraph mapping matching.
[0127] like Figure 5As shown in the candidate knowledge subset matching flow chart, the process knowledge graph type knowledge base includes all the knowledge required to complete the process design process, such as factual knowledge (part feature information, material information, etc.), procedural knowledge (case knowledge, example knowledge, etc.) and logical knowledge (rules, decision tables, etc.), providing rich knowledge resources for retrieval. Traditional knowledge matching using keyword matching and fuzzy retrieval generally has poor recognition ability for coarse-grained information, making it difficult to achieve accurate knowledge matching. Compared with traditional methods, the embodiment of the present application uses a large model to accurately parse coarse-grained, high-level scenarios and input information, and realizes the accurate construction of knowledge matching subgraphs. When mapping subgraphs, the query subgraph is mapped to the manufacturing process knowledge graph knowledge base using triples corresponding to known facts. The breadth-first algorithm is first used to extract candidate subgraphs, and then combined with the topological structure of the existing query graph, combined with typical design scenarios and design target information for screening and filtering, a candidate subgraph set is obtained to achieve accurate knowledge matching.
[0128] Step 4: Filter and sort subgraphs.
[0129] According to the matched candidate subgraph set, the entity semantic similarity and attribute similarity methods are used to calculate the similarity and combined with the comprehensive similarity calculation to perform screening and sorting. The entity semantic similarity calculation expression is:
[0130]
[0131] Sim(D1, D2) represents the semantic similarity between two entities, and Dis(D1, D2) represents the semantic distance between two entities. The expressions are as follows:
[0132]
[0133] d 1i and d 2i Respectively represent the values of the i-th dimension in vectors D1 and D2, and n is the vector dimension;
[0134] The entity attributes are composed of clear strings, and the similarity calculation expression is:
[0135]
[0136] J(X,Y) represents the similarity between sets X and Y, |X∩Y| represents the number of elements in the intersection of sets X and Y, and |X|, |Y| represent the number of elements in sets X and Y;
[0137] The expression of comprehensive similarity is:
[0138] S=w1Sim(D1,D2)+w2J(X,Y)
[0139] S is the comprehensive similarity, w1 is the weight of entity semantic similarity; w2 is the weight of entity attribute similarity, w1+w2=1.
[0140] Based on the above requirement example information, the specific knowledge push process example of the typical process route design scenario of the part is used for verification (in this embodiment, the entity semantic similarity weight and the attribute similarity weight are w1=0.7 and w2=0.3 respectively). Using the similarity calculation formula, the parts with comprehensive similarity results greater than the set threshold and the related process route design list are shown in Table 1.
[0141] Table 1
[0142]
[0143] Step 5: Knowledge push.
[0144] Figure 6 The following figure shows the implementation diagram of knowledge push in a typical process solution design scenario. Figure 7 The following figure shows the implementation diagram of knowledge push in a typical process route design scenario. Figure 8 The implementation diagram of knowledge push for typical process detailed design scenarios is shown, and the corresponding typical process design scenarios for parts include specific processes for scheme design, process route design, and process detailed design knowledge push. Based on the calculation results obtained in step 4, they are sorted in order from high to low according to the similarity. Compared with the traditional customized knowledge push process, the embodiment of the present application uses the existing capabilities of the fine-tuned large model combined with user demand information by designing prompt information to dynamically organize and push the optimal knowledge according to the user's requirements for the output form. The specific steps are:
[0145] (1) Obtain prompt information and confirm the knowledge demand form of the designer (graph structure output, table structure output, natural language output);
[0146] (2) Combined with the existing generation capabilities of the big model, the obtained knowledge is dynamically organized and pushed in the form required by the user. Among them, the output of the graph structure can use the big model combined with the HTML format to organize the knowledge and obtain the output in the form of a graph; the output in the form of table structure and natural language can be completed with the help of the generation capabilities of the big model itself.
[0147] In addition, this application also provides a part process design scenario knowledge push device, such as Fig. 9 As shown in the structural block diagram of the part process design scenario knowledge push device provided in the present application, the device specifically includes a memory 91 and a processor 92, and the memory 91 stores a computer program. When the computer program is executed by the processor 92, any one of the above-mentioned part process design scenario knowledge push methods is implemented.
[0148] In addition, the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements any of the above-mentioned part process design scenario knowledge push methods.
[0149] Computer readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0150] The professionals should also be further aware that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0151] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
[0152] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only the specific implementation method of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.
Claims
1. A method for pushing knowledge of part process design scenarios, characterized in that: include: Obtain task requirement information for part process design; Based on the constructed process knowledge matching model, the task requirement information is extracted and analyzed to obtain the scenario model information and the design requirement information; Based on the scenario model information and the design requirement information, scenario entity matching is performed from the constructed process knowledge graph knowledge base to obtain matching design case information; If there is no matching design case information, a knowledge matching subgraph is constructed; Combined with the constructed process knowledge graph knowledge base, mapping matching is performed based on the constructed knowledge matching subgraph, and the matched knowledge set is pushed to the user as push information.
2. The method for pushing knowledge of part process design scenarios according to claim 1 is characterized in that: The process of constructing the process knowledge matching model includes: The process knowledge matching model includes: a part design scenario model and a design requirement model; The construction of the part design scenario model includes: Define the scheme design scenario, process route design scenario and process detailed design scenario; Construct an ontology for each scenario, including identifying and defining scenario-related entities, attributes, and their relationships; Define the relationship between different scenarios and build a part design scenario model; The construction of the design requirement model includes: Identify the required information during the design process; Establishing a demand ontology, wherein the demand ontology includes basic information and design target information; For each type of requirement information, define the relevant entities, attributes and their relationships, and build a design requirement model.
3. The method for pushing knowledge of part process design scenarios according to claim 2 is characterized in that: Based on the constructed process knowledge matching model, the task requirement information is extracted and analyzed to obtain the scenario model information and design requirement information including: The design scenario is located through the predefined structure tree; based on the scenario model ontology corresponding to the design scenario, the relevant element information of the design scenario is extracted; Extract design requirement information from user input or project documents; The extracted information is input into the fine-tuned large model, and the information contained in the relevant ontology in the instantiated knowledge matching model is subjected to trunk extraction analysis to obtain the entities and relationships of the scenario model information and the design requirement information.
4. The method for pushing knowledge of part process design scenarios according to claim 3 is characterized in that: Based on the scenario model information and the design requirement information, scenario entity matching is performed from the constructed process knowledge graph knowledge base to obtain matching design case information, including: By parsing the entities and relationships of the scene model information and design requirement information, combined with the scene node entities and part entity information, and using the fine-tuned large model to generate retrieval statements for the graph database, typical cases are matched from the constructed process knowledge graph knowledge base to obtain matching design case information.
5. The method for pushing knowledge of part process design scenarios according to claim 3 is characterized in that: The constructing of the knowledge matching subgraph comprises: The entities and relationships of the scene model information and design requirement information are converted into standard triples, represented in RDF form, and a knowledge matching subgraph containing scene entities is constructed based on the resource attributes and relationships represented by the triples.
6. The method for pushing knowledge of part process design scenarios according to any one of claims 1 to 5, characterized in that: The pushing of the matched knowledge set as push information to the user comprises: A similarity calculation method is used to screen and sort the knowledge matching subgraphs; the similarity calculation includes entity semantic similarity, attribute similarity and comprehensive similarity; The calculation expression of entity semantic similarity is: Sim(D1, D2) represents the semantic similarity between two entities, and Dis(D1, D2) represents the semantic distance between two entities. The expressions are as follows: d 1i and d 2i Respectively represent the values of the i-th dimension in vectors D1 and D2, and n is the vector dimension; Entity attributes consist of clear strings, and the attribute similarity calculation expression is: J(X,Y) represents the matching degree between set X and set Y, |X∩Y| represents the number of elements in the intersection of sets X and Y, and |X|, |Y| represent the number of elements in sets X and Y; The expression of comprehensive similarity is: S=w1Sim(D1,D2)+w2j(X,Y) S is the comprehensive similarity, w1 is the weight of entity semantic similarity; w2 is the weight of entity attribute similarity, w1+w2=1.
7. The method for pushing knowledge of part process design scenarios according to claim 6 is characterized in that: The pushing of the matched knowledge set as push information to the user comprises: Based on user demand information and combined with the big model, the matched knowledge sets are pushed in order of similarity from high to low.
8. The method for pushing knowledge of part process design scenarios according to any one of claims 1 to 5, characterized in that: The pushing of the matched knowledge set as push information to the user comprises: Obtain prompt information and confirm the designer's demand form for knowledge collection; The knowledge set is processed by using a large model to generate a knowledge set in the form of designer's needs; The knowledge collection presented in the form of designer's needs is pushed to users as push information.
9. A device for pushing knowledge of part process design scenarios, characterized in that: It includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the method for pushing part process design scenario knowledge according to any one of claims 1 to 8 is implemented.
10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the method for pushing knowledge of a part process design scenario according to any one of claims 1-8 is implemented.
Citation Information
Cited By
Aircraft part manufacturing unit identification method based on rule matching
CN120067774A
