Steel structure welding process method and system based on knowledge graph

Abstract

The invention belongs to the related technical field of steel structure machining, and discloses a steel structure welding process method and system based on a knowledge graph, and the method comprises the steps that component information of a steel structure in a building and information of parts forming all the components are obtained, part information is established, and the steel structure is divided into component-part-part composition hierarchies; theoretical welding seam information of all welding seams of the steel structure is obtained, and matched welding process information of the welding seams is obtained based on the welding process knowledge graph according to the theoretical welding seam information; the welding sequence of the components is optimized according to the length of the welding path, and welding sequence information is obtained; and generating a welding process model by combining the welding sequence information and the matched welding process information of each welding seam. According to the method, the steel structure of the building is split, the theoretical welding seam information is generated, the welding process information is automatically matched and generated in combination with the welding process knowledge graph, and the efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field related to steel structure processing, and more specifically, relates to a steel structure welding process method and system based on a knowledge graph. Background Art

[0002] In the modern construction industry, steel structures are widely used due to their excellent mechanical properties, lightweight, high strength, and rapid construction speed, making them a crucial component of building industrialization. With the continuous expansion of building scale and increasing project complexity, the demand for automated and intelligent steel structure construction is becoming increasingly urgent. Traditional steel structure production methods rely heavily on manual labor, resulting in low production efficiency, high error rates, and potential safety hazards. Transformation and upgrading through technological innovation are urgently needed.

[0003] In recent years, Building Information Modeling (BIM) technology has gradually emerged, providing advanced data management and collaboration tools for steel structure design and construction. However, the current application of BIM models in welding processes remains insufficient, and many companies still rely on traditional manual welding operations. This practice not only leads to poor information transfer, but also increases worker workload and reduces construction efficiency.

[0004] The demand for welding automation is growing with advancements in welding robots and intelligent manufacturing technologies. While these robots play a significant role in improving steel structure production efficiency, existing welding robots often rely on pre-set paths, rely heavily on manual experience to develop welding patterns, and lack the ability to flexibly adapt to on-site changes.

[0005] In current welding operations, manual welding workers mostly rely on process cards, while automatic welding workstations rely on manually pre-defined process packages. Both have the problem of high manual dependence resulting in low efficiency. Summary of the Invention

[0006] In response to the above-mentioned defects or improvement needs of the existing technology, the present invention provides a steel structure welding process method and system based on a knowledge graph, which is used to solve the problem that the current steel structure welding operations are highly dependent on manual labor and lead to low efficiency. It aims to automatically generate a welding process model and guide welding based on data such as the workpiece model, and can convert data such as the model into specific welding process guidance, thereby significantly improving welding efficiency and quality, and promoting the development of steel structure construction to a higher level.

[0007] To achieve the above objectives, according to one aspect of the present invention, a steel structure welding process method based on a knowledge graph is provided, comprising: Steel structure hierarchical division: Obtain the component information of the steel structure in the building and the part information of each component, and establish component information based on the adjacent relationship of the parts, dividing the steel structure into a component-component-part hierarchy; Matching welding process information acquisition: Obtain theoretical weld information of each weld within and between components in the steel structure, and obtain matching welding process information for each weld based on the theoretical weld information and the pre-established welding process knowledge graph; Welding sequence information acquisition: Optimize the welding sequence of components based on the welding path length, and obtain welding sequence information based on the optimized component welding sequence; Welding process model generation and welding operation: combining the welding sequence information and the matching welding process information of each weld to generate a welding process model, and performing actual welding operations based on the welding process model.

[0008] According to the steel structure welding process method based on the knowledge graph provided by the present invention, the component information in the hierarchical division of the steel structure composition includes part composition, and also includes at least one of component type, component position, component size, component function and adjacent component information; part information includes part material and adjacent part information, and also includes at least one of part position, part size and part type; component information includes component position and part composition, and also includes at least one of component type, component size and adjacent component information; The hierarchical division of steel structure components also includes: dividing the building into multiple areas, and dividing the steel structure into component-part-part components for any area; accordingly, obtaining the matching welding process information of each weld for any area; obtaining the welding sequence information for any area; generating a welding process model for any area during welding operation.

[0009] According to the steel structure welding process method based on knowledge graph provided by the present invention, the welding process knowledge graph in the matching welding process information acquisition includes entity nodes and edges connecting entity nodes with matching relationships; the entity nodes include welding process nodes, welding equipment nodes, welding material nodes, welding parameter nodes and environmental parameter nodes, and also include at least one of weld type nodes, weld geometry parameter nodes and welding object material nodes; the matching relationships between the entity nodes include multiple of process equipment, process applicable welding materials, equipment applicable welding materials, process applicable welding parameters, process applicable environmental parameters, process applicable weld type, welding parameter corresponding weld geometry parameters, and process applicable object material.

[0010] According to the steel structure welding process method based on the knowledge graph provided by the present invention, the method for establishing the welding process knowledge graph includes: Collect welding related data and define entities and matching relationships; Extract knowledge of entities and matching relationships from welding-related materials, and connect the extracted entities through matching relationships to form triple structure information; The triple structure information is stored in the graph database to form a welding process knowledge graph.

[0011] According to the steel structure welding process method based on the knowledge graph provided by the present invention, the method for establishing the welding process knowledge graph further includes: Collect and obtain welding-related update information at preset intervals; Extract knowledge of entities and matching relationships from updated welding-related data to form updated triple structure information; The welding process knowledge graph is updated and maintained using the updated triple structure information.

[0012] According to the steel structure welding process method based on knowledge graph provided by the present invention, the theoretical weld information in the matching welding process information acquisition includes at least one of theoretical weld geometric parameters, theoretical weld type and weld application object material; wherein the theoretical weld geometric parameters include at least one of weld joint vector, weld length, weld angle, weld width, weld depth and weld cross-sectional area.

[0013] According to the steel structure welding process method based on knowledge graph provided by the present invention, when optimizing the welding sequence of components according to the welding path length in the welding sequence information acquisition, the optimization object is the component welding sequence vector , the optimization goal is to minimize the welding path length, welding path length ;in represents the component ranked as i, and d represents the spatial distance; Obtaining the welding sequence information also includes: analyzing the load-bearing level of each component in the building, and determining the welding sequence of the welds between the components in descending order of the load-bearing level.

[0014] According to the steel structure welding process method based on the knowledge graph provided by the present invention, the welding process model generation and welding operation also include: calculating the theoretical heat input of each weld according to the matching welding process information of each weld, adjusting the welding parameters until the heat input is within the first preset threshold range when the theoretical heat input exceeds the preset heat input threshold, obtaining the welding parameter information after the heat input of each weld is corrected; and generating a welding process model based on the welding parameter information after the heat input is corrected.

[0015] According to the steel structure welding process method based on the knowledge graph provided by the present invention, the welding process model is generated and during the actual welding operation, the welding parameters in the matching welding process information are corrected according to the deviation between the actual temperature and the theoretical temperature corresponding to the matching welding process information, and the actual welding operation is performed based on the welding parameters after the temperature correction; And / or, during the generation of the welding process model and the actual welding operation, the deviation between the actual weld information and the theoretical weld information is monitored. When the deviation exceeds a second preset threshold, the welding parameters in the matching welding process information are feedback-adjusted, and the actual welding operation is performed based on the feedback-corrected welding parameters.

[0016] According to another aspect of the present invention, a steel structure welding process system based on a knowledge graph is provided, the system including a memory and a processor, the memory storing a computer program, and the processor executing any one of the above-mentioned steel structure welding process methods based on the knowledge graph when executing the computer program.

[0017] In general, compared with the prior art, the above technical solutions conceived by the present invention provide a steel structure welding process method and system based on knowledge graph: 1. This paper proposes a hierarchical division of steel structure components into parts, components, and parts, and generates theoretical weld information. Combining this with a pre-built welding process knowledge graph, this approach enables automated matching and generation of welding process information. This significantly reduces the reliance on manual labor for welding processes and significantly improves the efficiency and accuracy of welding process model generation, providing support for the modernization of steel structure construction and significantly enhancing welding process efficiency. 2. The hierarchical structure of steel structures (components, parts, and components) allows for effective differentiation between primary welding (i.e., welding of welds within components) and secondary welding (i.e., welding of welds between components) during the welding process. This allows for more orderly welding of each weld, ensuring that the sequence of the welding process meets actual operational requirements. 3. Proposed corrections for heat input, temperature, and feedback adjustments to welding parameters. These take into account the effectiveness and accuracy of welding parameters from multiple perspectives and overcome the shortcomings of existing automatic welding systems that lack monitoring and feedback mechanisms, thereby ensuring better welding quality. 4. Combining building model data and other data with welding robots to build an automated, efficient, and intelligent welding process generation system has become an important direction for promoting the modernization of steel structure construction. Such a system can not only realize the visualization and informatization of the production process, but also improve the accuracy and safety of welding operations to meet increasingly diverse market demands. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1This is a flow chart of a steel structure welding process method based on a knowledge graph provided by an embodiment of the present invention; Figure 2 is an exemplary system architecture diagram of an embodiment of the present invention; Figure 3 A flowchart for splitting a BIM in-depth design model provided by an embodiment of the present invention; Figure 4 A flowchart for constructing a knowledge graph provided by an embodiment of the present invention; Figure 5 A flowchart of the knowledge graph-based component reorganization (PBOM) and welding sequence allocation provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0020] See also Figure 1 This embodiment provides a steel structure welding process method based on a knowledge graph, and the steel structure welding process method includes: Steel structure hierarchical division: Obtain the component information of the steel structure in the building and the part information of each component, and establish component information based on the adjacent relationship of the parts, dividing the steel structure into a component-component-part hierarchy; Matching welding process information acquisition: Obtain theoretical weld information of each weld within and between components in the steel structure, and obtain matching welding process information for each weld based on the theoretical weld information and the pre-established welding process knowledge graph; Welding sequence information acquisition: Optimize the welding sequence of components based on the welding path length, and obtain welding sequence information based on the optimized component welding sequence; Welding process model generation and welding operation: combining the welding sequence information and the matching welding process information of each weld to generate a welding process model, and performing actual welding operations based on the welding process model.

[0021] This embodiment proposes a method for automatically generating a steel structure welding process model and a welding operation process method. The method can obtain theoretical weld information for each weld through data such as a building model and combine it with a pre-constructed welding process knowledge graph. By performing process matching based on the theoretical weld information on the welding process knowledge graph, the matching welding process information can be automatically obtained, thereby achieving automated generation of welding processes, which is conducive to significantly improving the efficiency and accuracy of welding manufacturing processes. The welding process knowledge graph typically integrates welding-related materials such as welding manuals, industry specifications, process cards, and actual cases. This helps ensure that the generated process model meets industry standards and is flexible to adapt to different construction scenarios, providing support for the modernization of steel structure construction and significantly improving the efficiency and accuracy of the welding process.

[0022] Furthermore, the acquisition of component information and part information of steel structures in buildings, as well as the acquisition of theoretical weld information of each weld in the steel structure, can be obtained based on building industry data such as building BIM models, building drawings, and building laser scanning models, and the specific acquisition source is not limited. Preferably, the steel structure can be divided into hierarchical components based on the building BIM model, that is, based on the building BIM model, the component information of the steel structure in the building and the part information of each component are obtained, and the component information is established based on the adjacent relationship of the parts, dividing the steel structure into a component-component-part composition hierarchy; similarly, the theoretical weld information of each weld within the components and between the components in the steel structure can be obtained based on the building BIM model.

[0023] In some specific embodiments, taking obtaining required information based on a building BIM model as an example, refer to Figure 2 , a method for automatically generating a steel structure welding process model based on a BIM model is proposed, which includes the following main steps: First, the geometric data of the BIM detailed design model is analyzed and split based on a plug-in developed for Tekla, and the geometric hierarchy of the components is divided to provide a basis for subsequent welding process operations. Subsequently, a welding process knowledge graph is constructed using welding manuals, industry specifications, process cards, and actual case data to support the matching and selection of welding process information. Next, a component process BOM (i.e., PBOM) is generated and the component hierarchy is reorganized to optimize the welding process sequence. Then, by automatically identifying welding joints, the geometry and parameters of the theoretical weld in the BIM model are generated, and the key parameters required for welding, including welding process, equipment, welding materials, welding current, voltage, and speed, are matched. Finally, all welding process information is integrated into a complete welding process model to guide actual production and support real-time feedback optimization to ensure high quality and high efficiency of the welding process.

[0024] Specifically, the BIM detailed design model is first split and the welding process knowledge graph is constructed. In the initial stage of welding process design and process generation, the geometric structure of the BIM detailed design model must first be analyzed and split in detail. The BIM detailed design model usually contains a hierarchical bill of materials (EBOM) of components and parts, but due to the actual needs of the welding production process, it is also necessary to add a component level to the model to form a bill of materials (PBOM) based on the component-component-part hierarchy to clarify the hierarchy of welding operations. Through this hierarchical structure of components-components-parts, primary welds (i.e., welding of welds within components) and secondary welds (i.e., welding of welds between components) can be effectively distinguished in the welding process to ensure that the sequence of the welding process meets the needs of actual operations.

[0025] Through model analysis and decomposition, the system extracts BIM information about the steel structure from the BIM detailed design model. Simultaneously, by aggregating information from welding manuals, specifications, process cards, and other sources, it extracts key entity information, including welding processes, welding equipment, welding consumables, welding parameters (such as current, voltage, and welding speed), welding environment (such as temperature and humidity), and weld information. Each entity is defined as an entity node in the knowledge graph and interconnected through matching relationships (such as "usage," "applicable process," "applicable materials," and "influencing factors") to form a triple structure. This triple structure then forms the welding process knowledge graph. For example, the relationship between welding equipment and its corresponding welding consumables can be expressed as "Equipment A uses welding consumable B."

[0026] Among them, reference Figure 3 , the splitting and parsing of BIM geometric data are as follows: Since the building BIM model itself can be processed in Tekla software, a Tekla plug-in can be developed to split the BIM model. First, load the building BIM model in Tekla software. The model may contain one or more independent partition models, each partition model includes multiple specific components, and multiple regions are divided through geometric analysis of the BIM model. That is, the hierarchical division of steel structure components also includes: dividing the building into multiple regions, and for any region, dividing the steel structure into component-component-parts hierarchy; accordingly, in the matching welding process information acquisition, the matching welding process information of each weld is obtained for any region; in the welding sequence information acquisition, the welding sequence information is obtained for any region; and in the welding process model generation and welding operation, a welding process model is generated for any region.

[0027] Furthermore, component information is extracted within each partition model (i.e., within any region). Components can be grouped, typically by type, function, or geometric characteristics. By extracting different types of components separately, the steel structure model within the partition model can be split. Component information within the hierarchical division of the steel structure includes part composition (i.e., information about the individual parts that comprise the component), as well as at least one of component type, location, size, function, and information about adjacent components. Geometric information can be extracted for each component, including size, location, shape, and connection relationships. This geometric information serves as the foundational data for the subsequent generation of the welding process model.

[0028] The dimensions of a component can be expressed in the following ways. For example, the length, width, and height of component C (taking a rectangular body as an example) can be calculated as follows: ; If it is a simple shape, it can be described by standard geometry; complex geometry may require triangulation (such as the volume of the component). can be obtained by grid integration): ; in is the area of ​​the triangle, For the corresponding height, is the i-th triangle mesh.

[0029] The position of the component can be represented by the coordinates of the construction center point, that is, the center point position of the component for: ; Furthermore, each component can be assigned a unique ID and bound to the partition information to maintain consistency during subsequent welding process information generation and management. The partitioned component information is exported in a pre-defined format (such as an IFC or JSON file), preserving both the geometry and partition information. This exported file structure ensures component consistency and traceability across different systems, facilitating subsequent matching with the process knowledge graph.

[0030] Furthermore, after the construction splitting in the partition model is completed, the part information in the partition model can be obtained through the part composition of the component combined with the partition model. The part information includes part material and adjacent part information, and also includes at least one of part position, part size and part type; the part position can be represented by the position coordinates of the part center point, and the adjacent parts are other part information adjacent to the part and having a connection relationship. The part size can be represented similarly to the component size.

[0031] Furthermore, the partition model can be deeply analyzed for the adjacent relationship between parts, and several adjacent parts can be combined to form a component. That is, all parts that make up a component are grouped according to the adjacent relationship, and each group forms a component, thereby forming a component-component-part composition hierarchy of the partitioned steel structure model. That is, each component of the partitioned steel structure model is composed of multiple components, and each component includes multiple parts. Component information includes component position and part composition, and also includes at least one of component type, component size, and adjacent component information. Component position can be represented by the coordinates of the component center point position, and component size can be represented similarly to component size. Part composition refers to the part information that makes up the component. Adjacent component information refers to information about other components that are adjacent to the component and have a connection relationship.

[0032] In some specific embodiments, for building BIM models, component types can be divided into primary structural components and secondary structural components. Primary structural components may include beams (such as H-beams and I-beams), which primarily bear vertical loads and bending moments and are formed by welding components together (e.g., welding a web to a flange); columns (such as H-beams and box columns), which primarily transmit axial loads and horizontal forces and may involve welding multiple components (e.g., a box column welded from four steel plates); and trusses (such as roof trusses and pipe trusses), which are primarily composed of members (such as angle steel and round tubes) welded or bolted together at nodes and are used for long-span structures.

[0033] Secondary structural components may include supporting components (such as column supports, roof supports, etc.), which are mostly steel sections or round steel, used to enhance structural stability and need to be welded to the main components; node connection plate components, which are used for connection between components (such as beam-column connection plates) and are key force conversion components.

[0034] Part types may include: steel plate parts, such as rectangular steel plates (used for beam / column webs and flanges), special-shaped steel plates (node ​​plates, stiffeners); steel section parts, such as angle steels, channel steels, I-beams, H-beams, round tubes, square tubes, etc.; connection and construction parts, such as welded joint parts: the edges of steel plates after groove processing (used for butt welding and fillet welding); fastener-related parts, such as bolt hole plates (used for bolt-weld mixed connections), welds, etc.; special function parts; cast steel node parts, such as complex curved surface parts (such as cast steel parts at the intersection of multiple rods); support parts, such as column foot plates, anchor bolt connection pads (key parts connected to the foundation), etc.

[0035] Component types may include: H-shaped steel column components, which are formed by submerged arc welding of flange plates and web plates (part-level welding to form basic structural units); truss member components, which are welded with angle steel and node plates to form truss web members or chord members; beam-column node components, which are connected to the main components by secondary welding of components such as beam end connecting plates and column body stiffeners (such as node welding of steel columns and steel beams); box-type column assembly components, which are formed by secondary welding of multiple prefabricated components (such as side panels and partitions) to form a complete box-type column; functional components, such as crane beam splicing components, which are splicing sections of segmented crane beams, including splicing plates and welded joints as well as supporting node components, which are used to support ear plate components connected to the frame and require secondary welding and fixation.

[0036] The material of the parts can be steel models, such as carbon structural steel: Q235B (used for secondary parts, such as purlins and stiffeners); low-alloy high-strength structural steel: Q355B / C / D (used for main load-bearing parts, such as webs and flanges of beams / columns); weathering steel: Q355NH (used for parts in open-air environments, such as node plates of outdoor steel structures); stainless steel: 304 (used for corrosion-resistant parts, such as interior decorative steel structures); cast steel: GS-20Mn5 (used for cast steel parts of complex nodes), etc.

[0037] Component type, part type and part material are usually existing information designed in the BIM model. The component type can be designed according to the specific part composition and structural function.

[0038] Furthermore, the method for establishing the welding process knowledge graph includes: Collect welding related data and define entities and matching relationships; Extract knowledge of entities and matching relationships from welding-related materials, and connect the extracted entities through matching relationships to form triple structure information; The triple structure information is stored in the graph database to form a welding process knowledge graph.

[0039] Specifically, refer to Figure 4 ,When constructing the welding process knowledge graph, data collection and organization are carried out first, as follows: In the process of building a welding process knowledge graph, it is first necessary to comprehensively collect and organize various relevant materials. These materials mainly include welding manuals, welding specifications, process cards, industry standards, and actual welding cases.

[0040] A welding manual is a comprehensive document containing detailed information about various welding processes, providing the basic theory and application of each welding method. It typically includes a definition of the welding process, applicable materials, welding sequence, welding parameters, and their selection principles. For example, a manual might detail the application scope, equipment requirements, and practical procedures for gas shielded welding (MIG / MAG) and argon arc welding (TIG).

[0041] Welding specifications are guidance documents published by international and national standards organizations that outline welding process requirements and quality control standards. These specifications ensure the safety and reliability of welding and cover topics such as the classification of welded joints, the selection of welding consumables, and the evaluation criteria for welding procedures. For example, the ISO 3834 series of standards provides a framework for welding quality, while AWS D1.1 focuses on welding of steel structures.

[0042] A procedure sheet is a document generated for a specific welding operation, documenting the parameters and requirements for the welding process. These parameters include welding current, voltage, welding speed, welding materials, and more, typically presented in a table format. A procedure sheet provides specific guidance for welding operations and ensures a standardized welding process.

[0043] Then the triples of the knowledge graph are constructed, more specifically: After completing the information collection, the next step is to construct a knowledge graph, which is mainly achieved by defining entities and the relationships between them. In this embodiment, the welding process knowledge graph described in the matching welding process information acquisition includes entity nodes and edges connecting entity nodes with matching relationships; the entity nodes include welding process nodes, welding equipment nodes, welding material nodes, welding parameter nodes, and environmental parameter nodes, and also include at least one of weld type nodes, weld geometry parameter nodes, and welding object material nodes; the matching relationships between the entity nodes include multiple of the equipment used in the process, welding materials applicable to the process, welding materials applicable to the equipment, welding parameters applicable to the process, environmental parameters applicable to the process, weld types applicable to the process, weld geometry parameters corresponding to welding parameters, and objects applicable to the process.

[0044] The equipment used in the process can represent the matching relationship between the welding process and the corresponding welding equipment, and can be connected between the welding process node and the welding equipment node, indicating that the welding process can use the corresponding welding equipment; the welding materials applicable to the process represent the matching relationship between the welding process and the corresponding welding materials, and can be connected between the welding process node and the welding material node; the welding materials applicable to the equipment can represent the matching relationship between the welding equipment node and the welding material node; the welding parameters applicable to the process can represent the matching relationship between the welding process node and the welding parameter node; the environmental parameters applicable to the process can represent the matching relationship between the welding process node and the environmental parameter node; the weld type applicable to the process can represent the matching relationship between the welding process node and the weld type node; the welding parameters corresponding to the weld geometry parameters can represent the matching relationship between the welding parameter node and the weld geometry parameter node; the object applicable to the process can represent the matching relationship between the welding process node and the welding object material node.

[0045] For example, the main entity definitions are: (1) Welding Process: defines different types of welding processes, such as gas shielded welding, argon arc welding, laser welding, etc. (2) Welding Equipment: records the type of welding equipment used, including welding machines, welding guns, etc., with specific brand, model and performance parameters. (3) Welding Material: lists different types of welding materials, including welding wire, welding rods, filler metals, etc., and records their chemical composition and physical properties. (4) Welding Parameters: includes important parameters such as current, voltage, welding speed and welding time used in the welding process, which directly affect the quality of welding. (5) Welding Environment: considers external conditions during the welding process, such as temperature, humidity, air flow velocity, etc. These factors have a significant impact on the welding effect. (6) Weld information, such as at least one of the weld type node, weld geometry parameter node and weld object material node.

[0046] Relationship Definition: (1) Uses: Describes the relationship between welding processes and welding equipment. For example, a specific welding process requires a specific type of welding equipment. (2) Applicable Process: Refers to the welding process that a certain welding material can be applied to, ensuring that the material selection matches the welding method. (3) Applicable Material: The type of material that a certain welding equipment can be applied to, ensuring the compatibility of the equipment and materials. (4) Influencing Factors: The impact of environmental factors on the welding process, helping to understand the problems that may arise during the welding process and their solutions. (5) Corresponding weld effects, etc.

[0047] Furthermore, the construction of these entities and relationships can form the triple structure of the knowledge graph: ; For example, if welding process A uses welding equipment B, it can be expressed as: ; After that, the process of constructing the knowledge graph is as follows: Knowledge extraction is the first step in constructing the knowledge graph, mainly extracting relevant entities and relationships from the collected literature and materials through natural language processing (NLP) techniques.

[0048] First is data collection. For the selection of data sources, collect text data from professional books, standard documents, industry journals, and online databases in the welding field. These data sources usually contain rich information on welding techniques, material properties, and equipment. Ensure that the collected data is stored in a structured or semi-structured format, such as PDF, Word documents, HTML web pages, etc., for subsequent processing. Clean the collected text data, removing irrelevant information, advertisements, page numbers, and other miscellaneous items to improve the coherence and integrity of the text.

[0049] Furthermore, named entity recognition is the process of identifying entities with specific meanings in the text. Select a suitable NLP tool or library, such as SpaCy, NLTK, or Hugging Face Transformers, for subsequent entity recognition work.

[0050] Furthermore, preprocess the text. Clean and normalize the text, removing irrelevant characters and formats to keep the text clean. Decompose the text into words or phrases, including removing stop words (such as "of", "is", etc.) and punctuation marks. Perform stemming or lemmatization to unify the representation of entities.

[0051] Furthermore, prepare a labeled training dataset, which should contain labeled instances of entities such as welding processes, equipment, and materials. For example, clearly mark "MIG welding" as a welding process and "stainless steel" as a material. The loss function in the process of training the NER model can be expressed as the following formula: ; where N is the number of samples, is the true label (1 indicates that the sample is a positive example, 0 indicates a negative example). is the probability predicted by the model.

[0052] Furthermore, apply the trained model to the text to automatically identify welding processes, equipment, materials, etc., and generate a list of entities.

[0053] At the same time, the relationship extraction of the sentence where the entity is located is performed to identify the relationship between the entities. Dependency parsing technology is used to analyze the grammatical structure of the sentence to identify the relationship between the entities. For example, when analyzing the sentence "Argon is required to weld stainless steel with MIG", the relationship between the verb "use" and "welding process" and "material" is identified. Common relationship patterns are defined and these relationships are identified from the text through methods such as regular expressions. For example, if the sentence contains "use" and "need", the relationship between the welding process and the equipment or material can be inferred. Using the identified entities and relationships, a triple (entity 1, relationship, entity 2) is constructed. For example: Triplet 1: ("CO2 gas shielded welding", "use" "carbon dioxide"); Triplet 2: ("Q355B", "applicable" "CO2 gas shielded welding"); Furthermore, the extracted triples are stored in a graph database, such as Neo4j, to support highly associative data storage and complex query analysis. In the graph database, the data model is designed to reflect the structure of welding knowledge, including entity node types and relationship types.

[0054] The constructed triples are then imported into the database using the graph database's API or import tool. Each triple is converted into nodes and relationships, forming a graph structure. In a graph database, complex queries can be performed using query languages ​​such as Cypher. For example, one can query for welding processes and their parameters suitable for a specific material, thus supporting decision-making and optimizing welding processes.

[0055] Furthermore, the method for establishing the welding process knowledge graph also includes: Collect and obtain welding-related update information at preset intervals; Extract knowledge of entities and matching relationships from updated welding-related data to form updated triple structure information; The welding process knowledge graph is updated and maintained using the updated triple structure information.

[0056] Specifically, an automated literature retrieval system should be established to regularly collect the latest welding-related literature and standards from academic databases and industry websites. This can be achieved using web crawler technology to retrieve updated information from designated online databases. Information from new literature should be subjected to a knowledge extraction process, including NER and relation extraction, to extract and integrate the new information into the existing knowledge graph. The newly extracted information should be compared with the existing knowledge graph to identify duplicate and conflicting knowledge and merge or update it.

[0057] In some specific embodiments, the theoretical weld information in the matching welding process information acquisition includes at least one of theoretical weld geometric parameters, theoretical weld type, and weld application object material; wherein the theoretical weld geometric parameters include at least one of weld joint vector, weld length, weld angle, weld width, weld depth, and weld cross-sectional area.

[0058] Specifically, the coordinates of the weld joints are obtained from the building BIM model. The theoretical geometric parameters of the welds are then derived from these coordinates. This automatically generates welds based on the BIM model. In this step, the system enters the automatic weld generation phase. In this phase, the system leverages the geometric topology of the steel structure in the BIM model to automatically identify the locations of weld joints—the endpoints of the connection between two parts—and generates the weld geometry and parameters. By combining welding process knowledge with the generated theoretical weld information, the system automatically matches appropriate welding parameters (such as welding speed and current).

[0059] The weld geometry is generated as follows: The geometric generation of welds involves multiple aspects such as the identification of weld joints and the calculation of weld geometric parameters. First, the weld joints need to be identified. The geometric features of the weld joints are extracted from the BIM detailed design model. Assume that the two endpoints of the joint are and , their coordinates are: These two points usually correspond to the connection points of adjacent parts in a welded structure, and the system will automatically identify these joint points in the BIM model.

[0060] Furthermore, the weld joint vector is calculated. It can be expressed by the following formula: ; This vector reflects the direction and position of the weld joint and is the basis for calculating the weld length and angle.

[0061] Furthermore, the weld length is calculated. The calculation uses vector The modulus value is: ; The length of the weld is an important basis for determining welding parameters and directly affects the materials and time required for welding.

[0062] Furthermore, the weld angle is calculated. It is a key factor affecting welding quality, which can be determined by the weld vector and the reference surface normal vector. The angle is calculated as: ; in, It is the normal vector of the reference surface (such as the surface of the welded workpiece). The correct calculation of the angle can ensure the proper positioning of the weld, thereby improving the strength and stability of the weld.

[0063] Furthermore, define the weld width and depth ,width and depth Directly determines the cross-sectional area of ​​the weld The calculation formula is: ; The design of weld width, depth and cross-sectional area parameters aims to ensure that the weld has sufficient strength and toughness. The design and calculation can be carried out according to the specific location of the weld and the material and type of the application object according to industry standards. The acquisition of weld width, depth and cross-sectional area parameters can be calculated based on the BIM model, or calculated during the BIM model design phase and obtained as known information of the BIM model, with no specific restrictions. Weld types include butt welds, fillet welds, race welds, groove welds, etc., which can also be determined and obtained based on the specific conditions of the welds in the BIM model, or set during the BIM model design phase and obtained as known information of the BIM model, with no specific restrictions. The design and generation of the width and depth of each weld type can be calculated and obtained through the industry standard calculation process when the weld application object, i.e. the part, is determined. The specific examples are as follows: Weld width and depth are calculated based on groove design parameters. For butt welds (e.g., V-groove): width = groove opening width + 2 × gap; depth = plate thickness - blunt edge height. For fillet welds, theoretical depth = leg size × sin45° (isosceles fillet welds). In actual projects, leg size is often used as a direct representation. Parameters such as groove width, gap, and blunt edge height must be defined in the BIM model (either manually entered or generated through parametric modeling).

[0064] Furthermore, there should be at least one entity node in the welding process knowledge graph that corresponds to the theoretical weld information, that is, the same category of information, so that the corresponding information can be retrieved and matched in the knowledge graph. The specific matching of welding process information is as follows: After the weld is generated, the system needs to match the welding process information according to the constructed process knowledge graph. The matching welding process information includes welding process type, welding equipment, welding materials, welding parameters and environmental parameters. It can also include at least one of the weld type, weld geometry parameters and weld application object material. Welding parameters include welding current ,Voltage , welding speed Etc. are all key factors affecting welding quality.

[0065] Furthermore, when optimizing the welding sequence of components according to the welding path length in the welding sequence information acquisition, the optimization object is the component welding sequence vector , the optimization goal is to minimize the welding path length, welding path length ;in represents the component ranked as i, and d represents the spatial distance; refer to Figure 5 , that is, welding sequence optimization and path generation are also required, as follows: After completing the assembly of components, the next step is to optimize the welding sequence and path generation. First, the welding sequence vector needs to be defined. Let the welding sequence vector be The goal of optimizing the welding sequence is to reduce the moving path during welding, which can usually be expressed as Quantification. Among them, Representation components and components The welding path length between the two paths is: .

[0066] The above-mentioned component welding sequence optimization can determine the order of welding of the internal welds of the components, that is, the components ranked higher in the welding sequence vector are assembled and welded first, and after all the components are assembled and welded separately, the welding order of the welds between the components needs to be determined. This embodiment proposes that obtaining welding sequence information also includes: analyzing the load-bearing level of each component in the building BIM model, and determining the welding order of the welds between components in order from high to low load-bearing levels. That is, in any area, the load-bearing level of the components is analyzed based on the BIM model, and the secondary welds on the components with higher load-bearing levels are welded first, thereby completing the welding of the secondary welds between the components in sequence.

[0067] The load-bearing grade ranking of components can be obtained based on the position and force analysis of the components in the BIM model. For example, the first-level load-bearing components, i.e., components with high load-bearing grades, can be the key components of the main force path: such as the column base connection components, which directly transfer the column bottom load to the foundation, involving the welding of the base plate, anchor bolts and column body (secondary welding, complex force); beam-column rigid connection components, which are fully penetrated welded node components that bear the bending moment at the beam end (such as the combination of flange butt welds + web fillet welds); truss chord connection components, which are welded components between the node plate and the chord.

[0068] Secondary load-bearing components (secondary load-bearing or combined load-bearing components): such as the connection components between supports and frames, the welded components between ear plates and support rods (transmitting horizontal forces, secondary welding); component splicing components, the welded components of the connecting plates at the spliced ​​steel beams; box column partition components, the welded components between the internal partitions and side panels of box columns. Tertiary load-bearing components (structural or auxiliary load-bearing components): such as the connection components between purlins and main components, the welded or bolted components between purlin support plates and purlins (transmitting roof loads, secondary welding); stiffener components, the welded components of beam web stiffeners; and corner brace connection components, the welded components between corner braces and purlins (auxiliary anti-instability, bearing less force).

[0069] Furthermore, the welding process model generation and welding operation also include: calculating the theoretical heat input of each weld based on the matching welding process information of each weld, adjusting the welding parameters until the heat input is within the first preset threshold range when the theoretical heat input exceeds the preset heat input threshold, and obtaining the welding parameter information after the heat input of each weld is corrected; and generating a welding process model based on the welding parameter information after the heat input is corrected.

[0070] heat input It is an important parameter in the welding process, and its formula is: ; This formula calculates the heat transfer during welding via voltage, current and welding speed, which has a direct impact on the melting and solidification processes of the weld.

[0071] Furthermore, different first preset thresholds can be set for primary welds within components and secondary welds between components, allowing for separate heat input correction determinations. This embodiment takes into account the significant differences in geometric parameters and functions between primary and secondary welds, leading to significant differences in heat input. The heat input of secondary welds can be greater than that of primary welds. Based on this, preset thresholds can be set for separate heat input determinations. Specifically, the preset threshold set for secondary welds can be higher than the preset threshold set for primary welds to better reflect actual conditions and avoid overcorrection. Welding parameters corrected by heat input can help prevent excessive heat input from affecting welding performance.

[0072] Then the welding process model is generated as follows: After the weld geometry and welding parameters are determined, the system integrates this information into a complete welding process model. This model includes information such as weld geometry, welding sequence, required materials and equipment. In this step, the system integrates the welding sequence, process parameters, required materials and equipment for each weld into a three-dimensional model to form a complete welding process model. This model not only provides clear guidance for actual production, but also allows for optimization and adjustment based on real-time feedback during the subsequent welding process. For example, deviations in parameters such as temperature and speed may occur during the welding process. The system can continuously optimize the welding process by comparing it with the knowledge graph to achieve higher welding quality and production efficiency.

[0073] The welding process information is integrated as follows: Welding process model Contains complete information about each weld, including weld geometry, process parameters, and related equipment and materials. The following formula can be used to represent the composition of the welding process model: ; in: Indicates the length of the weld, Indicates the angle of the weld, Indicates the heat input during welding; Indicates welding current; Indicates welding voltage; Indicates welding speed, Indicates welding equipment; Represents welding consumables. By integrating the geometric information and process parameters extracted in the previous steps, each weld in the welding model will be associated with the required equipment and welding consumables.

[0074] Furthermore, during the generation of the welding process model and the actual welding operation, the welding parameters in the matched welding process information are corrected according to the deviation between the actual temperature and the theoretical temperature corresponding to the matched welding process information, and the actual welding operation is performed based on the welding parameters after the temperature correction; This embodiment takes into account that in actual welding, welding parameters are affected by changes in environmental conditions (such as temperature and humidity). Therefore, it is crucial to introduce a feedback mechanism into the process model to achieve real-time parameter optimization. Considering the influence of temperature, the adjustment of welding current and welding speed can be achieved using the following formula: ; in: is the initial welding current; is the initial welding speed; and are temperature adjustment coefficients, which can be obtained experimentally to reflect the sensitivity of the welding process to temperature changes; This is the optimal temperature for the welding process matched to the welding process knowledge graph. The introduction of a real-time feedback mechanism ensures that welding parameters can adapt to changes in the on-site environment, thereby improving welding quality.

[0075] And / or, during the generation of the welding process model and the actual welding operation, the deviation between the actual weld information and the theoretical weld information is monitored. When the deviation exceeds a second preset threshold, the welding parameters in the matching welding process information are feedback-adjusted, and the actual welding operation is performed based on the feedback-corrected welding parameters.

[0076] This embodiment takes into account the fact that existing automatic welding operations lack real-time monitoring and feedback mechanisms, making it difficult to ensure welding quality. Furthermore, feedback optimization of the welding process is proposed, as follows: In order to maintain the accuracy of the welding process model, the system needs to judge the process deviation by comparing the real-time monitoring parameters with the theoretical parameter values. For example, the deviation between the actual value and the theoretical value of at least one of the weld width, depth, length and cross-sectional area can be monitored. When the deviation exceeds the set threshold, The system will automatically provide feedback and update the welding parameters. The calculation can be expressed by the following formula: ; When the absolute value meets the conditions: ; At this time, the system can update the welding parameters based on empirical values. For example, when the actual value of the weld width is less than the theoretical value, the welding current or voltage can be increased, or the welding speed can be reduced to adjust the welding parameters. When the deviation exceeds the threshold, an alarm can be issued to manually adjust the welding parameters. There is no specific limitation.

[0077] When any weld is completed, the actual welding process information corresponding to that weld can be used as an instance to update the welding process knowledge graph. By updating the relevant welding process data items in the knowledge graph, the welding process model can be self-optimized as working conditions change. This mechanism ensures the adaptability and flexibility of the welding process, helping to improve welding quality and reduce rejection rates.

[0078] The resulting welding process model file is available in both IFC and JSON formats. The IFC file details the 3D geometry of each component, while the JSON file stores the BOM structure and additional process information. The two are linked via a unique ID, allowing each part and component in the IFC file to seamlessly match the information in the JSON file.

[0079] Furthermore, this embodiment provides a steel structure welding process system based on a knowledge graph, the system including a memory and a processor, the memory storing a computer program, and the processor executing any one of the above-mentioned steel structure welding process methods based on a knowledge graph when executing the computer program.

[0080] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A steel structure welding process method based on knowledge graph, characterized in that: include: Steel structure hierarchical division: Obtain the component information of the steel structure in the building and the part information of each component, and establish component information based on the adjacent relationship of the parts, dividing the steel structure into a component-component-part hierarchy; Matching welding process information acquisition: Obtain theoretical weld information of each weld within and between components in the steel structure, and obtain matching welding process information for each weld based on the theoretical weld information and the pre-established welding process knowledge graph; Welding sequence information acquisition: Optimize the welding sequence of components based on the welding path length, and obtain welding sequence information based on the optimized component welding sequence; Welding process model generation and welding operation: combining the welding sequence information and the matching welding process information of each weld to generate a welding process model, and performing actual welding operations based on the welding process model.

2. The steel structure welding process method based on knowledge graph according to claim 1, characterized in that: The component information in the hierarchical division of steel structure components includes part composition, and also includes at least one of component type, component position, component size, component function, and adjacent component information; part information includes part material and adjacent part information, and also includes at least one of part position, part size, and part type; component information includes component position and part composition, and also includes at least one of component type, component size, and adjacent component information; The hierarchical division of the steel structure also includes: dividing the building into multiple areas, and dividing the steel structure into component-part-part components for each area; accordingly, obtaining the matching welding process information for each weld in each area; and obtaining the welding sequence information for each area; Welding process model generation and welding operation generate welding process models for any area separately.

3. The steel structure welding process method based on knowledge graph according to claim 1, characterized in that: The welding process knowledge graph described in the matching welding process information acquisition includes entity nodes and edges connecting entity nodes with matching relationships; the entity nodes include welding process nodes, welding equipment nodes, welding material nodes, welding parameter nodes and environmental parameter nodes, and also include at least one of weld type nodes, weld geometry parameter nodes and welding object material nodes; the matching relationships between the entity nodes include multiple of process equipment, process applicable welding materials, equipment applicable welding materials, process applicable welding parameters, process applicable environmental parameters, process applicable weld type, welding parameter corresponding weld geometry parameters, and process applicable object material.

4. The steel structure welding process method based on knowledge graph according to claim 3 is characterized in that: The method for establishing the welding process knowledge graph includes: Collect welding related data and define entities and matching relationships; Extract knowledge of entities and matching relationships from welding-related materials, and connect the extracted entities through matching relationships to form triple structure information; The triple structure information is stored in the graph database to form a welding process knowledge graph.

5. The steel structure welding process method based on knowledge graph according to claim 4 is characterized in that: The method for establishing the welding process knowledge graph also includes: Collect and obtain welding-related update information at preset intervals; Extract knowledge of entities and matching relationships from updated welding-related data to form updated triple structure information; The welding process knowledge graph is updated and maintained using the updated triple structure information.

6. The steel structure welding process method based on knowledge graph according to claim 3 is characterized in that: The theoretical weld information in the matching welding process information acquisition includes at least one of theoretical weld geometric parameters, theoretical weld type and weld application object material; wherein the theoretical weld geometric parameters include at least one of weld joint vector, weld length, weld angle, weld width, weld depth and weld cross-sectional area.

7. The steel structure welding process method based on knowledge graph according to any one of claims 1 to 6, characterized in that: When optimizing the welding sequence of components according to the welding path length during welding sequence information acquisition, the optimization object is the component welding sequence vector. , the optimization goal is to minimize the welding path length, welding path length ;in represents the component ranked as i, and d represents the spatial distance; Obtaining the welding sequence information also includes: analyzing the load-bearing level of each component in the building, and determining the welding sequence of the welds between the components in descending order of the load-bearing level.

8. The steel structure welding process method based on knowledge graph according to any one of claims 1 to 5, characterized in that: The welding process model generation and welding operation also include: calculating the theoretical heat input of each weld based on the matching welding process information of each weld, adjusting the welding parameters until the heat input is within the first preset threshold range when the theoretical heat input exceeds the preset heat input threshold, and obtaining the welding parameter information after the heat input of each weld is corrected; and generating a welding process model based on the welding parameter information after the heat input is corrected.

9. The steel structure welding process method based on knowledge graph according to claim 8, characterized in that: Generating a welding process model and performing welding operations, during actual welding operations, correcting welding parameters in the matching welding process information according to the deviation between the actual temperature and the theoretical temperature corresponding to the matching welding process information, and performing actual welding operations based on the temperature-corrected welding parameters; And / or, during the generation of the welding process model and the actual welding operation, the deviation between the actual weld information and the theoretical weld information is monitored. When the deviation exceeds a second preset threshold, the welding parameters in the matching welding process information are feedback-adjusted, and the actual welding operation is performed based on the feedback-corrected welding parameters.

10. A steel structure welding process system based on knowledge graph, characterized in that: The system includes a memory and a processor, the memory stores a computer program, and the processor executes the steel structure welding process method based on the knowledge graph described in any one of claims 1 to 9 when executing the computer program.

Images