A Method and System for Constructing a Knowledge Graph of a BIM Model and Semantic Query Control

Abstract

The present invention provides a method and system for constructing a knowledge graph of a BIM model and semantic query control, which is characterized by including the following steps: BIM model data preparation and attribute extension, IFC data parsing and knowledge graph modeling rule definition, knowledge graph automatic construction tool development, storage and management of the Neo4j graph database, natural language processing and large model semantic understanding, Cypher query statement generation and IOT interface call, query and control interaction process. The method for constructing a knowledge graph of a BIM model and semantic query control provided by the present invention realizes unified query and control from building space information to the semantic level by converting the BIM model into a knowledge graph and utilizing natural language processing technology and large model capabilities.

Description

Technical Field

[0001] The present invention relates to the field of building information models, and particularly to a method and system for constructing a knowledge graph and semantic query control of a BIM model. Background Art

[0002] The technical field involved in the present invention is the integrated application of building information models and the Internet of Things, as well as semantic query and control technologies based on graph databases and knowledge graphs. The present invention is mainly applied to scenarios such as smart buildings, smart factories, and smart campuses. By converting the building space, equipment information, and relationships contained in the BIM model into a knowledge graph that can be semantically queried, the query and control of equipment status based on natural language or semantic descriptions can be realized.

[0003] Currently, in the field of building facility management or smart buildings, BIM technology has become the main means of digitally describing building, mechanical, electrical, plumbing, and spatial information. Usually, BIM data is stored and exchanged in standard formats such as IFC. However, traditional BIM data management mainly relies on files and proprietary databases. Although the IFC standard has defined rich attribute, relationship, and semantic information, in actual applications, IFC files are not suitable for efficient semantic retrieval and real-time interaction. At the same time, in order to achieve refined management and intelligent control, various types of equipment in modern buildings are connected to the Internet of Things system and associated with their status data and control instructions through unique ID identifiers. However, most current solutions require relatively cumbersome manual mapping and data retrieval operations between the BIM platform and the IOT platform, lacking high-level semantic abstraction and customizable query methods.

[0004] Existing background art solutions relatively close to the present invention include:

[0005] Semanticization of IFC data and SPARQL query: Some researchers use the IFC model as a basis, convert it into an OWL ontology, and then store it in a triple database through RDF triples and support SPARQL query. For example, the literature "Pauwels, P., Zhang, S., & Lee, Y. C. (2017). Semantic web technologies in AEC industry: A literature overview. Automation in Construction, 73, 145–165." reviews the existing research on using semantic web technologies to process IFC data. These methods have achieved semantic representation and query of IFC data to a certain extent, but are usually limited to the semantic models of RDF / OWL, and the query mainly relies on semantic web standard query languages such as SPARQL, lacking optimization for natural language dialogue scenarios and having limited support for the control of IOT devices.

[0006] System Docking Solution Based on the Integration of BIM and IOT: There are already some commercial or research software that record device information in the BIM platform and conduct data interaction with the IOT platform through external interfaces. For example, in existing device management systems or data visualization platforms, users can manually input the device ID in the BIM model and then query or control the corresponding device in an independent IOT management platform. However, these solutions often lack automated semantic mapping and natural language query capabilities. Users still need to switch between multiple platforms and remember or search for the device ID, rather than being able to directly use semantic descriptions (such as "the air conditioner status in the computer room") for quick retrieval and control.

[0007] Although the above background technology solutions have provided a certain basis for the association between BIM data and Internet of Things information, there are still the following deficiencies in actual applications:

[0008] Insufficient Semantic Level and Complex Retrieval Method: Existing semantic methods based on means such as IFCOWL can convert building information into a semantic web data structure and use SPARQL for query. However, SPARQL is not user-friendly for ordinary users. Users need to have a high technical threshold when performing queries, with a clear understanding of specific classes, attributes, and relationships. This makes it difficult to quickly locate the required device or space information in actual scenarios. In short, this method does not fully utilize the advantages of natural language processing and intelligent dialogue, resulting in a cumbersome query process and high costs.

[0009] The Connection between BIM and IOT Device Control is not Smooth Enough: Existing BIM and IOT integration solutions often require manually recording the device ID in the BIM model and querying or controlling the device in an independent IOT platform. When users perform spatial positioning and device status query or control, they need to first find the ID corresponding to the device through BIM tools and then switch to the IOT platform for relevant operations. This process lacks a unified semantic level abstraction and cannot directly use "room name" or "device function" as the query entry, resulting in a long operation chain and low intelligence level.

[0010] Lack of Direct Support for Natural Language Semantic Requirements: Existing technologies do not fully utilize natural language processing technology. Users cannot directly obtain or control relevant devices through simple natural language descriptions (such as "Please query the air conditioner status in the computer room" or "Please dim the lights in the meeting room"). This defect still requires professionals or external systems to convert natural language requirements into relevant technical query statements and control instructions in actual use, reducing the convenience and efficiency of human-computer interaction.

[0011] Therefore, it is necessary to provide a method for constructing a knowledge graph and semantic query control of a BIM model to solve the above technical problems. Summary of the Invention

[0012] The present invention provides a method for constructing a knowledge graph and semantic query control of a BIM model, which solves the problems of insufficient semanticization, cumbersome query and control processes, and lack of natural language interaction support in the integration of building information and Internet of Things data in the prior art.

[0013] To solve the above technical problems, a method and system for constructing a knowledge graph and semantic query control of a BIM model provided by the present invention include the following steps:

[0014] S1. BIM model data preparation and attribute extension: In the building information model design and management software, perform the following operations to prepare and extend BIM model data;

[0015] S11. PMID attribute definition: Define an additional attribute PMID for each device in Revit, and fill in the PMID of each device to make it correspond to the device ID in the IOT system;

[0016] S12. Spatial topology and device connection relationship definition: Define the spatial connection relationships among various spaces, pipelines, and devices in Revit, and ensure that the connection relationships among entities conform to the actual building structure and device layout through the built-in relationship management function of Revit;

[0017] S2. IFC data parsing and knowledge graph modeling rule definition: After exporting the prepared BIM model as an IFC format file, perform IFC data parsing and knowledge graph modeling;

[0018] S21. Use the xBIM parsing library: Use xBIM as the IFC data parsing tool to parse the entity, attribute, and relationship information in the IFC file;

[0019] S22. Define the knowledge graph structure: Node types, relationship types, and attribute mappings;

[0020] S3. Development of a knowledge graph automatic construction tool: Develop a set of automated tools to achieve efficient conversion from an IFC format BIM model to a knowledge graph;

[0021] S31. IFC parsing module: Use the xBIM parsing library to read the IFC file and extract all relevant entities, attributes, and their relationships;

[0022] S32. Data mapping and conversion engine: According to the pre-defined knowledge graph modeling rules, convert the parsed IFC entity types, attributes, and relationships into nodes and relationships in the Neo4j graph database;

[0023] S33. Batch Import Program: Use the Neo4j library to batch insert the converted data into the Neo4j graph database through the Bolt protocol, ensuring transaction management during the data import process to guarantee data consistency and integrity;

[0024] S4. Storage and Management of Neo4j Graph Database: Store the constructed knowledge graph in the Neo4j graph database;

[0025] S41. Node Storage: Entities such as various spaces, devices, pipelines, etc. are stored as nodes, including their attributes;

[0026] S42. Relationship Storage: Various space association relationships, device association relationships, and space - device relationships are stored as relationships, defining the connection methods between nodes;

[0027] S43. Indexes and Constraints: Create indexes for commonly queried fields to improve query performance, and set unique constraints to ensure that the PMID and IOT ID of each device are unique in the graph database;

[0028] S5. Natural Language Processing and Large - Model Semantic Understanding: To achieve the automatic generation of Cypher query statements from natural language, the system integrates a natural language processing module and a large model for semantic understanding;

[0029] S51. Semantic Parsing, Intent Recognition, Entity Extraction, and Context Understanding;

[0030] S6. Generation of Cypher Query Statements and Invocation of IOT Interfaces: According to the semantic parsing results, the system automatically generates corresponding Cypher query statements and realizes device status query or control through the IOT interface;

[0031] S61. Automatic Generation of Cypher Query Statements: Dynamically construct Cypher query statements based on the parsed intent and entity information;

[0032] S62. Invocation of IOT Interfaces;

[0033] S63. Result Feedback;

[0034] S7. Query and Control Interaction Process.

[0035] Preferably, the PMID attribute in S11 is used for the subsequent association between devices and the IOT system in the knowledge graph.

[0036] Preferably, the xBIM in S21 supports reading and operating on IFC files and can extract the required building element and their relationship data.

[0037] Preferably, the node types in S22 include the following steps:

[0038] S22a1, Space: Represents various spaces in a building, such as rooms and corridors;

[0039] S22a2, Equipment: Represents various equipment, such as air conditioners, lighting, and sensors;

[0040] S22a3, Pipeline: Represents the pipeline system in a building;

[0041] S22a4, Sensor: Represents sensors used to monitor the status of equipment;

[0042] S22a5, Controller: Represents controllers used to control equipment.

[0043] Preferably, the relationship types in S22 include the following steps:

[0044] S22b1, Spatial association relationship: Used to connect IfcBuildingElement entities, representing the aggregation relationship between spaces.

[0045] S22b2, Equipment association relationship:

[0046] IfcRelFlowControlElements: Components under an external controller are associated with the elements they control through this relationship.

[0047] IfcRelConnectsPortToElement: Components under an internal controller are connected to IfcFlowSegment or IfcFlowFitting through this relationship;

[0048] IfcRelConnectsSegments: Pipe segments are connected to other pipe segments or pipe fittings through this relationship;

[0049] IfcRelConnectsPorts: Pipe fittings are connected to other pipe fittings or terminals through this relationship;

[0050] IfcRelConnectsToPipe or IfcRelConnectsToPort: Terminals are connected to the system through these two relationships;

[0051] IfcRelConnectsToElement: Energy conversion equipment and fluid storage equipment are connected to other building elements through this relationship;

[0052] IfcRelConnectsToProcess: Fluid moving equipment and fluid processing equipment are connected to the flow process through this relationship;

[0053] S22b3, Spatial Equipment Relationship: IfcRelContainedInSpatialStructure is used to represent the relationship that an equipment belongs to a certain space.

[0054] Preferably, the attribute mapping in S22: Extract the attribute information of ID, name, location coordinates, function description, and IOT corresponding ID from the IFC entity, and map it to the node attributes in the knowledge graph.

[0055] Preferably, the semantic parsing in S51: The user inputs a query or control request through the natural language interface;

[0056] Intention Recognition and Entity Extraction: Use a large model to parse the natural language input by the user, and identify the key information of the operation type, target space, and target equipment type;

[0057] Context Understanding: Combine the context information to process the user's complex query requirements, such as adjusting the specific parameters of the equipment status.

[0058] Preferably, the result feedback in S63: Feed back the queried equipment status information or the execution result of the control instruction to the user through the natural language interface.

[0059] A knowledge graph construction and semantic query control system for a BIM model, adopting the knowledge graph construction and semantic query control method of the BIM model, characterized by including: a BIM model data preparation and attribute extension module, an IFC data parsing and knowledge graph modeling rule definition module, a knowledge graph automatic construction tool development module, a storage and management module of the Neo4j graph database, a natural language processing and large model semantic understanding module, a Cypher query statement generation and IOT interface call module, and a query and control interaction process module. The input and output end of the BIM model data preparation and attribute extension module is connected to the input end of the IFC data parsing and knowledge graph modeling rule definition module, the output end of the IFC data parsing and knowledge graph modeling rule definition module is connected to the input end of the knowledge graph automatic construction tool development module, the output end of the knowledge graph automatic construction tool development module is connected to the input end of the storage and management module of the Neo4j graph database, the output end of the storage and management module of the Neo4j graph database is connected to the input end of the natural language processing and large model semantic understanding module, the output end of the natural language processing and large model semantic understanding module is connected to the input end of the Cypher query statement generation and IOT interface call module, and the output end of the Cypher query statement generation and IOT interface call module is connected to the input end of the query and control interaction process module.

[0060] Compared with the related technologies, a method for constructing a knowledge graph and semantic query control of a BIM model provided by the present invention has the following beneficial effects:

[0061] The present invention provides a method for constructing a knowledge graph and semantic query control of a BIM model. By converting the BIM model into a knowledge graph and utilizing natural language processing technology and the capabilities of large models, unified query and control from building spatial information to the semantic level are achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic structural diagram of a preferred embodiment of a method for constructing a knowledge graph and semantic query control of a BIM model provided by the present invention;

[0063] Figure 2 It is a schematic structural diagram of the knowledge graph;

[0064] Figure 3 It is a flowchart of query and control;

[0065] Figure 4 It is a schematic diagram of IFC relationship mapping. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] The present invention will be further described below with reference to the drawings and embodiments.

[0067] Please refer to Figure 1 、 Figure 2 、 Figure 3 and Figure 4 , where Figure 1 It is a schematic structural diagram of a preferred embodiment of a method for constructing a knowledge graph and semantic query control of a BIM model provided by the present invention; Figure 2 It is a schematic structural diagram of the knowledge graph; Figure 3 It is a flowchart of query and control; Figure 4 It is a schematic diagram of IFC relationship mapping. A method and system for constructing a knowledge graph and semantic query control of a BIM model include the following steps:

[0068] S1. Preparation and attribute extension of BIM model data: In the building information model design and management software, perform the following operations to prepare and extend the BIM model data;

[0069] S11. Definition of PMID attributes: In Revit, define an additional attribute PMID for each device and fill in the PMID of each device to make it correspond to the device ID in the IOT system;

[0070] S12. Definition of Spatial Topology and Equipment Connection Relationships: Define the spatial connection relationships among various spaces, pipelines, and equipment in Revit, and ensure that the connection relationships among entities conform to the actual building structure and equipment layout through the built-in relationship management function of Revit;

[0071] S2. Definition of IFC Data Parsing and Knowledge Graph Modeling Rules: After exporting the prepared BIM model to an IFC format file, perform IFC data parsing and knowledge graph modeling;

[0072] S21. Use the xBIM (eXtensible Building Information Modeling) Parsing Library: Use xBIM (eXtensible Building Information Modeling) as an IFC data parsing tool to parse entity, attribute, and relationship information in the IFC file;

[0073] S22. Define the Knowledge Graph Structure: Node types, relationship types, and attribute mappings;

[0074] S3. Development of Knowledge Graph Automatic Construction Tools: Develop a set of automated tools to achieve efficient conversion from an IFC format BIM model to a knowledge graph;

[0075] S31. IFC Parsing Module: Use the xBIM (eXtensible Building Information Modeling) parsing library to read the IFC file and extract all relevant entities, attributes, and their relationships;

[0076] S32. Data Mapping and Conversion Engine: According to the pre-defined knowledge graph modeling rules, convert the parsed IFC entity types, attributes, and relationships into nodes and relationships in the Neo4j graph database;

[0077] S33. Batch Import Program: Use the Neo4j class library to batch insert the converted data into the Neo4j graph database through the Bolt protocol, ensure transaction management during the data import process, and guarantee data consistency and integrity;

[0078] S4. Storage and Management of the Neo4j Graph Database: Store the constructed knowledge graph in the Neo4j graph database;

[0079] S41. Node Storage: Entities such as various spaces, equipment, and pipelines are stored as nodes, including their attributes;

[0080] S42. Relationship Storage: Various spatial association relationships, equipment association relationships, and spatial equipment relationships are stored as relationships, defining the connection methods between nodes;

[0081] S43. Indexing and Constraints: Create indexes for frequently queried fields to improve query performance, and set unique constraints to ensure the uniqueness of PMID and IOT ID for each device in the graph database;

[0082] S5. Natural Language Processing and Large Model Semantic Understanding: To achieve the automatic generation of Cypher query statements from natural language, the system integrates a natural language processing module and a large model for semantic understanding;

[0083] S51. Semantic Parsing, Intent Recognition, Entity Extraction, and Context Understanding;

[0084] S6. Cypher Query Statement Generation and IOT Interface Invocation: According to the semantic parsing results, the system automatically generates corresponding Cypher query statements and realizes device status query or control through the IOT interface;

[0085] S61. Automatic Generation of Cypher Query Statements: Dynamically construct Cypher query statements based on the parsed intent and entity information;

[0086] S62. IOT Interface Invocation;

[0087] S63. Result Feedback;

[0088] S7. Query and Control Interaction Process.

[0089] IOT Interface Invocation in S62: Query Operation:

[0090] Use the queried IOT ID to obtain the real-time status information (such as temperature value) of the device through the API interface of the IOT platform.

[0091] Control Operation: Generate corresponding control commands according to the user's instructions and send control instructions through the API interface of the IOT platform. For example, adjust the air conditioner temperature.

[0092] The PMID attribute in S11 is used for the subsequent association of devices and the IOT system in the knowledge graph.

[0093] The xBIM (eXtensible Building Information Modeling) in S21 supports reading and operating IFC files and can extract the required building element and their relationship data.

[0094] The node types in S22 include the following steps:

[0095] S22a1. Space: Represents various spaces in the building, such as rooms and corridors;

[0096] S22a2, Equipment: Represents various types of equipment such as air conditioners, lighting, and sensors;

[0097] S22a3, Pipeline: Represents the pipeline system in a building;

[0098] S22a4, Sensor: Represents sensors used to monitor the status of equipment;

[0099] S22a5, Controller: Represents controllers used to control equipment.

[0100] The relationship types in S22 include the following steps:

[0101] S22b1, Spatial association relationship: Used to connect IfcBuildingElement entities, representing the aggregation relationship between spaces.

[0102] S22b2, Equipment association relationship:

[0103] IfcRelFlowControlElements: Components under an external controller are associated with the elements they control through this relationship.

[0104] IfcRelConnectsPortToElement: Components under an internal controller are connected to IfcFlowSegment or IfcFlowFitting through this relationship;

[0105] IfcRelConnectsSegments: Pipe segments are connected to other pipe segments or pipe fittings through this relationship;

[0106] IfcRelConnectsPorts: Pipe fittings are connected to other pipe fittings or terminals through this relationship;

[0107] IfcRelConnectsToPipe or IfcRelConnectsToPort: Terminals are connected to the system through these two relationships;

[0108] IfcRelConnectsToElement: Energy conversion equipment and fluid storage equipment are connected to other building elements through this relationship;

[0109] IfcRelConnectsToProcess: Fluid moving equipment and fluid processing equipment are connected to the flow process through this relationship;

[0110] S22b3, Spatial Equipment Relationship: IfcRelContainedInSpatialStructure is used to represent the relationship that an equipment belongs to a certain space.

[0111] The attribute mapping in S22: Extract the attribute information of ID (such as PMID, IFC GUID), name, location coordinates, function description, and IOT corresponding ID from the IFC entity, and map it to the node attributes in the knowledge graph.

[0112] The semantic parsing in S51: The user inputs a query or control request through the natural language interface (such as "Query the air conditioner status in the computer room" or "Turn off the lights in the meeting room");

[0113] Intention Recognition and Entity Extraction: Use the large model to parse the natural language input by the user, and identify the key information of the operation type (query or control), target space (such as computer room, meeting room), and target equipment type (such as air conditioner, light);

[0114] Context Understanding: Combine the context information to process the user's complex query requirements, such as adjusting the specific parameters of the equipment status.

[0115] The result feedback in S63: Feed back the queried equipment status information or the execution result of the control instruction to the user through the natural language interface.

[0116] The typical working process of the system is as follows:

[0117] User Request: The user submits a query or control request through the natural language interface, such as "Query the status of the temperature sensor in the computer room" or "Turn off the lights in the meeting room";

[0118] Semantic Parsing and Intention Recognition: The NLP module and the large model parse the user request to identify the operation type, target space, target equipment, and specific operation parameters;

[0119] Cypher Query Statement Generation: According to the parsing result, the system automatically generates the corresponding Cypher query statement to query the IOT ID of the target equipment through the Neo4j graph database;

[0120] IOT Interface Interaction:

[0121] Query Operation: Use the IOT ID to call the status query API of the IOT platform to obtain the real-time status information of the equipment;

[0122] Control Operation: Generate a control instruction and send the control command to the target equipment through the control API of the IOT platform.

[0123] Result feedback: Convert query results or control execution results into natural language responses and return them to the user.

[0124] Accurate IFC relationship mapping:

[0125] This paper defines in detail the specific relationships between various spaces, equipment, and pipelines in the BIM model, and accurately extracts these relationships through the xBIM parsing library to ensure that the structure of the knowledge graph is consistent with the actual building.

[0126] Automated knowledge graph construction:

[0127] By developing automated tools, complex IFC data can be efficiently converted into a knowledge graph in the Neo4j graph database, reducing manual intervention and improving the accuracy and efficiency of data conversion.

[0128] Seamless conversion from natural language to Cypher:

[0129] By leveraging the large model's powerful ability to understand natural language, users can achieve a seamless experience of querying and controlling through natural language, greatly improving the system's user-friendliness and intelligence.

[0130] Deep integration of BIM and IOT:

[0131] Unified management of BIM model data and IOT device information at the knowledge graph level establishes a close connection between space, equipment and control systems, supports cross-platform semantic query and control operations, and significantly improves the operation and maintenance management efficiency of smart buildings.

[0132] Compared with traditional technologies, this invention significantly improves efficiency and quality in the following aspects:

[0133] Improved query and control efficiency:

[0134] Background technology usually requires users to frequently switch between BIM platforms and IOT platforms, manually search for device IDs, and enter query instructions or control parameters. However, the present invention, through direct interaction between natural language and semantic knowledge graphs, only needs to enter sentences such as "query the air conditioning status of the computer room" or "lower the air conditioning temperature of the conference room by 2 degrees" to quickly obtain the device status or complete the control instruction. From the experimental test results, under the same test scenario, the query and control completion time of the present invention can be reduced by about 50% to 70% compared with the traditional manual operation method, which greatly improves the operating efficiency.

[0135] Data integration and maintenance costs are reduced:

[0136] In the traditional BIM and IOT data integration solution, it is necessary to manually add IOT device ID attributes to BIM components, and there is a lack of automated data relationship management. Through the IFC data parsing and automated knowledge graph construction tool of the present invention, a global device relationship graph based on Neo4j can be established in a short time. Compared with the traditional manual maintenance method, it can reduce the human cost of data processing by about 30% to 40%, and at the same time reduce the maintenance risks caused by data duplicate entry and ID matching errors, thereby improving data quality and consistency.

[0137] Improvement of usage experience and intelligence level:

[0138] Utilizing the natural language understanding ability of the large model, the present invention can seamlessly parse natural language queries and instructions, and is suitable for non-technical personnel and user groups who do not have knowledge of programming or professional graphical query languages. The learning cost of users for the system is significantly reduced. In the internal test, on-site operation and maintenance personnel with simple training can independently complete multiple query and control operations within less than 10 minutes when using this system for the first time, greatly reducing the learning threshold and operation difficulty. ?

[0139] Fast response and dynamic scalability:

[0140] Under test conditions, the present invention can obtain responses to query and control operations for various device types (such as HVAC, lighting, water pipes, sensors) within seconds. Compared with the traditional solution that relies on multiple platforms and multiple steps, the present invention provides a faster response time and dynamic expansion ability. As the types and quantities of IOT devices continue to grow, the system does not need to make a large number of modifications to the query process, and only needs to update the knowledge graph and large model prompts (Prompts) to quickly adapt to new query and control requirements.

[0141] Improvement of accuracy and traceability:

[0142] Through the unified knowledge graph structure and IFC standard relationship mapping rules, the present invention ensures the accuracy and traceability of data and relationships. In the test, for 100 randomly selected device query tasks, the present invention can correctly identify the device in the first query, and the accuracy rate of returning its IOT ID and status information exceeds 95%, which is about 20% higher than the traditional method of manually matching device IDs.

[0143] Intelligent decision-making support and future expansion:

[0144] Based on the semantic knowledge graph and natural language interaction capabilities established by the present invention, in the future, it can be further connected to intelligent decision-making modules and data analysis tools to achieve optimized control, automatic scheduling, and predictive maintenance of building environments and energy consumption. Through the rapid query analysis and remote control of sensor data in experimental scenarios, good results have been achieved in the energy consumption optimization of building air conditioning and lighting systems, and the comprehensive energy saving rate has been increased by about 5% to 10%.

[0145] In summary, the present invention has achieved remarkable results in terms of query and control efficiency, data maintenance cost, user experience, scalability, accuracy, and future intelligent decision-making and energy-saving optimization. Through theoretical derivation, test experimental data, and user experience feedback results.

[0146] Please refer to Figure 1 to learn about the overall process from BIM model data preparation, IFC data parsing, knowledge graph construction to natural language query and IOT control, as well as the relationships between each module.

[0147] Please refer to Figure 2 to learn about the representation forms of nodes such as space, equipment, pipelines, etc. and their associated relationships in the Neo4j graph database.

[0148] Please refer to Figure 3 to learn about the detailed description of each step from user requests to final feedback and their interaction relationships.

[0149] Please refer to Figure 4 to learn about the specific mapping methods of various IfcRel* relationships in IFC in the knowledge graph.

[0150] Through precise IFC data parsing, automated knowledge graph construction, semantic understanding of natural language, and its deep integration with the IOT system, the present invention realizes the efficient conversion from space to semantics, and significantly improves the intelligent management level of smart buildings.

[0151] Compared with related technologies, a method for constructing a knowledge graph and semantic query control of a BIM model provided by the present invention has the following beneficial effects:

[0152] The present invention provides a method for constructing a knowledge graph and semantic query control of a BIM model. By converting the BIM model into a knowledge graph and using natural language processing technology and the capabilities of large models, it realizes unified query and control from building space information to the semantic level.

[0153] A knowledge graph construction and semantic query control system for a BIM model, which adopts the knowledge graph construction and semantic query control method for the BIM model, includes: a BIM model data preparation and attribute extension module, an IFC data parsing and knowledge graph modeling rule definition module, a knowledge graph automatic construction tool development module, a storage and management module of a Neo4j graph database, a natural language processing and large model semantic understanding module, a Cypher query statement generation and IOT interface call module, and a query and control interaction process module. The input and output end of the BIM model data preparation and attribute extension module is connected to the input end of the IFC data parsing and knowledge graph modeling rule definition module. The output end of the IFC data parsing and knowledge graph modeling rule definition module is connected to the input end of the knowledge graph automatic construction tool development module. The output end of the knowledge graph automatic construction tool development module is connected to the input end of the storage and management module of the Neo4j graph database. The output end of the storage and management module of the Neo4j graph database is connected to the input end of the natural language processing and large model semantic understanding module. The output end of the natural language processing and large model semantic understanding module is connected to the input end of the Cypher query statement generation and IOT interface call module. The output end of the Cypher query statement generation and IOT interface call module is connected to the input end of the query and control interaction process module.

[0154] The above are only the embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A method and system for constructing a knowledge graph of a BIM model and semantic query control, characterized in that Including the following steps: S1. BIM Model Data Preparation and Attribute Extension: In the building information model design and management software, perform the following operations to prepare and extend BIM model data; S11. PMID Attribute Definition: In Revit, define an additional attribute PMID for each device and fill in the PMID of each device to make it correspond to the device ID in the IOT system; S12. Definition of Spatial Topology and Device Connection Relationships: In Revit, define the spatial connection relationships among various spaces, pipelines, and devices. Through the built-in relationship management function of Revit, ensure that the connection relationships among entities conform to the actual building structure and device layout; S2. IFC Data Parsing and Definition of Knowledge Graph Modeling Rules: After exporting the prepared BIM model as an IFC format file, perform IFC data parsing and knowledge graph modeling; S21. Use the xBIM Parsing Library: Use xBIM as an IFC data parsing tool to parse the entity, attribute, and relationship information in the IFC file; S22. Define the Knowledge Graph Structure: Node types, relationship types, and attribute mappings; S3. Development of Knowledge Graph Automatic Construction Tools: Develop a set of automated tools to achieve efficient conversion from the IFC format BIM model to the knowledge graph; S31. IFC Parsing Module: Use the xBIM parsing library to read the IFC file and extract all relevant entities, attributes, and their relationships; S32. Data Mapping and Conversion Engine: According to the pre-defined knowledge graph modeling rules, convert the parsed IFC entity types, attributes, and relationships into nodes and relationships in the Neo4j graph database; Scribed in the Neo4j graph database; S4. Storage and Management of the Neo4j Graph Database: Store the constructed knowledge graph in the Neo4j graph database; S41. Node Storage: Store various space, device, and pipeline entities as nodes, including their attributes; S42. Relationship Storage: Store various space association relationships, device association relationships, and space-device relationships as relationships, and define the connection methods between nodes; S43. Indexes and Constraints: Create indexes for commonly queried fields to improve query performance, and set unique constraints to ensure the uniqueness of the PMID and IOTID of each device in the graph database; S5. Natural Language Processing and Semantic Understanding of Large Models: To achieve the automatic generation of Cypher query statements from natural language, the system integrates a natural language processing module and a large model for semantic understanding; S51. Semantic Parsing, Intent Recognition, Entity Extraction, and Context Understanding; S6. Generation of Cypher Query Statements and Invocation of IOT Interfaces: According to the semantic parsing results, the system automatically generates corresponding Cypher query statements and realizes device status query or control through the IOT interface; S61, Automatic generation of Cypher query statements: Dynamically construct Cypher query statements based on the parsed intent and entity information; S62, IOT interface call; S63, Result feedback; S7, Query and control interaction process.

2. The method for constructing a knowledge graph and semantic query control of a BIM model according to claim 1, wherein The PMID attribute in S11 is used for the association between devices and the IOT system in the subsequent knowledge graph.

3. The method for constructing a knowledge graph and semantic query control of a BIM model according to claim 1, wherein, The xBIM in S21 supports reading and operating IFC files and can extract the required building element and their relationship data.

4. The method for constructing a knowledge graph and semantic query control of a BIM model according to claim 1, wherein, The node types in S22 include the following steps: S22a1, Space: Represents various spaces in the building; S22a2, Equipment: Represents various devices; S22a3, Pipeline: Represents the pipeline system in the building; S22a4, Sensor: Represents sensors used to monitor the device status; S22a5, Controller: Represents controllers used to control devices.

5. The knowledge graph construction and semantic query control method for the BIM model according to claim 1, characterized in that The relationship types in S22 include the following steps: S22b1, Spatial association relationship: Used to connect IfcBuildingElement entities, representing the aggregation relationship between spaces; S22b2, Equipment association relationship: IfcRelFlowControlElements: Components under an external controller are associated with the elements they control through this relationship; IfcRelConnectsPortToElement: Components under an internal controller are connected to IfcFlowSegment or IfcFlowFitting through this relationship; IfcRelConnectsSegments: Pipe segments are connected to other pipe segments or pipe fittings through this relationship; IfcRelConnectsPorts: Pipe fittings are connected to other pipe fittings or terminals through this relationship; IfcRelConnectsToPipe or IfcRelConnectsToPort: Terminals are connected to the system through these two relationships; IfcRelConnectsToElement: Energy conversion devices and fluid storage devices are connected to other building elements through this relationship; IfcRelConnectsToProcess: Fluid moving devices and fluid processing devices are connected to the flow process through this relationship; S22b3, Spatial equipment relationship: IfcRelContainedInSpatialStructure is used to represent the relationship that a device belongs to a certain space.

6. The method for constructing a knowledge graph of a BIM model and semantic query control according to claim 1, characterized in that, The attribute mapping in S22: Extract attribute information such as ID, name, location coordinates, function description, and IOT corresponding ID from IFC entities and map them to node attributes in the knowledge graph.

7. The method for constructing a knowledge graph and semantic query control of a BIM model according to claim 1, characterized in that, The semantic parsing in S51: The user inputs a query or control request through the natural language interface; Intent recognition and entity extraction: Use a large model to parse the natural language input by the user and identify key information such as the operation type, target space, and target device type; Context understanding: Combine context information to process the user's complex query requirements.

8. The method for constructing a knowledge graph and semantic query control of a BIM model according to claim 1, wherein Result feedback in S63: The queried device status information or the execution result of the control instruction is fed back to the user through the natural language interface.

9. A knowledge graph construction and semantic query control system for a BIM model, which adopts the knowledge graph construction and semantic query control method for a BIM model described in any one of claims 1 to 8, characterized in that, Including: BIM model data preparation and attribute extension module, IFC data parsing and knowledge graph modeling rule definition module, knowledge graph automatic construction tool development module, Neo4j graph database storage and management module, natural language processing and large model semantic understanding module, Cypher query statement generation and IOT interface call module, and query and control interaction process module. The input and output ends of the BIM model data preparation and attribute extension module are connected to the input end of the IFC data parsing and knowledge graph modeling rule definition module. The output end of the IFC data parsing and knowledge graph modeling rule definition module is connected to the input end of the knowledge graph automatic construction tool development module. The output end of the knowledge graph automatic construction tool development module is connected to the input end of the Neo4j graph database storage and management module. The output end of the Neo4j graph database storage and management module is connected to the input end of the natural language processing and large model semantic understanding module. The output end of the natural language processing and large model semantic understanding module is connected to the input end of the Cypher query statement generation and IOT interface call module. The output end of the Cypher query statement generation and IOT interface call module is connected to the input end of the query and control interaction process module.

Images