BIM and knowledge graph-based operation and maintenance information fusion and intelligent analysis method and system

Abstract

The application provides a kind of based on BIM and knowledge graph's operation and maintenance information fusion and intelligent analysis method and system, method includes following steps: obtaining BIM completion model and extracting component information, collecting and project related operation and maintenance document;BIM component information is carried out semantic recognition and classification, and is associated with component ID, carries out information extraction to operation and maintenance document, with BIM component ID as anchor point, document extraction information is associated with BIM component information;Predefine information graph architecture, automatically map the entity in structured operation and maintenance information dataset to the corresponding node type, automatically map the association between entities to the corresponding relationship type, generate operation and maintenance information graph;Based on the operation and maintenance information graph constructed, the graph data is reasoned and calculated, and intelligent analysis application service is provided.The method improves the intelligent level and efficiency of building operation and maintenance management, reduces operation and maintenance cost, and ensures the safe and reliable operation of building equipment.

Description

Technical Field

[0001] This invention belongs to the fields of building information modeling (BIM) technology and artificial intelligence technology, and specifically relates to a method and system for the fusion and intelligent analysis of operation and maintenance information based on BIM and knowledge graph. Background Technology

[0002] With the development of information technology, Building Information Modeling (BIM) technology has been widely applied in the field of building engineering. Currently, BIM technology plays a crucial role in the architectural design and construction phases, significantly improving the efficiency and quality of design and construction. However, once a project is completed and enters the operation and maintenance phase, the rich engineering information data in the BIM model is often not fully utilized, and the potential value of this information data is not fully realized during the operation and maintenance phase. Analysis reveals the following main problems currently faced in the building operation and maintenance phase:

[0003] 1. Information Disconnection: BIM as-built models contain both geometric and non-geometric information data. Due to inconsistent modeling standards and delivery requirements, the quality of information in BIM models varies greatly, often exhibiting issues such as missing information, incorrect attributes, and outdated data. Simultaneously, a large amount of operation and maintenance information, such as equipment manuals, maintenance records, and warranty certificates, exists in unstructured document form, disconnected from the component information in the BIM model. This results in a lack of effective correlation between the BIM model and operation and maintenance documents, making it difficult to form a complete building information view.

[0004] 2. Low information retrieval efficiency: During routine management, maintenance personnel need to manually search for component location and attribute information in the massive BIM model, and simultaneously cross-reference relevant maintenance requirements, historical records, and other information from numerous maintenance documents. This process relies entirely on manual operation, which is not only cumbersome and time-consuming, but also prone to overlooking critical information. Especially when dealing with emergency faults, the delay in information retrieval directly impacts response efficiency.

[0005] 3. Lack of intelligent analysis: Most existing BIM operation and maintenance platforms only realize the visualization of information and basic query functions, lacking the ability to deeply mine and intelligently analyze information. This makes operation and maintenance management still remain in the "passive response" stage, making it difficult to achieve "proactive prevention" and "intelligent auxiliary decision-making". Summary of the Invention

[0006] This invention provides a method and system for operation and maintenance information fusion and intelligent analysis based on BIM and knowledge graph, which improves the intelligence level and efficiency of building operation and maintenance management, reduces operation and maintenance costs, and ensures the safe and reliable operation of building equipment.

[0007] The technical solution of the present invention is as follows:

[0008] A method for operation and maintenance information fusion and intelligent analysis based on BIM and knowledge graph includes the following steps:

[0009] S1: Obtain the BIM as-built model and extract component information, which includes at least component ID and attribute data. At the same time, collect operation and maintenance documents related to the project and parse the unstructured data in the operation and maintenance documents, converting them into machine-readable text format.

[0010] S2: The BIM component information obtained in step S1 is semantically identified and classified using natural language processing technology and associated with the component ID. At the same time, information is extracted from the text-formatted operation and maintenance documents. Using the BIM component ID as the anchor point, the extracted information from the documents is associated with the BIM component information through a semantic matching algorithm. When inconsistencies are detected, the data source is selected for fusion according to the preset priority rules to generate a structured operation and maintenance information dataset containing multi-source information.

[0011] S3: Predefine an information graph architecture that includes multiple types of nodes and multiple types of relationships. Use a semantic matching-based mapping algorithm to automatically map entities in the structured operation and maintenance information dataset obtained in step S2 to the corresponding node types and automatically map the associations between entities to the corresponding relationship types to generate an operation and maintenance information graph containing semantic relationships.

[0012] S4: Based on the completed operation and maintenance information graph, it calls the large language model to perform reasoning and calculation on the graph data, and provides intelligent analysis application services for operation and maintenance management.

[0013] Furthermore, in the aforementioned method for fusion and intelligent analysis of operation and maintenance information based on BIM and knowledge graphs, in step S1:

[0014] The BIM component information also includes geometric information, spatial location, and material information; and / or,

[0015] The maintenance documentation includes one or more of the following: as-built drawings, equipment ledgers, product data sheets, system maintenance manuals, and warranty certificates; and / or,

[0016] Parsing image-based maintenance documents includes recognizing text content in images using optical character recognition (OCR) technology and converting it into machine-readable text format; and / or,

[0017] The extraction of component information includes parsing IFC files or calling BIM software application programming interfaces to extract component information from the BIM as-built model in IFC or RVT format.

[0018] Furthermore, in the BIM and knowledge graph-based operation and maintenance information fusion and intelligent analysis method, step S2, which involves extracting information from text-formatted operation and maintenance documents, specifically includes:

[0019] The text classification model is used to identify the type of operation and maintenance document, and the relationship extraction model is used to identify the entities and relationships between entities in the operation and maintenance document. The entities include equipment components, maintenance tasks and fault types, and the relationships between entities include execution cycles, associated components and processing methods.

[0020] Furthermore, in the BIM and knowledge graph-based operation and maintenance information fusion and intelligent analysis method, the semantic matching algorithm in step S2 includes:

[0021] Similarity calculation based on vector space model, path ranking algorithm based on knowledge graph, or semantic similarity matching model based on deep neural network.

[0022] Furthermore, in the BIM and knowledge graph-based operation and maintenance information fusion and intelligent analysis method, the preset priority rules in step S2 include:

[0023] Prioritize different data sources based on their update date, whether they have electronic signatures, document type hierarchy, or the authority of the source organization.

[0024] Data is fused based on information from high-priority data sources, and a data conflict log is generated to record conflict details and the basis for handling, which is then stored in the structured operation and maintenance information dataset.

[0025] Furthermore, in the BIM and knowledge graph-based operation and maintenance information fusion and intelligent analysis method, the predefined information graph architecture in step S3 includes:

[0026] Node types include at least device nodes, space nodes, system nodes, supplier nodes, maintenance task nodes, and fault type nodes;

[0027] The types of edge relationships include at least the "located" relationship representing spatial affiliation, the "belonging" relationship representing system composition, the "connection" relationship representing physical connection, the "supplied by" relationship representing the supply subject, the "having maintenance tasks" relationship representing the maintenance plan, the "may happen" relationship representing the possibility of failure, and the "has happened" relationship representing historical events.

[0028] Furthermore, in the BIM and knowledge graph-based operation and maintenance information fusion and intelligent analysis method, the semantic matching-based mapping algorithm in step S3 includes:

[0029] Semantically encode entity names and attributes using a pre-trained language model, calculate the semantic similarity between the entity to be mapped and predefined node types, and map the entity to the node type with the highest similarity.

[0030] The semantic relationships between entity pairs are identified through a relation extraction model, and the identified relationships are mapped to predefined relation types.

[0031] Furthermore, in the BIM and knowledge graph-based operation and maintenance information fusion and intelligent analysis method, the intelligent analysis application service in step S4 includes:

[0032] The intelligent retrieval service for operation and maintenance information responds to natural language queries, performs semantic matching retrieval in the operation and maintenance information map, locates target device nodes and related spatial nodes and maintenance task nodes, highlights the corresponding device locations in the BIM model, and outputs related operation and maintenance document links and historical records.

[0033] The maintenance early warning service periodically traverses equipment nodes based on the equipment attribute data and historical maintenance records stored in the operation and maintenance information map, calculates maintenance requirements according to the preset maintenance cycle, design life and last maintenance date, generates early warning notifications for equipment that meet the early warning conditions and pushes them to maintenance management personnel.

[0034] The fault impact analysis service responds to equipment fault signals by performing graph traversal based on connectivity in the operation and maintenance information map to identify affected upstream and downstream equipment. The scope of the fault impact is visualized in the BIM model. Simultaneously, based on the faulty equipment information, fault handling suggestions are retrieved and generated from the product data table and historical fault records associated with the map through a large language model combined with retrieval enhancement generation technology.

[0035] Furthermore, in the BIM and knowledge graph-based operation and maintenance information fusion and intelligent analysis method, the maintenance early warning service includes: feeding back the execution results of the early warning notification and the actual maintenance records to the operation and maintenance information graph, updating the "last maintenance date" and "historical operation and maintenance records" of the corresponding equipment nodes for calculation of the next maintenance requirement; and / or,

[0036] The fault impact analysis service includes: feeding back the actual fault handling process, handling results and impact assessment to the operation and maintenance information graph, adding or updating instance data of fault type nodes for subsequent retrieval and inference of similar faults.

[0037] A system for the fusion and intelligent analysis of operation and maintenance information based on BIM and knowledge graphs, comprising:

[0038] The data acquisition and preprocessing module is used to acquire the BIM as-built model and extract component information, which includes at least component ID and attribute data. At the same time, it collects operation and maintenance documents related to the project and parses the unstructured data in the operation and maintenance documents, converting it into a machine-readable text format.

[0039] The AI ​​information processing module is used to perform semantic recognition and classification of the BIM component information through natural language processing technology and associate it with the component ID. At the same time, it extracts information from the text-formatted operation and maintenance documents. Using the BIM component ID as the anchor point, it associates the extracted information from the documents with the BIM component information through a semantic matching algorithm. When inconsistencies are detected, it selects data sources for fusion according to preset priority rules to generate a structured operation and maintenance information dataset containing multi-source information.

[0040] The graph construction module is used to predefine an information graph architecture that includes multiple types of nodes and multiple types of relationships. It automatically maps entities in the structured operation and maintenance information dataset to the corresponding node types and automatically maps the associations between entities to the corresponding relationship types through a semantic matching mapping algorithm, thereby generating an operation and maintenance information graph containing semantic relationships.

[0041] The intelligent analysis application module is used to perform reasoning and calculation on the graph data based on the constructed operation and maintenance information graph, and to provide intelligent analysis application services for operation and maintenance management.

[0042] The beneficial effects of this invention are as follows:

[0043] This invention discloses a method for the fusion and intelligent analysis of operation and maintenance information based on BIM and knowledge graphs. By intelligently integrating BIM information with operation and maintenance documents, it constructs an operation and maintenance knowledge graph rich in semantic relationships and provides application services such as intelligent retrieval, maintenance early warning, and fault impact analysis. This method can significantly improve the intelligence level and efficiency of building operation and maintenance management, reduce operation and maintenance costs, and ensure the safe and reliable operation of building equipment. It has high industrial practical value and economic value.

[0044] This BIM and knowledge graph-based operation and maintenance information fusion and intelligent analysis method further breaks down the "data silos" in traditional BIM operation and maintenance by using BIM component IDs as anchors and combining semantic matching algorithms to cross-modal associate structured data from BIM models with unstructured text from operation and maintenance documents. This achieves deep fusion of multi-source information. Simultaneously, by automatically identifying information conflicts and automatically selecting authoritative data sources based on preset priority rules, the accuracy and reliability of the fused data are ensured. Attached Figure Description

[0045] Figure 1This is a flowchart of a method for the fusion and intelligent analysis of operation and maintenance information based on BIM and knowledge graphs according to the present invention;

[0046] Figure 2 This is a schematic diagram of an operation and maintenance information fusion and intelligent analysis system based on BIM and knowledge graph according to the present invention. Detailed Implementation

[0047] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become clearer from the following description and claims. It should be noted that the drawings are all in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0048] like Figure 1 As shown, this embodiment provides a method for operation and maintenance information fusion and intelligent analysis based on BIM and knowledge graph, including the following steps: S1-S4.

[0049] S1: Obtain the BIM as-built model and extract component information, which includes at least component ID and attribute data. At the same time, collect operation and maintenance documents related to the project, parse the unstructured data in the operation and maintenance documents, and convert them into machine-readable text format.

[0050] The component information is extracted from the BIM as-built model in IFC or RVT format by parsing the IFC file or calling the BIM software application programming interface. The component information includes component ID, geometric information, spatial location, material information, and predefined attribute data. Simultaneously, project-related maintenance documents are collected, including as-built drawings (PDF format), equipment ledgers (Excel format), product data sheets (PDF scans), system maintenance manuals (Word format), warranty certificates (PDF scans), etc. For image-based maintenance documents (e.g., scanned PDF documents), optical character recognition (OCR) technology is used to recognize the text content in the images and convert it into machine-readable text format.

[0051] For example, the BIM as-built model is stored in RVT format. By connecting to the BIM model via the Revit API, the element filter is called to obtain all mechanical equipment, electrical equipment, pipes and accessories, and other component elements. The component element attribute reading interface is used to obtain the component ID, component name, spatial coordinates, material information, and predefined attributes such as equipment model, rated power, and design life for each component.

[0052] S2: Using natural language processing technology, semantic recognition and classification are performed on the BIM component information obtained in step S1 to identify equipment information such as "fan coil unit", "air conditioning unit", and "water pump" and associate it with the corresponding component ID. For example, the information of AHU-01 air conditioning unit is identified: location on the first basement floor, size, name, system type, manufacturer, etc., and associated with its component ID such as 7654321.

[0053] Simultaneously, information is extracted from the maintenance documents converted to text format: the document type (such as product data sheets, maintenance manuals, warranty cards, etc.) is identified through a text classification model, and entities and relationships between entities are identified through a relationship extraction model. The entities include equipment components (such as fan coil unit FCU-101), maintenance tasks (such as filter cleaning), and fault types (such as motor burnout). The relationships between entities include execution cycles (such as the execution cycle of "filter replacement" being "every six months"), associated components, and processing methods (such as the processing method for "motor burnout" being "motor replacement").

[0054] Using BIM component IDs as anchors, semantic matching algorithms (such as BERT-based semantic similarity matching models) are used to associate document-extracted information with BIM component information. For example, if a document mentions "AHU-01 air conditioning unit," semantic matching can be used to find the component with BIM component ID "ID7654321," and the maintenance requirements and fault information in the document can be associated with that component.

[0055] When inconsistencies are detected, data sources are selected and merged according to preset priority rules to generate a structured operation and maintenance information dataset containing information from multiple sources. The preset priority rules include: determining the priority of different data sources based on the update date of the information source, whether it has an electronic signature, document type hierarchy, or the authority of the source organization; data merging is performed based on the information from the higher priority data source, and a data conflict log is generated to record conflict details and the basis for handling, which is then stored in the structured operation and maintenance information dataset.

[0056] For example, if the rated air volume of the equipment recorded in the BIM model is 6000 m³ / h, while the rated air volume in the equipment manual is 8000 m³ / h, according to the preset priority rule (the equipment manual is more up-to-date than the BIM model and is the manufacturer's official data, so it is more authoritative), the rated air volume of 8000 m³ / h in the equipment manual will be used for merging, and a data conflict log will be generated to record the conflict details and the basis for handling.

[0057] Step S2, using BIM component IDs as anchors and combining semantic matching algorithms, cross-modal association is performed between the structured data of the BIM model and the unstructured text of the operation and maintenance documents. This breaks down the "data silos" in traditional BIM operation and maintenance, achieving deep integration of multi-source information. Furthermore, by automatically identifying information conflicts and automatically selecting authoritative data sources based on preset priority rules, the accuracy and reliability of the integrated data are ensured.

[0058] S3: A predefined information graph architecture that includes multiple types of nodes and multiple types of relationships.

[0059] The node types include at least device nodes, space nodes, system nodes, supplier nodes, maintenance task nodes, and fault type nodes.

[0060] The relationship types of the edges include at least the "located" relationship representing spatial affiliation, the "belongs to" relationship representing system composition, the "connection" relationship representing physical connection, the "supplied by" relationship representing the supply entity, the "has maintenance task" relationship representing the maintenance plan, the "may happen" relationship representing the possibility of failure, and the "has occurred" relationship representing historical events.

[0061] A semantic matching-based mapping algorithm automatically maps entities in the structured operation and maintenance information dataset obtained in step S2 to their corresponding node types, and automatically maps the relationships between entities to their corresponding relation types. For example: Component ID 7654321, name AHU-01 air conditioning unit, rated air volume 8000m³ / h, located in the basement air conditioning room, belongs to the air supply system, connected to a certain duct, supplied by a certain supplier, and requires filter replacement every six months, etc.

[0062] Finally, an operation and maintenance information graph containing semantic relationships is generated.

[0063] Step S3 constructs an operations and maintenance information graph containing multiple types of nodes and relationships, organizing scattered information into a semantically related knowledge network. This supports relationship-based intelligent retrieval, significantly improving information search efficiency. Furthermore, a semantic matching-based mapping algorithm automatically maps structured operations and maintenance information to a predefined graph architecture, achieving automated construction of the operations and maintenance information graph without manual intervention, greatly reducing construction costs and barriers to entry.

[0064] S4: Based on the constructed operation and maintenance information graph, a large language model is invoked to perform reasoning and calculation on the graph data, providing intelligent analysis and application services for operation and maintenance management. These intelligent analysis and application services include intelligent operation and maintenance information retrieval services, maintenance early warning services, and fault impact analysis services.

[0065] The intelligent retrieval service for operation and maintenance information responds to natural language queries, performs semantic matching retrieval in the operation and maintenance information map, locates target device nodes and related spatial nodes and maintenance task nodes, highlights the corresponding device locations in the BIM model, and outputs related operation and maintenance document links and historical records.

[0066] For example, maintenance personnel can interact with the intelligent maintenance information retrieval service, requesting information such as "equipment information in the underground second-floor exhaust fan room 2," "supplier and contact number of AHU-01 air conditioning unit," and "maintenance cycle of the chiller unit." The system quickly retrieves the required maintenance information from the maintenance information map, displays it in the dialogue list, locates it in the BIM model, and provides the path to the retrieved document for secondary queries by maintenance personnel.

[0067] The maintenance early warning service periodically traverses the equipment nodes based on the equipment attribute data and historical maintenance records stored in the operation and maintenance information map. It calculates maintenance requirements according to the preset maintenance cycle, design life and last maintenance date, generates early warning notifications for equipment that meets the early warning conditions and pushes them to maintenance management personnel.

[0068] For example, a periodic scheduled task can be set to automatically calculate maintenance requirements based on equipment attributes (maintenance cycle, design life, etc.) and historical data (last maintenance date, running time, etc.) in the operation and maintenance information map. If the system finds that the chiller unit is close to 5 months since the last maintenance and the equipment has been running for nearly 2000 hours, it will meet the maintenance requirements in the equipment maintenance instructions. The system will then automatically generate an early warning work order and notify the operation and maintenance personnel.

[0069] In response to equipment failure signals, the failure impact analysis service performs graph traversal based on connectivity in the operation and maintenance information map to identify affected upstream and downstream equipment. The scope of failure impact is visualized in the BIM model. Simultaneously, based on the failure equipment information, the service retrieves and generates failure handling suggestions from the product data table and historical failure records associated with the map through a large language model combined with retrieval enhancement generation technology.

[0070] For example, when maintenance personnel report equipment failure, such as "AHU-01 air conditioning unit failure", the system performs graph traversal based on "connection" relationships in the operation and maintenance information map to find the upstream and downstream components (air ducts, power cabinets, equipment units, etc.) of the equipment node. The system then locates the potential affected area in the BIM model and, by integrating emergency plans and big data processing methods from the operation and maintenance documents, generates a fault handling report to assist operation and maintenance management personnel in making decisions.

[0071] Step S4, based on the operations and maintenance information graph, calls a large language model to perform reasoning and calculation on the graph data. Combining the structured knowledge of the graph with the generation capabilities of the language model, it provides application services such as intelligent retrieval, proactive early warning, and fault analysis, helping operations and maintenance managers quickly retrieve operations and maintenance information, improve efficiency, and reduce project losses and security risks.

[0072] The above method, by realizing the intelligent integration of BIM information and operation and maintenance documents, constructs an operation and maintenance knowledge graph rich in semantic relationships and provides intelligent analysis and application services, which can significantly improve the intelligence level and efficiency of building operation and maintenance management, reduce operation and maintenance costs, and ensure the safe and reliable operation of building equipment. It has high industrial practical value and economic value.

[0073] As a preferred implementation, the semantic matching algorithm in step S2 includes: similarity calculation based on a vector space model, path ranking algorithm based on a knowledge graph, or semantic similarity matching model based on a deep neural network (e.g., a semantic similarity matching model based on BERT). By employing multiple semantic matching algorithms such as vector space models, knowledge graph path ranking, or deep neural networks, accurate and intelligent association between BIM component information and operation and maintenance documents is achieved, significantly improving the accuracy and robustness of multi-source heterogeneous data fusion.

[0074] As a preferred implementation, the maintenance early warning service includes: feeding back the execution result of the early warning notification and the actual maintenance record to the operation and maintenance information graph, and updating the "last maintenance date" and "historical operation and maintenance record" of the corresponding device node for the calculation of the next maintenance requirement.

[0075] By feeding back the execution results of early warning notifications and actual maintenance records to the operation and maintenance information graph, the "last maintenance date" and "historical operation and maintenance records" of equipment nodes are dynamically updated. This enables the calculation of maintenance requirements to be based on the latest equipment status and historical data, thereby achieving continuous optimization of early warning rules and continuous improvement of early warning accuracy, forming a virtuous cycle of "early warning-processing-feedback-optimization".

[0076] As a preferred implementation, the fault impact analysis service includes: feeding back the actual fault handling process, handling results and impact assessment to the operation and maintenance information graph, adding or updating instance data of fault type nodes for subsequent retrieval and inference of similar faults.

[0077] By feeding back the actual fault handling process, results, and impact assessments to the operation and maintenance information graph, and adding or updating instance data for fault type nodes, the system can continuously accumulate fault handling experience. This provides richer and more accurate reference cases for subsequent retrieval and reasoning of similar faults, significantly improving the pertinence and practicality of fault handling suggestions, and enabling the fault knowledge base to enrich itself and evolve iteratively.

[0078] like Figure 2 As shown, this embodiment also provides an operation and maintenance information fusion and intelligent analysis system based on BIM and knowledge graph, including a data acquisition and preprocessing module, an AI information processing module, a graph construction module, and an intelligent analysis application module.

[0079] The data acquisition and preprocessing module is used to acquire the BIM as-built model and extract component information, which includes at least component ID and attribute data. At the same time, it collects operation and maintenance documents related to the project and parses the unstructured data in the operation and maintenance documents, converting it into a machine-readable text format.

[0080] The AI ​​information processing module is used to perform semantic recognition and classification of the BIM component information using natural language processing technology and associate it with the component ID. At the same time, it extracts information from the text-formatted operation and maintenance documents, uses the BIM component ID as the anchor point, and associates the extracted information from the documents with the BIM component information through a semantic matching algorithm. When inconsistencies are detected, it selects data sources for fusion according to preset priority rules to generate a structured operation and maintenance information dataset containing multi-source information.

[0081] The graph construction module is used to predefine an information graph architecture that includes multiple types of nodes and multiple types of relationships. It automatically maps entities in the structured operation and maintenance information dataset to their corresponding node types and automatically maps the associations between entities to their corresponding relationship types through a semantic matching mapping algorithm, thereby generating an operation and maintenance information graph containing semantic relationships.

[0082] The intelligent analysis application module is used to perform inference and calculation on the completed operation and maintenance information map by calling a large language model, providing intelligent analysis application services for operation and maintenance management. The intelligent analysis application module includes an intelligent operation and maintenance information retrieval service unit, a maintenance early warning service unit, and a fault impact analysis service unit.

[0083] The aforementioned system constructs a complete technical solution from data acquisition, intelligent processing, graph construction to analysis and application. It realizes the automated integration of BIM models and operation and maintenance documents, the graph-based organization of operation and maintenance knowledge, and intelligent reasoning and decision-making based on large language models. It significantly improves the automation and intelligence level of building operation and maintenance management and can be widely applied to operation and maintenance management scenarios of various buildings, including but not limited to commercial complexes, hospitals, schools, airports, data centers, and industrial parks.

[0084] The above description is merely a description of preferred embodiments of the present invention and is not intended to limit the scope of the present invention in any way. Any changes or modifications made by those skilled in the art based on the above disclosure shall fall within the protection scope of the claims.

Claims

1. A method for operation and maintenance information fusion and intelligent analysis based on BIM and knowledge graph, characterized in that, Includes the following steps: S1: Obtain the BIM as-built model and extract component information, which includes at least component ID and attribute data. At the same time, collect operation and maintenance documents related to the project and parse the unstructured data in the operation and maintenance documents, converting them into machine-readable text format. S2: The BIM component information obtained in step S1 is semantically identified and classified using natural language processing technology and associated with the component ID. At the same time, information is extracted from the text-formatted operation and maintenance documents. Using the BIM component ID as the anchor point, the extracted information from the documents is associated with the BIM component information through a semantic matching algorithm. When inconsistencies are detected, the data source is selected for fusion according to the preset priority rules to generate a structured operation and maintenance information dataset containing multi-source information. S3: Predefine an information graph architecture that includes multiple types of nodes and multiple types of relationships. Use a semantic matching-based mapping algorithm to automatically map entities in the structured operation and maintenance information dataset obtained in step S2 to the corresponding node types and automatically map the associations between entities to the corresponding relationship types to generate an operation and maintenance information graph containing semantic relationships. S4: Based on the completed operation and maintenance information graph, it calls the large language model to perform reasoning and calculation on the graph data, and provides intelligent analysis application services for operation and maintenance management.

2. The operation and maintenance information fusion and intelligent analysis method based on BIM and knowledge graph as described in claim 1, characterized in that, In step S1: The BIM component information also includes geometric information, spatial location, and material information; and / or, The maintenance documentation includes one or more of the following: as-built drawings, equipment ledgers, product data sheets, system maintenance manuals, and warranty certificates; and / or, Parsing image-based maintenance documents includes recognizing text content in images using optical character recognition (OCR) technology and converting it into machine-readable text format; and / or, The component information can be extracted by parsing IFC files or by calling the BIM software application programming interface to extract the component information from the BIM as-built model in IFC or RVT format.

3. The operation and maintenance information fusion and intelligent analysis method based on BIM and knowledge graph as described in claim 1, characterized in that, The information extraction process for the text-formatted maintenance document in step S2 specifically includes: The text classification model is used to identify the type of operation and maintenance document, and the relationship extraction model is used to identify the entities and relationships between entities in the operation and maintenance document. The entities include equipment components, maintenance tasks and fault types, and the relationships between entities include execution cycles, associated components and processing methods.

4. The operation and maintenance information fusion and intelligent analysis method based on BIM and knowledge graph as described in claim 1, characterized in that, The semantic matching algorithm in step S2 includes: Similarity calculation based on vector space model, path ranking algorithm based on knowledge graph, or semantic similarity matching model based on deep neural network.

5. The operation and maintenance information fusion and intelligent analysis method based on BIM and knowledge graph as described in claim 1, characterized in that, The preset priority rules in step S2 include: Prioritize different data sources based on their update date, whether they have electronic signatures, document type hierarchy, or the authority of the source organization. Data is fused based on information from high-priority data sources, and a data conflict log is generated to record conflict details and the basis for handling, which is then stored in the structured operation and maintenance information dataset.

6. The method for operation and maintenance information fusion and intelligent analysis based on BIM and knowledge graph as described in claim 1, characterized in that, The predefined information graph architecture in step S3 includes: Node types include at least device nodes, space nodes, system nodes, supplier nodes, maintenance task nodes, and fault type nodes; The types of edge relationships include at least the "located" relationship representing spatial affiliation, the "belonging" relationship representing system composition, the "connection" relationship representing physical connection, the "supplied by" relationship representing the supply subject, the "having maintenance tasks" relationship representing the maintenance plan, the "may happen" relationship representing the possibility of failure, and the "has happened" relationship representing historical events.

7. The method for operation and maintenance information fusion and intelligent analysis based on BIM and knowledge graph as described in claim 1, characterized in that, The semantic matching-based mapping algorithm in step S3 includes: Semantically encode entity names and attributes using a pre-trained language model, calculate the semantic similarity between the entity to be mapped and predefined node types, and map the entity to the node type with the highest similarity. The semantic relationships between entity pairs are identified through a relation extraction model, and the identified relationships are mapped to predefined relation types.

8. The method for operation and maintenance information fusion and intelligent analysis based on BIM and knowledge graph as described in claim 1, characterized in that, The intelligent analysis application service in step S4 includes: The intelligent retrieval service for operation and maintenance information responds to natural language queries, performs semantic matching retrieval in the operation and maintenance information map, locates target device nodes and related spatial nodes and maintenance task nodes, highlights the corresponding device locations in the BIM model, and outputs related operation and maintenance document links and historical records. The maintenance early warning service periodically traverses equipment nodes based on the equipment attribute data and historical maintenance records stored in the operation and maintenance information map, calculates maintenance requirements according to the preset maintenance cycle, design life and last maintenance date, generates early warning notifications for equipment that meet the early warning conditions and pushes them to maintenance management personnel. The fault impact analysis service responds to equipment fault signals by performing graph traversal based on connectivity in the operation and maintenance information map to identify affected upstream and downstream equipment. The scope of the fault impact is visualized in the BIM model. Simultaneously, based on the faulty equipment information, fault handling suggestions are retrieved and generated from the product data table and historical fault records associated with the map through a large language model combined with retrieval enhancement generation technology.

9. The method for operation and maintenance information fusion and intelligent analysis based on BIM and knowledge graph as described in claim 8, characterized in that, The maintenance early warning service includes: feeding back the execution results of the early warning notification and the actual maintenance records to the operation and maintenance information graph, updating the "last maintenance date" and "historical operation and maintenance records" of the corresponding device nodes for calculation of the next maintenance requirement; and / or, The fault impact analysis service includes: feeding back the actual fault handling process, handling results and impact assessment to the operation and maintenance information graph, adding or updating instance data of fault type nodes for subsequent retrieval and inference of similar faults.

10. A system for the fusion and intelligent analysis of operation and maintenance information based on BIM and knowledge graphs, characterized in that: include: The data acquisition and preprocessing module is used to acquire the BIM as-built model and extract component information, which includes at least component ID and attribute data. At the same time, it collects operation and maintenance documents related to the project and parses the unstructured data in the operation and maintenance documents, converting it into a machine-readable text format. The AI ​​information processing module is used to perform semantic recognition and classification of the BIM component information through natural language processing technology and associate it with the component ID. At the same time, it extracts information from the text-formatted operation and maintenance documents. Using the BIM component ID as the anchor point, it associates the extracted information from the documents with the BIM component information through a semantic matching algorithm. When inconsistencies are detected, it selects data sources for fusion according to preset priority rules to generate a structured operation and maintenance information dataset containing multi-source information. The graph construction module is used to predefine an information graph architecture that includes multiple types of nodes and multiple types of relationships. It automatically maps entities in the structured operation and maintenance information dataset to the corresponding node types and automatically maps the associations between entities to the corresponding relationship types through a semantic matching mapping algorithm, thereby generating an operation and maintenance information graph containing semantic relationships. The intelligent analysis application module is used to perform reasoning and calculation on the graph data based on the constructed operation and maintenance information graph, and to provide intelligent analysis application services for operation and maintenance management.

Images