Component alignment and knowledge graph construction method and system based on BIM model

Abstract

This invention discloses a method and system for component alignment and knowledge graph construction based on a BIM model. The method includes: S1, inputting data from multiple stages such as design and construction into the BIM model as raw data; S2, extracting the attributes, geometry, and internal relationships of each entity in the raw data, and constructing a relationship graph for each stage; S3, using the source graph as a reference, finding the unique corresponding entity for each entity in the target graph, and outputting a global alignment mapping set; S4, updating the global alignment mapping set according to changes in BIM data. This invention directly utilizes the predefined complete relationships within the BIM model at each stage as core constraints to construct a fused knowledge graph that can guide, verify, and continuously maintain the alignment process, thereby achieving accurate, consistent, and sustainable association of BIM components at different stages at the structural and semantic levels.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional intelligent spatial data registration, specifically to a method and system for component alignment and knowledge graph construction based on BIM models. Background Technology

[0002] The design of foundation pit excavation, slope protection, underground structures, and earthwork engineering is gradually shifting from "two-dimensional drawings + experience-based judgment" to "three-dimensional digital twins + intelligent decision-making." Component alignment and knowledge graph construction, as two core technological pillars supporting this transformation, are still far below the stringent requirements of actual engineering for "safety redundancy, construction efficiency, and cost control." Related technological bottlenecks are prevalent in engineering practice and exhibit significant industry-specific characteristics. During construction, "component alignment" is not simply about geometric overlap, but rather the determination and verification of spatial positions under multi-dimensional constraints of structural parameters, construction procedures, geological conditions, and safety distances. Its pain points are mainly reflected in four aspects: "insufficient accuracy, inadequate linkage, incomplete scenarios, and lack of standards."

[0003] 1. The reliance on a single alignment benchmark makes it difficult to adapt to complex construction conditions.

[0004] Most existing alignment technologies use an absolute coordinate system or a fixed reference surface (such as ±0.000, the design ground elevation) as the sole benchmark, which can only achieve alignment between components and the static design surface. However, in actual engineering, earthwork excavation is characterized by layering, segmentation, and dynamic retreat, with slope gradients constantly changing with construction progress, and uneven settlement and displacement of the geological soil. For example, in the alignment of pile foundation construction and foundation pit support structure, existing methods cannot update the alignment parameters of support piles, anchor cables, and steel supports in real time with the dynamic adjustment of the excavation surface. This leads to deviations between the "design alignment position" and the "actual construction position" at different excavation stages, which may result in insufficient embedment depth of the support structure, conflict between anchor cables and the excavation surface, and may also cause over-excavation of soil and slope instability risks, directly violating the core requirements for foundation stability and structural safety.

[0005] 2. The fragmented alignment of multi-disciplinary components lacks cross-group collaborative logic.

[0006] Existing alignment technologies are mostly limited to isolated alignment of single component types or single disciplines, failing to establish cross-group associative alignment logic. For example, aligning the axis of a diaphragm wall with the pile position of a pile bank does not consider the spatial nesting relationship between the two and the soil excavation boundary; aligning the pile cap with the pile foundation does not consider the uneven distribution of the foundation soil layers. This fragmentation leads to hidden problems of "seemingly aligned, but actually conflicting" between components, increasing construction rework and safety hazards. Therefore, we adopt a BIM model and apply the dispersed component entities, attributes, and their multi-dimensional relationships (such as spatial topology, system composition, and functional associations) in the BIM model to reorganize and express them according to the "entity-relationship-entity" triple structure, thereby forming a structured semantic network that provides a data foundation for in-depth analysis and decision-making. When local data changes occur, the scope of impact can be located based on the relationship network, and the alignment status of related entities can be collaboratively recalculated and maintained for consistency, thus supporting the continuous evolution of the alignment map throughout its entire lifecycle. Summary of the Invention

[0007] This invention addresses the problems in existing technologies by disclosing a method and system for component alignment and knowledge graph construction based on BIM models. Addressing the heterogeneous, complexly correlated, and continuously evolving characteristics of BIM data at different stages, this invention does not rely on external computation or repetitive reasoning. Instead, it directly utilizes predefined complete relationships within the BIM models at each stage as core constraints to construct a fused knowledge graph that can guide, verify, and continuously maintain the alignment process. This achieves accurate, consistent, and sustainable association of BIM components at different stages at both structural and semantic levels.

[0008] This invention is achieved through the following technical solution:

[0009] This invention first provides a method for component alignment and knowledge graph construction based on BIM models, including the following steps:

[0010] S1. Input the data from the design and construction phases into the BIM model as raw data;

[0011] S2. Analyze the data of each stage in the original data, extract the features of each entity, and construct an internal relationship graph for each stage based on the extracted features. The graph of the design stage is denoted as the source graph. The maps used during the construction phase are recorded as target maps. ;

[0012] S3, Source Map As a baseline, for the target map For each entity in the graph, find its unique counterpart in the source graph and output a global alignment map set;

[0013] S4. When the BIM data changes at any stage, the BIM model automatically determines the set of changed nodes and its affected edge set, and defines the incremental calculation range; step S3 is repeated only within the incremental calculation range, and the latest global alignment mapping set is output.

[0014] As a further step, in S2, the method for constructing a relational graph within each corresponding stage includes:

[0015] S21. Construct an internal relationship graph. , ,

[0016] in, This represents a set of nodes, whose elements are entities in the BIM model corresponding to the current stage.

[0017] Represents a set of relation types used to describe various predefined semantic relationships between entities;

[0018] Let represent the set of directed edges, and ;

[0019] S22. Record the design phase diagrams as the source diagrams. ,but , Represents the set of nodes in the source graph; Let represent the set of directed edges in the source graph, and ;

[0020] The map of the target stage is denoted as the target map. ; Represents the set of nodes in the target graph; Let represent the set of directed edges in the target graph, and Source map and target map Shared Relationship Type Set .

[0021] As a further step, the specific method of S3 includes:

[0022] S31. For the target entity Extract its geometric features, semantic features, and topological features, among which, ;

[0023] S32, for the target entity In the source map Filter potential matching source entities In the source map Spatial-semantic collaborative retrieval is performed to generate its final candidate matching set. ;

[0024] S33. Criteria for determining relation consistency constraints;

[0025] S34. Determine the target entity through relation consistency verification. The only candidate, and form candidate alignment pairs;

[0026] S35. Gather all the finally determined alignment pairs to form a global alignment map set. As direct output.

[0027] As a further step, the specific method of S32 includes:

[0028] S321, Based on target entity Global coordinates in the source map Retrieve all objects with a distance less than a threshold from the spatial index. The entities generate a spatial candidate set. ;

[0029] S322. Based on a predefined cross-stage type mapping table, from the spatial candidate set... Filter out entities matching the target entity Type-compatible entities form a semantic candidate set. ;

[0030] S323, on semantic candidate sets Each candidate entity in Perform a traversal in the source graph. Retrieve all edges that are connected to predefined relation edges First-degree neighbors that are directly connected are added to the candidate set. (Initially an empty set), after traversal, this candidate set will be... and Merge (i.e., take the union of the two), and remove duplicates based on the unique identifier (id) of the nodes: if two nodes have the same id, they are considered the same entity, and only one is kept, thus generating the final candidate matching set. .

[0031] As a further step, the specific method of S33 includes:

[0032] Candidate alignment pairs The requirement for it to be valid is: if and Alignment, then exist Local relational context Must with exist Local relational context It is structurally compatible;

[0033] The exist Local relational context For: with A one-hop neighbor subgraph centered on the center, containing Self, All and Directly connected nodes and the edges connecting these nodes;

[0034] Similarly, exist Local relational context For: with A one-hop neighbor subgraph centered on the center, containing Self, All and Directly connected nodes and the edges connecting these nodes;

[0035] The compatibility requirements will The neighbor nodes in the mapping are mapped to Find compatible nodes and verify them. For each relation edge in the graph, between its mapped node pairs, There exist semantically corresponding relation edges in each of them. This makes... The observed relation edges and their types, in We can find semantically corresponding relational edges as support;

[0036] In this step, All represent the target entity. Both represent matching source entities. All represent source maps. All represent target maps.

[0037] As a further step, the specific method of S34 includes:

[0038] S341. For each candidate pair ,in Extract respectively exist neutralization exist Local relational context and Determine the consistency of the execution relationship constraints;

[0039] S342, By searching arrive The structural mapping is used to verify the compatibility of the two in relational structures: for Each neighbor node in Find the corresponding node with type compatibility in the middle; check Each of the following is For the endpoints of the relation edge, the nodes at both ends are mapped to Between corresponding nodes, are there relationship edges of the same type or semantic compatibility? If the proportion of successfully matched relationship edges is not less than the fault tolerance threshold... If the two structures are found to be compatible, the verification passes and the process proceeds directly to S35. If the verification fails, the process proceeds to S343.

[0040] S343. Activate the collaborative decision-making algorithm to determine a unique match, and calculate each candidate... Overall rating And based on a comprehensive score highest As the final alignment object, an alignment pair is formed. ;

[0041] Overall score The calculation formula is:

[0042] (1);

[0043] in, The preset weighting coefficients, The topology compactness score evaluates the alignment coherence of the neighboring nodes of both sides after alignment.

[0044] (2);

[0045] in, Indicates quantization candidate alignment pairs The degree of consistency in the overall relationship graph structure.

[0046] Indicates in Local relation neighborhood In China, it has already been with neighborhood The number of effectively aligned neighbor node pairs established by the middle node;

[0047] Indicates in neighborhood In theory, it can be with neighborhood The total number of neighboring nodes that are compatible with the node type;

[0048] Attribute similarity score, used for evaluation and Similarity of key attributes:

[0049] (3);

[0050] This represents the preset weights used to balance the importance of geometric and semantic attributes. The final score is normalized to the [0,1] interval.

[0051] For geometric attribute similarity, For key semantic attribute similarity,

[0052] Choose overall rating highest As the final alignment object, an alignment pair is formed. .

[0053] As a further step, the global alignment mapping set Including all confirmed cross-stage entity equivalence pairs, formally defined as:

[0054] and and Pointing to the same object in the real world ;

[0055] in, This is called an alignment pair.

[0056] As a further step, the incremental calculation range in S4 is denoted as , ,in, To represent a set of changed nodes, it refers to the entity nodes that have undergone add, delete, or modify operations; To influence the edge set, it refers to all edges in the global alignment map set that are related to... Predefined relation edges that directly connect nodes in the middle.

[0057] This invention also provides a component alignment and knowledge graph construction system based on BIM models, including:

[0058] Data source input module: Used to input data from the design and construction phases into the BIM model as raw data;

[0059] Multi-stage relation graph construction module: used to extract node attributes and relationships from the raw data through a parser, and construct a relation graph for each stage;

[0060] Relationship Consistency Consistency Alignment Module: Used to find the unique corresponding entity in the source graph for each entity in the target graph and output a global alignment mapping set;

[0061] Dynamic update module: Used to update the global alignment mapping set in real time based on changes in the BIM model.

[0062] As a further improvement, the relation consistency constraint alignment module includes a relation consistency constraint entity alignment unit and a relation consistency verification unit. The relation consistency constraint entity alignment unit is used to filter candidate pairs of target entities, and the relation consistency verification unit is used to determine the unique candidate pair of target entities.

[0063] The features and beneficial effects of this invention are as follows:

[0064] (1) This invention directly utilizes the clearly defined semantic and topological relationships within the BIM models at each stage to construct a unified fusion knowledge graph as an alignment framework. By enforcing relationship consistency verification, it ensures that alignment results at any different stage can continue the original design intent and engineering logic. Simultaneously, a relationship consistency verification mechanism is designed and implemented. This mechanism not only relies on geometric or attribute similarity but also requires candidate alignment results to satisfy structural compatibility within their respective stage relationship subgraphs, ensuring that alignment decisions are consistent with the overall design intent and engineering logic, thereby achieving accurate and reliable alignment at the semantic level. This invention transforms alignment from a one-time task into a continuously maintainable process. By defining incremental alignment domains and local realignment mechanisms, the system can automatically lock the scope of impact and quickly update the alignment status with minimal computational cost when the model changes at any stage, while always maintaining the continuity and consistency of the global relationship network. This allows the alignment results to dynamically evolve throughout the entire project lifecycle, providing reliable support for data-driven real-time decision-making.

[0065] (2) In view of the heterogeneous, complex and continuously evolving system characteristics of BIM data at different stages, this invention proposes an intelligent alignment method "with the existing relational network as a constraint framework". This method does not rely on external calculation or repeated reasoning, but directly uses the predefined complete relations in the BIM model at each stage as the core constraint to construct a fused knowledge graph that can guide, verify and continuously maintain the alignment process, thereby achieving accurate, consistent and sustainable association of BIM components at different stages at the structural and semantic levels.

[0066] (3) This invention directly connects to BIM data sources that already contain rich predefined relationships. Through a lightweight parser mapper, these predefined components, attributes, and their multi-dimensional relationships are directly converted into a structured knowledge graph. The nodes and relationship edges in this graph are directly inherited from the results confirmed in the design or analysis phase, thereby ensuring its authority and accuracy as an alignment benchmark framework.

[0067] (4) This invention uses global relationship consistency as the core criterion for alignment decision-making. Unlike traditional "point-to-point" matching that relies solely on geometric or attribute similarity, the alignment engine of this invention introduces a pre-constructed fusion relationship graph as a global constraint when performing entity matching. During the matching process, not only is the feature similarity between candidate entities calculated, but more importantly, relationship consistency verification is performed. This mechanism ensures that the alignment result, as a whole, can perfectly embed and maintain the design intent and associated logic at different stages, fundamentally solving the problems of semantic distortion and logical contradiction in the alignment result.

[0068] (5) This invention treats alignment as a dynamically maintainable process. When the input data undergoes local changes, the system can automatically identify the affected domain and, based on a complete relational network, trigger re-alignment calculations only for entities within the affected region and their associated entities. This incremental update mechanism strictly follows the constraints of the original relational network, ensuring the continuity and consistency of the alignment results between the old and new versions in the global relational logic. This allows the final generated fused knowledge graph to continuously and stably reflect the latest engineering status, providing a reliable data foundation for intelligent applications based on the alignment results. Attached Figure Description

[0069] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0070] Figure 1 This is a flowchart of the component alignment and knowledge graph construction method based on the BIM model according to an embodiment of the present invention;

[0071] Figure 2 This is a detailed schematic diagram of the internal relationship map of the design stage or construction stage as described in the embodiments of the present invention;

[0072] Figure 3 This is an overall schematic diagram of the internal relationship map of the design stage or construction stage as described in the embodiments of the present invention. Detailed Implementation

[0073] To facilitate understanding of the present invention, a more comprehensive description of the present invention will be given below, and embodiments of the present invention will be provided, but this does not limit the scope of the present invention.

[0074] Definitions:

[0075] BIM (Building Information Modeling) is a three-dimensional digital technology-based method for managing the entire lifecycle of a building. It integrates geometric and non-geometric information (such as materials, costs, schedules, etc.) of a building project through parametric modeling to achieve data collaboration and sharing across the design, construction, and operation and maintenance stages.

[0076] Component alignment: Component alignment refers to the process of establishing and maintaining a one-to-one, semantically defined mapping relationship between digital component instances representing the same physical or functional entity in BIM models across different stages of a building project's lifecycle, including design, construction, and operation and maintenance. This process uses the inherent relationship network within each stage's model as a structured constraint, ensuring the accuracy and uniqueness of the mapping relationship through consistency checks of spatial, semantic, and relational contexts. Simultaneously, this process possesses incremental update capabilities, responding to local changes in the model at specific stages. Through constrained recalculation, it efficiently updates and maintains overall consistency across stage mapping relationships, thereby supporting multi-stage data association and traceability based on BIM.

[0077] Converting BIM models into machine-understandable and reasonable knowledge graphs is a key step in realizing intelligent applications. This conversion process aims to extract the scattered component entities, attributes, and their diverse relationships (such as spatial topology, system composition, and functional associations) from the BIM, and reorganize and express them according to the "entity-relationship-entity" triple structure, thereby forming a structured semantic network that provides a data foundation for in-depth analysis and decision-making. However, at different stages of a project (such as design, construction, and operation and maintenance), different BIM models or data representations with different focuses will be generated due to the different goals, participants, and usage requirements at each stage. For example, the design model focuses on geometry and aesthetics, the construction model needs to integrate process and schedule information, and the operation and maintenance model focuses on equipment attributes and maintenance records. Although these models describe the same entity, they often differ in geometric accuracy, attribute detail, and semantic structure. Therefore, to achieve the integration and collaboration of data throughout the entire lifecycle, it is necessary to accurately align BIM entities at different stages to establish semantic mapping and association relationships across different stages and sources. In addition, this invention designs an efficient dynamic maintenance mechanism. When a local change in data is detected, the system can quickly locate the affected entity and its associated domain based on the existing relationship network, and perform collaborative realignment only on that local area, thereby achieving intelligent and incremental updates to the alignment results. This has significant theoretical innovation significance and broad engineering application prospects.

[0078] like Figures 1 to 3 As shown, this invention provides a method for component alignment and knowledge graph construction based on a BIM model, including the following steps:

[0079] S1. Input data from multiple stages such as design and construction into the BIM model as raw data;

[0080] S2. Analyze the data from each stage of the original data, extracting the attributes, geometry, and internal relationships of each entity. Based on the extracted data, construct a relationship graph for each stage.

[0081] S3, Source Map As a baseline, for the target map Each entity in the search finds its in The unique corresponding entity in the map, and output the global alignment map set. ;

[0082] S4. When BIM data is updated at any stage, the BIM model automatically determines the set of change nodes. and its influence edge set Define the scope of incremental calculation Only in Within the specified range, the alignment process of S3 is repeated. This process reuses existing, unaffected alignment results for efficient local computation and consistency adjustments. Finally, the latest global alignment map set is updated and output. 新 This enables sustainable and precise maintenance of cross-stage entity relationships.

[0083] This invention directly utilizes the clearly defined semantic and topological relationships within the BIM models at each stage to construct a unified, integrated knowledge graph as an alignment framework. By enforcing relationship consistency checks, it ensures that alignment results at any different stage maintain the original design intent and engineering logic. Simultaneously, a relationship consistency check mechanism is designed and implemented. This mechanism not only relies on geometric or attribute similarity but also requires candidate alignment results to satisfy structural compatibility within their respective stage's relationship subgraphs, ensuring that alignment decisions are consistent with the overall design intent and engineering logic, thereby achieving accurate and reliable alignment at the semantic level. This invention transforms alignment from a one-time task into a continuously maintainable process. By defining incremental alignment domains and local realignment mechanisms, the system can automatically lock the scope of impact and quickly update the alignment status with minimal computational cost when the model changes at any stage, while always maintaining the coherence and consistency of the global relationship network. This allows alignment results to dynamically evolve throughout the project's entire lifecycle, providing reliable support for data-driven real-time decision-making.

[0084] In some embodiments, the method for constructing an internal relationship graph of a corresponding stage includes:

[0085] S21. Construct an internal relationship graph G. ,

[0086] in, This represents a set of nodes, whose elements are entities in the BIM model corresponding to the current stage.

[0087] Represents a set of relation types used to describe various predefined semantic relationships between entities (such as ConnectsTo, Contains, IsPartOf).

[0088] Let represent the set of directed edges, and .

[0089] Wherein, the directed edge in the set of directed edges is denoted as e.

[0090] , indicating from node To the node There exists a relation of type r∈R. ;

[0091] S22. Record the design phase diagrams as the source diagrams. ,but , Represents the set of nodes in the source graph; Let represent the set of directed edges in the source graph, and The map of the target phase (such as construction and operation and maintenance) is recorded as the target map. ,but . Represents the set of nodes in the target graph; Let represent the set of directed edges in the target graph, and Source map and target map Shared Relationship Type Set .

[0092] like Figure 2 and Figure 3 As shown in the diagram, purple circles represent nodes, orange circles represent attribute sets, and blue circles represent attributes. These relationship graphs fully preserve the authority relationship constraints within each stage, providing a structured contextual basis for subsequent cross-stage alignment.

[0093] In some embodiments, S3 uses the source map As a baseline, for the target map Each entity in the search finds its in The unique corresponding entity in the map, and output the global alignment map set. The specific methods of delivery include:

[0094] S31. Feature Extraction and Candidate Set Generation: For target entities (representing the set of nodes in the target graph), extracting its geometric features (geometric coordinates, triangular faces), semantic features (name, key attributes), and topological features (in... (Neighbor distribution in the data).

[0095] S32, for the target entity In the source map Filter potential matching source entities In the source map Spatial-semantic collaborative retrieval is performed to generate its final candidate matching set. Specifically, it includes:

[0096] S321, Based on target entity Global coordinates in the source map Retrieve all objects with a distance less than a threshold from the spatial index. The entities generate a spatial candidate set. ;

[0097] S322. Based on a predefined cross-stage type mapping table, from the spatial candidate set... Filter out entities matching the target entity Type-compatible entities form a semantic candidate set. ;

[0098] The cross-stage type mapping table is a predefined table that records the semantic equivalence or derivation relationships of component types at different stages. This table ensures effective semantic matching even when there are stage differences in component type naming or classification standards. The mapping table is shown in Table 1.

[0099] Table 1

[0100]

[0101] S323, on semantic candidate sets Each candidate entity in Perform a traversal in the source graph. Retrieve all entities that are connected to the source entity through predefined relation edges. First-degree neighbors that are directly connected are added to the candidate set. (Initially an empty set), after traversal, this candidate set will be... and Merge (i.e., take the union of the two), and remove duplicates based on the unique identifier (id) of the nodes: if two nodes have the same id, they are considered the same entity, and only one is kept, thus generating the final candidate matching set. .

[0102] S33. Establish consistency constraint conditions for relationships:

[0103] Candidate alignment pairs The requirement for it to be valid is: if and Alignment, then exist Local relational context Must with exist Local relational context It is structurally compatible.

[0104] The exist Local relational context For A one-hop neighbor subgraph centered on the center, containing Self, All and Directly connected nodes and the edges connecting these nodes.

[0105] Similarly, exist Local relational context For A one-hop neighbor subgraph centered on the center, containing Self, All and Directly connected nodes and the edges connecting these nodes.

[0106] The compatibility requirements will The neighbor nodes in the mapping are mapped to Find compatible nodes and verify them. For each relation edge in the graph, between its mapped node pairs, There exist semantically corresponding relation edges in each of them. This makes... The observed relation edges and their types, in The semantically corresponding relational edges can be found as support; this constraint ensures that the alignment does not disrupt the inherent design logic and physical relationship.

[0107] This application provides a highly reliable contextual constraint framework for entity alignment. Traditional methods typically require re-parseing and calculating entity relationships from the original structure, a complex process prone to losing high-level semantics. This invention, however, directly interfaces with BIM data sources that already contain rich predefined relationships. Through a lightweight parser mapper, it directly transforms these predefined components, attributes, and their diverse relationships into a structured knowledge graph. The nodes and relationship edges in this graph directly inherit from results confirmed during the design or analysis phase, thus ensuring its authority and accuracy as an alignment benchmark framework. This method skips the tedious and error-prone relationship calculation or reasoning process, providing a stable, reliable, and semantically rich logical constraint foundation for subsequent alignment steps.

[0108] S34. Specific methods for verifying relational consistency include:

[0109] S341. For each candidate pair ,in Extract respectively exist neutralization exist Local relational context and Determine the consistency of the execution relationship constraints;

[0110] S342, By searching arrive The structural mapping is used to verify the compatibility of the two in relational structures: for Each neighbor node in Find the corresponding node with type compatibility in the middle; check Each of the following is For the endpoints of the relation edge, the nodes at both ends are mapped to Between corresponding nodes, are there any edges of the same type or semantic compatibility? If the proportion of successfully matched edges is not less than the fault tolerance threshold... If both structures are compatible, the verification passes. There is only one candidate entity Upon successful verification, proceed directly to S35;

[0111] S343, If one There are multiple candidate entities If the verification is successful, the collaborative decision-making algorithm is activated to determine a unique match, and each candidate entity is calculated. Overall rating And based on a comprehensive score highest As the final alignment object, an alignment pair is formed. .

[0112] Overall score The calculation formula is:

[0113] (1);

[0114] in, The preset weighting coefficients, The topology compactness score evaluates the alignment coherence of the neighboring nodes of both sides after alignment.

[0115] (2);

[0116] in, Indicates quantization candidate alignment pairs The degree of consistency in the overall relationship graph structure.

[0117] Indicates in Local relation neighborhood In China, it has already been with neighborhood The number of effectively aligned neighbor node pairs established by the middle node.

[0118] Indicates in neighborhood In theory, it can be with neighborhood The total number of neighboring nodes that are compatible with the node type.

[0119] Attribute similarity score, used for evaluation and Similarity of key attributes between the two:

[0120] (3);

[0121] This represents the preset weights used to balance the importance of geometric and semantic attributes. The final score is normalized to the [0,1] interval.

[0122] Geometric attribute similarity, used for calculation and The axial bounding boxes of the two are similar in size in the three dimensions of length, width, and height.

[0123] For example, using the complement of the relative error: The average of each dimension is calculated. This refers to the dimension of the target component in a specified dimension. The dimension values ​​of the source candidate components in the same dimension.

[0124] This refers to the similarity of key semantic attributes, used to select stable and discriminative attributes, such as `material` and `systemType`. Methods such as cosine similarity can be used on string descriptions, for example... .in, The concrete grade values ​​represent the target component and the source component. The value represents the wall thickness, and F represents the fire resistance rating. If the ratings are the same, the value is 1.0; otherwise, it is 0.

[0125] Substitute the calculation results of formulas (2) and (3) into formula (1) and select the comprehensive score. highest As the final alignment object, an alignment pair is formed. .

[0126] S35. Generate a global alignment map set: Gather all the finally determined alignment pairs to form a global alignment map set. Output directly.

[0127] Global Alignment Map Set Including all confirmed cross-stage entity equivalence pairs, formally defined as:

[0128] and and Pointing to the same object in the real world ;

[0129] in, This is called an alignment pair, and the equivalence relation can be denoted as `correspondsTo`.

[0130] This application uses global relational consistency as the core criterion for alignment decisions. Unlike traditional "point-to-point" matching that relies solely on geometric or attribute similarity, the alignment engine of this invention introduces a pre-constructed fusion relation graph as a global constraint when performing entity matching. During the matching process, not only is the feature similarity between candidate entities calculated, but more importantly, relational consistency verification is performed: that is, verifying that if target stage entity A is aligned with source stage entity B, then the local relational network (such as connection and containment relationships) of A in its own model must be structurally compatible with the predefined relational network of B in the source stage model. This mechanism ensures that the alignment result, as a whole, can perfectly embed and maintain the design intent and associated logic of different stages, fundamentally solving the problems of semantic distortion and logical contradictions in the alignment result.

[0131] In some embodiments, when a change occurs in the BIM model at a certain stage, the directly changed components and their directly connected neighboring components in the relationship graph are identified, together constituting the incremental calculation scope. , ,in, To represent a set of changed nodes, it refers to the entity nodes that have undergone add, delete, or modify operations. To influence the edge set, it refers to all edges in the global alignment map set that are related to... Predefined relational edges that directly connect nodes in the middle. The nodes connected by these edges constitute the local range that needs to be re-verified and aligned.

[0132] To address the challenge of continuous iteration between BIM models and site data, this invention treats alignment as a dynamically maintainable process. When local changes occur in the input data, the system automatically identifies the affected domain and, based on a complete relational network, triggers re-alignment calculations only for entities within the affected area and their associated entities. This incremental update mechanism strictly adheres to the constraints of the original relational network, ensuring the consistency and coherence of the alignment results between the old and new versions in terms of global relational logic. This allows the final generated fused knowledge graph to continuously and stably reflect the latest engineering status, providing a reliable data foundation for intelligent applications based on the alignment results.

[0133] In this application, All represent the target entity. Both represent matching source entities. All represent source maps. All represent target maps.

[0134] This invention provides a method for component alignment and knowledge graph construction based on BIM models. Its core lies in abandoning the traditional approach of recalculating or completing relationships from raw data, and creatively proposing a new alignment paradigm "with a relational network as a constraint framework." Unlike existing methods, this invention directly utilizes the complete spatial topology and semantic relationships pre-calculated and existing in the BIM model, constructing them into a strongly constrained knowledge graph that serves as the core verification framework. During the alignment process, it verifies in real time whether candidate matches are compatible with the existing relational network, thereby driving and correcting decisions. This achieves accuracy, semantic consistency, and dynamic maintainability in cross-modal entity matching.

[0135] Example

[0136] The No. 3 Tunnel construction project has established a design model (No. 3 Tunnel-Design.json) and a refined construction model (No. 3 Tunnel-Construction.json). Both models contain core structural components and some equipment components, but the component IDs, geometric details, and some attributes differ. The goal of this embodiment is to generate a global alignment mapping set for the components between these two models.

[0137] A1. Predefined core parameters:

[0138] Spatial proximity threshold (λ): set according to engineering precision, for example, 0.5 meters for structural components and 2.0 meters for equipment such as air ducts.

[0139] Cross-stage type mapping table: predefined semantic equivalence rules, such as {"BasicWall": "ShearWall", "StructuralColumn": "CastInPlaceColumn"}.

[0140] Relationship consistency tolerance threshold (θ): The upper limit of the proportion of missing local relations that can be allowed, for example, set to 0.3.

[0141] Decision weight (α): A parameter used to weigh topological structure against attribute similarity, for example, 0.6.

[0142] A2. Specific implementation steps:

[0143] The design model and construction model are analyzed separately. Based on information such as the linked nodes of the components, two independent internal relationship graphs for each stage are constructed: Source Graph (Design) and Target Map (Construction). The nodes in the graph are components, and the edges are predefined relationships such as "connection" and "containment" within the model.

[0144] A21. Examples of components used in the design phase are as follows:

[0145] A211, Design Wall :

[0146] Unique ID: D_Wall_001;

[0147] Component type: BasicWall;

[0148] Component Name: Basic Wall: W-1:C40:2600 (Indicates that the number is W-1, the concrete strength is C40, and the thickness is 2600 mm).

[0149] This component is directly associated with the three components D_Beam_001, D_Column_001, and D_Slab_001. It provides the 3D coordinates of 8 vertices: [

[0150] {"X": 1.3989357, "Y": 1.3, "Z": -4.715},

[0151] {"X":-1.3989357, "Y": 1.3, "Z": 4.715},

[0152] {"X": 1.3989357, "Y": 1.3, "Z": 4.715},

[0153] {"X":-1.3989357, "Y": 1.3, "Z": -4.715},

[0154] {"X":-1.3989357, "Y":-1.3, "Z": 4.715},

[0155] {"X":-1.3989357, "Y":-1.3, "Z": -4.715},

[0156] {"X": 1.3989357, "Y":-1.3, "Z": -4.715},

[0157] {"X": 1.3989357, "Y":-1.3, "Z": 4.715}

[0158] A cuboid is defined in the local coordinate system. The transformation matrix contains rotation and scaling components: [-1.0,0.0, 0.0, 0.0, 0.0, -1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, ...], and translation components: [495166.2219305616,4319129.367084531,-11.364999847412108,1.0].

[0159] A212, Design Wall :

[0160] Unique ID: D_Wall_002;

[0161] Component type: BasicWall;

[0162] Component Name: Basic Wall: W-2:C45:700 (Indicates that the number is W-2, the concrete strength is C45, and the thickness is 700 mm).

[0163] This component is directly associated with the three components D_Beam_002, D_Column_002, and D_Wall_003. It provides the 3D coordinates of 8 vertices: [

[0164] {"X": 1.1544762, "Y": 0.35, "Z": -4.715},

[0165] {"X":-1.1544762, "Y": 0.35, "Z": 4.715},

[0166] {"X": 1.1544762, "Y": 0.35, "Z": 4.715},

[0167] {"X":-1.1544762, "Y": 0.35, "Z": -4.715},

[0168] {"X":-1.1544762, "Y":-0.35, "Z": 4.715},

[0169] {"X":-1.1544762, "Y":-0.35, "Z": -4.715},

[0170] {"X": 1.1544762, "Y":-0.35, "Z": -4.715},

[0171] {"X": 1.1544762, "Y":-0.35, "Z": 4.715}

[0172] A cuboid is defined in the local coordinate system. The transformation matrix contains rotation and scaling components: [1.0,0.0, 0.0, 0.0, 0.0, -1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, ...], and translation components: [495135.73022672953,4319130.816903769,-11.364999847412108,1.0].

[0173] A213, Design Wall :

[0174] Unique ID: D_Wall_003;

[0175] Component type: BasicWall;

[0176] Component Name: Basic Wall: W-3:C35:1000 (indicating number W-3, concrete strength C35, thickness 1000 mm);

[0177] This component is directly associated with the three components D_Beam_003, D_Column_003, and D_Wall_002. It provides the 3D coordinates of 8 vertices: [

[0178] {"X": 3.8177214, "Y": 0.5, "Z": -3.29},

[0179] {"X":-3.8177214, "Y": 0.5, "Z": 3.29},

[0180] {"X": 3.8177214, "Y": 0.5, "Z": 3.29},

[0181] {"X":-3.8177214, "Y": 0.5, "Z": -3.29},

[0182] {"X":-3.8177214, "Y":-0.5, "Z": 3.29},

[0183] {"X":-3.8177214, "Y":-0.5, "Z": -3.29},

[0184] {"X": 3.8177214, "Y":-0.5, "Z": -3.29},

[0185] {"X": 3.8177214, "Y":-0.5, "Z": 3.29}

[0186] A cuboid is defined in the local coordinate system. The transformation matrix contains rotation and scaling components: [1.0,0.0, 0.0, 0.0, 0.0, -1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, ...], and translation components: [495153.688489231,4319146.2169920765,-11.340000038146973,1.0].

[0187] A22. Components during the construction phase :

[0188] Unique ID: C_Wall_001;

[0189] Component type: Shear wall;

[0190] Component name: Shear wall: WQ-1:C45:2620 (indicating that the number is WQ-1, the concrete strength is C45, and the thickness is 2620 mm).

[0191] This component is directly associated with the three components C_Beam_001, C_Column_001, and C_Slab_001. It provides the 3D coordinates of 8 vertices: [

[0192] {"X": 1.3989357, "Y": 1.31, "Z": -4.715},

[0193] {"X": -1.3989357, "Y": 1.31, "Z": 4.715},

[0194] {"X": 1.3989357, "Y": 1.31, "Z": 4.715},

[0195] {"X": -1.3989357, "Y": 1.31, "Z": -4.715},

[0196] {"X": -1.3989357, "Y": -1.31, "Z": 4.715},

[0197] {"X": -1.3989357, "Y": -1.31, "Z": -4.715},

[0198] {"X": 1.3989357, "Y": -1.31, "Z": -4.715},

[0199] {"X": 1.3989357, "Y": -1.31, "Z": 4.715}

[0200] A cuboid is defined in the local coordinate system. The transformation matrix contains rotation and scaling components: [-0.999, 0.01, 0.0, 0.0, -0.01, -0.999, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, ...], and translation components: [

[0201] 495166.2219305616, 4319129.367084531, -11.364999847412108, 1.0 ];

[0203] A23, with target map Target components Example: C_Wall_001 Shear Wall:WQ-1:C45:2620)

[0204] Candidate generation: extraction The geometric coordinates and semantic types are used for spatial retrieval (threshold λ) and semantic filtering (type mapping table) in the design graph. Rapidly locate the initial candidate set .

[0205] Relationship verification: For each candidate ,extract and Each local relation neighborhood subgraph and The consistency constraint of the relation is used for verification to determine... Can the relational schema in [the context of the relational schema] be [used]? The structure accommodates, that is, allows for omissions within the range of θ.

[0206] 0162. Uniqueness Decision: If more than one candidate passes the verification, calculate the comprehensive score of each candidate. The one with the highest score is selected as the sole alignment target.

[0207] According to formula (1) It is necessary to calculate the topology compactness score. Similarity score of attributes .

[0208] According to formula (2) Calculate the topology density score ( ):

[0209] Score: Of the three neighbors, two have existing alignment relationships with... Neighbor alignment is consistent → = 2 / 3 ≈ 0.6667, the specific process is as follows:

[0210] The three neighbors of (construction wall C_Wall_001) are C_Beam_001, C_Column_001, and C_Slab_001; The three neighbors of (design wall D_Wall_001) are D_Beam_001, D_Column_001, and D_Slab_001.

[0211] Among them, "there are 2 existing alignment relationships with" "Neighbor alignment coherence" means that in the existing alignment mapping, there exist two pairs of neighbor alignment relationships:

[0212] 1. C_Beam_001 ↔ D_Beam_001 (The construction beam is aligned with the design beam);

[0213] 2.C_Column_001 ↔ D_Column_001 (Construction column and design column are aligned);

[0214] The third neighbor, C_Slab_001, failed the verification and there is no alignment relationship between it and D_Slab_001. Therefore, only 2 neighbors achieve "alignment coherence", and the score is 2 / 3.

[0215] Score: neighbors and Neighbors are not aligned and coherent → = 0;

[0216] Score: neighbors and Neighbors are not aligned and coherent → = 0;

[0217] According to formula (3) Calculate attribute similarity score ( ):

[0218] Calculation of similarity of length, width, and height dimensions of the bounding box along the axis :

[0219] Calculate the geometric similarity of the four elements in terms of length, width, and height. And take the average. This refers to the dimension of the target component in a specified dimension. The dimension values ​​of the source candidate components in the same dimension.

[0220] The dimension value of the source candidate component in the length dimension. The dimension value of the source candidate component in the width dimension. The dimension value of the source candidate component in the height dimension. This refers to the dimension of the target component in the length dimension. This refers to the dimension of the target component in the width dimension. The dimension of the target component in the height dimension.

[0221] Basic wall: W-1:C40:2600 in length dimension 1.3989357 - (-1.3989357) = 2.7978714;

[0222] Basic wall: W-1:C40:2600 in width dimension 1.3 - (-1.3) = 2.6;

[0223] Basic wall: W-1:C40:2600 in height dimension 4.715 - (-4.715) = 9.43;

[0224] Basic wall: W-2:C45:700 in length dimension 1.1544762 - (-1.1544762) = 2.3089524;

[0225] Basic wall: W-2:C45:700 in width dimension 0.35 - (-0.35) = 0.7;

[0226] Basic wall: W-2:C45:700 in height dimension 4.715 - (-4.715) = 9.43;

[0227] Basic wall: W-3:C35:1000 in length dimension 3.8177214 - (-3.8177214) = 7.6354428;

[0228] Basic wall: W-3:C35:1000 in width dimension 0.5 - (-0.5) = 1.0;

[0229] Basic wall: W-3:C35:1000 in height dimension 3.29 - (-3.29) = 6.58;

[0230] Shear wall: WQ-1:C45:2620 in length dimension 1.3989357 - (-1.3989357) = 2.7978714;

[0231] Shear wall: WQ-1:C45:2620 in width dimension 1.31 - (-1.31) = 2.62;

[0232] Shear wall: WQ-1:C45:2620 in height dimension 4.715 - (-4.715) = 9.43;

[0233] Compared to basic wall: W-1:C40:2600 and shear wall: WQ-1:C45:2620:

[0234] according to ,

[0235] Geometric similarity of length: ;

[0236] Geometric similarity of width: ;

[0237] High geometric similarity: ;

[0238] Geometric property similarity between basic wall: W-1:C40:2600 and shear wall: WQ-1:C45:2620 .

[0239] Compared to basic wall: W-2: C45: 700 and shear wall: WQ-1: C45: 2620:

[0240] according to ,

[0241] Geometric similarity of length: ;

[0242] Geometric similarity of width:

[0243] High geometric similarity: ;

[0244] Geometric property similarity between basic wall: W-2:C45:700 and shear wall: WQ-1:C45:2620 = .

[0245] Compared to basic wall: W-3:C35:1000 and shear wall: WQ-1:C45:2620:

[0246] according to ,

[0247] Geometric similarity of length: ;

[0248] Geometric similarity of width: ;

[0249] High geometric similarity: ;

[0250] Geometric property similarity between basic wall: W-3:C35:1000 and shear wall: WQ-1:C45:2620 = .

[0251] Key semantic attribute similarity Comparing three key attributes—concrete strength, fire resistance rating, and wall thickness—the strengths of the following basic wall types are quantified as follows: Basic wall:W-1:C40:2600 (C40), Basic wall:W-2:C45:700 (C45), and Basic wall:W-3:C35:1000 (C35) are quantified as 40, 45, and 35 respectively. The strengths of basic wall types W-1:C40:2600, W-2:C45:700, and W-3:C35 are also quantified as follows. The fire ratings of shear wall :1000 and shear wall :WQ-1:C45:2620 are the same, and the similarity is set to 1.0. The similarity between shear wall :WQ-1:C45:2620 and the three key attributes of basic wall :W-1:C40:2600, basic wall :W-2:C45:2600, basic wall :W-2:C45:2600, and basic wall :W-3:C35:2650 is calculated and averaged.

[0252] according to ,in, The concrete grade values ​​represent the target component and the source component. The value represents the wall thickness, and F represents the fire resistance rating. If the ratings are the same, the value is 1.0; otherwise, it is 0.

[0253] The result is: Basic wall: W-1:C40:2600 =40, =2600, F=1;

[0254] Basic wall: W-2:C45:700 =45, =700, F=1;

[0255] Basic wall: W-3:C35:1000 =35, =1000, F=1;

[0256] Shear wall: WQ-1:C45:2620 =45, =2620, F=1;

[0257] ;

[0258] ;

[0259] ;

[0260] Set the weights for geometric attribute similarity (e.g., β=0.7) and semantic attribute similarity (e.g., 1-β=0.3) and calculate...

[0261] The scores are as follows:

[0262] According to formula (3) =β× + (1-β)× ;

[0263] =0.7× + 0.3× = 0.7×0.9974+0.3×0.9604=0.9863;

[0264] The scores are as follows:

[0265] According to formula (3) =β× + (1-β)× ;

[0266] = 0.7× + 0.3× = 0.7×0.6974+0.3×0.7557=0.7148;

[0267] The scores are as follows:

[0268] According to formula (3) =β× + (1-β)× ;

[0269] =0.7× + 0.3× = 0.7×0.4819+0.3×0.7198=0.5532;

[0270] Substitute the values ​​obtained above into formula (1). calculate:

[0271] Preset weighting coefficients It is 0.6;

[0272] =0.6667, =0.9863, S1= 0.6×0.6667 + 0.4×0.9863=0.7945;

[0273] =0, =0.7148, S2= 0.6×0 + 0.4×0.7148=0.2859;

[0274] =0, =0.5532, S3= 0.6×0 + 0.4×0.5532=0.2213.

[0275] 0.9863 > 0.2859 > 0.2213, The one with the highest overall score was selected as the construction wall. The only alignment object.

[0276] Output mapping set: The finalized alignment pairs Add to global alignment map set Traversal Repeat the above process for all components to complete batch alignment.

[0277] A24. When the construction model undergoes partial changes:

[0278] Determine the influence domain: Identify the directly altered components and their directly connected neighboring components in the relationship graph, which together constitute the incremental calculation scope. .

[0279] Local realignment: only when Within the specified range, re-execute the alignment process for A23.

[0280] Update the mapping set: Incrementally update the global alignment mapping set based on the local realignment results. To maintain its up-to-dateness and consistency.

[0281] Through the above steps, this invention ultimately outputs a continuously updated global alignment mapping set. This set accurately records the one-to-one correspondence between design components and construction components, providing a core foundation for cross-stage data association, progress tracking, and information integration. This embodiment verifies the effectiveness and practicality of the solution in terms of accuracy, efficiency, and dynamic adaptability.

[0282] A component alignment and knowledge graph construction system based on a BIM model, comprising:

[0283] Data source input module: Used to input data from multiple stages such as design and construction into the BIM model as raw data;

[0284] Multi-stage relation graph construction module: used to extract node attributes and relationships from the raw data through a parser, and construct a relation graph for each stage;

[0285] Relationship Consistency Consistency Alignment Module: Used to find the unique corresponding entity in the source graph for each entity in the target graph and output a global alignment mapping set;

[0286] Dynamic update module: Used to update the global alignment mapping set in real time based on changes in the BIM model.

[0287] The dataset of the BIM model is represented as follows:

[0288] (1) Node ID field id: uniquely identifies each node.

[0289] (2) Linkednodes field: Records a list of IDs of other nodes that are directly connected to the current node.

[0290] (3) Data type field dataType: identifies the type of data.

[0291] (4) Node name field name: Identifies the name of the node.

[0292] (5) The shapegeometry field indicates whether a node contains geometric data. It includes the vertex (vertices) field, the face (faces) field, and the transform (transform) field. The coordinates of all vertices are extracted from the vertex (vertices) field, and each vertex is represented as a 3D vector. The vertex indices of each face are then extracted from the face (faces) field. Each face defines a polygon, and the face is formed by connecting the vertices corresponding to these vertex indices. Finally, the 4x4 transform matrix is ​​extracted from the transform field. Each vertex is transformed from the local coordinate system to the global coordinate system.

[0293] The relation consistency constraint alignment module includes a relation consistency constraint entity alignment unit and a relation consistency verification unit. The relation consistency constraint entity alignment unit is used to filter candidate pairs, and the relation consistency verification unit is used to determine a unique candidate pair.

[0294] The process of the BIM model-based component alignment and knowledge graph construction system is as follows:

[0295] enter:

[0296] data1: Phase 1 BIM data

[0297] data2: Phase Two BIM Data

[0298] Output:

[0299] alignment_mappings: Set of alignment maps

[0300] start

[0301] / / Step 1: Data Preprocessing (Multi-stage Relationship Graph Construction Module)

[0302] Network 1 G_1, Network 2 G_2 = Data Preprocessing (data1, data2)

[0303] / / Step 2: Initial Alignment (Relation Consistency Consistency Alignment Module)

[0304] Current alignment map set M = Cross-stage alignment (G_1, G_2)

[0305] / / Step 3: Check for updates

[0306] If update_data is not empty:

[0307] / / Dynamic updates (dynamic update module)

[0308] M = Dynamically update(current alignment map set M, update_data)

[0309] Finish

[0310] Return to M

[0311] End the program.

[0312] The present invention also provides a computer device applicable to a component alignment and knowledge graph construction method based on a BIM model, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the component alignment and knowledge graph construction method based on a BIM model as proposed in the above embodiments.

[0313] The present invention also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements a component alignment and knowledge graph construction method based on a BIM model as proposed in the above embodiments.

[0314] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0315] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0316] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0317] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0318] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0319] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for component alignment and knowledge graph construction based on a BIM model, characterized in that: Includes the following steps: S1. Input the data from the design and construction phases into the BIM model as raw data; S2. Analyze the data of each stage in the original data, extract the features of each entity, and construct an internal relationship graph for each stage based on the extracted features. The graph of the design stage is denoted as the source graph. The maps used during the construction phase are recorded as target maps. ; S3, Source Map As a baseline, for the target map For each entity in the graph, find its unique counterpart in the source graph and output a global alignment map set; S4. When the BIM data changes at any stage, the BIM model automatically determines the set of changed nodes and its affected edge set, and defines the incremental calculation range; step S3 is repeated only within the incremental calculation range, and the latest global alignment mapping set is output.

2. The component alignment and knowledge graph construction method based on BIM model according to claim 1, characterized in that: In S2, the method for constructing a relationship graph within each corresponding stage includes: S21. Construct an internal relationship graph. , , in, This represents a set of nodes, whose elements are entities in the BIM model corresponding to the current stage. Represents a set of relation types used to describe various predefined semantic relationships between entities; Let represent the set of directed edges, and ; S22, Source Map Represented as: , Represents the set of nodes in the source graph; Let represent the set of directed edges in the source graph, and ; target map ; Represents the set of nodes in the target graph; Let represent the set of directed edges in the target graph, and Source map and target map Shared Relationship Type Set .

3. The component alignment and knowledge graph construction method based on BIM model according to claim 1, characterized in that: The specific method of S3 includes: S31. For the target entity Extract its geometric features, semantic features, and topological features, among which, ; S32, for the target entity In the source map Filter potential matching source entities In the source map Spatial-semantic collaborative retrieval is performed to generate its final candidate matching set. ; S33. Criteria for determining relation consistency constraints; S34. Determine the target entity through relation consistency verification. The only candidate, and form candidate alignment pairs; S35. Gather all the finally determined alignment pairs to form a global alignment map set. As direct output.

4. The component alignment and knowledge graph construction method based on BIM model according to claim 3, characterized in that: The specific method of S32 includes: S321, Based on target entity Global coordinates in the source map Retrieve all objects with a distance less than the threshold from the spatial index. The entities generate a spatial candidate set. ; S322. Based on a predefined cross-stage type mapping table, from the spatial candidate set... Filter out entities matching the target entity Type-compatible entities form a semantic candidate set. ; S323, on semantic candidate sets Each candidate entity in Perform a traversal in the source graph. Retrieve all edges that are connected to predefined relation edges First-degree neighbors that are directly connected are added to the candidate set. After traversing the set, this candidate set will be... With candidate set The nodes are merged and deduplicated based on their unique identifiers. If two nodes have the same unique identifier, they are considered the same entity, and only one is retained, ultimately generating the final candidate matching set. .

5. The component alignment and knowledge graph construction method based on BIM model according to claim 3, characterized in that: The specific method of S33 includes: Candidate alignment pairs The requirement for it to be valid is: if and Alignment, then exist Local relational context Must with exist Local relational context It is structurally compatible; The exist Local relational context For: with A one-hop neighbor subgraph centered on the center, containing Self, All and Directly connected nodes and the edges connecting these nodes; Similarly, exist Local relational context For: with A one-hop neighbor subgraph centered on the center, containing Self, All and Directly connected nodes and the edges connecting these nodes; The compatibility requirements will The neighbor nodes in the mapping are mapped to Find compatible nodes and verify them. For each relation edge in the graph, between its mapped node pairs, There exist semantically corresponding relation edges in each, such that The observed relation edges and their types, in We can find semantically corresponding relational edges as support; In this step, All represent the target entity. Both represent matching source entities. All represent source maps. All represent target maps.

6. The component alignment and knowledge graph construction method based on BIM model according to claim 3, characterized in that: The specific method of S34 includes: S341. For each candidate pair ,in Extract respectively exist neutralization exist Local relational context and Determine the consistency of the execution relationship constraints; S342, By searching arrive The structural mapping is used to verify the compatibility of the two in relational structures: for Each neighbor node in Find the corresponding node with type compatibility in the middle; check Each of the following is For the endpoints of the relation edge, the nodes at both ends are mapped to Between corresponding nodes, are there relationship edges of the same type or semantic compatibility? If the proportion of successfully matched relationship edges is not less than the fault tolerance threshold... If the two structures are found to be compatible, the verification passes and the process proceeds directly to S35. If the verification fails, the process proceeds to S343. S343. Activate the collaborative decision-making algorithm to determine a unique match, and calculate each candidate... Overall rating And based on a comprehensive score highest As the final alignment object, forming an alignment pair ; Overall score The calculation formula is: （1）； in, The preset weighting coefficients, The topology compactness score evaluates the alignment coherence of the neighboring nodes of both sides after alignment. （2）； in, Indicates quantization candidate alignment pairs The degree of consistency in the overall relational graph structure; Indicates in Local relation neighborhood In China, it has already been with neighborhood The number of effectively aligned neighbor node pairs established by the middle node; Indicates in neighborhood In theory, it can be with neighborhood The total number of neighboring nodes that are compatible with the node type; Attribute similarity score, used for evaluation and Similarity of key attributes between the two: （3）； This represents the preset weights used to balance the importance of geometric and semantic attributes; the final score is normalized to the [0,1] interval. For geometric attribute similarity, For key semantic attribute similarity, Choose overall rating highest As the final alignment object, forming an alignment pair .

7. The component alignment and knowledge graph construction method based on BIM model according to claim 3, characterized in that: The global alignment mapping set Including all confirmed cross-stage entity equivalence pairs, formally defined as: and and Pointing to the same object in the real world ; in, This is called an alignment pair.

8. The component alignment and knowledge graph construction method based on BIM model according to claim 1, characterized in that: The incremental calculation range in S4 is denoted as , ,in, To represent a set of changed nodes, it refers to the entity nodes that have undergone add, delete, or modify operations; To influence the edge set, it refers to all edges in the global alignment map set that are related to... Predefined relation edges that directly connect nodes in the middle.

9. A component alignment and knowledge graph construction system based on BIM models, characterized in that: include: Data source input module: Used to input data from the design and construction phases into the BIM model as raw data; Multi-stage relation graph construction module: used to extract node attributes and relationships from the raw data through a parser, and construct a relation graph for each stage; Relationship Consistency Consistency Alignment Module: Used to find the unique corresponding entity in the source graph for each entity in the target graph and output a global alignment mapping set; Dynamic update module: Used to update the global alignment mapping set in real time based on changes in the BIM model.

10. The component alignment and knowledge graph construction system based on BIM model according to claim 9, characterized in that: The relation consistency constraint alignment module includes a relation consistency constraint entity alignment unit and a relation consistency verification unit. The relation consistency constraint entity alignment unit is used to filter candidate pairs of target entities, and the relation consistency verification unit is used to determine the unique candidate pair of target entities.

Images