Heterogeneous data fusion structure primitive dynamic generation system and method for building information provision

Abstract

The invention relates to the technical field of building information processing, in particular to a heterogeneous data fusion structure primitive dynamic generation system and method for building information provisioning, the system comprises a multi-source analysis module, a semantic analysis module, a primitive generation module, a linkage update module, a layout optimization module and an output interface module, firstly, the multi-source analysis module unifies a building design data format; secondly, a semantic analysis module extracts features and constructs a structure knowledge graph; then, a primitive generation module creates parameterized primitives based on the atlas; the linkage updating module monitors changes and ensures synchronous updating of the primitives; the layout optimization module detects conflicts and automatically adjusts the layout; finally, the output interface module imports the optimized primitives into a CAD environment; according to the system, the design efficiency is remarkably improved, the detail drawing time is shortened by 85%, the error rate is reduced by 90% or above, the design quality is effectively improved, and creativity of designers is released.

Description

Technical Field

[0001] The present invention relates to the technical field of building information processing, and in particular to a system and method for dynamically generating structural elements by integrating heterogeneous data for building information improvement. Background Art

[0002] In the collaborative workflow of architectural and structural design, after completing the building design, architects need to provide building data (referred to as "submission") to structural engineers, who then perform structural layout and calculations based on this data. The traditional process from submission to structural drawing suffers from multiple issues: difficulty integrating heterogeneous data, inefficient manual drawing of structural elements, difficulty in quickly responding to building changes, and insufficient standardization of drawings.

[0003] In the existing technology, for example, Chinese patent CN109542984B discloses a "heterogeneous data fusion method for building structure testing." This method primarily performs heterogeneous data fusion at the data receiving level, converting data from different collectors into a unified JSON format and storing it in a relational database. While this method solves the problem of unified storage of heterogeneous data, it still has significant shortcomings: it lacks intuitive visual presentation capabilities and cannot accurately match the spatial location of collected data with architectural drawings; it lacks intelligent data analysis and pattern recognition capabilities and cannot automatically detect data anomalies; it fails to directly integrate with design documents such as CAD drawings or BIM models; data changes do not automatically trigger updates to structural models or drawings; and it cannot automatically generate structural detailed drawings based on collected data. These shortcomings result in the process from building funding to structural drawing still requiring a lot of manual intervention, which is inefficient and prone to errors.

[0004] Therefore, there is an urgent need for a system and method that can realize the automatic generation and dynamic update of building capital data into structural elements to improve the efficiency and accuracy of building structure design. Summary of the Invention

[0005] The purpose of the present invention is to provide a system and method for dynamically generating structural elements by integrating heterogeneous data for building investment. Through intelligent analysis and unified modeling of multi-source heterogeneous data, deep semantic analysis and construction of structural knowledge graphs, parameter-driven intelligent generation of structural elements, event-driven linkage update and version control of elements, as well as intelligent conflict detection and adaptive layout optimization, the system can realize the automatic generation and dynamic response from building investment data to structural elements, thereby greatly improving the efficiency and accuracy of building structure design.

[0006] The present invention proposes a heterogeneous data fusion structural element dynamic generation system for building investment, including:

[0007] A multi-source parsing module, configured to receive architectural design data and parse and convert the architectural design data into a unified intermediate representation;

[0008] A semantic parsing module, connected to the multi-source parsing module data, is used to receive the unified intermediate representation, extract building physical features, and construct a structural knowledge graph;

[0009] A graphic element generation module, connected to the semantic parsing module, for generating parameterized structural graphic elements based on the structural knowledge graph;

[0010] A linkage update module, data-connected to the primitive generation module, configured to monitor change events of the parameterized structural primitives, determine the scope of change impact, and trigger linkage updates of related primitives;

[0011] A layout optimization module, data-connected to the linkage update module, configured to detect spatial conflicts between the parameterized structural elements and perform adaptive layout optimization based on preset rules;

[0012] The output interface module is connected to the layout optimization module data and is used to output the optimized parameterized structural primitives to the target CAD environment.

[0013] Preferably, the multi-source analysis module includes:

[0014] Data receiving unit, used to receive heterogeneous data including CAD drawings, BIM models and structural test data;

[0015] a format identification unit, configured to identify a data format type of the heterogeneous data;

[0016] A parsing and conversion unit, configured to call a corresponding parser to perform data parsing according to the data format type;

[0017] A unified modeling unit that transforms the parsed data into a unified intermediate representation that follows a predefined schema.

[0018] Preferably, the semantic parsing module includes:

[0019] a feature extraction unit, configured to extract building physical features from the unified intermediate representation;

[0020] A deep learning unit for identifying building floor plans and functional zoning using a pre-trained neural network model;

[0021] a semantic mapping unit for mapping physical features into structural semantic entities;

[0022] The knowledge graph unit is used to construct a structural knowledge graph containing structural entities and their relationships.

[0023] Preferably, the primitive generation module includes:

[0024] Parametric template library, used to store parametric templates of various structural elements;

[0025] Code constraint unit, used to load structural design codes and convert them into constraint conditions;

[0026] A graphic element instantiation unit, configured to select an appropriate parameterized template based on the structural knowledge graph and instantiate it into a specific graphic element;

[0027] Detail recursive unit is used to recursively refine the primitives and generate multi-level expressions from conceptual design to detailed construction.

[0028] Preferably, the linkage update module includes:

[0029] An event monitoring unit, used to monitor attribute change events of parameterized structural elements;

[0030] A dependency analysis unit, configured to maintain a dependency graph between graphic elements and determine the scope of impact of a change based on the dependency graph;

[0031] An update propagation unit, used to trigger the linkage update of related primitives according to a predefined propagation strategy;

[0032] The version control unit is used to record the change history of graphic elements and support comparison and backtracking between different versions.

[0033] Preferably, the layout optimization module includes:

[0034] Conflict detection unit, used to detect spatial overlap, annotation intersection and specification conflicts between parametric structural elements;

[0035] An optimization strategy unit is used to select a corresponding processing strategy according to the conflict type;

[0036] A layout adjustment unit, used for adaptively adjusting the primitive layout based on a multi-objective optimization algorithm;

[0037] Annotation optimization unit, used to dynamically adjust annotation position and display mode according to the density and importance of graphics elements.

[0038] Preferably, the dependency analysis unit determines the change impact scope by:

[0039] Calculate the influence coefficient I(A→d) between the changed element A and the dependent element d, where:

[0040] I(A→d)=w×f(property, d.dependentProperties),

[0041] Where w is the weight coefficient, f is the correlation function, property is the changed property, and d.dependentProperties is the set of related properties of the dependent element d;

[0042] When I(A→d) is greater than a preset threshold, it is determined that the primitive d is affected by the change and needs to be updated.

[0043] Preferably, the conflict detection unit performs hierarchical processing according to the severity of the conflict:

[0044] For slight overlaps, only the primitive boundaries are adjusted to maintain geometric properties;

[0045] For moderate overlap, perform primitive scaling combined with position fine-tuning;

[0046] For severe overlap, trigger element splitting or layer separation processing.

[0047] Preferably, the semantic mapping unit adopts a hybrid mapping strategy based on rules and probability:

[0048] When the matching confidence is higher than 90%, the deterministic mapping rule is directly applied;

[0049] When the matching confidence is between 70% and 90%, a multi-rule voting decision is adopted;

[0050] When the matching confidence is lower than 70%, a manual confirmation mechanism is triggered to improve the mapping accuracy.

[0051] The method for dynamically generating structural elements by fusion of heterogeneous data for building investment includes:

[0052] Receiving architectural design data, and parsing and converting the architectural design data into a unified intermediate representation;

[0053] Extracting building physical features from the unified intermediate representation and constructing a structural knowledge graph;

[0054] Generate parameterized structural primitives based on the structural knowledge graph;

[0055] Monitoring the change events of the parametric structural elements, determining the scope of change impact, and triggering the linkage update of related elements;

[0056] Detecting spatial conflicts between the parametric structural elements and performing adaptive layout optimization based on preset rules;

[0057] Export the optimized parametric structural elements to the target CAD environment.

[0058] The beneficial effects of the present invention are:

[0059] 1. Significantly improve design efficiency: The time required to draw conventional structural details is reduced by 85%, freeing up designers’ creativity;

[0060] 2. Significantly reduce error rates: Automatic conflict detection eliminates more than 90% of human errors and improves design quality;

[0061] 3. Real-time response: Architectural changes can be quickly transmitted to structural design, reducing design change coordination time by 73% and shortening project cycles by 20%;

[0062] 4. Maintain a unified drawing style: The consistency of drawing styles among different designers has increased by 95%, enhancing the corporate brand image;

[0063] 5. Support design collaboration: Multiple designers can work simultaneously on a unified data basis, achieving true parallel design;

[0064] 6. Improve data value: Establish the connection between building and structural data to lay the foundation for subsequent full life cycle management. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 This is the overall architecture diagram of the heterogeneous data fusion structural element dynamic generation system for building investment improvement of the present invention;

[0066] Figure 2 It is a schematic structural diagram of the multi-source resolution module of the present invention;

[0067] Figure 3 It is a structural diagram of the semantic parsing module of the present invention;

[0068] Figure 4 It is a structural diagram of the graphic element generation module of the present invention;

[0069] Figure 5 It is a structural diagram of the linkage update module of the present invention;

[0070] Figure 6 It is a structural diagram of the layout optimization module of the present invention;

[0071] Figure 7 This is a schematic diagram of the event-driven primitive linkage update mechanism of the present invention;

[0072] Figure 8 It is a schematic diagram of the spatial conflict detection and resolution strategy of the present invention;

[0073] Figure 9 It is a flow chart of the method for dynamically generating structural elements by integrating heterogeneous data for building investment improvement according to the present invention. DETAILED DESCRIPTION

[0074] Please refer to the attached Figure 1-9The preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention and are not used to limit the present invention.

[0075] like Figure 1 As shown, the heterogeneous data fusion structural element dynamic generation system for building investment provided by the present invention includes a multi-source parsing module 1, a semantic parsing module 2, an element generation module 3, a linkage update module 4, a layout optimization module 5 and an output interface module 6.

[0076] The multi-source parsing module 1 is used to receive architectural design data and parse it into a unified intermediate representation. The architectural design data can be heterogeneous data such as CAD drawings, BIM models, or structural test data. The unified intermediate representation uses the unified Building Structure Description Language (BSDL) format defined in XML Schema as a standardized carrier for heterogeneous data.

[0077] Semantic Parsing Module 2 is connected to the data of Multi-Source Parsing Module 1 to receive the unified intermediate representation, extract the building's physical features, and construct a structural knowledge graph. This module uses deep learning technology to identify the building's floor plan layout and functional zoning, mapping physical graphic features into structural semantic entities and establishing a relationship network between building elements.

[0078] The primitive generation module 3 is connected to the semantic parsing module 2 and is used to generate parametric structural primitives based on the structural knowledge graph. This module contains a rich library of parametric primitive templates. It can select appropriate templates based on the structural knowledge graph and instantiate them into specific primitives, while ensuring that the primitives comply with structural design specifications.

[0079] Linked Update Module 4, connected to Element Generation Module 3, monitors changes to parametric structural elements, determines the scope of impact, and triggers linked updates for related elements. This module employs an event-driven mechanism to maintain a dependency graph between elements, ensuring that design changes are accurately propagated to all relevant elements.

[0080] Layout Optimization Module 5, connected to Linked Update Module 4, detects spatial conflicts between parametric structural elements and performs adaptive layout optimization based on pre-set rules. This module detects and resolves issues such as spatial overlap, annotation intersections, and specification conflicts between elements, ensuring the clarity and compliance of the final drawing.

[0081] The output interface module 6 is connected to the layout optimization module 5 and is used to output the optimized parametric structural elements to the target CAD environment. This module supports mainstream CAD software, ensuring that the generated structural elements can be seamlessly integrated into the designer's work environment.

[0082] Each module transmits information through data connections, forming a complete data processing chain. The overall system workflow is: receive architectural design data → parse and convert it into a unified representation → semantically analyze and build a knowledge graph → parametrically generate structural elements → monitor changes for linked updates → detect conflicts and optimize layout → output to the CAD environment.

[0083] like Figure 2 As shown, the multi-source parsing module 1 includes a data receiving unit 11 , a format identifying unit 12 , a parsing and converting unit 13 and a unified modeling unit 14 .

[0084] The data receiving unit 11 is used to receive heterogeneous data, including CAD drawings, BIM models, and structural test data. This unit supports multiple data input methods, including local file import, network transmission, and database access. For CAD drawings, formats such as DWG and DXF are supported; for BIM models, formats such as RVT and IFC are supported; and for structural test data, various sensor data formats are supported.

[0085] The format recognition unit 12 is used to identify the data format type of heterogeneous data. This unit automatically identifies the data format through various methods, such as file header information, extension, and content characteristics, without the need for manual user specification. For data in non-standard formats, this unit also includes a heuristic recognition algorithm to make intelligent judgments based on data structure characteristics.

[0086] The parsing and conversion unit 13 is used to parse the data by calling the corresponding parser based on the data format. This unit includes multiple specialized parsers, such as DWG, RVT, and JSON, each optimized for processing specific data formats. For CAD data, the parser can identify CAD elements such as layers, linetypes, and block references; for BIM data, the parser can extract information such as component types, attributes, and spatial relationships.

[0087] The unified modeling unit 14 converts the parsed data into a unified intermediate representation that adheres to a predefined schema. This unit uses the unified Building Structure Description Language (BSDL) defined in XML Schema as a unified representation format, which offers excellent scalability and interoperability. A complete XML Schema structure: The BSDL should contain definitions of core entity types in the architectural and structural domains (e.g., BuildingElement, Structure, Space), attribute definitions (e.g., geometry, materials, loads), and relationship definitions (e.g., supports, connections, etc.).

[0088] Mapping relationship rule set: mapping rules for different data sources, for example:

[0089] CAD mapping: entity recognition rules based on layer, line type, and block definitions;

[0090] BIM mapping: entity attribute mapping rules based on IFC standards;

[0091] Experimental data mapping: association rules between sensor data and physical locations;

[0092] This mapping is usually achieved through a specific parser, such as a DWG parser that maps CAD layer structures to BSDL elements, or an IFC parser that maps BIM components to BSDL elements.

[0093] BSDL defines a rich set of building structural element types, such as beams, columns, walls, and slabs, as well as their properties and relationships. During conversion, the unit preserves the key features of the original data while unifying the data structure and representation.

[0094] The core structure of the unified intermediate representation is as follows:

[0095] <BuildingElementid="BE001"type="Column">

[0096] <GeometryInfox="3200"y="5400"rotation="0" / >

[0097] <properties>

[0098] <Propertyname="Section"value="400x400" / >

[0099] <Propertyname="Material"value="C30" / >

[0100] <Propertyname="Load"value="950kN" / >

[0101] < / properties>

[0102] <relations>

[0103] <Relationtype="Support"target="BE002" / >

[0104] <Relationtype="Connect"target="BE003" / >

[0105] < / relations>

[0106]

[0107] The multi-source parsing module 1 realizes the reception, recognition, parsing and unified representation of various heterogeneous data through the collaborative work of the above four units, laying the foundation for subsequent semantic parsing and primitive generation.

[0108] like Figure 3 As shown, the semantic parsing module 2 includes a feature extraction unit 21, a deep learning unit 22, a semantic mapping unit 23 and a knowledge graph unit 24.

[0109] The feature extraction unit 21 is used to extract building physical features from the unified intermediate representation. This unit uses a geometric feature recognition algorithm to extract geometric information such as the shape, position, and size of building elements, as well as non-geometric information such as materials and loads. For walls, this unit extracts their axis, thickness, height, and opening information; for columns, their center point location, cross-sectional shape, and dimensions; and for beams, their start and end points, cross-sectional shape, and dimensions.

[0110] Deep Learning Unit 22 utilizes a pretrained neural network model to identify building floor plans and functional zoning. This unit employs a convolutional neural network based on a ResNet-50 variant, adapted for the task of building floor plan recognition through transfer learning. Trained on a large number of building floor plans, the model accurately identifies different functional areas, such as residential, office, and commercial, and infers corresponding structural design requirements based on the characteristics of these functional zoning areas.

[0111] The semantic mapping unit 23 is responsible for mapping physical features into structural semantic entities. This unit employs a hybrid mapping strategy based on rules and probabilities, converting geometric features into semantic entities with structural meaning using a series of predefined mapping rules. For example, if four equally spaced point-like objects are identified within an enclosed space, they might be mapped into a "column grid" semantic entity.

[0112] Example of semantic mapping rules:

[0113] Rule-S01:IF[Graphic Type = Closed Polygon] AND[Line Type = Wall] AND[Location Feature = Through Multiple Layers]

[0114] THEN [structural semantics = shear wall] WITH [confidence = 0.85]

[0115] Rule-S02:IF [Shape Type = Rectangle] AND [Size Ratio ≈ 1:1] AND [Location = Wall Intersection]

[0116] THEN [structural semantics = construction column] WITH [confidence = 0.92].

[0117] The bridging between the deep learning unit 22 and the semantic mapping unit 23 is achieved in the following manner:

[0118] Feature extraction intermediate layer: The output of the deep learning unit 22 should not be used directly for rule mapping, but should be transformed by the feature extraction unit 21. The feature extraction unit extracts physical features such as geometry and spatial relationships from the unified intermediate representation (BSDL), while obtaining semantic labels and confidence from the deep learning unit.

[0119] Fusion mechanism: Two inputs are fused through feature vectors:

[0120] CombinedFeature=α*GeometryFeature+β*SemanticFeature

[0121] Among them, α and β are weight coefficients, which are dynamically adjusted according to the reliability of different features.

[0122] Output normalization: The converted features should conform to the input format of the semantic mapping unit (23), including normalized geometric attributes, spatial relations, and preliminary semantic labels.

[0123] This bridging mechanism ensures that image recognition results can be effectively combined with the rule system.

[0124] The knowledge graph unit 24 is used to construct a structural knowledge graph containing structural entities and their relationships. This unit uses graph database technology to store and manage the structural knowledge graph, supporting complex relationship queries and reasoning. In the knowledge graph, nodes represent structural entities (such as beams, columns, and walls), and edges represent relationships between entities (such as support, connection, and inclusion). By constructing the structural knowledge graph, the system can understand the complex topological relationships between building elements, providing a semantic foundation for subsequent structural element generation.

[0125] Through the collaborative work of the above four units, the semantic parsing module 2 realizes the mapping transformation from physical graphic features to structural semantic entities, constructs a structural knowledge graph with rich semantic information, and enables the system to have the ability to "understand" the architectural design intention.

[0126] like Figure 4 As shown, the primitive generation module 3 includes a parameterized template library 31 , a specification constraint unit 32 , a primitive instantiation unit 33 and a detail recursion unit 34 .

[0127] The parametric template library 31 stores parametric templates for various structural elements. This library includes a rich collection of structural element templates, such as columns, beams, walls, slabs, and joint details. Each template defines a series of adjustable parameters and their constraints. For example, a column template defines parameters such as cross-sectional shape, size, material, and reinforcement method; a beam template defines parameters such as cross-sectional shape, size, span, and load. These parameters can be adjusted according to specific design requirements, enabling flexible element generation.

[0128] Example of parametric primitive definition:

[0129] ColumnElement:

[0130] Parameters:

[0131] section_width: range [300-800] mm, step length 50 mm;

[0132] section_height: range [300-800] mm, step length 50 mm;

[0133] concrete_grade: enumeration [C25,C30,C35,C40,C45,C50];

[0134] rebar_config: enumeration [4Φ16,4Φ18,4Φ20,8Φ18,8Φ20,8Φ22];

[0135] Constraints:

[0136] section_width == section_height (square column constraint);

[0137] IFconcrete_grade>C40THENrebar_configIN[8Φ18,8Φ20,8Φ22];

[0138] Visualization:

[0139] MainShape:Rectangle(section_width,section_height);

[0140] Annotation:TextBlock("${section_width}×{section_height}",position="center");

[0141] RebarIndicator:SymbolType.REBAR_CONFIG[rebar_config]

[0142] The Code Constraint Unit 32 is used to load structural design codes and convert them into constraints. This unit includes built-in digital representations of various structural design codes, such as the Code for Design of Concrete Structures and the Code for Seismic Design of Buildings. These codes are converted into a series of constraints, such as minimum reinforcement ratios, maximum beam height-to-span ratios, and seismic structural requirements, to ensure that the generated structural elements comply with the relevant code requirements.

[0143] The primitive instantiation unit 33 is used to select appropriate parameterized templates based on the structural knowledge graph and instantiate them into specific primitives. This unit first analyzes the entities and their relationships in the structural knowledge graph to determine the primitive type to be generated. Then, based on the attributes and relationships of the entities, it selects the most suitable template from the parameterized template library. Finally, based on the specific design requirements and regulatory constraints, it sets the template parameter values ​​and instantiates it into a specific structural primitive.

[0144] The detail recursion unit 34 is used to recursively refine the elements, generating a multi-level representation from conceptual design to detailed construction. This unit employs a recursive refinement strategy, first generating the main structural elements (such as column grids and frames), then gradually refining them into more detailed components (such as beams and columns of specific dimensions), and finally generating detailed structural joints (such as beam-column joints and wall-beam connections). This multi-level representation allows designers to view structural elements at different levels of detail as needed while maintaining consistent associations between the elements.

[0145] Through the collaborative work of the above four units, the primitive generation module 3 realizes the intelligent generation from the structural knowledge graph to the parametric structural primitives, ensuring that the generated primitives meet the design needs and specifications and have multi-level detailed expressions.

[0146] like Figure 5 As shown, the linkage update module 4 includes an event monitoring unit 41 , a dependency analysis unit 42 , an update propagation unit 43 and a version control unit 44 .

[0147] The event monitoring unit 41 is used to monitor attribute change events of parametric structural elements. This unit adopts the observer design pattern, registering an event listener for each structural element and capturing changes in element attributes in real time. The monitored attributes include geometric attributes (such as position and size) and non-geometric attributes (such as material and reinforcement). When a designer modifies element attributes or the architectural design changes, this unit can promptly capture the change event and pass it to the dependency analysis unit for processing.

[0148] The dependency analysis unit 42 is used to maintain a dependency graph between graph elements and determine the scope of impact of changes based on the dependency graph. This unit constructs a directed acyclic graph (DAG) to represent the dependencies between graph elements, where nodes represent graph elements and edges represent dependency relationships. When a graph element is changed, the unit uses a graph traversal algorithm to identify all other graph elements that depend on it and determine the scope of impact of the change.

[0149] The dependency analysis unit determines the scope of change impact by calculating the impact coefficient I(A→d) between the changed element A and the dependent element d, where:

[0150] I(A→d)=w×f(property,d.dependentProperties),

[0151] Where w is the weight coefficient, determined by the importance of the dependency; f is the correlation function, which measures the degree of association between the changed property and the properties of the dependent element; property is the changed property, and d.dep endentProperties is the set of properties related to the dependent element d. When I(A→d) is greater than a preset threshold, element d is determined to be affected by the change and needs to be updated.

[0152] The update propagation unit 43 is used to trigger the coordinated update of related primitives according to a predefined propagation strategy. Based on the impact range determined by the dependency analysis unit, this unit uses a topological sorting algorithm to determine the order of primitive updates to avoid update conflicts and circular dependencies. For each primitive requiring an update, the unit calls the corresponding update processing function and adjusts the primitive's properties based on the change content and dependencies. A recursive update strategy is used during the update process to ensure that the impact of the change is propagated throughout the entire dependency chain.

[0153] The version control unit 44 records the history of changes to graphics elements and supports comparison and rollback of different versions. This unit uses an incremental storage strategy, recording the content, time, and operator of each change, building a complete change history tree. Designers can browse the change history, compare differences between versions, and roll back to previous versions when necessary. This version control feature enables design teams to track the evolution of designs, understand the reasons behind design decisions, and restore previous design states when necessary.

[0154] The linkage update module 4 realizes the event-driven linkage update mechanism of structural elements through the collaborative work of the above four units, ensuring that design changes can be accurately and effectively propagated to all related elements, while retaining a complete change history, providing the design team with powerful version management capabilities.

[0155] like Figure 6 As shown, the layout optimization module 5 includes a conflict detection unit 51 , an optimization strategy unit 52 , a layout adjustment unit 53 and an annotation optimization unit 54 .

[0156] The conflict detection unit 51 is used to detect spatial overlap, annotation intersections, and specification conflicts between parametric structural elements. This unit uses spatial indexing techniques (such as R-trees) to accelerate spatial conflict detection, efficiently identifying overlapping areas between elements. It also detects intersection conflicts between annotations and inconsistencies between elements and specification requirements. Conflict detection results are graded by severity and provide detailed conflict information to facilitate subsequent processing.

[0157] The optimization strategy unit 52 is used to select a corresponding handling strategy based on the conflict type. This unit has multiple built-in conflict resolution strategies, selecting the most appropriate approach based on the conflict type, severity, and context. Different handling strategies are employed for different conflict types: spatial overlap conflicts are resolved by adjusting the position or size of the element; annotation intersection conflicts are resolved by rearranging the annotation positions; and specification conflicts are resolved by prompting the designer or automatically adjusting parameters.

[0158] Example of a conflict resolution policy map:

[0159] Spatial overlap conflicts:

[0160] Slight overlap (<10%): adjust the primitive boundaries to maintain geometric characteristics;

[0161] Moderate overlap (10-30%): primitive scaling + position fine-tuning;

[0162] Severe overlap (>30%): trigger splitting or layer separation;

[0163] Label conflicts:

[0164] Implemented a force-directed annotation layout algorithm;

[0165] Priority: Structural safety related > Dimensioning > Material description > Auxiliary information;

[0166] Conflict of regulations:

[0167] - Trigger standard prompts and generate three-level processing suggestions:

[0168] a. Automatic correction (low risk adjustment);

[0169] b. Manual confirmation (medium risk adjustment);

[0170] c. Design change request (high-risk issue);

[0171] The layout adjustment unit 53 is used to adaptively adjust the primitive layout based on a multi-objective optimization algorithm. This unit models the primitive layout problem as a multi-objective optimization problem, taking into account multiple objectives such as space utilization, readability, aesthetics, and compliance with regulations. The optimization process uses heuristic algorithms (such as simulated annealing and genetic algorithms) to find the layout solution with the best overall performance while satisfying the constraints. For complex layout problems, this unit also adopts a divide-and-conquer strategy to decompose the overall layout into local layout sub-problems and gradually optimize and solve them.

[0172] The annotation optimization unit 54 dynamically adjusts annotation placement and display based on the density and importance of the element. This unit utilizes a force-directed annotation layout algorithm, treating annotations as charged particles and automatically calculating optimal annotation placement by simulating the interaction of electric field forces. Furthermore, this unit dynamically adjusts annotation content and display based on element importance and contextual information, simplifying annotations in information-dense areas and detailing them at key points to ensure overall readability and information accuracy.

[0173] The layout optimization module 5 realizes intelligent conflict detection and adaptive layout optimization of structural elements through the collaborative work of the above four units, ensuring that the generated structural drawings are clear and easy to read, meet the requirements of the specifications, and have good visual aesthetics.

[0174] like Figure 7 As shown, this embodiment details the working principle of the dependency analysis unit in the linkage update module. The dependency analysis unit determines the scope of change impact through the following steps:

[0175] 1. Construct a graph dependency graph G, where nodes represent graph elements and directed edges represent dependency relationships.

[0176] 2. When element A changes, the ChangeEvent(A,property,oldValue,newValue) event is triggered.

[0177] 3. The system queries the dependency graph G and obtains the set of all primitives D = {B, C, D...} that directly depend on A.

[0178] 4. For each element d∈D, calculate the change impact coefficient I(A→d):

[0179] I(A→d)=w×f(property,d.dependentProperties),

[0180] Where w is the weight coefficient, which is determined by the importance of the dependency relationship and has a typical value range of [0.1, 1.0]. f is the correlation function, which measures the degree of association between the changed property and the related properties of the dependent element and has a value range of [0, 1]. property is the changed property, and d.dependentProperties is the set of related properties of the dependent element d.

[0181] The calculation method of the correlation function f is:

[0182]

[0183] sim(property, p) is a property similarity function that measures the correlation between two properties. Its value range is [0, 1]. Property represents the property being changed, and p represents the related property in the dependent primitive that may be affected. For the property position, similarity is inversely proportional to spatial distance; for the property size, similarity is related to the size ratio; and for the property material, similarity depends on the dependency relationship between the material properties.

[0184] 5. When I(A→d)>threshold (threshold is usually set to 0.3~0.5), it is determined that the element d is affected by the change and needs to be updated.

[0185] 6. For the elements that need to be updated, recursively trigger their ChangeEvent to further propagate the change impact to the dependency graph.

[0186] 7. Record all changes to the ChangeLog and build a change tree T for version control.

[0187] Through this dependency analysis method based on impact coefficients, the system can accurately calculate the impact of design changes, avoid unnecessary updates, and improve the efficiency and accuracy of linked updates. Furthermore, through the accumulation of change records, the system can learn and optimize the impact coefficient calculation model, continuously improving the intelligence of change propagation.

[0188] To ensure that the scale of attribute comparison is consistent, attribute standardization should be performed:

[0189] For numerical attributes, they should be normalized to the interval [0,1];

[0190] For categorical attributes, they should be converted to one-hot encoding;

[0191] For structural attributes, structural similarity algorithms should be used;

[0192] Reasonable threshold setting: The threshold of 0.3-0.5 is selected based on practical experience and system balance point: below 0.3: too many irrelevant elements are updated, which puts a heavy burden on the system; above 0.5: some elements that should be updated may be missed;

[0193] In actual implementation, dynamic thresholds should be set according to different attribute types:

[0194] threshold=baseThreshold*attrTypeWeight*relationWeight;

[0195] Among them, different attribute types (position, size, material) have different weights, and different dependencies (support, connection) also have different weights.

[0196] like Figure 8 As shown, this embodiment details the working principle of the conflict detection unit in the layout optimization module. The conflict detection unit performs graded processing according to the severity of the conflict:

[0197] 1. Spatial conflict detection: Use R-tree spatial indexing technology to accelerate spatial area queries and calculate the overlapping area ratio p between primitives:

[0198]

[0199] Wherein, E1 and E2 are two primitives to be detected, Area(E) represents the area of ​​the primitive, and Area(E1∩E2) represents the overlapping area of ​​the two primitives.

[0200] Reason for choosing the denominator: The original intention of choosing a smaller area as the denominator was to pay special attention to the protection of small components, because small components would be completely invisible if covered by large components.

[0201] Dealing with large area differences: Regarding the issue you mentioned, where large area differences can result in even minor overlaps being considered serious conflicts, the following improvement measures should be adopted in actual implementation:

[0202] Area ratio adjustment: When the area ratio of two primitives is greater than the preset threshold, the correction formula is applied:

[0203]

[0204] Area ratio adjustment: When the area ratio of two primitives is greater than the preset threshold, the correction formula is applied: where α is the adjustment index (usually between 0.3-0.5), which is used to smooth out extreme differences. Dual judgment mechanism: Calculate the IOU value at the same time and make a comprehensive judgment with the original p value:

[0205]

[0206] Component priority: Set priority based on component type. Important components (such as structural columns) should be considered as serious conflicts even if they are slightly obscured.

[0207] 2. According to the size of the overlap ratio p, spatial conflicts are divided into three levels:

[0208] Slight overlap: p<10%, using a boundary adjustment strategy to keep the main geometric characteristics of the primitives unchanged.

[0209] Moderate overlap: 10% ≤ p < 30%. A strategy of combining primitive scaling with position fine-tuning is used to reduce overlap while keeping the primitives recognizable.

[0210] Severe overlap: p ≥ 30%, triggering element splitting or layer separation, prompting the designer to perform manual intervention if necessary.

[0211] 3. For annotation conflicts, a force-guided layout algorithm is used: the annotations are treated as charged particles, and the attractive and repulsive forces are set:

[0212] F attract (d) = k a ×d,

[0213]

[0214] Where d is the distance between two annotations, k a is the attraction coefficient, k r is the repulsive force coefficient. The final position of the annotation is determined by iteratively solving the force balance equation:

[0215] F total =∑F attract +∑F repel =0,

[0216] 4. For regulatory conflicts, three levels of action will be taken based on severity:

[0217] Low-risk adjustments: The system automatically makes corrections without user confirmation, such as simple annotation formatting adjustments.

[0218] Medium-risk adjustments: The system generates correction suggestions that require user confirmation before execution, such as minor adjustments to non-critical dimensions.

[0219] High-risk issues: The system generates design change requests that require designers to reassess and make decisions, such as non-compliance issues with key structural parameters.

[0220] Through this hierarchical processing strategy, the system can intelligently handle conflicts of varying types and severity, minimizing manual intervention and improving design efficiency while ensuring design quality. For complex conflict situations, the system records solutions and gradually builds a conflict resolution knowledge base, continuously improving the intelligence and efficiency of conflict handling.

[0221] This embodiment describes in detail the hybrid mapping strategy adopted by the semantic mapping unit in the semantic parsing module.

[0222] The semantic mapping unit adopts a hybrid mapping strategy based on rules and probability:

[0223] 1. Rule Mapping: Map physical features to structural semantic entities based on predefined mapping rules. Rules take a condition-result format, where the condition describes the combination pattern of physical features, and the result specifies the corresponding semantic entity type and confidence level.

[0224] 2. Probabilistic Mapping: For physical features that cannot be mapped using deterministic rules, probabilistic models are used to calculate various possible semantic mappings and their probability distributions. Probabilistic models are based on techniques such as Bayesian networks or random forests and can handle uncertainty and ambiguity.

[0225] 3. Hybrid decision-making strategy:

[0226] When the matching confidence is higher than 90%, the deterministic mapping rule is directly applied;

[0227] When the matching confidence is between 70% and 90%, a multi-rule voting decision is adopted;

[0228] When the matching confidence is lower than 70%, a manual confirmation mechanism is triggered to improve the mapping accuracy.

[0229] 4. Multi-rule voting decision: When multiple rules apply simultaneously, the system calculates the weighted voting score S:

[0230]

[0231] Among them, R match is the set of matching rules, w r is the weight of rule r, conf r (entity) is the confidence of rule r on the semantic entity entity. The semantic entity with the highest score is selected as the mapping result.

[0232] 5. Feedback learning: The system records the user's confirmation or correction operations on the mapping results, dynamically adjusts the rule weights and probability model parameters, and realizes the continuous optimization of the mapping strategy.

[0233] Through this hybrid mapping strategy, the semantic mapping unit can flexibly handle a variety of complex mapping situations, ensuring mapping accuracy while minimizing manual intervention and improving the system's automation and intelligence. The system can adapt to new architectural design patterns or structural entity types through rule expansion and model retraining, demonstrating excellent scalability and learning capabilities.

[0234] Based on the application examples, this system also implements a detailed set of automatic structural element generation rules to convert the recognized architectural drawing information into standardized structural design elements. These rules reflect the intelligence and specialization of the system, as follows:

[0235] Architectural drawing recognition: The system automatically identifies key elements in architectural drawings, including structural frame columns, building elevations, opening lines, functional rooms, and interior walls. These recognition results serve as the basis for generating structural element data.

[0236] Structural element generation rules: Based on the recognition results, the system automatically generates structural elements according to the following standards:

[0237] Generation of frame beams and secondary beams:

[0238] Identify the location of structural frame columns and match them to the structural column layer;

[0239] Measure the distance L between frame columns and automatically generate the frame beam section;

[0240] Frame beam standard: text marking BXH (B represents beam width, H represents beam height), H = 1 / 10L, B = 1 / 2H;

[0241] Constraints: beam height not less than 400mm, beam width not less than 200mm;

[0242] According to the building boundary line and the opening line position, adjust the frame beam position to make it flush with the boundary line and the opening line;

[0243] Secondary beam standard: text mark bxh (b represents beam width, h represents beam height), h = H-50mm, b = B-100mm;

[0244] Secondary beam constraint: h not less than 400mm, b not less than 200mm;

[0245] Determination of structural elevation:

[0246] Structural elevation = building elevation - 50mm;

[0247] According to the building elevation distribution, calculate the structural elevation values ​​of different areas;

[0248] Generate differentiated fill styles for different elevation areas;

[0249] Draw boundary lines at drop-board locations;

[0250] Plate thickness calculation:

[0251] Plate thickness = 1 / 40 plate span;

[0252] Automatically generate plate thickness annotation;

[0253] Component marking specifications:

[0254] Strictly follow the "Concrete Structure Construction Drawing Plan Overall Representation Method Drawing Rules and Construction Details" (22G) standard

[0255] Ensure that the location, style and content of the markings comply with the specifications.

[0256] This set of rules demonstrates the system's core capability to automatically transform architectural design information into structural design information, significantly improving design efficiency while ensuring standardized and consistent structural design. By integrating these rules with building recognition results through intelligent algorithms, the system achieves a high degree of automation from building funding to structural drawing production, effectively addressing the efficiency and consistency issues inherent in the traditional design process.

[0257] like Figure 9 As shown, the method for dynamically generating heterogeneous data fusion structural elements for building investment provided by the present invention includes the following steps:

[0258] Step S1: Receive architectural design data and parse and convert the architectural design data into a unified intermediate representation.

[0259] In the specific implementation, it first receives heterogeneous data including CAD drawings, BIM models and structural test data, then identifies the data format type, calls the corresponding parser to parse the data according to the data format type, and finally converts the parsed data into a unified intermediate representation that follows the predefined architecture.

[0260] Step S2: Extract the building physical features in the unified intermediate representation and construct a structural knowledge graph.

[0261] In the specific implementation, the building physical features are first extracted from the unified intermediate representation, and then the pre-trained neural network model is used to identify the building floor plan and functional zoning. The physical features are then mapped into structural semantic entities, and finally a structural knowledge graph containing structural entities and their relationships is constructed.

[0262] Step S3: Generate parameterized structural primitives based on the structural knowledge graph.

[0263] In the specific implementation, first select the appropriate element template from the parametric template library, then set the template parameters according to the entity attributes and relationships in the structural knowledge graph, then apply the specification constraints to ensure that the element meets the design specifications, and finally instantiate it into a specific parametric structural element and perform recursive refinement.

[0264] Step S4: Monitor the change events of the parametric structural elements, determine the scope of the change impact, and trigger the linkage update of related elements.

[0265] In the specific implementation, we first monitor the change events of the primitive attributes, then calculate the impact range of the change based on the dependency graph, then trigger the linkage update of related primitives according to the predefined propagation strategy, and finally record the change history for version control.

[0266] Step S5: Detecting spatial conflicts between parameterized structural elements and performing adaptive layout optimization based on preset rules.

[0267] In the specific implementation, the spatial overlap, annotation intersection and specification conflict between the graphics elements are first detected, and then the corresponding processing strategy is selected according to the type and severity of the conflict. Then, a multi-objective optimization algorithm is used to adjust the graphics element layout, and finally the annotation position and display method are optimized to ensure the clarity and standardization of the drawings.

[0268] Step S6: Output the optimized parametric structural primitives to the target CAD environment.

[0269] During the specific implementation, according to the requirements of the target CAD environment, the optimized parametric structural elements are converted into the corresponding CAD object format and output to the CAD environment to ensure that the elements can be seamlessly integrated into the designer's working environment.

[0270] Through the above steps, the method of the present invention realizes the automatic generation and dynamic update of building capital data to structural elements, greatly improving the efficiency and accuracy of building structure design, and solving the problems of traditional methods such as difficulty in integrating heterogeneous data, low efficiency of manual drawing, and difficulty in quickly responding to building changes.

[0271] A comparison chart of the application effects of the present invention in a certain office building project. The project is a 15-story office building with a construction area of ​​approximately 20,000 square meters and a frame-shear wall structure.

[0272] Before the application of the system of the present invention, it took 5 working days from building funding to preliminary structural layout, 15 working days to draw detailed structural drawings, and an average of 3 working days to respond to building changes. The average number of errors and conflicts discovered during the design process was 2.5 per 100 square meters.

[0273] After applying the system of the present invention, it only takes 0.5 working days from building funding to preliminary structural layout, and only 2 working days to automatically generate structural detailed drawings. The average response time for building changes is shortened to 0.5 working days, and design errors and conflicts are reduced to less than 0.3 per 100 square meters.

[0274] The advantages of this system are particularly evident when building plans undergo multiple changes. Traditionally, each architectural change requires structural engineers to redraw numerous drawings, which is time-consuming and prone to errors. However, with this system, architectural changes are automatically transmitted to the structural design, and the system automatically updates the affected structural elements, significantly reducing manual work and the potential for error.

[0275] Overall, the present invention achieved remarkable results in this project, including an 85% increase in design efficiency, an 88% reduction in error rate, an 83% reduction in design change response time, and a 20% reduction in project cycle, fully verifying the practical value and innovativeness of the present invention.

[0276] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. The heterogeneous data fusion structural element dynamic generation system for building investment is characterized by: include: A multi-source parsing module, configured to receive architectural design data and parse and convert the architectural design data into a unified intermediate representation; A semantic parsing module, connected to the multi-source parsing module data, is used to receive the unified intermediate representation, extract building physical features, and construct a structural knowledge graph; A graphic element generation module, connected to the semantic parsing module, for generating parameterized structural graphic elements based on the structural knowledge graph; A linkage update module, data-connected to the primitive generation module, configured to monitor change events of the parameterized structural primitives, determine the scope of change impact, and trigger linkage updates of related primitives; A layout optimization module, data-connected to the linkage update module, configured to detect spatial conflicts between the parameterized structural elements and perform adaptive layout optimization based on preset rules; The output interface module is connected to the layout optimization module data and is used to output the optimized parameterized structural primitives to the target CAD environment.

2. The system according to claim 1, wherein: The multi-source analysis module includes: Data receiving unit, used to receive heterogeneous data including CAD drawings, BIM models and structural test data; a format identification unit, configured to identify a data format type of the heterogeneous data; A parsing and conversion unit, configured to call a corresponding parser to perform data parsing according to the data format type; A unified modeling unit that transforms the parsed data into a unified intermediate representation that follows a predefined schema.

3. The system according to claim 1, wherein: The semantic parsing module includes: a feature extraction unit, configured to extract building physical features from the unified intermediate representation; A deep learning unit for identifying building floor plans and functional zoning using a pre-trained neural network model; a semantic mapping unit for mapping physical features into structural semantic entities; The knowledge graph unit is used to construct a structural knowledge graph containing structural entities and their relationships.

4. The system according to claim 1, wherein: The primitive generation module includes: Parametric template library, used to store parametric templates of various structural elements; Code constraint unit, used to load structural design codes and convert them into constraint conditions; A graphic element instantiation unit, configured to select an appropriate parameterized template based on the structural knowledge graph and instantiate it into a specific graphic element; Detail recursive unit is used to recursively refine the primitives and generate multi-level expressions from conceptual design to detailed construction.

5. The system according to claim 1, wherein: The linkage update module includes: An event monitoring unit, used to monitor attribute change events of parameterized structural elements; A dependency analysis unit, configured to maintain a dependency graph between graphic elements and determine the scope of impact of a change based on the dependency graph; An update propagation unit, used to trigger the linkage update of related primitives according to a predefined propagation strategy; The version control unit is used to record the change history of graphic elements and support comparison and backtracking between different versions.

6. The system according to claim 1, wherein: The layout optimization module includes: Conflict detection unit, used to detect spatial overlap, annotation intersection and specification conflicts between parametric structural elements; An optimization strategy unit is used to select a corresponding processing strategy according to the conflict type; A layout adjustment unit, used for adaptively adjusting the primitive layout based on a multi-objective optimization algorithm; Annotation optimization unit, used to dynamically adjust annotation position and display mode according to the density and importance of graphics elements.

7. The system according to claim 5, characterized in that The dependency analysis unit determines the impact scope of the change by: Calculate the influence coefficient I(A→d) between the changed element A and the dependent element d, where: I(A→d)=w×f(property,d.dependentProperties, Where w is the weight coefficient, f is the correlation function, property is the changed property, and d.dependentProperties is the set of related properties of the dependent element d; When I(A→d) is greater than a preset threshold, it is determined that the primitive d is affected by the change and needs to be updated.

8. The system according to claim 6, wherein: The conflict detection unit performs graded processing according to the severity of the conflict: For slight overlaps, only the primitive boundaries are adjusted to maintain geometric properties; For moderate overlap, perform primitive scaling combined with position fine-tuning; For severe overlap, trigger element splitting or layer separation processing.

9. The system according to claim 3, wherein: The semantic mapping unit adopts a hybrid mapping strategy based on rules and probability: When the matching confidence is higher than 90%, the deterministic mapping rule is directly applied; When the matching confidence is between 70% and 90%, a multi-rule voting decision is adopted; When the matching confidence is lower than 70%, a manual confirmation mechanism is triggered to improve the mapping accuracy.

10. A method for dynamically generating structural elements by integrating heterogeneous data for building investment, using the system according to any one of claims 1 to 9, characterized in that: include: Receiving architectural design data, and parsing and converting the architectural design data into a unified intermediate representation; Extracting building physical features from the unified intermediate representation and constructing a structural knowledge graph; Generate parameterized structural primitives based on the structural knowledge graph; Monitoring the change events of the parametric structural elements, determining the scope of change impact, and triggering the linkage update of related elements; Detecting spatial conflicts between the parametric structural elements and performing adaptive layout optimization based on preset rules; Export the optimized parametric structural elements to the target CAD environment.

Images