Revit modeling method and system based on BIM (Building Information Modeling) data large model

Abstract

The invention relates to a BIM data large model-based Revit modeling method and system, and the method comprises the steps: building a modeling model according to a multi-modal training data set, and the modeling model comprises a feature extraction module, a rule base module, a topological relation module and a rule mapping module. And in response to instruction information of a modeling demand input by a user, extracting structured feature parameters through a feature extraction module. And matching the structured feature parameters with the building rules in the rule base module to obtain the building rules meeting matching conditions. And based on the building rule, obtaining a component topological relation through a topological relation module. And based on the component topological relation, generating a building model meeting modeling requirements through a rule mapping module. The problem that in the prior art, a Revit modeling method based on a BIM data large model is low in automatic modeling efficiency is solved.

Description

Technical Field

[0001] This application relates to the field of building information modeling technology, and in particular to a Revit modeling method and system based on large BIM data models. Background Technology

[0002] Currently, BIM (Building Information Modeling) requires adherence to numerous building rules during the modeling process, resulting in a lot of repetitive work and time-consuming complex model construction.

[0003] Traditional methods require users to input detailed feature parameters for modeling and necessitate repeated adjustments to the model according to architectural rules. Therefore, Revit modeling methods based on large BIM data models suffer from low efficiency in automated modeling. Summary of the Invention

[0004] This application provides a Revit modeling method and system based on large BIM data models, which at least solves the problem of low automated modeling efficiency in existing Revit modeling methods based on large BIM data models.

[0005] In a first aspect, embodiments of this application provide a Revit modeling method based on a large BIM data model, the method comprising:

[0006] A modeling model is constructed based on a multimodal training dataset, wherein the modeling model includes a feature extraction module, a rule base module, a topological relationship module, and a rule mapping module;

[0007] In response to receiving user input instructions regarding modeling requirements, the feature extraction module extracts structured feature parameters.

[0008] The structured feature parameters are matched with the building rules in the rule base module to obtain building rules that meet the matching conditions;

[0009] Based on the building rules, the component topology relationships are obtained through the topology relationship module;

[0010] Based on the aforementioned component topology, a building model that meets the aforementioned modeling requirements is generated through the rule mapping module.

[0011] In one embodiment, constructing a modeling model based on a multimodal training dataset includes:

[0012] A multimodal training dataset was constructed based on historical BIM models, building rules, and user correction records.

[0013] Data cleaning and multimodal transformation are performed on the multimodal training dataset to obtain structured data; wherein, the multimodal transformation is used to convert various types of data into data with the same structure through preset standardized transformation rules;

[0014] The structured data is trained using a deep learning architecture to obtain a modeling model.

[0015] In one embodiment, training the structured data using a deep learning architecture to obtain a modeling model includes:

[0016] Based on the BIM model data in the structured data, the spatial geometric features of the components are extracted through a three-dimensional point cloud neural network, wherein the spatial geometric features include component size, spatial coordinates and shape parameters;

[0017] Based on the component connection relationships in the structured data, a graph neural network is used to construct the topological connection relationships between components, wherein the topological connection relationships include structural support relationships and spatial adjacency relationships;

[0018] Based on the building rules in the structured data, the building rules are parsed using a natural language processing model to establish a mapping relationship between the building rules and the BIM model;

[0019] The spatial geometric features, the topological connectivity, and the mapping relationship are concatenated, and the features are fused using a multilayer perceptron to obtain a modeling model.

[0020] In one embodiment, the step of extracting structured feature parameters through the feature extraction module in response to receiving instruction information from the user regarding modeling requirements includes:

[0021] The user input modeling requirements are segmented based on natural language processing, and key design elements are extracted through entity recognition model. The key design elements include building type, functional requirements and performance indicators.

[0022] Based on the knowledge graph of the building domain, the key design elements are mapped to modeling parameters, wherein the mapping includes mapping the building type to standard functional zoning and area ratio, mapping the functional requirements to spatial layout constraints, and mapping the performance indicators to construction requirements.

[0023] The modeling parameters are analyzed using graph neural networks to identify missing key parameters, and users are guided to supplement and improve them through an interactive interface.

[0024] The improved modeling parameters are then converted into structured feature parameters.

[0025] In one embodiment, matching the structured feature parameters with building rules in the rule base module to obtain building rules that meet the matching conditions includes:

[0026] Based on the building type identifier in the structured feature parameters, retrieve the candidate rule set for the corresponding category from the rule base module;

[0027] Based on the functional requirements and performance indicators in the structured feature parameters, the semantic matching degree between the functional requirements and performance indicators and the candidate rule set is evaluated by a similarity algorithm.

[0028] Obtain the building rules that satisfy the preset semantic matching degree.

[0029] In one embodiment, obtaining the component topology relationship based on the building rules through the topology relationship module includes:

[0030] Based on the architectural, structural, and mechanical and electrical component parameters in the building rules, the corresponding component library is called through the Revit API to obtain components including walls, floors, beams, columns, water and electricity pipes, and building equipment;

[0031] By analyzing the connection and position rules of components in the architectural rules using a graph neural network, the topological relationships of the components, including connection and position relationships, are obtained.

[0032] In one embodiment, based on the component topology relationship, a target component topology relationship is obtained through a rule mapping module, and a building model that meets the modeling requirements is generated based on the target component topology relationship, including:

[0033] The building rules are encoded into rule feature vectors using a pre-trained Transformer encoder;

[0034] Calculate the feature matching degree between the rule feature vector and the component topological relationship;

[0035] Based on the feature matching degree, a building model that meets the modeling requirements is generated.

[0036] In one embodiment, after generating a building model that meets the modeling requirements, the method further includes:

[0037] The generated building model is subjected to conflict detection, including geometric conflict detection, rule conflict detection, and inter-professional conflict detection;

[0038] The detected conflicts are visualized, including highlighting the colliding components, marking the violated specifications, and identifying the locations of mismatched professional interfaces;

[0039] Based on the conflict type, a correction plan is generated, including providing an avoidance path optimization suggestion if the conflict type is a geometric conflict; recommending design parameters that conform to the specifications if the conflict type is a rule conflict; and coordinating the interface dimensions of each discipline if the conflict type is an interdisciplinary conflict.

[0040] The modified building model is subjected to conflict detection. If the detection result shows no conflict, a conflict-free building model is obtained. If the detection result shows a conflict, conflict correction is performed until the conflict is eliminated.

[0041] In one embodiment, after generating a building model that meets the modeling requirements, the method further includes:

[0042] In response to receiving an adjustment instruction from the user for the building model, the adjustment area and associated building components are identified;

[0043] Based on the adjustment instructions, the building components are optimized. The optimization includes calculating the spatial topology of the affected area, adjusting the parameters of the associated electromechanical and structural components, and updating the functional zoning layout.

[0044] Multi-dimensional conflict detection is performed on the optimized building model, including geometric conflict detection, rule conflict detection, and inter-professional conflict detection.

[0045] The building model and optimized data that meet the conflict detection criteria are used as feedback data and stored in the multimodal training dataset.

[0046] Secondly, embodiments of this application provide a Revit modeling system based on a large BIM data model. The system includes a model building module, a structured feature parameter module, a matching module, a component topology relationship module, and a building model generation module.

[0047] The model building module is used to build a modeling model based on a multimodal training dataset. The modeling model includes a feature extraction module, a rule base module, a topological relationship module, and a rule mapping module.

[0048] The structured feature parameter module is used to extract structured feature parameters through the feature extraction module in response to the instruction information received from the user to input modeling requirements.

[0049] The matching module is used to match the structured feature parameters with the building rules in the rule base module to obtain building rules that meet the matching conditions.

[0050] The component topology relationship module obtains the component topology relationship based on the building rules.

[0051] The building model generation module generates a building model that meets the modeling requirements based on the component topology relationship and through the rule mapping module.

[0052] Thirdly, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a Revit modeling method based on a large BIM data model as described in the first aspect above.

[0053] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements a Revit modeling method based on a large BIM data model as described in the first aspect above.

[0054] The Revit modeling method and system based on large BIM data provided in this application embodiment have at least the following technical effects.

[0055] The modeling model, trained using multimodal data, achieves fully automated modeling from requirement input to model generation. First, the feature extraction module accurately converts user natural language requirements into structured parameters, solving the problem of low efficiency in manual parameter setting. Second, the rule base module quickly identifies optimal building rules through a multi-dimensional matching algorithm, significantly improving retrieval efficiency. Next, the topology relationship module automatically derives component connection relationships based on graph neural networks, improving the accuracy of spatial relationship construction. Finally, the rule mapping module generates target topology component relationships that meet all specification requirements through multi-disciplinary collaboration, and generates a building model that meets the aforementioned modeling requirements based on these target topology component relationships. This improves the efficiency of automated modeling and addresses the problem of low automated modeling efficiency in existing Revit modeling methods based on large BIM data models.

[0056] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0057] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0058] Figure 1 This is a flowchart of a Revit modeling method based on a large BIM data model;

[0059] Figure 2 This is a flowchart illustrating step S101 according to an exemplary embodiment;

[0060] Figure 3 This is a flowchart illustrating step S102 according to an exemplary embodiment;

[0061] Figure 4 This is a block diagram of a Revit modeling system based on a large BIM data model, according to one embodiment.

[0062] Figure 5 This is a structural block diagram of an electronic device according to an exemplary embodiment. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0064] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0065] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0066] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0067] BIM technology has been widely applied to the entire lifecycle management of buildings, but modeling is still mainly done manually using tools such as Revit. In recent years, related technological developments have included rule-driven modeling, parametric design tools, and AI-assisted design.

[0068] Because the modeling rules rely entirely on manual coding by engineers, the preset rules often cannot fully cover complex scenarios such as special structural nodes (e.g., irregular curtain wall connections) or integrated pipeline layouts. This results in numerous conflicts in the generated models at the details, requiring repeated manual checks and corrections, which not only increases the workload of later modifications by 30%-50% but also extends the project delivery cycle.

[0069] The existing template library is built based on historical project data. When a new version of the "Building Fire Protection Code" or green building standard is released, the system cannot automatically identify the code changes (such as adjustments to the fire compartment area), which leads to compliance risks in the generated models.

[0070] In the traditional process, design schemes, cost accounting, performance simulation and other links are isolated from each other, and each adjustment requires re-integration of data from various disciplines.

[0071] To address the aforementioned shortcomings, a data-driven BIM large-scale model architecture is proposed, which achieves the following: The model is trained using massive amounts of BIM data, automatically learning modeling rules and design logic. Inputting natural language or parameter requirements directly generates a compliant model. Based on feedback data, the model self-optimizes, forming a closed loop of "requirement input - model generation - iterative optimization".

[0072] Therefore, Revit modeling methods based on large BIM data models suffer from low efficiency in automated modeling. In light of this, this application provides a Revit modeling method and system based on large BIM data models.

[0073] In this document, it should be understood that the terms used may be technical means used to implement part of the present invention or other summary technical terms. For example, the terms may include:

[0074] BIM: Building Information Modeling, a digital model that includes both geometric and non-geometric information.

[0075] Revit: A BIM modeling software developed by Autodesk, based on a parametric family library modeling platform.

[0076] Large models: Deep learning models with over 10 billion parameters, possessing powerful semantic understanding and generation capabilities.

[0077] Firstly, embodiments of this application provide a Revit modeling method based on a large BIM data model. Figure 1 This is a flowchart of a Revit modeling method based on a large BIM data model, such as... Figure 1 As shown, a Revit modeling method based on a large BIM data model includes:

[0078] Step S101: Construct a modeling model based on the multimodal training dataset. The modeling model includes a feature extraction module, a rule base module, a topological relationship module, and a rule mapping module.

[0079] Step S102: In response to receiving the instruction information of the user input modeling requirements, extract structured feature parameters through the feature extraction module.

[0080] Step S103: Match the structured feature parameters with the building rules in the rule base module to obtain building rules that meet the matching conditions.

[0081] Step S104: Based on building rules, obtain the component topology relationship through the topology relationship module.

[0082] Step S105: Based on the component topology relationship, generate a building model that meets the modeling requirements through the rule mapping module.

[0083] In summary, this application provides a Revit modeling method and system based on large BIM data models. By training the model with a large amount of BIM data, it automatically learns modeling rules and design logic. Inputting natural language or parameter requirements directly generates a compliant model. The model self-optimizes based on feedback data, forming a closed loop of "requirement input - model generation - iterative optimization." This improves the efficiency of automated modeling and solves the problem of low automated modeling efficiency in existing Revit modeling methods based on large BIM data models.

[0084] Figure 2 This is a flowchart illustrating step S101 according to an exemplary embodiment, as follows: Figure 2 As shown, step S101 specifically includes the following steps:

[0085] Step S1011: Construct a multimodal training dataset based on historical BIM models, building rules, and user correction records.

[0086] Optionally, first, collect three types of core data: historical BIM models (.rvt files), building rules (such as design specification texts), and user correction records, to construct a multimodal training dataset.

[0087] Step S1011 integrates three types of heterogeneous data: historical BIM models, building rules, and user correction records, ensuring the comprehensiveness and representativeness of the training data. This effectively solves the problems of single data types and inconsistent quality in traditional detection methods, enabling the model to learn from rich architectural expertise and correction experience.

[0088] Step S1012: Perform data cleaning and multimodal transformation based on the multimodal training dataset to obtain structured data; wherein, multimodal transformation is used to convert various types of data into data with the same structure through preset standardized transformation rules.

[0089] Optionally, firstly, perform data cleaning: use the Revit API to batch parse .rvt files, remove redundant components, and repair damaged geometric topological relationships; extract key rule clauses (such as wall thickness and beam and column size standards) from the design specification text using regular expressions; and organize user correction records into a comparison dataset of "error model - corrected model".

[0090] Secondly, multimodal conversion is performed: the BIM model is converted into the IFC standard format, and geometric coordinates and topological adjacency matrix are extracted; the design specification text is converted into a structured rule tree using NLP technology; user correction records are marked with semantic tags (such as "size conflict correction" and "component missing supplement").

[0091] Step S1012 employs the Revit API and regular expressions for professional data cleaning, effectively improving data quality. Multimodal data standardization is achieved through IFC transformation and NLP processing, providing structured input for subsequent machine learning.

[0092] Step S1013: Train the structured data using a deep learning architecture to obtain a modeling model. Specifically, this includes:

[0093] Based on BIM model data from structured data, spatial geometric features of components are extracted using a 3D point cloud neural network. These features include component dimensions, spatial coordinates, and shape parameters. Optionally, a PointNet++-based 3D point cloud network is constructed, taking the component vertex data from the BIM model as input and outputting geometric feature vectors containing spatial location and shape parameters. For example, when modeling beam components in a residential project, the network can learn the distribution patterns of the beam's length, width, and height.

[0094] Based on the component connection relationships in the structured data, a graph neural network is used to construct the topological connection relationships between components. These topological connection relationships include structural support relationships and spatial adjacency relationships. Optionally, a graph neural network (GNN) is used, with components as nodes and connection relationships as edges, to train and identify the support constraint relationships between walls, beams, and columns. In the training of the office building model, the GNN can accurately capture the fixed connection topological logic between the curtain wall and the columns.

[0095] Based on the building rules in the structured data, a natural language processing model is used to parse the building rules and establish a mapping relationship between the building rules and the BIM model. Optionally, a Transformer encoder is designed to process the design specification text and associate the rule clauses with the features of the BIM model through an attention mechanism. For example, the specification "stair tread height should be 150-175mm" is matched and trained with the dimensions of stair components in the model.

[0096] Spatial geometric features, topological connectivity relationships, and mapping relationships are concatenated, and the features are fused using a multilayer perceptron to obtain a modeling model. A multimodal fusion module is constructed, which concatenates geometric features, topological relationship vectors, and rule embedding vectors and inputs them into a multilayer perceptron (MLP). The model parameters are optimized using a cross-entropy loss function to achieve cross-modal knowledge fusion.

[0097] Step S1013 uses a 3D point cloud neural network (such as PointNet++) to extract high-precision geometric features of components, solving the problems of low efficiency and large errors in traditional manual measurement. Secondly, the graph neural network (GNN) models component nodes and connecting edges, automatically learning complex topological relationships (such as accurate identification of curtain wall connecting nodes), overcoming the deficiency of insufficient coverage of manually defined connection rules. Next, the Transformer model parses building code text and establishes a dynamic mapping with BIM features (such as automatically associating stair tread height specifications with model parameters), improving the accuracy of rule application. Finally, a multilayer perceptron fuses geometric, topological, and rule features to form a modeling model with cross-modal understanding capabilities.

[0098] It is worth noting that the above training process generates a modeling model with the ability to reason according to modeling rules. Unlabeled BIM model fragments can also be input into the test set, and the model can predict the types of missing components (e.g., inferring the location of doors and windows based on adjacent walls) and correct conflicting dimensions (e.g., automatically adjusting beam height parameters that exceed specifications). Using the modeling model as initial parameters, fine-tuning can be performed for specific projects (e.g., hospital buildings), allowing for rapid adaptation to new scenarios with a small amount of labeled data, significantly improving modeling efficiency.

[0099] Figure 3 This is a flowchart illustrating step S102 according to an exemplary embodiment, as follows: Figure 3 As shown, step S102, in response to receiving user input indicating modeling requirements, extracts structured feature parameters through the feature extraction module. Specifically, this includes the following steps:

[0100] Step S1021: Perform word segmentation on the user input modeling requirements based on natural language processing, and extract key design elements through entity recognition model. The key design elements include building type, functional requirements and performance indicators.

[0101] Optionally, first, the user requirement (e.g., "Design a 20-story office building with LEED Gold certification") is received. Then, an NLP tool (e.g., spaCy) is used to segment the requirement and identify key entities. For example, entities such as "office building" (building type), "20 stories" (functional requirement), and "LEED Gold certification" (performance indicator) are extracted from "Design a 20-story office building with LEED Gold certification".

[0102] Step S1022: Based on the knowledge graph of the building domain, map key design elements into modeling parameters. The mapping includes mapping building type into standard functional zoning and area ratio, mapping functional requirements into spatial layout constraints, and mapping performance indicators into structural requirements.

[0103] Optionally, by combining BIM knowledge graphs, entities can be mapped to computable parameters. For example, "office building" can be associated with standard functional zoning (office areas, meeting rooms, rest areas, etc.) and area ratios; "LEED Gold certification" corresponds to a set of rules such as energy-saving design specifications and material environmental protection requirements.

[0104] Step S1023: Analyze the modeling parameters through graph neural networks, identify the missing key parameters, and guide the user to supplement and improve them through an interactive interface.

[0105] Optionally, the modeling parameters can be analyzed using graph neural networks to check for missing key information (such as unspecified building dimensions or vague functional requirements). If any are found, the user can be guided to supplement the information through an interactive interface, ultimately forming structured requirement data, such as a requirement vector containing dimensions like building type, number of floors, area, functional requirements, and certification standards.

[0106] Step S1024: Convert the supplemented and improved modeling parameters into structured feature parameters.

[0107] Optionally, for example, to meet the requirement of "20-story office building, LEED Gold certification", the system outputs structured data containing multi-dimensional parameters such as building type, number of floors, area allocation, energy consumption target, and material requirements, driving the model to automatically generate a Revit building model that conforms to the standards, significantly improving modeling efficiency and compliance.

[0108] Step S102 transforms the requirements from natural language to structured parameters. First, NLP technology accurately identifies key elements in the user's requirements. Second, based on a building knowledge graph, semantic elements are mapped to computable parameters (such as specific energy-saving indicators corresponding to LEED certification). Next, a graph neural network analyzes the completeness of the parameters, indicating missing information (such as automatically guiding supplementation when floor height is not specified). Finally, structured parameters containing multi-dimensional features (such as integrating spatial layout and material properties) are generated, providing accurate input for subsequent automated modeling. This reduces the traditional requirement analysis, which takes hours, to minutes, further improving modeling efficiency.

[0109] In one embodiment, step S103 involves matching the structured feature parameters with building rules in the rule base module to obtain building rules that meet the matching conditions. Specifically, this includes:

[0110] Based on the building type identifier in the structured feature parameters, retrieve the candidate rule set for the corresponding category from the rule base module;

[0111] Based on the functional requirements and performance indicators in the structured feature parameters, the semantic matching degree between the functional requirements and performance indicators and the candidate rule set is evaluated by a similarity algorithm.

[0112] Obtain architectural rules that satisfy the preset semantic matching degree.

[0113] Optionally, a rule base is constructed first. This rule base consists of design specifications, historical project experience, industry standards, etc., stored in the form of a knowledge graph. Nodes represent various rules (e.g., "the core area of ​​a high-rise office building should account for 15%-25% of its total area"), and edges represent the relationships between rules (e.g., the relationship between area percentage and the number of floors). Secondly, a similarity-based matching algorithm is used to calculate the semantic similarity between the requirement data and the rule nodes. For example, for the requirement of a "20-story office building," the matching degree with design rules for office buildings with different numbers of floors in the rule base is calculated, and the most relevant set of rules is selected. Simultaneously, combined with the requirement of "LEED Gold certification," further selection of materials and construction rules that meet environmental standards is made.

[0114] Step S103 improves the compliance of modeling design through rule base matching. First, candidate rule sets are quickly filtered based on building type to narrow down the search scope; second, semantic similarity algorithms (such as cosine similarity) are used to accurately match functional requirements and performance indicators (such as environmentally friendly material rules corresponding to LEED Gold certification); finally, a set of compliant rules is output, avoiding the risk of modeling conflicts. Step S103 also uses a knowledge graph to associate rules and parameters to ensure compliance with the latest industry standards.

[0115] Step S104: Based on building rules, obtain the component topology relationships through the topology relationship module. Specifically, this includes:

[0116] Based on the architectural, structural, and MEP component parameters in the building rules, the corresponding component library is called through the Revit API to obtain components including walls, floors, beams, columns, water and electricity pipes, and building equipment.

[0117] Optionally, parametric modeling can be used to drive Revit family library components with rule-based parameters. For example, floor elevation parameters can be set based on "20th floor" to generate wall and window layouts for areas such as offices and meeting rooms according to functional requirements. Component instances can be automatically created and associated parameters set by calling the Revit API via Python.

[0118] By analyzing the connection and position rules of components in architectural rules using graph neural networks, the topological relationships of components, including connection and position relationships, are obtained.

[0119] Optionally, based on topological relationship rules learned by graph neural networks (GNNs), the connection relationships between components can be automatically constructed. For example, correctly connecting floor slabs with beams and columns generates a complete building structural system.

[0120] Step S104 automatically instantiates standard components using the Revit API based on matched architectural rule parameters (such as floor height). A pre-trained graph neural network is then used to analyze component connection rules (such as the stress relationships between beam-column joints).

[0121] In one embodiment, step S105 involves generating a building model that meets the modeling requirements based on the component topology relationship using a rule mapping module. Specifically, this includes:

[0122] The building rules are encoded into rule feature vectors using a pre-trained Transformer encoder;

[0123] Calculate the feature matching degree between the rule feature vector and the component topological relationship;

[0124] Based on feature matching degree, a building model that meets the modeling requirements is generated.

[0125] Optionally, firstly, a Transformer architecture is used to process architectural rule texts. Specifically, architectural rule texts such as the "Code for Fire Protection Design of Buildings" are segmented and semantically embedded, and the context of the clauses is preserved through positional encoding. A multi-head attention mechanism is used to construct a rule feature matrix, where each feature vector contains three dimensions of information: clause type, parameter threshold, and constraint strength. Secondly, the feature parameters of the component topological relationships are matched with the rule feature vectors to obtain feature parameters that meet the standard feature matching degree. Based on the feature parameters, the feature parameters of each component and the feature parameters between each component are determined (e.g., the previous staircase width of 1.05m is determined to be 1.2m). Based on the determined feature parameters of each component and the feature parameters between each component, an architectural model that meets the modeling requirements is generated.

[0126] Step S105 uses a Transformer encoder to accurately parse the semantics of building code provisions. Next, precise parameters are generated through feature matching. Finally, all components are ensured to simultaneously meet geometric compatibility and code compliance requirements, improving the model's compliance.

[0127] In one embodiment, after generating a building model that meets the modeling requirements, the method further includes:

[0128] The generated building model is subjected to conflict detection, including geometric conflict detection, rule conflict detection, and interdisciplinary conflict detection.

[0129] Optionally, geometric conflict detection: Utilizes spatial geometric algorithms to check for collisions between components (such as pipes and beams). It determines if overlapping areas exist by calculating the 3D bounding boxes of the components. Rule conflict detection: Compares model parameters with design code rules, such as checking if the width of evacuation routes complies with fire safety regulations. If conflicts are found, the conflict location and the violated rule clauses are marked. Interdisciplinary conflict detection: Through IFC data parsing, analyzes whether the interfaces between architectural, structural, and MEP (Mechanical, Electrical, and Cemented Materials) models match, for example, checking if air conditioning pipes are consistent with the location of pre-reserved openings in the building. Detected conflicts are visualized, including highlighting colliding components, marking violated code clauses, and indicating mismatched interdisciplinary interfaces.

[0130] Based on the conflict type, a correction plan is generated, including providing avoidance path optimization suggestions if the conflict type is geometric; recommending design parameters that conform to the specifications if the conflict type is rule conflict; and coordinating the interface dimensions of each discipline if the conflict type is interdisciplinary conflict.

[0131] The modified building model is subjected to conflict detection. If the detection result shows no conflict, a conflict-free building model is obtained. If the detection result shows a conflict, conflict correction is performed until the conflict is eliminated.

[0132] Optionally, high-quality modeling can be achieved through collision detection and correction. First, a 3D bounding box algorithm is used to quickly identify physical collisions between components (such as pipelines penetrating beams). Second, design compliance is verified through comparison with a regulatory database (e.g., automatic alarm for insufficient evacuation route width), mitigating risks. Third, professional interfaces are analyzed through IFC data flow analysis (e.g., discrepancies between building pre-reserved holes and MEP pipeline locations), reducing interdisciplinary collaboration issues. Targeted correction solutions are generated (e.g., adjusting pipeline elevation, expanding passage dimensions), and an iterative detection-correction closed loop ensures the final model is conflict-free, making it suitable for complex projects involving multiple disciplines, such as large-scale complexes.

[0133] In one embodiment, after generating a building model that meets the modeling requirements, the method further includes:

[0134] In response to receiving user instructions to adjust the building model, the system identifies the adjustment area and associated building components.

[0135] Based on the adjustment instructions, the building components are optimized, including calculating the spatial topology of the affected area, adjusting the parameters of related electromechanical and structural components, and updating the functional zoning layout.

[0136] Multi-dimensional conflict detection is performed on the optimized building model, including geometric conflict detection, rule conflict detection, and inter-professional conflict detection.

[0137] Building models and optimized data that meet the conflict detection criteria are used as feedback data and stored in the multimodal training dataset.

[0138] Optionally, continuous optimization of the building model can be achieved through user interaction. When users adjust the building layout (e.g., changing an operating room to a circular layout), the affected component range is identified, related professional parameters are coordinated (e.g., simultaneously adjusting the routing of MEP pipelines and the location of structural openings), and multi-dimensional conflict detection ensures that the adjusted model is fully compliant. The optimization results serve as feedback data to enrich the training set, improving the model's efficiency when handling similar needs in subsequent iterations. This forms a closed-loop learning mechanism of "needs input - model generation - iterative optimization," effectively solving the problem that traditional BIM tools cannot accumulate design experience.

[0139] In summary, the Revit modeling method based on a large BIM data model provided in this application embodiment achieves fully automated modeling from requirement input to model generation through a modeling model built using multimodal data training. First, the feature extraction module accurately converts user natural language requirements into structured parameters, solving the problem of low efficiency in manual parameter setting. Second, the rule base module quickly identifies the optimal building rules through a multi-dimensional matching algorithm, greatly improving retrieval efficiency. Next, the topology relationship module automatically derives component connection relationships based on graph neural networks, improving the accuracy of spatial relationship construction. Finally, the rule mapping module generates target topology component relationships that meet all specification requirements through multi-disciplinary collaboration, and generates a building model that meets modeling requirements based on these target topology component relationships. This improves the efficiency of automated modeling and solves the problem of low automated modeling efficiency in existing Revit modeling methods based on large BIM data models.

[0140] Secondly, embodiments of this application provide a Revit modeling system based on a large BIM data model. Figure 4 This is a block diagram illustrating a Revit modeling system based on a large BIM data model, according to one embodiment. For example... Figure 4 As shown, the system includes a model building module, a structured feature parameter module, a matching module, a component topology relationship module, and a building model generation module.

[0141] The model building module is used to build a modeling model based on the multimodal training dataset. The modeling model includes a feature extraction module, a rule base module, a topological relationship module, and a rule mapping module.

[0142] The structured feature parameter module is used to extract structured feature parameters through the feature extraction module in response to user input instructions for modeling requirements.

[0143] The matching module is used to match structured feature parameters with building rules in the rule base module to obtain building rules that meet the matching conditions.

[0144] The component topology module, based on building rules, obtains the component topology relationships.

[0145] The building model generation module generates building models that meet modeling requirements based on component topology relationships and through the rule mapping module.

[0146] In summary, this application provides a Revit modeling system based on large BIM data models. By training the model with a large amount of BIM data, it automatically learns modeling rules and design logic. Inputting natural language or parameter requirements directly generates a compliant model. The model self-optimizes based on feedback data, forming a closed loop of "requirement input - model generation - iterative optimization." This improves the efficiency of automated modeling and solves the problem of low automated modeling efficiency in existing Revit modeling methods based on large BIM data models.

[0147] It should be noted that the Revit modeling system based on a large BIM data model provided in this embodiment is used to implement the above-described implementation methods, and details already described will not be repeated. As used above, the terms "module," "unit," and "subunit" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the above embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0148] Thirdly, embodiments of this application provide an electronic device, Figure 5 This is a block diagram illustrating an electronic device according to an exemplary embodiment. (e.g.) Figure 5 As shown, the electronic device may include a processor 51 and a memory 52 storing computer program instructions.

[0149] Specifically, the processor 51 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0150] The memory 52 may include a large-capacity storage device for data or instructions. For example, and not limitingly, the memory 52 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 52 may include removable or non-removable (or fixed) media. Where appropriate, the memory 52 may be internal or external to a data processing device. In a particular embodiment, the memory 52 is non-volatile memory. In a particular embodiment, the memory 52 includes read-only memory (ROM) and random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable read-only ROM (PROM), an erasable programmable read-only ROM (EPROM), an electrically erasable programmable read-only ROM (EEPROM), an electrically alterable read-only ROM (EAROM), or flash memory, or a combination of two or more of these. Where appropriate, the RAM can be Static Random-Access Memory (SRAM) or Dynamic Random-Access Memory (DRAM). DRAM can be Fast Page Mode Dynamic Random-Access Memory (FPMDRAM), Extended Data Out Dynamic Random-Access Memory (EDODRAM), Synchronous Dynamic Random-Access Memory (SDRAM), etc.

[0151] The memory 52 can be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor 51.

[0152] The processor 51 reads and executes the computer program instructions stored in the memory 52 to implement any of the Revit modeling methods based on BIM data large models in the above embodiments.

[0153] In one embodiment, a Revit modeling device based on a large BIM data model may further include a communication interface 53 and a bus 50. Wherein, as Figure 5 As shown, the processor 51, memory 52, and communication interface 53 are connected through bus 50 and complete communication with each other.

[0154] The communication interface 53 is used to enable communication between the various modules, devices, units, and / or equipment in the embodiments of this application. The communication port 53 can also enable data communication with other components such as external devices, image / data acquisition devices, databases, external storage, and image / data processing workstations.

[0155] Bus 50 includes hardware, software, or both, that couples together components of a Revit modeling device based on a large BIM data model. Bus 50 includes, but is not limited to, at least one of the following: Data Bus, Address Bus, Control Bus, Expansion Bus, and Local Bus. For example, and not as a limitation, bus 50 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 50 may include one or more buses. Although specific buses are described and illustrated in the embodiments of this application, this application considers any suitable bus or interconnection.

[0156] Fourthly, embodiments of this application provide a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements a Revit modeling method based on a large BIM data model provided in the first aspect.

[0157] The readable storage medium may be more specifically adopted, including but not limited to: portable disk, hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical storage device, magnetic storage device, or any suitable combination thereof.

[0158] In a possible implementation, the present invention can also be implemented as a program product comprising program code, which, when the program product is run on a terminal device, causes the terminal device to perform steps to implement the Revit modeling method based on a large BIM data model provided in the first aspect.

[0159] The program code for executing the present invention can be written in any combination of one or more programming languages. The program code can be executed entirely on the user device, partially on the user device, as a standalone software package, partially on the user device and partially on a remote device, or entirely on a remote device.

[0160] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0161] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A Revit modeling method based on a large BIM data model, characterized in that, The method includes: A modeling model is constructed based on a multimodal training dataset, wherein the modeling model includes a feature extraction module, a rule base module, a topological relationship module, and a rule mapping module; In response to receiving user input instructions regarding modeling requirements, the feature extraction module extracts structured feature parameters. The structured feature parameters are matched with the building rules in the rule base module to obtain building rules that meet the matching conditions; Based on the building rules, the component topology relationships are obtained through the topology relationship module; Based on the aforementioned component topology, a building model that meets the aforementioned modeling requirements is generated through the rule mapping module.

2. The Revit modeling method based on a large BIM data model according to claim 1, characterized in that, The step of constructing a modeling model based on a multimodal training dataset includes: A multimodal training dataset was constructed based on historical BIM models, building rules, and user correction records. Data cleaning and multimodal transformation are performed on the multimodal training dataset to obtain structured data; wherein, the multimodal transformation is used to convert various types of data into data with the same structure through preset standardized transformation rules; The structured data is trained using a deep learning architecture to obtain a modeling model.

3. A Revit modeling method based on a large BIM data model according to claim 2, wherein the modeling model is obtained by training the structured data using a deep learning architecture, comprising: Based on the BIM model data in the structured data, the spatial geometric features of the components are extracted through a three-dimensional point cloud neural network, wherein the spatial geometric features include component size, spatial coordinates and shape parameters; Based on the component connection relationships in the structured data, a graph neural network is used to construct the topological connection relationships between components, wherein the topological connection relationships include structural support relationships and spatial adjacency relationships; Based on the building rules in the structured data, the building rules are parsed using a natural language processing model to establish a mapping relationship between the building rules and the BIM model; The spatial geometric features, the topological connectivity, and the mapping relationship are concatenated, and the features are fused using a multilayer perceptron to obtain a modeling model.

4. The Revit modeling method based on a large BIM data model according to claim 1, characterized in that, The step of responding to the instruction information received from the user to input modeling requirements, and extracting structured feature parameters through the feature extraction module, includes: The user input modeling requirements are segmented based on natural language processing, and key design elements are extracted through entity recognition model. The key design elements include building type, functional requirements and performance indicators. Based on the knowledge graph of the building domain, the key design elements are mapped to modeling parameters, wherein the mapping includes mapping the building type to standard functional zoning and area ratio, mapping the functional requirements to spatial layout constraints, and mapping the performance indicators to construction requirements. The modeling parameters are analyzed using graph neural networks to identify missing key parameters, and users are guided to supplement and improve them through an interactive interface. The improved modeling parameters are then converted into structured feature parameters.

5. The Revit modeling method based on a large BIM data model according to claim 1, characterized in that, The step of matching the structured feature parameters with the building rules in the rule base module to obtain building rules that meet the matching conditions includes: Based on the building type identifier in the structured feature parameters, retrieve the candidate rule set for the corresponding category from the rule base module; Based on the functional requirements and performance indicators in the structured feature parameters, the semantic matching degree between the functional requirements and performance indicators and the candidate rule set is evaluated by a similarity algorithm. Obtain the building rules that satisfy the preset semantic matching degree.

6. The Revit modeling method based on a large BIM data model according to claim 1, characterized in that, The process of obtaining component topology relationships based on the building rules through the topology relationship module includes: Based on the architectural, structural, and mechanical and electrical component parameters in the building rules, the corresponding component library is called through the Revit API to obtain components including walls, floors, beams, columns, water and electricity pipes, and building equipment; By analyzing the connection and position rules of components in the architectural rules using a graph neural network, the topological relationships of the components, including connection and position relationships, are obtained.

7. The Revit modeling method based on a large BIM data model according to claim 1, characterized in that, Based on the aforementioned component topology, a building model conforming to the aforementioned modeling requirements is generated through the rule mapping module, including: The building rules are encoded into rule feature vectors using a pre-trained Transformer encoder; Calculate the feature matching degree between the rule feature vector and the component topological relationship; Based on the feature matching degree, a building model that meets the modeling requirements is generated.

8. The Revit modeling method based on a large BIM data model according to claim 1, characterized in that, After generating a building model that meets the modeling requirements, the method further includes: The generated building model is subjected to conflict detection, including geometric conflict detection, rule conflict detection, and inter-professional conflict detection; The detected conflicts are visualized, including highlighting the colliding components, marking the violated specifications, and identifying the locations of mismatched professional interfaces; Based on the conflict type, a correction plan is generated, including providing an avoidance path optimization suggestion if the conflict type is a geometric conflict; recommending design parameters that conform to the specifications if the conflict type is a rule conflict; and coordinating the interface dimensions of each discipline if the conflict type is an interdisciplinary conflict. The modified building model is subjected to conflict detection. If the detection result shows no conflict, a conflict-free building model is obtained. If the detection result shows a conflict, conflict correction is performed until the conflict is eliminated.

9. A Revit modeling method based on a large BIM data model according to claim 1, characterized in that, After generating a building model that meets the modeling requirements, the method further includes: In response to receiving an adjustment instruction from the user for the building model, the adjustment area and associated building components are identified; Based on the adjustment instructions, the building components are optimized. The optimization includes calculating the spatial topology of the affected area, adjusting the parameters of the associated electromechanical and structural components, and updating the functional zoning layout. Multi-dimensional conflict detection is performed on the optimized building model, including geometric conflict detection, rule conflict detection, and inter-professional conflict detection. The building model and optimized data that meet the conflict detection criteria are used as feedback data and stored in the multimodal training dataset.

10. A Revit modeling system based on a large BIM data model, characterized in that, The system includes a model building module, a structured feature parameter module, a matching module, a component topology relationship module, and a building model generation module. The model building module is used to build a modeling model based on a multimodal training dataset. The modeling model includes a feature extraction module, a rule base module, a topological relationship module, and a rule mapping module. The structured feature parameter module is used to extract structured feature parameters through the feature extraction module in response to the instruction information received from the user to input modeling requirements. The matching module is used to match the structured feature parameters with the building rules in the rule base module to obtain building rules that meet the matching conditions. The component topology relationship module obtains the component topology relationship based on the building rules. The building model generation module generates a building model that meets the modeling requirements based on the component topology relationship and through the rule mapping module.

Images