Building information model element classification method based on artificial intelligence

Through artificial intelligence-based methods, combining three-dimensional data and attribute data, image description and stitching models are created to classify BIM elements, solving the problems of high cost, low accuracy and low efficiency caused by manual judgment in the existing technology, and realizing the Efficient BIM element classification.

CN119066467AInactive Publication Date: 2024-12-03BEIJING YUNJIANXIN TECH CO LTD

Patent Information

Application Number
CN202411106537.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-13
Publication Date
2024-12-03
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

In the prior art, BIM element classification relies on manual judgment, resulting in high cost, low accuracy and low efficiency, low classification accuracy of automatic judgment methods is not high, and incomplete classification.

Method used

Using an artificial intelligence-based method, combining three-dimensional data and attribute data, BIM elements are classified using neural networks by creating and training BIM element classification models, including image description models and stitching models.

Benefits of technology

The accuracy and efficiency of BIM element classification is significantly improved, with a classification accuracy of more than 95%, which is better than pure three-dimensional data and pure attribute data models.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119066467A_ABST
    Figure CN119066467A_ABST
Patent Text Reader

Abstract

The invention discloses a building information model element classification method based on artificial intelligence, and the method comprises the steps: determining BIM modeling software based on a file format; from the BIM modeling software, BIM element data used for training are exported; determining a BIM element category of the BIM modeling software; a BIM element classification model is created and trained, and the BIM element classification model comprises an image description BIM element classification model supporting image input or a splicing BIM element classification model supporting image and text input; and utilizing the BIM element classification model to classify BIM element data. According to the building information model element classification method based on artificial intelligence, the three-dimensional data and the attribute data are supported to complete BIM element classification at the same time, so that the BIM element information can be fully used, the classification accuracy is improved, and different software and hardware environments and capability requirements can be supported.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of building information model applications, and in particular relates to a building information model element classification method based on artificial intelligence. Background Art

[0002] As one of the key technologies for the informatization development of the construction industry, Building Information Modeling (BIM) provides the construction industry with a series of standards, methods and tools for the creation, storage, transmission, delivery and application of engineering data.

[0003] BIM elements refer to the components of buildings / structures with actual shapes and engineering properties in the building information model data. BIM elements have certain category characteristics. For example, the EF table (element / function) in the Uniclass classification system and Table 21 (elements) in the OmniClass coding system are both classification standards for BIM elements. In addition, the definition of classes in the IFC standard is also a classification method, that is, all classes derived from IfcElement are BIM elements. BIM elements are core building data, and their related data generally include three-dimensional, attributes and associations. BIM element categories can support basic capabilities such as data retrieval, statistics, and association, and thus support numerous BIM application processes. Therefore, the accuracy of BIM element categories is very critical.

[0004] At present, the categories of BIM elements are basically determined manually, such as the model family selected by the modeler during the modeling process, or the classification code added to the model of the building element manually after the modeling is completed. Such methods not only require a lot of labor costs, but are also prone to classification errors or incompleteness.

[0005] In recent years, people have studied the method of automatically determining the category of BIM elements. The three-dimensional data of point clouds or images from multiple perspectives are used to classify BIM elements, which solves a lot of labor costs. However, the method of automatically determining the category of BIM elements still has problems such as low classification accuracy, incomplete classification and low efficiency. Summary of the invention

[0006] In order to address the deficiencies of the prior art, the purpose of this application is to provide a building information model element classification method based on artificial intelligence, which uses three-dimensional data and attribute data to complete BIM element classification and improve the accuracy and efficiency of classification.

[0007] To achieve the above purpose, the present application also provides an artificial intelligence-based building information model element classification method, including:

[0008] Determine the BIM modeling software based on the file format;

[0009] Exporting BIM element data for training from the BIM modeling software;

[0010] Determine the BIM element categories of the BIM modeling software;

[0011] Creating and training a BIM element classification model, wherein the BIM element classification model includes an image description BIM element classification model supporting image input or a spliced ​​BIM element classification model supporting image and text input;

[0012] The BIM element data is classified using the BIM element classification model.

[0013] Furthermore, the step of determining the BIM modeling software based on the file format also includes:

[0014] For the unique file format of BIM modeling software, the BIM modeling software can be directly determined;

[0015] For generic files, the modeling software that generated the file is determined based on the file content or source.

[0016] Furthermore, the step of exporting each BIM element data for training from the BIM modeling software further includes:

[0017] Obtain BIM files generated by all identified BIM modeling software;

[0018] Extract the 3D data and attribute data of BIM elements from the acquired BIM files;

[0019] Store the 3D data and attribute data of all BIM elements as exported intermediate format files;

[0020] Using the exported intermediate format file, a fixed-viewing angle rendering image is generated for each BIM element.

[0021] Furthermore, the step of storing the three-dimensional data and attribute data of all BIM elements as exported intermediate format files further includes:

[0022] Store the 3D data of each BIM element as a glTF format file;

[0023] The property data of each BIM element is stored as a JSON format file.

[0024] Furthermore, the step of generating a fixed-viewing angle rendering image for each BIM element using the exported intermediate format file further includes:

[0025] Use blender to load the glTF format file of each BIM element, and render the images separately at a fixed camera perspective to generate a rendered image with a fixed perspective.

[0026] Furthermore, the step of creating and training an image description BIM element classification model supporting image input further includes:

[0027] Generate description text for each fixed-view rendering image of a BIM element;

[0028] Integrate the description texts of the fixed-view rendering images of all BIM elements into component description texts;

[0029] Integrate the obtained component description text with the BIM element attribute text to obtain the description text integration result of the BIM element;

[0030] The descriptive text integration results of the BIM elements are classified to obtain the classification results of the BIM elements of the BIM modeling software.

[0031] Furthermore, the step of creating and training a splicing BIM element classification model supporting image and text input also includes:

[0032] Combine the fixed-view rendering images into an image representation tensor;

[0033] Process attribute data into an attribute representation tensor;

[0034] The image representation tensor and the attribute representation tensor are classified to obtain a classification result of the BIM elements of the BIM modeling software.

[0035] Furthermore, the step of classifying the image representation tensor and the attribute representation tensor to obtain a classification result of the BIM element of the BIM modeling software further includes:

[0036] After the image representation tensor and the attribute representation tensor are concatenated, they are passed through a neural network layer that supports classification to obtain a classification result.

[0037] To achieve the above-mentioned purpose, the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is used to execute the computer program stored in the memory to implement the steps of the artificial intelligence-based building information model element classification method as described above.

[0038] To achieve the above objectives, the present application also provides a computer-readable storage medium, in which a computer program is stored. The computer program is loaded and executed by a processor to implement the steps of the artificial intelligence-based building information model element classification method as described above.

[0039] The artificial intelligence-based building information model element classification method provided in this application has the following beneficial effects compared with the prior art:

[0040] In view of the unique classification model of each modeling software, it supports the use of 3D data and attribute data to complete BIM element classification, so that BIM element information can be fully used and the upper limit of classification accuracy is improved. At the same time, image description model and splicing model are proposed to support different software and hardware environments and capability requirements.

[0041] After being fully trained with data of more than 50,000 BIM elements from actual projects, the classification model proposed in the present invention has been verified to have a classification accuracy of more than 95%, and is significantly greater than classification models using pure three-dimensional data and pure attribute data.

[0042] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent from the description, or may be understood by practicing the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings are used to provide a further understanding of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings:

[0044] Figure 1 A flowchart of a method for classifying building information model elements based on artificial intelligence according to an embodiment of the present application;

[0045] Figure 2 A flowchart for training a BIM element classification model according to an image description of the present application;

[0046] Figure 3 A flow chart for training a splicing BIM element classification model according to the present application;

[0047] Figure 4 Schematic diagram of the structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] The preferred embodiments of the present application are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application and are not used to limit the present application.

[0049] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not intended to limit the scope of protection of the present application.

[0050] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0051] It should be noted that the concepts of “first”, “second”, etc. may be mentioned in this application and are only used to distinguish different devices, components or parts, and are not used to limit the order or interdependence of the functions performed by these devices, components or parts.

[0052] It should be noted that the modifications of "one" and "plurality" mentioned in this application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more". "Plurality" should be understood as two or more.

[0053] In an embodiment of the present application, a method for classifying building information model elements based on artificial intelligence is provided, comprising the following steps:

[0054] Based on the file format, determine the BIM modeling software that generated the file;

[0055] Export BIM element data for training from the determined BIM modeling software;

[0056] Determine the BIM element categories of the BIM modeling software;

[0057] Build and train the BIM element classification model for the determined BIM modeling software;

[0058] The BIM elements of BIM modeling software are classified using the BIM element classification model.

[0059] Figure 1 The following is a flow chart of the method for classifying building information model elements based on artificial intelligence according to an embodiment of the present application. Figure 1 , the artificial intelligence-based building information model element classification method of this application is described in detail.

[0060] First, in step 101, based on the file format, the BIM modeling software that generates the file is determined.

[0061] In the embodiment of the present application, for the unique file format of the BIM modeling software, the BIM modeling software can be directly judged; for the IFC standard or other common file formats, the modeling software that generates the file is judged according to the file content or source.

[0062] Preferably, IfcApplication in the IFC file is extracted. Wherein, BIM modeling software includes but is not limited to Autodesk Revit, Bentley Microstation, Tekla Structures, and Graphisoft Archicad.

[0063] In step 102, BIM element data for training is exported from the determined BIM modeling software.

[0064] In the embodiment of the present application, the BIM element data for training is exported from the determined BIM modeling software, including:

[0065] 1) Prepare a series of BIM files with significantly different contents generated by the determined BIM modeling software.

[0066] Preferably, collect the final version RVT files created using Autodesk Revit in various engineering projects.

[0067] 2) Extract the three-dimensional data and attribute data of BIM elements from BIM files.

[0068] Preferably, for RVT files, use the Autodesk Revit plug-in to load the files, and use the secondary development interface to read the three-dimensional data (including geometric data and material data) and attribute data of all BIM elements;

[0069] Preferably, for an IFC file, an IFC parsing library is used to read all IfcElements to obtain their geometric data, material data, and associated attribute data.

[0070] 3) Export the 3D data and attribute data storage of all BIM elements to an intermediate format.

[0071] Preferably, the three-dimensional data of each BIM element is stored as a glTF format file, and the attribute data is stored as a JSON format file.

[0072] 4) Generate several fixed-view rendering images for each BIM element.

[0073] Preferably, use blender to load the glTF format file generated by each BIM element in the previous step, and render the image at 8 fixed camera perspectives respectively to generate 8 fixed perspective rendering images.

[0074] In step 103, a BIM element category of the BIM modeling software is determined.

[0075] In the embodiment of the present application, the categories of BIM elements include but are not limited to: IFC class name, Uniclass code, and OmniClass code.

[0076] In the embodiment of the present application, the method used to determine the categories of all BIM elements includes:

[0077] 1) Directly use manual means to determine the categories of all BIM elements;

[0078] 2) Use automated methods to determine the categories of some BIM elements, and then use manual review to correct the automatically determined category results and supplement the categories of some BIM elements that cannot be automatically determined.

[0079] In the embodiment of the present application, the automation means that can be used are related to the specific form of the category. If the IFC class name is used, the IFC can be exported using the BIM modeling software, and then the IFC Class of each BIM element can be determined by a unique identifier; if the Uniclass code is used, a mapping table between the category of the BIM element in the BIM modeling software and the Uniclass code can be prepared, and the Uniclass code can be automatically added to each BIM element.

[0080] In step 104, a BIM element classification model of the determined BIM modeling software is constructed and trained.

[0081] In the embodiment of the present application, the BIM element classification model of the constructed BIM modeling software includes: an image description BIM element classification model and a splicing BIM element classification model.

[0082] The image description BIM element classification model supports image input. The multi-angle component images are processed into description text by a pre-trained picture description model (such as BLIP, BLIP2 or VisualGLM, etc.), and then summarized into component description text by a large prediction model (such as ChatGLM); thereafter, the component description text is concatenated with the attribute text, and the classification result label is obtained through a pre-trained text classification model (such as fasttext, BERT, etc.).

[0083] The spliced ​​BIM element classification model supports both image and text input. The multi-angle component images are processed through the image processing layer and the ResNet network with the fully connected (FC) layer removed to obtain the image representation tensor. The attribute text is processed through the text2vec layer to obtain the attribute representation tensor. After the image representation tensor and the attribute representation tensor are spliced, they pass through the new FC layer to obtain the classification result label.

[0084] Figure 2 To describe the BIM element classification model training flow chart according to the image of this application, the following will refer to Figure 2 , the image description BIM element classification model of this application is described in detail for the classification process of BIM elements of each BIM modeling software:

[0085] First, in step 201, a description text is generated for a fixed-view rendering image of each BIM element;

[0086] Preferably, this embodiment can use image description models such as BLIP, BLIP2, VisualGLM, etc. to generate description text for the fixed-viewing angle rendering image of each BIM element.

[0087] In step 202, the description texts of all rendered images are integrated into one description text (component description text);

[0088] Preferably, this embodiment can use a large language model to integrate the description texts of all rendered images into one paragraph of description text.

[0089] In step 203, the component description text obtained in steps 1) and 2) is integrated and spliced ​​with the BIM element attribute text to obtain a description text integration result of the BIM element;

[0090] Preferably, this embodiment integrates the component description text with the BIM element attribute text in the following manner:

[0091] Add a "Description" attribute in the BIM element properties, and its value is the component description text.

[0092] In step 204, the description text integration results of the BIM elements are classified to obtain the classification results of the BIM elements of the BIM modeling software.

[0093] Preferably, this embodiment integrates the description text of the BIM elements and inputs them into a text classification model such as fasttext and BERT for classification to obtain the classification results of the BIM elements.

[0094] Figure 3 For the training flow chart of the splicing BIM element classification model according to this application, the following will refer to Figure 3, the splicing BIM element classification model of this application is used to classify the BIM elements of each BIM modeling software in detail:

[0095] First, in step 301, the rendered images of each fixed perspective are merged into a representation tensor tensor (image representation tensor);

[0096] Preferably, use the opencv library to read and merge the image matrix, then use transforms to convert the image matrix into a tensor, unify the size, complete normalization, use this tensor as input, use the ResNet pre-trained model with the fully connected layer removed, and output the image representation tensor (1-dimensional tensor).

[0097] In step 302, the attribute data is processed into an attribute representation tensor (1-dimensional tensor);

[0098] Preferably, the attribute JSON text is converted into a vector using the text2vec pre-trained model, and then directly converted into a 1-dimensional tensor.

[0099] In step 303, the obtained tensor is classified to obtain the classification result of the BIM elements of the BIM modeling software.

[0100] Preferably, in this embodiment, the above two 1-dimensional tensors are concatenated and then passed through a neural network layer that supports classification (such as a fully connected layer) to obtain a classification result.

[0101] In an embodiment of the present application, the BIM element data of the designated modeling software is used to train one of the above-mentioned image description BIM element classification model and the splicing BIM element classification model to obtain a usable BIM element classification model.

[0102] In step 105, the BIM elements are classified using the usable BIM element classification model.

[0103] In the embodiment of the present invention, for the BIM elements to be classified, the classification is completed through the following steps:

[0104] 1) Using the method of step 102, export the BIM element data to be classified;

[0105] 2) For each BIM element, the exported BIM element data is input into the BIM element classification model obtained in step 104 to obtain a BIM element classification result.

[0106] In an embodiment of the present application, an electronic device is also provided. Figure 4 is a schematic diagram of the structure of an electronic device according to an embodiment of the present application, such as Figure 4 As shown, the electronic device of the present application includes a processor 401 and a memory 402, wherein:

[0107] The memory 402 stores a computer program. When the computer program is read and executed by the processor 401, the computer program executes the steps in the embodiment of the method for classifying building information model elements based on artificial intelligence as described above.

[0108] In an embodiment of the present application, a computer-readable storage medium is further provided, in which a computer program is stored, wherein the computer program is configured to execute the steps in the embodiment of the artificial intelligence-based building information model element classification method when running.

[0109] In this embodiment, the above-mentioned computer-readable storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store computer programs.

[0110] Those skilled in the art can understand that the above description is only a preferred embodiment of the present application and is not intended to limit the present application. Although the present application is described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions recorded in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for classifying building information model elements based on artificial intelligence, comprising: Determine the BIM modeling software based on the file format; Exporting BIM element data for training from the BIM modeling software; Determine the BIM element categories of the BIM modeling software; Creating and training a BIM element classification model, wherein the BIM element classification model includes an image description BIM element classification model supporting image input or a spliced ​​BIM element classification model supporting image and text input; The BIM element data is classified using the BIM element classification model.

2. The method for classifying building information model elements based on artificial intelligence according to claim 1, characterized in that: The step of determining the BIM modeling software based on the file format also includes: For the unique file format of BIM modeling software, the BIM modeling software can be directly determined; Based on the generic file, the modeling software that generated the file is determined based on the file content or source.

3. The method for classifying building information model elements based on artificial intelligence according to claim 1, characterized in that: The step of exporting each BIM element data for training from the BIM modeling software further includes: Obtain BIM files generated by all identified BIM modeling software; Extract the 3D data and attribute data of BIM elements from the acquired BIM files; Store the 3D data and attribute data of all BIM elements as exported intermediate format files; Using the exported intermediate format file, a fixed-viewing angle rendering image is generated for each BIM element.

4. The method for classifying building information model elements based on artificial intelligence according to claim 3 is characterized in that: The step of storing the three-dimensional data and attribute data of all BIM elements as an exported intermediate format file also includes: Store the 3D data of each BIM element as a glTF format file; The property data of each BIM element is stored as a JSON format file.

5. The method for classifying building information model elements based on artificial intelligence according to claim 4, characterized in that: The step of generating a fixed-viewing angle rendering image for each BIM element using the exported intermediate format file further includes: Use blender to load the glTF format file of each BIM element, and render the images separately at a fixed camera perspective to generate a rendered image with a fixed perspective.

6. The method for classifying building information model elements based on artificial intelligence according to claim 1, characterized in that: The step of creating and training an image description BIM element classification model that supports image input also includes: Generate description text for each fixed-view rendering image of a BIM element; Integrate the description texts of the fixed-view rendering images of all BIM elements into component description texts; Integrate the obtained component description text with the BIM element attribute text to obtain the description text integration result of the BIM element; The descriptive text integration results of the BIM elements are classified to obtain the classification results of the BIM elements of the BIM modeling software.

7. The method for classifying building information model elements based on artificial intelligence according to claim 1, characterized in that: The step of creating and training a splicing BIM element classification model supporting image and text input also includes: Combine the fixed-view rendering images into an image representation tensor; Process attribute data into an attribute representation tensor; The image representation tensor and the attribute representation tensor are classified to obtain a classification result of the BIM elements of the BIM modeling software.

8. The method for classifying building information model elements based on artificial intelligence according to claim 7, characterized in that: The step of classifying the image representation tensor and the attribute representation tensor to obtain a classification result of the BIM element of the BIM modeling software further includes: After the image representation tensor and the attribute representation tensor are concatenated, they are passed through a neural network layer that supports classification to obtain a classification result.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor is used to execute the computer program stored in the memory to implement the steps of the artificial intelligence-based building information model element classification method described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, which is loaded and executed by a processor to implement the steps of the artificial intelligence-based building information model element classification method according to any one of claims 1 to 8.

Citation Information

Patent Citations

  • BIM-based data processing method and apparatus for prefabricated building process design

    CN108427810A

  • BIM rapid modeling method based on computer vision

    CN112613097A

  • Three-dimensional building model component classification method and device

    CN113554065A

  • Image classification method and device, computer equipment and medium

    CN114741581A

  • BIM graphic engine loading method and device based on WEB end

    CN114969586A

Cited By

  • Multi-engineering design tool collaborative design information providing management method and device

    CN119441145A

  • A collaborative design information submission management method and device for multiple engineering design tools

    CN119441145B