Pre-training artificial intelligence method for automatically generating BIM, electronic equipment and storage medium

Abstract

The invention discloses a pre-training artificial intelligence method for automatically generating BIM, electronic equipment and a storage medium, and belongs to the field of building intelligence. Real features and spatial features of five types of building components are extracted from the BIM to construct a BIM graph, residual quantization is performed on coded representation after graph neural network coding, the coded representation after residual quantization is reversely added and sent into a residual network decoder to restore the real features of the coded representation, and token sequence representation of a building is obtained from the coded representation. And inputting the token sequence of the building into the pre-training network model to learn component arrangement in the building. The obtained pre-training network model generates a building description sequence, a decoder decodes real features, and modeling is completed through a BIM modeling script program. According to the method, the BIM automatic generation time is remarkably shortened, expansion is easy, continuous updating and learning are facilitated, the generated BIM can be displayed on multiple platforms, people can check and modify the generated BIM conveniently, and the method has wide application prospects.

Description

Technical Field

[0001] The present invention relates to a pre-training artificial intelligence method, electronic equipment and storage medium for automatically generating BIM, and belongs to the field of building intelligentization. Background Art

[0002] Building Information Modeling (BIM) technology has significantly transformed the entire lifecycle of the construction industry, improving design and construction efficiency and reducing construction and maintenance costs. However, automated BIM modeling remains a challenge that needs to be addressed.

[0003] Currently, traditional BIM modeling relies primarily on manual operations using specialized software (such as Revit and ArchiCAD), supplemented by design drawings, specifications, and 2D CAD software. The modeling process requires manual input of geometry and attribute information, as well as the integration of models from multiple disciplines. This is not only inefficient and error-prone, but also makes collaborative work complex and cumbersome. Automating these repetitive and tedious modeling processes would significantly reduce BIM modeling time, allowing engineers to focus more on other critical architectural design tasks.

[0004] Therefore, the present invention aims to propose a feasible solution for automatically generating BIM (Building Information Model) to solve the problems of low efficiency, error-proneness, and complex collaboration in the traditional BIM modeling process, while providing support for intelligent building design. Summary of the Invention

[0005] Purpose: This invention designs a pre-trained artificial intelligence method, electronic device, and storage medium for automatically generating BIM. The method comprises three components: building information extraction, network model training, and BIM modeling. This method not only learns building component layouts for architectural design but also automatically completes BIM modeling based on predicted building description sequences, transforming traditional manual BIM modeling into a fully automated process. It also transforms difficult-to-describe building component features (both physical and spatial) into trainable tokens, enabling learning of architectural styles and generation of description sequences. This method significantly improves modeling efficiency, reduces the errors and time costs associated with manual operations, and provides relevant assistance for architectural design.

[0006] Technical solution: To solve the above technical problems, the technical solution adopted by the present invention is:

[0007] First, a pre-trained artificial intelligence method for automatically generating BIM, specifically comprising:

[0008] The BIM model is cleaned and relevant building component information is extracted, wherein the relevant building component information includes real characteristics and spatial characteristics of the building components.

[0009] The real features of building components are discretized to obtain the discretized real features.

[0010] Graph data is constructed based on the discretized real features and the spatial features of building components. The graph data is input into the graph neural network, residual quantization module and decoder connected in sequence, and the building component description vocabulary, as well as the trained graph neural network, residual quantization module and decoder are output.

[0011] A description sequence of building components is constructed based on relevant building component information and a building component description vocabulary.

[0012] The pre-trained model is trained using the description sequence of the building components to obtain a trained pre-trained model.

[0013] Obtain the information of the building components to be generated, obtain the vocabulary sequence of the building components to be generated through the trained graph neural network and residual quantization module, input the vocabulary sequence of the building components to be generated into the trained pre-trained model, and output the description sequence of the building components to be generated.

[0014] The description sequence of the building component to be generated is quantized and deeply summed up, and then input into the trained decoder to output the real features of the building component to be generated.

[0015] Generate BIM modeling based on the real characteristics of the building components to be generated.

[0016] Optionally, the real characteristics of the building components include: types, orientations, sizes and positions of columns, beams, walls, doors and windows.

[0017] The spatial characteristics of the building components include the adjacent relationships among columns, beams, walls, doors and windows.

[0018] Optionally, the discretization operation specifically includes:

[0019] Taking the minimum difference in the size of building components as the resolution, the position of the building components is adjusted to the positive range in the three-dimensional coordinate system, and the real features of the discretized building components are constructed based on the resolution and the adjusted position.

[0020] Optionally, the graph neural network includes three layers of SAGEConv convolutional layers connected sequentially.

[0021] The quantization depth in the residual quantization module is set to 3.

[0022] The decoder includes 5 layers of residual networks connected in sequence.

[0023] Optionally, constructing a description sequence of building components based on relevant building component information and a building component description vocabulary specifically includes:

[0024] According to the relevant building component information, corresponding vocabulary is obtained from the building component description vocabulary table.

[0025] Arrange the vocabulary in the order of column, beam, wall, door, and window. Arrange the vocabulary of the same type in ascending order according to the coordinate values ​​of the Z, Y, and X axes of the three-dimensional Cartesian coordinate system.

[0026] Then, the beginning word and the ending word are added to the first and last position of the sequence respectively to obtain the description sequence of the building components.

[0027] Optionally, the pre-training model adopts a Transformer-OnlyDecoder model with a 3layers-4heads attention mechanism.

[0028] Optionally, the step of obtaining information of the building component to be generated and obtaining a vocabulary sequence of the building component to be generated through a trained graph neural network and a residual quantization module may specifically include:

[0029] The information of the building components to be generated is obtained as the real features of some building components, the real features of some building components are input into the trained graph neural network and residual quantization module, and the vocabulary sequence of the building components to be generated is output.

[0030] The information of the building component to be generated is a starting word, and the starting word is used as a word sequence of the building component to be generated.

[0031] Optionally, generating a BIM model based on the real features of the building components to be generated specifically includes:

[0032] Create IFC entities corresponding to columns, beams, walls, doors, and windows, and obtain the category, orientation, size, and position based on the real characteristics of the building components to be generated. Generate IFC statements corresponding to the IFC entities to define the geometric and spatial characteristics of the building components and complete BIM modeling.

[0033] The Z-axis positions of column and wall building components are divided into intervals and counted, and the intervals with the number of data points exceeding the threshold are screened out. The number of floors and heights are divided according to the number of intervals and the minimum value in the interval to obtain the information of each floor in BIM modeling.

[0034] In a second aspect, a computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements a pre-trained artificial intelligence method for automatically generating BIM as described in any one of the first aspects.

[0035] According to a third aspect, a computer device includes:

[0036] Memory, used to store instructions.

[0037] A processor is configured to execute the instructions so that the computer device performs the operations of a pre-trained artificial intelligence method for automatically generating a BIM as described in any one of the first aspects.

[0038] Beneficial Effects: This invention provides a pre-trained artificial intelligence method, electronic device, and storage medium for automatically generating BIM. By automating the entire process, this method significantly improves modeling efficiency and reduces the errors and time costs associated with manual operations. It generates preliminary BIM designs based on predicted building description sequences, helping designers quickly optimize their designs. With data accumulation and optimization, the method can be continuously improved to accommodate various BIM modeling requirements. The method is robust, streamlined, and effective, with broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of a pre-training artificial intelligence method for automatically generating BIM according to the present invention.

[0040] Figure 2 It is a schematic diagram of BIM information extraction and result visualization of the present invention.

[0041] Figure 3 It is a schematic diagram of the spatial feature determination algorithm in the present invention.

[0042] Figure 4 It is a schematic diagram of the floor division algorithm of the present invention.

[0043] Figure 5 It is an input pattern diagram of two generation tasks of the pre-training model of the present invention.

[0044] Figure 6 It is a schematic diagram of the BIM model generated by the present invention. DETAILED DESCRIPTION

[0045] The following is a clear and complete description of the technical solutions in the examples of the present invention, in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative work are within the scope of protection of the present invention.

[0046] The present invention will be further described below with reference to specific embodiments.

[0047] Example 1:

[0048] This embodiment introduces a pre-trained artificial intelligence method for automatically generating a BIM (Building Information Model). This method covers BIM information processing, network model training, and BIM automatic modeling procedures. The three are integrated into a continuously learning BIM automatic modeling system. The method is as follows:

[0049] Step 1: Clean the acquired BIM model and extract relevant building component information, where the relevant building component information includes the real characteristics (category, orientation, size, location) and spatial characteristics (adjacency relationship between components).

[0050] This step is used to obtain the five types of building components that this method focuses on, which helps to obtain a more concise ifc file and improve the efficiency of building component information extraction.

[0051] Furthermore, the step 1 specifically includes:

[0052] Step 1.1: Get the real characteristics of the building components:

[0053] Clean up the building components that are not considered in the BIM model to obtain the cleaned BIM model.

[0054] Export the cleaned BIM model to the corresponding IFC (Industry Foundation Classes) file format and extract the real features of the components according to the semantic rules of IFC.

[0055] The real features of the components are extracted according to the semantic rules of IFC, including accessing IfcColumn, IfcBeam, IfcWallStandardCase, IfcDoor and IfcWindow entities in the file through the designed real feature extraction program to obtain relevant information of columns, beams, walls, doors and windows respectively. For columns and walls, the dimensions are obtained by traversing their dimension statements (such as IfcRectangleProfileDef, IfcArbitraryClosedProfileDef, IfcPolyLine, and IfcClosedShell), and the position and orientation are determined based on the IfcAxis2Placement3D statement. For beams, the dimensions are obtained through dimension statements (such as IfcRectangleProfileDef and IfcConnectedFaceSet), and the position and orientation are reversely deduced based on the IfcAxis2Placement3D statement. For doors and windows, the dimensions are obtained through the OverallWidth, OverallHeight, IfcPolyLine, and IfcClosedShell statements, and their position and orientation are determined through the IfcAxis2Placement3D statement. A CSV file containing the true characteristics of the building components is then obtained.

[0056] Step 1.2: Obtain the spatial characteristics between building components:

[0057] Access BIM model related information by accessing the API and implementing external commands (IExternalCommand) to obtain the bounding box of the building component and add the bounding box expansion value ,in, ( is a manually set bounding box expansion value) to expand the bounding box, and determine whether it is (Volumn is the amount of cross-overlap of the bounding box after the building component is enlarged) to determine whether the building components are adjacent and obtain the corresponding CSV file, such as Figure 3 shown.

[0058] Step 2: Distribute and analyze the real characteristics of the building components in step 1, calculate the minimum difference in the size of the building components, and define the BIM modeling space resolution. , adjust the extracted overall building position coordinates to the positive range in the three-dimensional coordinate system. That is, define the BIM modeling space range according to the maximum value of the size and position attributes of the building components: The size and position characteristics of the building components after displacement and discretization are used as the real features after discretization.

[0059] This step not only preserves the characteristics of the BIM itself to the greatest extent, but also makes the decoder in step 3.3 more accurate in predicting the real features.

[0060] Step 3: The discretized real features are encoded through the graph neural network encoder, and then sent to the decoder network after passing through the residual quantization module to obtain the predicted real features of the building components. These predicted real features are used as modeling parameters of the building components and for BIM modeling in step 6.

[0061] This step uses an encoder-decoder training method to input the real features of the building components and the spatial feature graph from step 2. The encoded high-dimensional features of the building components are re-described by the residual quantization module and finally fed into the decoder to obtain the predicted values ​​of the real features of the building components. The parameters of the encoder-decoder neural network are trained using cross-loss entropy. After training, the building component description vocabulary (token list) is obtained from the residual quantization module.

[0062] Furthermore, the step 3 specifically includes:

[0063] Step 3.1: Encode the true features of building components to obtain high-dimensional features. Describe building components in a higher dimensional way:

[0064] The graph neural network consists of 3 layers of SAGEConv convolutional layers; the depth is 3 residual quantization depths; and the decoder consists of a 5-layer residual network. Specifically, the first layer of SAGEConv convolutional layer maps the input feature dimension from 15 to 32, the second layer maps the feature dimension from 32 to 64, and the third layer maps the feature dimension from 64 to 128.

[0065] Step 3.2: The quantization depth in the residual quantization module is set to 3. The high-dimensional features encoded by the graph neural network are subjected to residual quantization to obtain the vocabulary (tokens). The vocabulary (tokens) in the vocabulary are updated using the average exponential moving method.

[0066]

[0067]

[0068] in, represents the vocabulary, represents the residual quantization operation, Represents the quantitative features obtained after the qth residual quantization, Represents the distance in the vocabulary after the residual quantization operation Recent vocabulary index. Indicates that the corresponding word embedding is taken out according to the word index, and the residual feature obtained last time is obtained ( ) minus the distance Recent word embeddings Get the quantized features after the qth residual .

[0069]

[0070] Finally, the features of the re-described building components are obtained by summing the residual depth. .

[0071] Step 3.3: The decoder consists of a 5-layer residual network, which uses the re-described building component features obtained in step 3.2 , which is then fed into the decoder. Each residual network layer uses residual connections and convolution operations to predict the true characteristics of the building components. The decoder's prediction of the true characteristics of the building components is based on the output value range and accuracy of the predicted value within the BIM modeling space and spatial resolution defined in step 2.

[0072] Step 4: The BIM-related building component information extracted in step 1 is mapped to the vocabulary obtained in step 3 to obtain a description sequence of the building components. The (vocabulary) tokens in the description sequence of each building component are arranged in a specific order: for different building components, the order of column -> beam -> wall -> door -> window is adopted. For the same component, the priority is arranged from small to large based on the three-dimensional Cartesian coordinate system with ZYX. The start word (SOS) and end word (EOS) are added to the first and last positions of the sequence respectively. This is used as the training corpus for the pre-training model.

[0073] Step 5: Use the Transformer-Only Decoder model with a 3-layers-4-heads attention mechanism as the pre-trained model. Its input is: the building component descriptions obtained by comparing the vocabulary obtained in step 3.2 and sorting them according to the rules described in step 4 to obtain the building description sequence (self-supervised learning method - both the input corpus and the real building description sequence). Based on the prior distribution of the input description sequence, the content of the subsequent building component description sequence is predicted, and the complete predicted building description sequence is output. Specifically:

[0074] Training: The building description sequence is fed into the pre-trained model to learn the layout of building components. The model parameters are updated based on the cross-loss entropy between the predicted output building sequence and the real building description sequence as the error backpropagation function;

[0075] The loss function of the pre-trained model is as follows:

[0076]

[0077]

[0078] in: represents the total number of building components, Indicates the number of words in the vocabulary, represents the residual depth, represents the probability distribution of the predicted words in the vocabulary, Indicates the The building components in Layer quantization depth prediction vocabulary , Represents the smoothed probability distribution tensor obtained by smoothing the one-hot tensor of the vocabulary in the real sequence. Indicates smooth operation, Indicates the first The building components in Layer quantization depth vocabulary index.

[0079] Generate: Set two task generation modes: incomplete BIM completion and random generation.

[0080] Among them, the incomplete BIM completion generation: the real features of some building components are used as input, and after passing through the encoder and quantization modules mentioned above, partial building description sequence vocabulary is obtained, and then input into the pre-trained model to obtain a complete building description sequence.

[0081] Random Generation: Given a starting word (SOS) as input to the pre-trained model, a randomly generated sequence of complete building descriptions is obtained.

[0082] Step 6: The building description sequence predicted by the pre-trained model is summed according to the quantized depth and then input into the decoder to decode the real features of the building components. BIM modeling is completed based on the decoded real features. The specific BIM modeling process includes: creating a building project, defining building units and model views, modeling building components, and dividing floor information.

[0083] As a preferred solution, the step 6 specifically includes:

[0084] Step 6.1: Modeling of building components:

[0085] The modeling method for building components includes five types: columns, beams, walls, doors, and windows. The modeling of each component begins by creating a corresponding ifc_entity (solid) and specifying its category. Subsequently, the corresponding IFC statements are generated to define the component's geometric and spatial characteristics, combining the decoded size, position, and orientation information. Specifically, the modeling of columns, beams, and walls primarily involves IfcRectangleProfileDef (rectangular profile), IfcExtrudedAreaSolid (extruded body), and IfcAxis2Placement3D (three-dimensional coordinate) statements. Their dimensional parameters (such as XDim, YDim, Depth, OuterCurve) are directly derived from the decoded component size values. Door and window modeling additionally includes IfcArbitraryClosedProfileDef (polygonal profile definition), IfcDoorPanelProperties (door panel properties), IfcDoorLiningProperties (door frame properties), IfcWindowPanelProperties (window panel properties), and IfcWindowLiningProperties (window frame properties). These sizing parameters are also derived from decoded values, while style parameters (such as Operation_type) are set to a specific style based on the component length (for doors: SINGLE_SWING_LEFT, DOUBLE_DOOR_SLIDING; for windows: DOUBLE_PANEL_VERTICAL, TRIPLE_PANEL_VERTICAL, SINGLE_PANEL). The orientation of door and window components is checked, and a preliminary match is generated to generate a list of potential wall pairs. The Euclidean distance between this list of potential wall pairs and the door is then calculated. , the expression is as follows:

[0086]

[0087] in Indicates the location of door and window components. Indicates the position of the wall, and calculates the minimum Euclidean distance between doors and windows and each wall , divide the walls to which the door and window components belong, then create an ifc_entity based on their size and position and specify the category as IfcOpeningElement (opening space), and create an IfcRelVoidsElement (space belonging) statement to describe the association between the door and the wall to which it belongs.

[0088] For the location and orientation properties of all the above building components, the Location and RefDirection parameter values ​​in the IfcAxis2Placement3D statement are generated based on the decoded coordinates and orientation properties.

[0089] Step 6.2: Building floor information division, such as Figure 4 As shown:

[0090] After completing component modeling in step 6.1, the Z-axis positions of the columns and walls are divided into intervals and counted. Intervals with a data point count exceeding a threshold (9% of the total number of building components) are selected. The number of intervals and the minimum value within each interval are used to divide the number of floors and heights. This generates the ObjectPlacement (location), Representation (property representation), and LongName (floor naming) parameter values ​​in the IfcBuildingStorey (floor definition) statement. Finally, the IfcRelContainedInSpatialStructure (space belonging) statement is generated to associate the building components with the corresponding floors. Through these steps, a complete BIM modeling process is completed, from building component modeling to floor information division.

[0091] Example 2:

[0092] This embodiment introduces a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it implements a pre-trained artificial intelligence framework for automatically generating a BIM (Building Information Model) as described in any one of the embodiments 1.

[0093] Example 3:

[0094] This embodiment introduces a computer device, including:

[0095] Memory, used to store instructions.

[0096] A processor is used to execute the instructions so that the computer device performs an operation of automatically generating a pre-trained artificial intelligence framework for BIM (Building Information Model) as described in any one of Example 1.

[0097] Example 4:

[0098] This embodiment introduces the specific implementation process of the method of the present invention as follows: Figure 1 As shown:

[0099] Clean up the acquired BIM (remove components that are not considered and will interfere with information extraction, such as curtain walls, non-standard walls, circular columns, etc.).

[0100] First, based on the .NET development program, the bounding box overlap in the BIM component is determined to obtain spatial features. Then, the corresponding IFC file of the BIM is exported. The real features (category, orientation, size, location) of columns, beams, walls, doors, and windows are extracted through the written information extraction script program and the corresponding CSV file is generated. Taking the information extraction of a BIM building component as an example, the information extraction results are as follows: Figure 2 shown.

[0101] Pseudo code for extracting true features of building components (Pyhton):

[0102] Step 1: Initialize the path and file name

[0103] Step 1.1: Set the IFC file path and name.

[0104] Step 1.2: Set the CSV file path and name.

[0105] Step 2: Define the helper function

[0106] Step 2.1: Define the direction vector normalization function: `RegulateAndRespond_direct`.

[0107] Step 2.2: Define the size update function:

[0108] Step 2.2.1: Define `update_dimensions_with_outercurve`.

[0109] Step 2.2.2: Define `update_dimensions`.

[0110] Step 2.3: Define the geometry representation processing function: `calculate_dimensions`.

[0111] Step 3: Processing building components (columns, beams, walls, doors, windows)

[0112] Step 3.1: For each component type:

[0113] Step 3.1.1: Initialize the relevant list (name, direction, position, size, category).

[0114] Step 3.1.2: Get a list of components of this type.

[0115] Step 3.1.3: Iterate over each component:

[0116] Step 3.1.3.1: Calculate the size and position based on the geometry representation type (Clipping, MappedRepresentation, Brep, etc.).

[0117] Step 3.1.3.2: Call the helper function to process the direction vector and size.

[0118] Step 3.1.3.3: Add the result to the corresponding list.

[0119] Step 4: Recombining Features

[0120] Step 4.1: Combine the features of all the building blocks into a single list: `building_list`.

[0121] Step 4.2: Sort by x, y, z coordinates: `building_list`.

[0122] Step 4.3: Move the coordinates back to the origin.

[0123] Step 5: Generate CSV file

[0124] Step 5.1: Set the CSV file header.

[0125] Step 5.2: Open the CSV file and write the header.

[0126] Step 5.3: Iterate over `building_list` and write the characteristics of each building into a CSV file.

[0127] Pseudo code for extracting spatial features of building components (C#):

[0128] Step 1: Initialize variables

[0129] Step 1.1: Define the filter: OrFilter.

[0130] Step 1.2: Filter elements: selectedElements.

[0131] Step 1.3: Initialize the list: adjacencyList.

[0132] Step 2: Iterate over selectedElements

[0133] Step 2.1: For each element:

[0134] Step 2.1.1: Get the bounding box: box1.

[0135] Step 2.1.2: Expand the bounding box: Box1 (call the method ExpandBoundingBox).

[0136] Step 2.1.3: Loop through selectedElements again:

[0137] Step 2.1.3.1: If the current element is not the same element:

[0138] Step 2.1.3.1.1: Get the bounding box: box2.

[0139] Step 2.1.3.1.2: Expand the bounding box: Box2 (call the method ExpandBoundingBox).

[0140] Step 2.1.3.1.3: Check if the bounding boxes overlap (call method DoBoundingBoxesOverlap):

[0141] Step 2.1.3.1.3.1: If the bounding boxes overlap:

[0142] Step 2.1.3.1.3.1.1: Calculate the overlapping volume: volume. If the volume meets the conditions: add the result to adjacencyList.

[0143] Step 3: Process the results

[0144] Step 3.1: Sort and deduplicate adjacencyList (call method IdsSort_with_Deduplication).

[0145] Step 3.2: Write the results to a CSV file (call the WriteToCSV method).

[0146] Secondly, the extracted building component information is input into the encoding-decoding module to train the vocabulary, and the final vocabulary size is 67*128.

[0147] Map the building component information to the vocabulary (tokens) to complete the learning of building component layout.

[0148] First, according to the quantization depth Converted into a building description sequence, each building component is described by three words. In this sequence, components are arranged in the order of column -> beam -> wall -> door -> window. Components are arranged in ascending order of priority using the 3D Cartesian coordinate system, ZYX. SOS (start token) and EOS (end token) are added to the first and last positions in the sequence, respectively.

[0149] Secondly, the building description sequence is used as the training corpus for the pre-training model to learn the building component layout.

[0150] BIM automated modeling of building description sequences based on pre-trained model predictions can be completed in just 10 seconds from prediction to BIM modeling.

[0151] First, the task of generating building sequences for the pre-trained model is divided into two types: 1) generating a building description sequence by randomly inputting a start symbol (SOS); 2) encoding and quantizing the incomplete (partial) building components and then splicing them with the start symbol in the above arrangement order as the output to obtain the generated building description sequence, such as Figure 5 As shown. The predicted building description sequence is quantized according to the depth The sum is then input into the decoder to obtain the true features after decoding.

[0152] Secondly, the real characteristics of the building components are obtained based on the BIM modeling script, and the corresponding IFC statement description is generated to obtain the generated IFC file, completing the BIM parametric modeling.

[0153] Taking a BIM parametric modeling as an example, the results are as follows Figure 6 shown.

[0154] Some pseudocode for BIM modeling (Python):

[0155] Step 1: Generate building components

[0156] Step 1.1: Create building component dictionaries: walls_dict, doors_dict, windows_dict, column_dict, beam_dict.

[0157] Step 1.2: Traverse each element in trans_result (including columns, beams, walls, doors, and windows):

[0158] Step 1.2.1: Get the element category class_key.

[0159] Step 1.2.2: Create the corresponding building component entity according to the class_key, set its geometric representation and position, and then add it to the corresponding dictionary.

[0160] Step 2: Create floors and assign heights

[0161] Step 2.1: Create the floor height list floor_list, the floor entity dictionary floor_dic and the floor height information dictionary floor_h_dic.

[0162] Step 2.2: Divide the floor intervals according to the height information, create an entity for each floor and set the height information.

[0163] Step 3: Assign floors to buildings

[0164] Step 3.1: Declare building as a building entity.

[0165] Step 3.2: Add all floor entities to the building.

[0166] Step 4: Traverse the transformed results and assign entities to floors

[0167] Step 4.1: Declare trans_result as the transformed result list.

[0168] Step 4.2: Traverse each element in trans_result:

[0169] Step 4.2.1: Generate the corresponding IFC entity according to the element category and set the geometric representation and position.

[0170] Step 4.2.2: Assign entities to corresponding floors based on height information.

[0171] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0172] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0173] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0174] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0175] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A pre-trained artificial intelligence method for automatically generating BIM, characterized by: Specifically include: Cleaning the BIM model and extracting relevant building component information, including the real characteristics and spatial characteristics of the building components; Discretize the real features of building components to obtain the discretized real features; Graph data is constructed based on the discretized real features and the spatial features of building components. The graph data is input into the sequentially connected graph neural network, residual quantization module, and decoder, which outputs a vocabulary of building component descriptions and the trained graph neural network, residual quantization module, and decoder. Construct a description sequence of building components based on relevant building component information and a building component description vocabulary; The pre-trained model is trained using the description sequence of the building components to obtain a trained pre-trained model; Obtain information about the building components to be generated, obtain a vocabulary sequence of the building components to be generated through the trained graph neural network and residual quantization module, input the vocabulary sequence of the building components to be generated into the trained pre-trained model, and output a description sequence of the building components to be generated; The description sequence of the building component to be generated is quantized and deeply summed, and then input into the trained decoder to output the real features of the building component to be generated; Generate BIM modeling based on the real characteristics of the building components to be generated.

2. The pre-training artificial intelligence method for automatically generating BIM according to claim 1, characterized in that: The real characteristics of the building components include: the type, orientation, size and position of columns, beams, walls, doors and windows; the spatial characteristics of the building components include: the adjacent relationship between columns, beams, walls, doors and windows.

3. The pre-training artificial intelligence method for automatically generating BIM according to claim 1, characterized in that: The discretization operation specifically includes: using the minimum difference in the size of the building components as the resolution, adjusting the position of the building components to a positive value interval in the three-dimensional coordinate system, and constructing the real features of the discretized building components based on the resolution and the adjusted positions.

4. The pre-training artificial intelligence method for automatically generating BIM according to claim 1, characterized in that: The graph neural network includes three layers of SAGEConv convolutional layers connected in sequence; the quantization depth in the residual quantization module is set to 3; the decoder includes five layers of residual networks connected in sequence.

5. The pre-training artificial intelligence method for automatically generating BIM according to claim 1, characterized in that: The step of constructing a description sequence of building components based on the relevant building component information and the building component description vocabulary specifically includes: Obtain corresponding vocabulary from the building component description vocabulary table according to relevant building component information; Arrange the words in the order of column, beam, wall, door, and window. Arrange the words of the same type in ascending order of the coordinate values ​​of the three-dimensional Cartesian coordinate system Z, Y, and X axes. Then, the beginning word and the ending word are added to the first and last position of the sequence respectively to obtain the description sequence of the building components.

6. The pre-training artificial intelligence method for automatically generating BIM according to claim 1, characterized in that: The pre-training model adopts the Transformer-Only Decoder model with a 3layers-4heads attention mechanism.

7. The pre-training artificial intelligence method for automatically generating BIM according to claim 1, characterized in that: The method of obtaining information of the building components to be generated and obtaining a vocabulary sequence of the building components to be generated through a trained graph neural network and a residual quantization module specifically includes: The information of the building components to be generated is obtained as the real features of some building components, the real features of some building components are input into the trained graph neural network and residual quantization module, and the vocabulary sequence of the building components to be generated is output; The information of the building component to be generated is a starting word, and the starting word is used as a word sequence of the building component to be generated.

8. The pre-training artificial intelligence method for automatically generating BIM according to claim 1, characterized in that: Generating BIM modeling based on the real features of the building components to be generated specifically includes: Create IFC entities corresponding to columns, beams, walls, doors, and windows, and obtain the category, orientation, size, and position based on the real characteristics of the building components to be generated. Generate IFC statements corresponding to the IFC entities to define the geometric and spatial characteristics of the building components and complete BIM modeling; The Z-axis positions of column and wall building components are divided into intervals and counted, and the intervals with the number of data points exceeding the threshold are screened out. The number of floors and heights are divided according to the number of intervals and the minimum value in the interval to obtain the information of each floor in BIM modeling.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, it implements a pre-training artificial intelligence method for automatically generating BIM as described in any one of claims 1 to 8.

10. A computer device, characterized in that: include: a memory for storing instructions; A processor is configured to execute the instructions so that the computer device performs the operations of a pre-training artificial intelligence method for automatically generating BIM as described in any one of claims 1 to 8.

Images