BIM building scheme intelligent generation method and device based on artificial intelligence

Abstract

The invention discloses a BIM (Building Information Modeling) building scheme intelligent generation method and device based on artificial intelligence. The method comprises the following steps: firstly, acquiring initial user input parameters, and forming a parameterized constraint model with design specification and building standard data; obtaining user input sub-parameters corresponding to each room, inputting the user input sub-parameters into a spatial layout generation model, respectively generating initial layout schemes, and forming an initial layout scheme set; verifying the layout scheme set through space collision detection and a rule constraint correction strategy to obtain a verified layout scheme set; and if it is determined that the initial user input parameter corresponds to the single house type, generating a single house model based on the initial user input parameter, and rendering the single house model to obtain a rendered house model. According to the embodiment of the invention, initial user input parameters, design specifications and building standard data can be combined and input into the pre-trained spatial layout generation model to generate the initial layout scheme set, and after related verification of spatial collision detection and rule constraint correction, rapid rendering is carried out, and a visual model is output for display.

Description

Technical Field

[0001] This invention relates to the field of building design automation technology, and in particular to a method and apparatus for intelligent generation of BIM building schemes based on artificial intelligence. Background Technology

[0002] With the digital transformation of the construction industry, BIM (Building Information Modeling) technology has been widely used in the entire lifecycle management of buildings. However, traditional BIM design still mainly relies on manual modeling and layout, lacking intelligent assistance and automatic optimization capabilities. For example, patent CN115795638A provides a residential area layout optimization method based on genetic algorithms and neural networks, and patent CN107729664A proposes a BIM design method using rooms as the smallest graphic element, which has certain effects in information processing and macro-layout optimization. However, these solutions mostly focus on the overall layout of building levels or modeling of single components, failing to achieve layout scheme generation based on the fusion of unit type or building parameters with intelligent automatic layout. Moreover, existing BIM model layout automatic optimization and the generation of renderings generally rely on high-performance local rendering capabilities, or the rendered images generated by AI have poor consistency with the actual unit space, failing to achieve the generation of renderings based on three-dimensional semantic constraints. Summary of the Invention

[0003] This invention provides an intelligent generation method and apparatus for BIM building schemes based on artificial intelligence, aiming to solve the problem that in the existing technology, BIM design still mainly relies on manual modeling and layout, and cannot generate layout schemes by integrating with intelligent automatic layout based on unit type or building parameters.

[0004] In a first aspect, embodiments of the present invention provide an intelligent generation method for BIM building schemes based on artificial intelligence, comprising: In response to the intelligent generation command of BIM building scheme sent by the user terminal, the initial user input parameters corresponding to the intelligent generation command of BIM building scheme are obtained and combined with the preset design specifications and building standard data to form a parametric constraint model; wherein, the initial user input parameters include at least a set of user input sub-parameters, and each set of user input sub-parameters includes at least the floor plan data, the functional requirements description text of the floor plan, and the layout parameters, and each set of user input sub-parameters corresponds to the building parameters of a building room; The user input sub-parameters corresponding to each room in the parameterized constraint model are obtained and input into the pre-trained spatial layout generation model to generate an initial layout scheme corresponding to each room and form an initial layout scheme set; wherein, the initial layout scheme includes at least the layout points of furniture type components, the layout points of decoration type components, the layout points of home appliance type components and the layout points of electromechanical type components. The initial layout scheme set is subjected to spatial collision detection and rule constraint verification through a preset spatial collision detection and rule constraint correction strategy to obtain the verified layout scheme set. If it is determined that the initial user input parameters correspond to a single-unit apartment type, then a single-unit apartment model is generated based on the verified layout scheme set; The single-unit apartment model is visualized and rendered based on a preset model rendering strategy to obtain and display the rendered apartment model.

[0005] Secondly, embodiments of the present invention also provide an intelligent generation device for BIM building schemes based on artificial intelligence, comprising: The parametric constraint model acquisition unit is used to respond to the BIM building scheme intelligent generation instruction sent by the user terminal, acquire the initial user input parameters corresponding to the BIM building scheme intelligent generation instruction, and form a parametric constraint model with the preset design specifications and building standard data; wherein, the initial user input parameters include at least a set of user input sub-parameters, and each set of user input sub-parameters includes at least the floor plan data, the functional requirements description text of the floor plan, and the layout parameters, and each set of user input sub-parameters corresponds to the building parameters of a building room; The initial layout scheme set generation unit is used to obtain the user input sub-parameters corresponding to each room in the parameterized constraint model and input them into the pre-trained spatial layout generation model to generate an initial layout scheme corresponding to each room and form an initial layout scheme set; wherein, the initial layout scheme includes at least the layout points of furniture type components, the layout points of decoration type components, the layout points of home appliance type components and the layout points of electromechanical type components. The scheme verification unit is used to perform spatial collision detection and rule constraint verification on the initial layout scheme set through a preset spatial collision detection and rule constraint correction strategy, so as to obtain the verified layout scheme set. The first model generation unit is used to generate a single-unit model based on the verified layout scheme set if it is determined that the initial user input parameters correspond to a single-unit type. The first model rendering unit is used to perform visualization model rendering on the single unit type model based on a preset model rendering strategy, and to obtain and display the rendered unit type model.

[0006] Thirdly, embodiments of the present invention also provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect above.

[0007] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, can implement the method described in the first aspect above.

[0008] This invention provides an artificial intelligence-based method and apparatus for intelligent generation of BIM building schemes. The method includes: responding to a BIM building scheme intelligent generation command sent by a user terminal; acquiring initial user input parameters corresponding to the BIM building scheme intelligent generation command; and forming a parametric constraint model with preset design specifications and building standard data; wherein the initial user input parameters include at least a set of user input sub-parameters, and each set of user input sub-parameters includes at least floor plan data, functional requirement description text of the floor plan, and layout parameters, and each set of user input sub-parameters corresponds to the building parameters of a building room; acquiring the user input sub-parameters corresponding to each room in the parametric constraint model and inputting them into a pre-trained spatial layout... The system generates an initial layout scheme for each room, forming an initial layout scheme set. This initial layout scheme includes at least the layout points for furniture, decorative components, home appliances, and electromechanical components. A preset spatial collision detection and rule constraint correction strategy is used to perform spatial collision detection and rule constraint verification on the initial layout scheme set to obtain a verified layout scheme set. If the initial user input parameters correspond to a single-unit apartment type, a single-unit apartment model is generated based on the verified layout scheme set. A preset model rendering strategy is used to render the single-unit apartment model, resulting in a rendered apartment model for display. This embodiment of the invention combines initial user input parameters, design specifications, and building standard data into a pre-trained spatial layout generation model to generate an initial layout scheme set. After passing relevant verifications through spatial collision detection and rule constraint correction, a visual model is quickly rendered and displayed. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1A schematic diagram illustrating an application scenario of the AI-based intelligent generation method for BIM building schemes provided in this embodiment of the invention; Figure 2 A flowchart illustrating the intelligent generation method for BIM building schemes based on artificial intelligence provided in an embodiment of the present invention; Figure 3 A schematic diagram of the first sub-process of the AI-based intelligent generation method for BIM building schemes provided in an embodiment of the present invention; Figure 4 A schematic diagram of the second sub-process of the AI-based intelligent generation method for BIM building schemes provided in an embodiment of the present invention; Figure 5 A schematic diagram of the spatial layout generation model in the AI-based intelligent generation method for BIM building schemes provided in this embodiment of the invention; Figure 6 A schematic diagram of the third sub-process of the AI-based intelligent generation method for BIM building schemes provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the fourth sub-process of the AI-based intelligent generation method for BIM building schemes provided in an embodiment of the present invention. Figure 8 A schematic block diagram of an AI-based intelligent generation device for BIM building schemes provided in an embodiment of the present invention; Figure 9 A schematic block diagram of a computer device provided for an embodiment of the present invention. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0013] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0014] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0015] Please also refer to Figure 1 and Figure 2 ,in Figure 1 This is a schematic diagram of a scenario illustrating the intelligent generation method for BIM building schemes based on artificial intelligence, according to an embodiment of the present invention. Figure 2 This is a flowchart illustrating the intelligent generation method for BIM building schemes based on artificial intelligence provided in an embodiment of the present invention. Figure 1 As shown, the AI-based intelligent generation method for BIM building schemes provided in this embodiment of the invention is applied to server 10, and server 10 is communicatively connected to user terminal 20. Figure 2 As shown, the method includes the following steps S110-S150.

[0016] S110. In response to the BIM building scheme intelligent generation instruction sent by the user terminal, obtain the initial user input parameters corresponding to the BIM building scheme intelligent generation instruction, and form a parametric constraint model with the preset design specifications and building standard data.

[0017] The initial user input parameters include at least one set of user input sub-parameters, and each set of user input sub-parameters includes at least floor plan data, functional requirements description text, and layout parameters. Each set of user input sub-parameters corresponds to the architectural parameters of a building room.

[0018] In this embodiment, the technical solution is described using a server as the execution entity. An AI-based BIM building scheme intelligent generation platform is deployed on the server. This platform embeds a BIM system for users to design architectural models online. After logging into the platform using user login information (such as username and password), users can input initial user input parameters through various methods on the user interface (e.g., initial text import, voice input, input via user interface input boxes). For example, if the design requirement is to generate layout schemes for all rooms in a single apartment, user input sub-parameters need to be set for each room in that apartment. Each sub-parameter includes at least the apartment floor plan data (including room boundaries, dimensions, and functional attributes), functional requirement descriptions, and layout parameters. After completing the initial input, clicking the virtual button for intelligently generating the building scheme on the user interface triggers the BIM building scheme intelligent generation command. Subsequently, the AI-based BIM building scheme intelligent generation platform will automatically call the preset design specifications and building standard data, which together with the initial user input parameters form a parametric constraint model. This model is used to guide the subsequent layout scheme generation and module splicing process based on AI technology, ensuring that the generated results comply with the design specifications.

[0019] S120. Obtain the user input sub-parameters corresponding to each room in the parameterized constraint model and input them into the pre-trained spatial layout generation model to generate an initial layout scheme corresponding to each room and form an initial layout scheme set.

[0020] The initial layout scheme includes at least the layout points of furniture type components, the layout points of decorative type components, the layout points of home appliance type components, and the layout points of electromechanical type components.

[0021] In this embodiment, the spatial layout generation model can specifically employ a feature hybrid mechanism combining a convolutional neural network (CNN) and a Transformer model architecture to simultaneously extract spatial geometric features and room functional semantics. Specifically, the CNN in the spatial layout generation model encodes local geometric information such as room boundaries, dimensions, and shapes, while the Transformer model models the relationships between rooms, functional semantics, and global spatial constraints. Through this hybrid structure of the spatial layout generation model, an AI-based BIM building scheme intelligent generation platform can automatically generate an initial layout scheme that conforms to spatial rules and includes at least the layout points of furniture-type components, decorative-type components, appliance-type components, and electromechanical-type components.

[0022] In one embodiment, such as Figure 3 As shown, step S120 includes: S121. For each room in the parameterized constraint model, the user input sub-parameters are obtained from the user input sub-parameters, including the floor plan data, functional requirements description text, layout parameters, and design specifications and building standards data, to form the initial input features. S122. Input the initial input features into the spatial layout generation model to generate the initial layout scheme; S123. The initial layout scheme set is formed by the initial layout schemes corresponding to all rooms in the parameterized constraint model.

[0023] In this embodiment, for each room in the parameterized constraint model, specific floor plan data, functional requirement description text, layout parameters, and design specifications and building standards data are obtained to form initial input features and input into the spatial layout generation model to generate an initial layout scheme for each room. Then, the initial layout schemes corresponding to all rooms included in the parameterized constraint model are combined to form the initial layout scheme set, thereby achieving rapid acquisition of the initial layout scheme set.

[0024] In one embodiment, such as Figure 4 As shown, step S122 includes: S1221. The structured apartment data in the initial input features is encoded through the first encoding layer in the spatial layout generation model to obtain the first encoding result; S1222. The unstructured house type data in the initial input features is encoded through the second encoding layer in the spatial layout generation model to obtain the second encoding result; S1223. Encode the design specifications, building standard data and layout parameters in the initial input features through the third coding layer in the spatial layout generation model to obtain the third coding result; S1224. The first encoding result is extracted using the convolutional neural network in the spatial layout generation model to obtain the first apartment type features; S1225. The second and third encoding results are extracted using the Transformer model in the spatial layout generation model to obtain the second apartment type features. S1226. The first apartment type feature and the second apartment type feature are fused through the cross-modal attention fusion layer in the spatial layout generation model to obtain the current fused feature; S1227. Obtain multiple grid layout allocation results corresponding to the current fusion feature through the decoder output layer in the spatial layout generation model to form the initial layout scheme.

[0025] In this embodiment, the spatial layout generation model deployed on the server includes at least a first encoding layer, a second encoding layer, a third encoding layer, a convolutional neural network, a Transformer model, a cross-modal attention fusion layer, and a decoder output layer. A schematic diagram of the specific spatial layout generation model structure is shown below. Figure 5 As shown.

[0026] For example, the structured apartment layout data in the initial input features specifically includes apartment floor plan data. The two-dimensional coordinates of the room boundaries in the apartment floor plan data are converted into a rasterized feature map. The resolution of each grid in the rasterized feature map is, for example, the side length of a grid is 0.1 meters, so a 40-square-meter square room corresponds to 20*20 grids. In an H*W*D1 rasterized feature map, H represents the total number of grids that the room is cut in the vertical direction, W represents the total number of grids that the room is cut in the horizontal direction, and D1 is equal to a preset first positive integer value (e.g., if D1=2, then the attributes included by D1 include at least wall type and lighting coefficient). The size parameters and functional attributes in the apartment floor plan data are one-hot encoded to obtain a K*D2 structured feature vector (where K = the total number of room types; if D2=3, then the attributes included include at least one-hot function code, standardized area, standardized maximum length and width, and priority weight).

[0027] The unstructured apartment type data in the initial input features specifically includes apartment type functional requirement description text. By inputting the apartment type functional requirement description text into the second encoding layer, it is sequentially segmented and converted into word vectors, and then combined into semantic vectors to obtain the second encoding result.

[0028] When the third encoding layer encodes the design specifications, building standard data, and layout parameters in the initial input features, it specifically applies quantitative and semantic constraints to the design specifications and building standard data. For example, a room area of ​​10-20㎡ is standardized to [0.25, 0.5] after quantitative constraint processing. The text "load-bearing walls cannot be demolished" is encoded as a text prompt. The layout parameters are encoded as weight vectors (e.g., master bedroom priority 0.9, living room priority 0.8). After completing the above encoding, an M*D3 constraint feature vector is output, where M equals the total number of constraints. If D3=3, the corresponding attributes include at least constraint type, text prompt, and constraint weight vector.

[0029] When the convolutional neural network in the spatial layout generation model extracts features from the first encoding result, the convolutional neural network is mainly used to mine the spatial topological features of the apartment type (such as regional connectivity, lighting distribution, and wall layout). By performing feature mining and processing on the first encoding result, a first apartment type feature is obtained to correspond to a spatial feature map.

[0030] When extracting features from the second and third encoding results using the Transformer model in the spatial layout generation model, the Transformer model is mainly used to capture semantic dependencies (such as demand associations and room priorities) and cross-modal associations. The second and third encoding results need to be concatenated into a fusion sequence and then input into the Transformer model for processing to obtain a global sequence feature that integrates semantics, functions, and constraints, which is then used as the second unit type feature.

[0031] Subsequently, the first and second apartment features are aligned, fused, and dimension-unified by the cross-modal attention fusion layer (such as a dual attention fusion mechanism) in the spatial layout generation model to obtain the current fused features.

[0032] Finally, since the current fused feature contains complete information about the type of component (such as furniture, decoration, home appliances, or electromechanical equipment) suitable for deployment in each grid, the decoder output layer segments the current fused feature into the component type probability distribution corresponding to each grid, thus obtaining an initial layout scheme for a room. It can be seen that, through the above method, intelligent generation of a specific initial layout scheme for each room in a building is achieved.

[0033] In this process, a standardized room module and component library must be pre-built on the server, defining the room as the smallest design unit. Through modular division, data models of various types of components, such as structure, decoration, and electromechanical components, are formed, providing basic data support for subsequent intelligent layout and combination generation.

[0034] S130. The initial layout scheme set is subjected to spatial collision detection and rule constraint verification through a preset spatial collision detection and rule constraint correction strategy to obtain the verified layout scheme set.

[0035] In this embodiment, after the initial layout scheme set is generated, spatial collision detection and rule constraint correction are required for the initial layout scheme of each room to obtain the verified layout scheme of each room and form the verified layout scheme set.

[0036] In one embodiment, such as Figure 6 As shown, step S130 includes: S131. For each initial layout scheme in the initial layout scheme set, the initial layout scheme is subjected to spatial collision detection and corresponding adjustment through the spatial collision detection sub-strategy in the spatial collision detection and rule constraint correction strategy to obtain the initial verified layout scheme. S132. The initial layout scheme is validated and adjusted according to the rule constraint correction sub-strategy in the spatial collision detection and rule constraint correction strategy to obtain the validated layout scheme corresponding to the initial layout scheme.

[0037] In this embodiment, taking the spatial collision detection and rule constraint correction strategy for detecting and correcting spatial collisions among components distributed in each grid of a room as an example, the initial layout scheme is first adjusted using the spatial collision detection sub-strategy within the spatial collision detection and rule constraint correction strategy to obtain the initial verified layout scheme. For example, in this room, component A occupies an area composed of multiple grids, and component B occupies an area composed of multiple grids. If the two components overlap spatially, resulting in a spatial collision, adjustments are needed. These adjustments are performed based on the set adjustment priority and the feasible spatial direction, using translation, rotation, or size switching operations to eliminate the spatial collision between the two components.

[0038] Subsequently, the rule constraint correction sub-strategy within the spatial collision detection and rule constraint correction strategy is used to perform rule constraint verification and corresponding adjustments on the initial verified layout scheme, resulting in a verified layout scheme corresponding to the initial layout scheme. For example, the minimum clear distance, safety interval, and dimensional specifications of each component in the initial verified layout scheme are verified to meet preset design standards, and the verified layout scheme of the room is obtained after the verification passes. It is evident that the above method achieves rapid spatial collision detection and rule constraint correction of the initial layout scheme.

[0039] In one embodiment, such as Figure 7 As shown, step S131, which involves performing spatial collision detection and corresponding adjustments on the initial layout scheme using the spatial collision detection sub-strategy within the spatial collision detection and rule constraint correction strategy to obtain the initial verified layout scheme, includes: S1311. Obtain the distribution area and corresponding geometric bounding box of each component included in the initial layout scheme; S1312. For the bounding box of each component, detect the collision relationship between it and the bounding boxes of adjacent components and the room boundary. S1313 If a collision relationship is detected between the bounding box of a component and the bounding box of an adjacent component or the boundary of a room, the component to be adjusted is determined according to the preset component adjustment priority, and the component to be adjusted is translated, rotated or sized according to the feasible direction of the room space to obtain the initial verified layout scheme.

[0040] In this embodiment, referring to the example above, component A and component B occupy areas composed of multiple grids in the room. If these two components overlap, resulting in a spatial collision, and if component A's adjustment priority is determined to be higher than component B's, then component A's position needs to be adjusted first. At this point, it is determined whether component A has any space for translational adjustment in each direction within the room's area; that is, whether adjacent grids correspond to empty areas without any deployed components. This determines the feasible spatial directions of component A within the room. Then, translation, rotation, or size switching operations are performed based on the feasible spatial directions of the component to be adjusted, thereby eliminating the collision between the two components. Once all components in the room have passed collision detection, the initial verified layout scheme is obtained. It can be seen that the above method achieves rapid spatial collision detection and adjustment of the initial layout scheme.

[0041] In one embodiment, step S132 includes: The rule-constrained correction sub-strategy is used to perform rule-constrained verification and corresponding adjustments on the initial verified layout scheme so that each component in the room meets the conditions of the preset design standard, and a verified layout scheme corresponding to the initial layout scheme is obtained.

[0042] In this embodiment, when specifically performing rule constraint verification and corresponding adjustments on the initial verified layout scheme using the rule constraint correction sub-strategy, the above example is still followed. If component A and component B in the room have eliminated their collision relationship (for example, after component A is shifted 10cm to the left), it is also necessary to check whether component A and component B meet the conditions of the preset design standard. For example, the conditions of the preset design standard limit the minimum clear distance and safety interval between components and between components and the room boundary, as well as the size specifications corresponding to each type of component. After all components in the room meet the conditions of the preset design standard, it means that it has passed the rule constraint verification, and the current layout scheme can be used as the verified layout scheme corresponding to the initial layout scheme. Of course, if component A and component B do not meet the conditions of the preset design standard, for example, if the horizontal distance between them does not exceed the minimum clear distance, component A can be further shifted laterally to make the horizontal distance between it and component B exceed the minimum clear distance. It can be seen that, through the above method, rapid rule constraint verification and adjustment of the initial verified layout scheme is achieved.

[0043] It is important to note that when the building frame or unit layout parameters corresponding to a room change, the system automatically recalculates the position and orientation of the affected components based on their attachment relationships, positioning benchmarks, and directional constraints. Specific adjustments include: 1) updating the component's position in the new frame coordinates based on the offset, alignment, or proportional relationship between the component and the wall or axis; 2) updating the component's orientation according to the binding rules between the component and the room's walls, functional spaces, or reference directions; and 3) after completing the position and orientation updates, performing local spatial verification and automatically fine-tuning components that still have conflicts. Through these methods, a dynamic layout optimization effect is achieved where components automatically adjust as the frame changes.

[0044] S140. If it is determined that the initial user input parameters correspond to a single-unit apartment type, then a single-unit apartment model is generated based on the verified layout scheme set.

[0045] In this embodiment, if the server determines that the initial user input parameters in step S110 are input parameters for automatically generating a BIM building scheme for a set of single-unit apartment models, then after generating a verified layout scheme for each user input sub-parameter in the initial user input parameters, a single-unit apartment model can be formed by combining it with the initial BIM model corresponding to each room.

[0046] S150. Based on the preset model rendering strategy, the single-unit apartment model is visualized and rendered to obtain the rendered apartment model and display it.

[0047] In this embodiment, a web-based 3D engine is deployed on the server to implement the model. This model serves as the processing model corresponding to the rendering strategy, performing visualization rendering on the individual apartment model to obtain a rendered apartment model with both visual display and AI-generated renderings. Specifically, the generated individual apartment model undergoes wireframe and semantic-level rendering, encoding information such as spatial structure, room functions, and facade materials in a multi-channel manner. Subsequently, the text-based image model within the web-based 3D engine is invoked, using the rendering result as geometric and semantic constraint input to guide the generation of a rendered apartment model that conforms to the architectural style and apartment layout. As can be seen, this rendering method, relying on plain text prompts while using a real 3D model as the generation constraint, achieves a unity of design consistency and generation realism. Furthermore, the final rendering strictly conforms to the apartment layout and spatial proportions. The web-based 3D engine also supports rapid rendering of multiple styles and scheme comparison.

[0048] In one embodiment, the method further includes the following after step S130: If it is determined that the initial user input parameters correspond to the building type, then based on the verified layout scheme set, the corresponding single unit model is generated, and based on the preset single unit splicing rules and spatial constraints, the building combination model is generated. The building combination model is visualized and rendered based on a preset model rendering strategy to obtain and display the rendered building combination model.

[0049] In this embodiment, if the server determines that the initial user input parameters in step S110 are input parameters for automatically generating BIM architectural schemes for the building combination model of the entire building, then after generating a verified layout scheme for each user input sub-parameter in the initial user input parameters, and generating individual unit models by combining the initial BIM model corresponding to each room and the combination relationship between rooms, the individual unit models are then combined based on the individual unit splicing rules and spatial constraints also included in the initial user input parameters to generate the building combination model. The visualization rendering of the building combination model can refer to the rendering process for the individual unit models, except that the entire process renders multiple individual unit models.

[0050] In this system, the geometric boundaries, connection points, and floor height parameters of each individual unit model are known. The target combination form (such as a straight line, L-shape, U-shape, etc.) between individual unit models is determined by the unit model splicing rules and spatial constraints. The connection and adjacency relationships between each individual unit model and the building model are also defined. The system uses unit model splicing rules and spatial constraints to detect whether each individual unit model and the building model are correctly spliced, whether there are overlaps or conflicts, and whether misalignments or breaks occur. Corrections are made according to preset building rules, ensuring that the automatically generated building combination model meets standards in terms of geometric relationships, logical relationships, and spatial connectivity, forming a usable and reasonable building combination scheme, and ensuring that the connection relationships between models are correct and without overlap or misalignment.

[0051] Afterwards, users can fine-tune parameters such as the number of floors, offset distance, and spacing ratio in the parametric interface corresponding to the building combination model; the server will update the 3D visualization model of the rendered building combination model in real time and link it with the apartment layout to achieve instant feedback on local adjustments.

[0052] After step S150, or after the step of performing visualization model rendering on the building combination model based on the preset model rendering strategy to obtain and display the rendered building combination model, the system can also automatically perform real-time verification of the unit type parameters, component dimensions and layout relationships of the individual unit type model or building combination model according to the pre-built standard component library and design specifications; automatically output the index data and component list that meet the specifications to ensure the feasibility and consistency of the design results, and support subsequent design review and engineering docking.

[0053] As can be seen, the implementation of this method can combine initial user input parameters, design specifications and building standard data into a pre-trained spatial layout generation model to generate an initial layout scheme set, and after passing relevant verifications such as spatial collision detection and rule constraint correction, quickly render and output a visual model for display.

[0054] Figure 8 This is a schematic block diagram of an intelligent BIM building scheme generation device based on artificial intelligence, provided in an embodiment of the present invention. Figure 8 As shown, corresponding to the above-described AI-based intelligent generation method for BIM building schemes, the present invention also provides an AI-based intelligent generation device 100 for BIM building schemes. This AI-based intelligent generation device 100 includes a unit for executing the above-described AI-based intelligent generation method for BIM building schemes. Please refer to... Figure 8The AI-based BIM building scheme intelligent generation device 100 includes: a parametric constraint model acquisition unit 110, an initial layout scheme set generation unit 120, a scheme verification unit 130, a first model generation unit 140, and a first model rendering unit 150.

[0055] The parametric constraint model acquisition unit 110 is used to respond to the BIM building scheme intelligent generation instruction sent by the user terminal, acquire the initial user input parameters corresponding to the BIM building scheme intelligent generation instruction, and form a parametric constraint model with the preset design specifications and building standard data.

[0056] The initial user input parameters include at least one set of user input sub-parameters, and each set of user input sub-parameters includes at least floor plan data, functional requirements description text, and layout parameters. Each set of user input sub-parameters corresponds to the architectural parameters of a building room.

[0057] In this embodiment, the technical solution is described using a server as the execution entity. An AI-based BIM building scheme intelligent generation platform is deployed on the server. This platform embeds a BIM system (Building Information Modeling) for users to design building model schemes online. After logging into the platform using user login information (such as username and password), users can input initial user input parameters through various methods on the user interface (e.g., initial text import, voice input, input via user interface input boxes). For example, if the design requirement is to generate layout schemes for all rooms in a single apartment, user input sub-parameters need to be set for each room in that apartment. Each sub-parameter includes at least the apartment floor plan data (including room boundaries, dimensions, and functional attributes), functional requirement descriptions, and layout parameters. After completing the initial input, clicking the virtual button for intelligently generating the building scheme on the user interface triggers the BIM building scheme intelligent generation command. Subsequently, the AI-based BIM building scheme intelligent generation platform will automatically call the preset design specifications and building standard data, which together with the initial user input parameters form a parametric constraint model. This model is used to guide the subsequent layout scheme generation and module splicing process based on AI technology, ensuring that the generated results comply with the design specifications.

[0058] The initial layout scheme set generation unit 120 is used to obtain the user input sub-parameters corresponding to each room in the parameterized constraint model and input them into the pre-trained spatial layout generation model to generate an initial layout scheme corresponding to each room and form an initial layout scheme set.

[0059] The initial layout scheme includes at least the layout points of furniture type components, the layout points of decorative type components, the layout points of home appliance type components, and the layout points of electromechanical type components.

[0060] In this embodiment, the spatial layout generation model can specifically employ a feature hybrid mechanism combining a convolutional neural network (CNN) and a Transformer model architecture to simultaneously extract spatial geometric features and room functional semantics. Specifically, the CNN in the spatial layout generation model encodes local geometric information such as room boundaries, dimensions, and shapes, while the Transformer model models the relationships between rooms, functional semantics, and global spatial constraints. Through this hybrid structure of the spatial layout generation model, an AI-based BIM building scheme intelligent generation platform can automatically generate an initial layout scheme that conforms to spatial rules and includes at least the layout points of furniture-type components, decorative-type components, appliance-type components, and electromechanical-type components.

[0061] In one embodiment, the initial layout scheme generation unit 120 is specifically used for: For each room in the parameterized constraint model, the user input sub-parameters are obtained, including the floor plan data, functional requirements description text, layout parameters, and design specifications and building standards data, to form the initial input features. The initial input features are input into the spatial layout generation model to generate the initial layout scheme; The initial layout scheme set is composed of the initial layout schemes corresponding to all rooms in the parameterized constraint model.

[0062] In this embodiment, for each room in the parameterized constraint model, specific floor plan data, functional requirement description text, layout parameters, and design specifications and building standards data are obtained to form initial input features and input into the spatial layout generation model to generate an initial layout scheme for each room. Then, the initial layout schemes corresponding to all rooms included in the parameterized constraint model are combined to form the initial layout scheme set, thereby achieving rapid acquisition of the initial layout scheme set.

[0063] In one embodiment, the initial layout scheme generation unit 120 is further specifically used for: The structured apartment data in the initial input features are encoded through the first encoding layer in the spatial layout generation model to obtain the first encoding result; The unstructured apartment data in the initial input features are encoded through the second encoding layer in the spatial layout generation model to obtain the second encoding result; The design specifications, building standard data, and layout parameters in the initial input features are encoded through the third encoding layer in the spatial layout generation model to obtain the third encoding result. The first apartment type features are obtained by extracting features from the first encoding result through the convolutional neural network in the spatial layout generation model. The second apartment type features are obtained by extracting features from the second encoding result and the third encoding result using the Transformer model in the spatial layout generation model. The first apartment feature and the second apartment feature are fused by the cross-modal attention fusion layer in the spatial layout generation model to obtain the current fused feature; The decoder output layer in the spatial layout generation model obtains multiple grid layout allocation results corresponding to the current fusion feature to form the initial layout scheme.

[0064] In this embodiment, the spatial layout generation model deployed on the server includes at least a first encoding layer, a second encoding layer, a third encoding layer, a convolutional neural network, a Transformer model, a cross-modal attention fusion layer, and a decoder output layer. A schematic diagram of the specific spatial layout generation model structure is shown below. Figure 5 As shown.

[0065] For example, the structured apartment layout data in the initial input features specifically includes apartment floor plan data. The two-dimensional coordinates of the room boundaries in the apartment floor plan data are converted into a rasterized feature map. The resolution of each grid in the rasterized feature map is, for example, the side length of a grid is 0.1 meters, so a 40-square-meter square room corresponds to 20*20 grids. In an H*W*D1 rasterized feature map, H represents the total number of grids that the room is cut in the vertical direction, W represents the total number of grids that the room is cut in the horizontal direction, and D1 is equal to a preset first positive integer value (e.g., if D1=2, then the attributes included by D1 include at least wall type and lighting coefficient). The size parameters and functional attributes in the apartment floor plan data are one-hot encoded to obtain a K*D2 structured feature vector (where K = the total number of room types; if D2=3, then the attributes included include at least one-hot function code, standardized area, standardized maximum length and width, and priority weight).

[0066] The unstructured apartment type data in the initial input features specifically includes apartment type functional requirement description text. By inputting the apartment type functional requirement description text into the second encoding layer, it is sequentially segmented and converted into word vectors, and then combined into semantic vectors to obtain the second encoding result.

[0067] When the third encoding layer encodes the design specifications, building standard data, and layout parameters in the initial input features, it specifically applies quantitative and semantic constraints to the design specifications and building standard data. For example, a room area of ​​10-20㎡ is standardized to [0.25, 0.5] after quantitative constraint processing. The text "load-bearing walls cannot be demolished" is encoded as a text prompt. The layout parameters are encoded as weight vectors (e.g., master bedroom priority 0.9, living room priority 0.8). After completing the above encoding, an M*D3 constraint feature vector is output, where M equals the total number of constraints. If D3=3, the corresponding attributes include at least constraint type, text prompt, and constraint weight vector.

[0068] When the convolutional neural network in the spatial layout generation model extracts features from the first encoding result, the convolutional neural network is mainly used to mine the spatial topological features of the apartment type (such as regional connectivity, lighting distribution, and wall layout). By performing feature mining and processing on the first encoding result, a first apartment type feature is obtained to correspond to a spatial feature map.

[0069] When extracting features from the second and third encoding results using the Transformer model in the spatial layout generation model, the Transformer model is mainly used to capture semantic dependencies (such as demand associations and room priorities) and cross-modal associations. The second and third encoding results need to be concatenated into a fusion sequence and then input into the Transformer model for processing to obtain a global sequence feature that integrates semantics, functions, and constraints, which is then used as the second unit type feature.

[0070] Subsequently, the first and second apartment features are aligned, fused, and dimension-unified by the cross-modal attention fusion layer (such as a dual attention fusion mechanism) in the spatial layout generation model to obtain the current fused features.

[0071] Finally, since the current fused feature contains complete information about the type of component (such as furniture, decoration, home appliances, or electromechanical equipment) suitable for deployment in each grid, the decoder output layer segments the current fused feature into the component type probability distribution corresponding to each grid, thus obtaining an initial layout scheme for a room. It can be seen that, through the above method, intelligent generation of a specific initial layout scheme for each room in a building is achieved.

[0072] In this process, a standardized room module and component library must be pre-built on the server, defining the room as the smallest design unit. Through modular division, data models of various types of components, such as structure, decoration, and electromechanical components, are formed, providing basic data support for subsequent intelligent layout and combination generation.

[0073] The scheme verification unit 130 is used to perform spatial collision detection and rule constraint verification on the initial layout scheme set through a preset spatial collision detection and rule constraint correction strategy, so as to obtain the verified layout scheme set.

[0074] In this embodiment, after the initial layout scheme set is generated, spatial collision detection and rule constraint correction are required for the initial layout scheme of each room to obtain the verified layout scheme of each room and form the verified layout scheme set.

[0075] In one embodiment, the scheme verification unit 130 is specifically used for: For each initial layout scheme in the initial layout scheme set, spatial collision detection and corresponding adjustment are performed on the initial layout scheme through the spatial collision detection and rule constraint correction strategy to obtain the initial verified layout scheme. The initial layout scheme is validated and adjusted according to the rule constraint correction sub-strategy in the spatial collision detection and rule constraint correction strategy to obtain the validated layout scheme corresponding to the initial layout scheme.

[0076] In this embodiment, taking the spatial collision detection and rule constraint correction strategy for detecting and correcting spatial collisions among components distributed in each grid of a room as an example, the initial layout scheme is first adjusted using the spatial collision detection sub-strategy within the spatial collision detection and rule constraint correction strategy to obtain the initial verified layout scheme. For example, in this room, component A occupies an area composed of multiple grids, and component B occupies an area composed of multiple grids. If the two components overlap spatially, resulting in a spatial collision, adjustments are needed. These adjustments are performed based on the set adjustment priority and the feasible spatial direction, using translation, rotation, or size switching operations to eliminate the spatial collision between the two components.

[0077] Subsequently, the rule constraint correction sub-strategy within the spatial collision detection and rule constraint correction strategy is used to perform rule constraint verification and corresponding adjustments on the initial verified layout scheme, resulting in a verified layout scheme corresponding to the initial layout scheme. For example, the minimum clear distance, safety interval, and dimensional specifications of each component in the initial verified layout scheme are verified to meet preset design standards, and the verified layout scheme of the room is obtained after the verification passes. It is evident that the above method achieves rapid spatial collision detection and rule constraint correction of the initial layout scheme.

[0078] In one embodiment, the step of performing spatial collision detection and corresponding adjustments on the initial layout scheme through the spatial collision detection sub-strategy in the spatial collision detection and rule constraint correction strategy to obtain the initial verified layout scheme includes: Obtain the distribution area and corresponding geometric bounding box of each component included in the initial layout scheme; For each component's bounding box, collision relationships with the bounding boxes of adjacent components and the room boundary are detected. If a collision is detected between the bounding box of a component and the bounding box of an adjacent component or the boundary of a room, the component to be adjusted is determined according to the preset component adjustment priority. The component to be adjusted is then translated, rotated or sized according to the feasible direction of the room space to obtain the initial verified layout scheme.

[0079] In this embodiment, referring to the example above, component A and component B occupy areas composed of multiple grids in the room. If these two components overlap, resulting in a spatial collision, and if component A's adjustment priority is determined to be higher than component B's, then component A's position needs to be adjusted first. At this point, it is determined whether component A has any space for translational adjustment in each direction within the room's area; that is, whether adjacent grids correspond to empty areas without any deployed components. This determines the feasible spatial directions of component A within the room. Then, translation, rotation, or size switching operations are performed based on the feasible spatial directions of the component to be adjusted, thereby eliminating the collision between the two components. Once all components in the room have passed collision detection, the initial verified layout scheme is obtained. It can be seen that the above method achieves rapid spatial collision detection and adjustment of the initial layout scheme.

[0080] In one embodiment, the step of performing rule constraint correction sub-strategy on the initial verified layout scheme and making corresponding adjustments through the rule constraint correction sub-strategy in the spatial collision detection and rule constraint correction strategy to obtain a verified layout scheme corresponding to the initial layout scheme includes: The rule-constrained correction sub-strategy is used to perform rule-constrained verification and corresponding adjustments on the initial verified layout scheme so that each component in the room meets the conditions of the preset design standard, and a verified layout scheme corresponding to the initial layout scheme is obtained.

[0081] In this embodiment, when specifically performing rule constraint verification and corresponding adjustments on the initial verified layout scheme using the rule constraint correction sub-strategy, the above example is still followed. If component A and component B in the room have eliminated their collision relationship (for example, after component A is shifted 10cm to the left), it is also necessary to check whether component A and component B meet the conditions of the preset design standard. For example, the conditions of the preset design standard limit the minimum clear distance and safety interval between components and between components and the room boundary, as well as the size specifications corresponding to each type of component. After all components in the room meet the conditions of the preset design standard, it means that it has passed the rule constraint verification, and the current layout scheme can be used as the verified layout scheme corresponding to the initial layout scheme. Of course, if component A and component B do not meet the conditions of the preset design standard, for example, if the horizontal distance between them does not exceed the minimum clear distance, component A can be further shifted laterally to make the horizontal distance between it and component B exceed the minimum clear distance. It can be seen that, through the above method, rapid rule constraint verification and adjustment of the initial verified layout scheme is achieved.

[0082] It is important to note that when the building frame or unit layout parameters corresponding to a room change, the system automatically recalculates the position and orientation of the affected components based on their attachment relationships, positioning benchmarks, and directional constraints. Specific adjustments include: 1) updating the component's position in the new frame coordinates based on the offset, alignment, or proportional relationship between the component and the wall or axis; 2) updating the component's orientation according to the binding rules between the component and the room's walls, functional spaces, or reference directions; and 3) after completing the position and orientation updates, performing local spatial verification and automatically fine-tuning components that still have conflicts. Through these methods, a dynamic layout optimization effect is achieved where components automatically adjust as the frame changes.

[0083] The first model generation unit 140 is used to generate a single-unit model based on the verified layout scheme set if it is determined that the initial user input parameters correspond to a single-unit type.

[0084] In this embodiment, if it is determined in the server that the initial user input parameters in the parameterized constraint model acquisition unit 110 are input parameters for automatically generating BIM building schemes for a set of single-unit apartment models, then after generating a verified layout scheme for each user input sub-parameter in the initial user input parameters, a single-unit apartment model can be formed by combining it with the initial BIM model corresponding to each room.

[0085] The first model rendering unit 150 is used to perform visualization model rendering on the single unit model based on a preset model rendering strategy, and to obtain and display the rendered unit model.

[0086] In this embodiment, a web-based 3D engine is deployed on the server to implement the model. This model serves as the processing model corresponding to the rendering strategy, performing visualization rendering on the individual apartment model to obtain a rendered apartment model with both visual display and AI-generated renderings. Specifically, the generated individual apartment model undergoes wireframe and semantic-level rendering, encoding information such as spatial structure, room functions, and facade materials in a multi-channel manner. Subsequently, the text-based image model within the web-based 3D engine is invoked, using the rendering result as geometric and semantic constraint input to guide the generation of a rendered apartment model that conforms to the architectural style and apartment layout. As can be seen, this rendering method, relying on plain text prompts while using a real 3D model as the generation constraint, achieves a unity of design consistency and generation realism. Furthermore, the final rendering strictly conforms to the apartment layout and spatial proportions. The web-based 3D engine also supports rapid rendering of multiple styles and scheme comparison.

[0087] In one embodiment, the AI-based BIM building scheme intelligent generation device 100 further includes: The second model generation unit is used to generate corresponding single-unit models based on the verified layout scheme set and generate building combination models based on preset single-unit splicing rules and spatial constraints if it is determined that the initial user input parameters correspond to building types. The second model rendering unit is used to perform visualization model rendering on the building combination model based on a preset model rendering strategy, and to obtain and display the rendered building combination model.

[0088] In this embodiment, if the initial user input parameters in the parameterized constraint model acquisition unit 110 are determined in the server to be input parameters for automatically generating BIM architectural schemes for the building combination model of the entire building, then after generating a verified layout scheme for each user input sub-parameter in the initial user input parameters, and generating individual unit models by combining the initial BIM model corresponding to each room and the combination relationship between rooms, the individual unit models are then combined based on the individual unit splicing rules and spatial constraints also included in the initial user input parameters to generate the building combination model. The visualization rendering of the building combination model can refer to the rendering process of the individual unit models, except that the entire process renders multiple individual unit models.

[0089] In this system, the geometric boundaries, connection points, and floor height parameters of each individual unit model are known. The target combination form (such as a straight line, L-shape, U-shape, etc.) between individual unit models is determined by the unit model splicing rules and spatial constraints. The connection and adjacency relationships between each individual unit model and the building model are also defined. The system uses unit model splicing rules and spatial constraints to detect whether each individual unit model and the building model are correctly spliced, whether there are overlaps or conflicts, and whether misalignments or breaks occur. Corrections are made according to preset building rules, ensuring that the automatically generated building combination model meets standards in terms of geometric relationships, logical relationships, and spatial connectivity, forming a usable and reasonable building combination scheme, and ensuring that the connection relationships between models are correct and without overlap or misalignment.

[0090] Afterwards, users can fine-tune parameters such as the number of floors, offset distance, and spacing ratio in the parametric interface corresponding to the building combination model; the server will update the 3D visualization model of the rendered building combination model in real time and link it with the apartment layout to achieve instant feedback on local adjustments.

[0091] After the single-unit model is visualized and rendered based on the preset model rendering strategy, and the rendered unit model is displayed, or after the building combination model is visualized and rendered based on the preset model rendering strategy, and the rendered building combination model is displayed, the unit parameters, component dimensions and layout relationships of the single-unit model or building combination model can be automatically verified in real time according to the pre-built standard component library and design specifications; the indicator data and component list that meet the specifications are automatically output to ensure the feasibility and consistency of the design results, and to support subsequent design review and engineering docking.

[0092] As can be seen, the implementation of this device can combine initial user input parameters, design specifications and building standard data into a pre-trained spatial layout generation model to generate an initial layout scheme set, and after passing relevant verifications such as spatial collision detection and rule constraint correction, quickly render and output a visual model for display.

[0093] The aforementioned modular building component library's intelligent recommendation device can be implemented as a computer program, which can, for example... Figure 9 It runs on the computer device shown.

[0094] Please see Figure 9 , Figure 9 This is a schematic block diagram of a computer device provided in an embodiment of the present invention. This computer device integrates any of the AI-based intelligent generation devices for BIM building schemes provided in this embodiment of the present invention.

[0095] See Figure 9The computer device 400 includes a processor 402, a memory, and a network interface 405 connected via a system bus 401. The memory may include a storage medium 403 and internal memory 404.

[0096] The storage medium 403 may store an operating system 4031 and a computer program 4032. The computer program 4032 includes program instructions that, when executed, cause the processor 402 to perform an intelligent generation method for BIM building schemes based on artificial intelligence.

[0097] The processor 402 provides computing and control capabilities to support the operation of the entire computer device.

[0098] The internal memory 404 provides an environment for the computer program 4032 in the storage medium 403 to run. When the computer program 4032 is executed by the processor 402, the processor 402 can execute the above-mentioned intelligent generation method of BIM building scheme based on artificial intelligence.

[0099] This network interface 405 is used for network communication with other devices. Those skilled in the art will understand that... Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the computer device to which the present invention is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0100] The processor 402 is used to run the computer program 4032 stored in the memory to implement the above-mentioned intelligent generation method of BIM building scheme based on artificial intelligence.

[0101] It should be understood that, in this embodiment of the invention, the processor 402 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0102] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0103] Therefore, the present invention also provides a computer-readable storage medium. This computer-readable storage medium stores a computer program, wherein the computer program includes program instructions. When executed by a processor, the program instructions cause the processor to perform the aforementioned artificial intelligence-based intelligent generation method for BIM building schemes.

[0104] The storage medium can be any computer-readable storage medium that can store program code, such as a USB flash drive, external hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0105] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0106] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0107] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0108] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0109] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for intelligently generating BIM building schemes based on artificial intelligence, characterized in that, include: In response to the intelligent generation command of BIM building scheme sent by the user terminal, the initial user input parameters corresponding to the intelligent generation command of BIM building scheme are obtained and combined with the preset design specifications and building standard data to form a parametric constraint model; wherein, the initial user input parameters include at least a set of user input sub-parameters, and each set of user input sub-parameters includes at least the floor plan data, the functional requirements description text of the floor plan, and the layout parameters, and each set of user input sub-parameters corresponds to the building parameters of a building room; The user input sub-parameters corresponding to each room in the parameterized constraint model are obtained and input into the pre-trained spatial layout generation model to generate an initial layout scheme corresponding to each room and form an initial layout scheme set; wherein, the initial layout scheme includes at least the layout points of furniture type components, the layout points of decoration type components, the layout points of home appliance type components and the layout points of electromechanical type components. The initial layout scheme set is subjected to spatial collision detection and rule constraint verification through a preset spatial collision detection and rule constraint correction strategy to obtain the verified layout scheme set. If it is determined that the initial user input parameters correspond to a single-unit apartment type, then a single-unit apartment model is generated based on the verified layout scheme set; The single-unit apartment model is visualized and rendered based on a preset model rendering strategy to obtain and display the rendered apartment model.

2. The method according to claim 1, characterized in that, The process involves obtaining the user input sub-parameters corresponding to each room in the parameterized constraint model and inputting them into a pre-trained spatial layout generation model to generate an initial layout scheme corresponding to each room, forming an initial layout scheme set, including: For each room in the parameterized constraint model, the user input sub-parameters are obtained, including the floor plan data, functional requirements description text, layout parameters, and design specifications and building standards data, to form the initial input features. The initial input features are input into the spatial layout generation model to generate the initial layout scheme; The initial layout scheme set is composed of the initial layout schemes corresponding to all rooms in the parameterized constraint model.

3. The method according to claim 2, characterized in that, The step of inputting the initial input features into the spatial layout generation model to generate the initial layout scheme includes: The structured apartment data in the initial input features are encoded through the first encoding layer in the spatial layout generation model to obtain the first encoding result; The unstructured apartment data in the initial input features are encoded through the second encoding layer in the spatial layout generation model to obtain the second encoding result; The design specifications, building standard data, and layout parameters in the initial input features are encoded through the third encoding layer in the spatial layout generation model to obtain the third encoding result. The first apartment type features are obtained by extracting features from the first encoding result through the convolutional neural network in the spatial layout generation model. The second apartment type features are obtained by extracting features from the second encoding result and the third encoding result using the Transformer model in the spatial layout generation model. The first apartment feature and the second apartment feature are fused by the cross-modal attention fusion layer in the spatial layout generation model to obtain the current fused feature; The decoder output layer in the spatial layout generation model obtains multiple grid layout allocation results corresponding to the current fusion feature to form the initial layout scheme.

4. The method according to claim 1, characterized in that, The step of performing spatial collision detection and rule constraint verification on the initial layout scheme set through a preset spatial collision detection and rule constraint correction strategy to obtain a verified layout scheme set includes: For each initial layout scheme in the initial layout scheme set, spatial collision detection and corresponding adjustment are performed on the initial layout scheme through the spatial collision detection and rule constraint correction strategy to obtain the initial verified layout scheme. The initial layout scheme is validated and adjusted according to the rule constraint correction sub-strategy in the spatial collision detection and rule constraint correction strategy to obtain the validated layout scheme corresponding to the initial layout scheme.

5. The method according to claim 4, characterized in that, The step of performing spatial collision detection and corresponding adjustments on the initial layout scheme through the spatial collision detection and rule constraint correction strategy to obtain the initial verified layout scheme includes: Obtain the distribution area and corresponding geometric bounding box of each component included in the initial layout scheme; For each component's bounding box, collision relationships with the bounding boxes of adjacent components and the room boundary are detected. If a collision is detected between the bounding box of a component and the bounding box of an adjacent component or the boundary of a room, the component to be adjusted is determined according to the preset component adjustment priority. The component to be adjusted is then translated, rotated or sized according to the feasible direction of the room space to obtain the initial verified layout scheme.

6. The method according to claim 4, characterized in that, The step of performing rule constraint correction sub-strategy on the initial verified layout scheme and making corresponding adjustments through the rule constraint correction sub-strategy in the spatial collision detection and rule constraint correction strategy to obtain the verified layout scheme corresponding to the initial layout scheme includes: The rule-constrained correction sub-strategy is used to perform rule-constrained verification and corresponding adjustments on the initial verified layout scheme so that each component in the room meets the conditions of the preset design standard, and a verified layout scheme corresponding to the initial layout scheme is obtained.

7. The method according to any one of claims 1-6, characterized in that, After the step of performing spatial collision detection and rule constraint verification on the initial layout scheme set using a preset spatial collision detection and rule constraint correction strategy to obtain a verified layout scheme set, the method further includes: If it is determined that the initial user input parameters correspond to the building type, then based on the verified layout scheme set, the corresponding single unit model is generated, and based on the preset single unit splicing rules and spatial constraints, the building combination model is generated. The building combination model is visualized and rendered based on a preset model rendering strategy to obtain and display the rendered building combination model.

8. An intelligent BIM building scheme generation device based on artificial intelligence, characterized in that, include: The parametric constraint model acquisition unit is used to respond to the BIM building scheme intelligent generation instruction sent by the user terminal, acquire the initial user input parameters corresponding to the BIM building scheme intelligent generation instruction, and form a parametric constraint model with the preset design specifications and building standard data; wherein, the initial user input parameters include at least a set of user input sub-parameters, and each set of user input sub-parameters includes at least the floor plan data, the functional requirements description text of the floor plan, and the layout parameters, and each set of user input sub-parameters corresponds to the building parameters of a building room; The initial layout scheme set generation unit is used to obtain the user input sub-parameters corresponding to each room in the parameterized constraint model and input them into the pre-trained spatial layout generation model to generate an initial layout scheme corresponding to each room and form an initial layout scheme set; wherein, the initial layout scheme includes at least the layout points of furniture type components, the layout points of decoration type components, the layout points of home appliance type components and the layout points of electromechanical type components. The scheme verification unit is used to perform spatial collision detection and rule constraint verification on the initial layout scheme set through a preset spatial collision detection and rule constraint correction strategy, so as to obtain the verified layout scheme set. The first model generation unit is used to generate a single-unit model based on the verified layout scheme set if it is determined that the initial user input parameters correspond to a single-unit type. The first model rendering unit is used to perform visualization model rendering on the single unit type model based on a preset model rendering strategy, and to obtain and display the rendered unit type model.

9. A computer device, characterized in that, The computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the intelligent generation method of BIM building scheme based on artificial intelligence as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which includes program instructions that, when executed by a processor, can implement the AI-based intelligent generation method for BIM building schemes as described in any one of claims 1-7.

Images