BIM intelligent building modeling method and system based on large language model

Abstract

The invention relates to the technical field of BIM building modeling processing, and discloses a BIM intelligent building modeling method and system based on a large language model, and the system comprises a BIM building modeling object analysis module, a BIM building modeling instruction analysis module, and a BIM building modeling execution module. Performing BIM building modeling object three-dimensional model intelligent search according to a building modeling object type analysis result in combination with an intelligent recognition algorithm and different types of building modeling object three-dimensional model information stored based on big data, and realizing scientific and efficient generation of a BIM building modeling object three-dimensional structure based on a natural language; bIM building modeling control instruction intelligent matching is carried out according to the modeling demand text data of the position, shape and size relation between a building modeling object and a building modeling scene in combination with an intelligent search algorithm and standard-stored building modeling control instructions with different modeling demands; bIM building modeling control instructions are intelligently and accurately selected based on natural languages and design concepts of designers.

Description

Technical Field

[0001] The present invention relates to the technical field of BIM building modeling processing, and specifically to a BIM intelligent building modeling method and system based on a large language model. Background Art

[0002] BIM building modeling technology is an engineering management method based on three-dimensional digital technology. By integrating information throughout the building life cycle, including processes such as design, construction, and operation and maintenance, it realizes efficient collaboration and data sharing for projects; BIM building modeling technology includes core technical advantages such as visualization and information integration, simulation and optimization, and multi-disciplinary collaboration; among them, visualization and information integration means that the BIM model is based on three-dimensional graphics, integrating geometric information and non-geometric information to form a dynamic database. This feature significantly improves the communication efficiency of design intentions and provides data support for construction management; simulation and optimization means supporting complex scenario simulations (such as sunlight analysis, energy consumption assessment) to assist in scientific decision-making; automatically detecting design conflicts, optimizing material utilization and structural safety, and reducing construction rework rates; multi-disciplinary collaboration means coordinating the design of disciplines such as architecture, structure, and mechanical and electrical through a unified platform; the existing BIM building modeling process still requires manual building modeling by designers or based on architectural CAD drawings, and cannot intelligently complete the building entity modeling process based on the natural language information of designers and the image information of modeling design concepts, reducing the efficiency and quality of BIM building modeling.

[0003] The Chinese patent application with publication number CN114880877A discloses a BIM modular modeling method and a BIM modular modeling system. The modeling method completes the model process analysis of BIM through Rhino model establishment, model refinement, model expansion, process selection, optimization layout, dimension collection, loss rate calculation, and prefabrication production. However, the above technical solutions cannot intelligently complete the building entity modeling process based on the natural language information of designers and the modeling concept image information. Summary of the Invention

[0004] (I) Technical Problems to be Solved To solve the problem that the existing BIM building modeling process still requires manual building modeling by designers or based on architectural CAD drawings, and cannot intelligently complete the building entity modeling process based on the natural language information of designers and the image information of modeling design concepts, reducing the efficiency and quality of BIM building modeling, and achieve the purpose of scientifically analyzing the types of building modeling objects, accurately searching for three-dimensional model information of building modeling objects, intelligently analyzing building modeling control instruction information, and independently and efficiently executing building modeling operations.

[0005] (II) Technical Solutions The present invention is realized through the following technical solutions: A BIM intelligent building modeling method based on a large language model, the method comprising the following steps: S1. Collect characteristic text data of building modeling objects; S2. Perform BIM building modeling object type analysis and processing based on the characteristic text data of the building modeling objects and the characteristic text data of different types of building modeling objects, and generate building modeling object type analysis data; S3. Perform three-dimensional model search processing of BIM building modeling objects according to the building modeling object type analysis data and the three-dimensional model data of different types of building modeling objects, and generate three-dimensional model data of target building modeling objects; S4. Collect three-dimensional image data of building modeling objects, three-dimensional model data of building modeling scenes, three-dimensional image data of building modeling scenes, and three-dimensional image data of the positional shape and size relationship between building modeling objects and building modeling scenes; S5. Perform recognition and processing of modeling requirement text information on the spatial position and size of building modeling objects and building modeling scenes and the shape and size of building modeling objects based on the three-dimensional image data of building modeling objects, the three-dimensional image data of building modeling scenes, and the three-dimensional image data of the positional shape and size relationship between building modeling objects and building modeling scenes, and generate modeling requirement text data on the positional shape and size relationship between building modeling objects and building modeling scenes; S6. Perform BIM building modeling control instruction screening and search processing according to the modeling requirement text data on the positional shape and size relationship between building modeling objects and building modeling scenes and the building modeling control instruction data of different modeling requirements, and generate target building modeling control instruction data; S7. Construct BIM building modeling summary data and execute BIM building modeling operations.

[0006] Preferably, the operation steps for collecting the characteristic text data of building modeling objects are as follows: S11. Online collect, through a voice input module, the characteristic description text information of designers about specific BIM building modeling objects, and generate characteristic text data of building modeling objects .

[0007] Preferably, the operation steps for performing BIM building modeling object type analysis and processing based on the characteristic text data of the building modeling objects and the characteristic text data of different types of building modeling objects, and generating building modeling object type analysis data are as follows: S21. Establish a collection of characteristic text data of different types of building modeling objects , ; where represents the The characteristic text data of different types of BIM building modeling objects corresponding to the types of BIM building modeling objects, represents the maximum value of the number of types of BIM building modeling objects; the types of BIM building modeling objects include walls, floors, doors and windows, roofs, stairs, railings, beams, columns, slabs, shear walls, pile foundations, pipes, valves, and pumps; the characteristic text data of different types of BIM building modeling objects represents the standard building modeling object characteristic text information set for different types of BIM building modeling objects; S22. Use an intelligent large language model to process the characteristic text data of the building modeling objects with the set of characteristic text data of different types of BIM building modeling objects in the set of characteristic text data of different types of BIM building modeling objects to perform building modeling feature character matching, search for the characteristic text data of different types of BIM building modeling objects matching the characteristic text data of the building modeling objects to obtain the corresponding building modeling object type text information, and construct building modeling object type analysis data , and the intelligent large language model includes any one of DeepSeekV3, Alibaba Tongyi Qianwen Qwen2.5, and Tencent Hunyuan Turbo.

[0008] Preferably, according to the building modeling object type analysis data and the three-dimensional model data of different types of BIM building modeling objects, the operation steps for searching the three-dimensional model of BIM building modeling objects are as follows: S31. Establish a set of three-dimensional model data of different types of BIM building modeling objects , where represents the three-dimensional model data of different types of BIM building modeling objects corresponding to the th type of BIM building modeling object type, and the characteristic text data of different types of BIM building modeling objects represents the standard three-dimensional entity data of building modeling for different types of BIM building modeling objects; S32. Perform building modeling object type keyword matching on the building modeling object type analysis data and the three-dimensional model data of different types of BIM building modeling objects in the set of three-dimensional model data of different types of BIM building modeling objects to search for the three-dimensional model data of different types of BIM building modeling objects corresponding to the building modeling object type analysis data , and construct the three-dimensional model data of the target BIM building modeling object , and execute the generation of the three-dimensional model data of the target BIM building modeling object ​The specific steps are as follows: S321, collecting three-dimensional model data of different types of building modeling objects All the three-dimensional model data of the different types of building modeling objects are collected in the search space Initialize as a honey source ; Initialize the honey source at the same time The quantity is , the current number of iterations t and the maximum number of iterations T and the honey source Initial position , initial position The calculation formula is as follows: Three-dimensional model data sets of different types of building modeling objects Randomly generated in the search space, express The random number inside, and Respectively represent the three-dimensional model data sets of different types of building modeling objects The upper and lower bounds in the search space of ; S322, initial position Assign hired bees. At the beginning of the search, hired bees are assigned to the three-dimensional model data sets of different types of building modeling objects. The search space consists of all the three-dimensional model data of the different types of building modeling objects Initial position of the composition Around, according to the formula , in the three-dimensional model data set of different types of building modeling objects Searching the search space generates analytical data related to the building modeling object type Matching three-dimensional model data of the different types of building modeling objects New honey source location ;in Indicated in Randomly select a nectar source that is not equal to i from the nectar sources. Represents a randomly selected nectar source Three-dimensional model data sets of different types of building modeling objects The position in the search space of Represents a random number in the range [-1,1]; S323, based on the location of the new nectar source The fitness value is used to determine the retained honey source according to the greedy selection method, and retain the analysis data related to the building modeling object type. Matching three-dimensional model data of the different types of building modeling objects ; S324. The employed bees share the nectar source information, and the follower bees search for the 3D model data of different types of building modeling objects in the search space of the set according to the probability value. The follower bees search for the 3D model data of different types of building modeling objects based on the search results shared by the employed bees, and search out the 3D model data of different types of building modeling objects that match the building modeling object type analysis data. Calculate the probability that the nectar source found by the employed bees is followed according to the probability formula. The probability calculation formula is as follows: where represents the probability that the new nectar source location is found by the employed bees and followed, represents the fitness value of the new nectar source location shared by the employed bees by the follower bees; S325. The follower bees search in the same way as the employed bees, and search out the 3D model data of different types of building modeling objects that match the building modeling object type analysis data in the search space of the set according to the greedy selection method. Nectar source and keep it; S326. During the random search of the scout bees in the search space of the 3D model data set of different types of building modeling objects , if the nectar source location does not find a better nectar source after t iterations of search reaching the iteration threshold , this nectar source location will be abandoned, and the corresponding employed bee will become a scout bee. The calculation formula for the scout bee to randomly generate a new nectar source location in the search space is as follows: where represents generating a new nectar source after iterations at the location in the search space of the 3D model data set of different types of building modeling objects, represents generating a new nectar source after t iterations at the location in the search space of the 3D model data set of different types of building modeling objects; S327. When the algorithm meets the maximum number of iterations T, output the 3D model data of different types of building modeling objects that match the building modeling object type analysis data. , and generate the three-dimensional model data of the target building modeling object through data identification .

[0009] Preferably, the operation steps for collecting the three-dimensional image data of the building modeling object, the three-dimensional model data of the building modeling scene, the three-dimensional image data of the building modeling scene, and the three-dimensional image data of the position, shape, and size relationship between the building modeling object and the building modeling scene are as follows: S41. Import the three-dimensional model data of the target building modeling object into the BIM building modeling software for operation, and cooperate with the screenshot software to collect the axonometric drawing of the target building modeling object, and generate the three-dimensional image data of the building modeling object ; The BIM building modeling software includes any one of Autodesk Revit, ArchiCAD, and Bentley Systems, and the screenshot software includes any one of the screenshot tools built in Windows, Snipasteh, and Jinzhou Screenshot Software; Online obtain the three-dimensional entity information of the building modeling scene as the modeling design and assembly object of the target building modeling object through the BIM building modeling software, and generate the three-dimensional model data of the building modeling scene ; Online obtain the axonometric drawing of the building modeling scene corresponding to the three-dimensional model data of the building modeling scene in the BIM building modeling software through the screenshot software, and generate the three-dimensional image data of the building modeling scene ; Online synchronously collect the positioning dimension design concept of the designer regarding the spatial position between the target building modeling object and the building modeling scene, and the shape and size design concept of the target building modeling object through the brain-computer interface device, form the three-dimensional image information of the position dimension annotation and shape dimension annotation of the building modeling object, and generate the three-dimensional image data of the position, shape, and size relationship between the building modeling object and the building modeling scene , and the brain-computer interface device includes any one of the Neuralink brain-computer interface, the North Brain No. 1 system, and the mixed reality brain-computer interface system.

[0010] Preferably, the operation steps for identifying and processing the modeling requirement text information of the spatial position dimensions between the building modeling object and the building modeling scene and the shape dimensions of the building modeling object based on the three-dimensional image data of the building modeling object, the three-dimensional image data of the building modeling scene, and the three-dimensional image data of the position, shape, and size relationship between the building modeling object and the building modeling scene, and generating the modeling requirement text data of the position, shape, and size relationship between the building modeling object and the building modeling scene are as follows: S51. The three-dimensional image data of the building modeling object , the three-dimensional image data of the building modeling scene The three-dimensional image data of the relationship between the position, shape, and size of the building modeling object and the building modeling scene are respectively imported into the intelligent large language model. The intelligent large language model identifies and processes the spatial position and size text information of the target building modeling object and the building modeling scene, as well as the shape and size text information of the target building modeling object based on the image information, and generates the modeling requirement text data of the relationship between the position, shape, and size of the building modeling object and the building modeling scene . The modeling requirement text data of the relationship between the position, shape, and size of the building modeling object and the building modeling scene represents the positioning and size text information of the target building modeling object and the building modeling scene, and the shape and size text information of the target building modeling object

[0011] Preferably, the operation steps for screening and searching the BIM building modeling control instructions based on the modeling requirement text data of the relationship between the position, shape, and size of the building modeling object and the building modeling scene and the building modeling control instruction data with different modeling requirements are as follows S61. Establish a set of building modeling control instruction data with different modeling requirements , where represents the building modeling control instruction data with different modeling requirements corresponding to the th type of building modeling requirement feature combination type. The building modeling requirement feature combination type represents an index data type formed mainly by the data combination of the positioning and size text information of the target building modeling object and the building modeling scene, as well as the shape and size text information of the target building modeling object, and is used to search for the building modeling control instructions; the building modeling control instruction data with different modeling requirements represents the building modeling control instruction information of the BIM building modeling software set according to different building modeling requirement feature combination types S62. Use the Rabin-Karp search algorithm to match the modeling requirement text data of the relationship between the position, shape, and size of the building modeling object and the building modeling scene with the building modeling control instruction data with different modeling requirements in the set of building modeling control instruction data with different modeling requirements to search for the building modeling control instruction data with different modeling requirements corresponding to the modeling requirement text data of the relationship between the position, shape, and size of the building modeling object and the building modeling scene , and identify the target building modeling control instruction data through data . .

[0012] Preferably, the operation steps for constructing the BIM building modeling summary data and executing the BIM building modeling operation are as follows S71. Combine the three-dimensional model data of the target building modeling object and the target building modeling control instruction data to construct BIM building modeling summary data through data combination identification , where ; S72. Import the three-dimensional model data of the target building modeling object in the BIM building modeling summary data into the BIM building modeling software to run in the same building modeling space as the three-dimensional model data of the building modeling scene . The BIM building modeling software performs modeling operations on the spatial positions of the target building modeling object and the building modeling scene, as well as the spatial shape of the target building modeling object, according to the target building modeling control instruction data .

[0013] A BIM intelligent building modeling system based on a large language model is used to implement the above-mentioned BIM intelligent building modeling method based on a large language model. The system includes a BIM building modeling object analysis module, a BIM building modeling instruction analysis module, and a BIM building modeling execution module; The BIM building modeling object analysis module includes a building modeling object feature information collection unit, a different type building modeling object feature information storage unit, a building modeling object type analysis unit, a different type building modeling object three-dimensional model storage unit, and a building modeling object three-dimensional model search unit; The building modeling object feature information collection unit collects building modeling object feature text data through a voice input module; the different type building modeling object feature information storage unit is used to store different type building modeling object feature text data; the building modeling object type analysis unit performs BIM building modeling object type analysis processing on the building modeling object feature text data in combination with the intelligent large language model and different type building modeling object feature text data to generate building modeling object type analysis data; the different type building modeling object three-dimensional model storage unit is used to store different type building modeling object three-dimensional model data; the building modeling object three-dimensional model search unit performs BIM building modeling object three-dimensional model search processing based on the building modeling object type analysis data and different type building modeling object three-dimensional model data to generate the three-dimensional model data of the target building modeling object; ​The BIM building modeling instruction analysis module includes a three-dimensional image information acquisition unit for building modeling objects, a three-dimensional model acquisition unit for building modeling scenes, a three-dimensional image information acquisition unit for building modeling scenes, a three-dimensional image information acquisition unit for the position, shape, and size of building modeling objects and building modeling scenes, a text information generation unit for the modeling requirements of the position, shape, and size of building modeling objects and building modeling scenes, a storage unit for building modeling control instructions with different modeling requirements, and a building modeling control instruction screening unit; The three-dimensional image information acquisition unit for building modeling objects acquires three-dimensional image data of building modeling objects through screenshot software; the three-dimensional model acquisition unit for building modeling scenes acquires three-dimensional model data of building modeling scenes through BIM building modeling software; the three-dimensional image information acquisition unit for building modeling scenes acquires three-dimensional image data of building modeling scenes through screenshot software; the three-dimensional image information acquisition unit for the position, shape, and size of building modeling objects and building modeling scenes acquires three-dimensional image data of the relationship between the position, shape, and size of building modeling objects and building modeling scenes through a brain-computer interface device; the text information generation unit for the modeling requirements of the position, shape, and size of building modeling objects and building modeling scenes performs recognition and processing on the text information of the modeling requirements for the spatial position and size of building modeling objects and building modeling scenes and the shape and size of building modeling objects based on the three-dimensional image data of building modeling objects, the three-dimensional image data of building modeling scenes, and the three-dimensional image data of the relationship between the position, shape, and size of building modeling objects and building modeling scenes in combination with an intelligent large language model, and generates text data of the modeling requirements for the relationship between the position, shape, and size of building modeling objects and building modeling scenes; the storage unit for building modeling control instructions with different modeling requirements is used to store data of building modeling control instructions with different modeling requirements; the building modeling control instruction screening unit performs screening and search processing on BIM building modeling control instructions based on the text data of the modeling requirements for the relationship between the position, shape, and size of building modeling objects and building modeling scenes and the data of building modeling control instructions with different modeling requirements, and generates target building modeling control instruction data; The BIM building modeling execution module includes a BIM building modeling information construction unit and a BIM building modeling operation execution unit; The BIM building modeling information construction unit constructs BIM building modeling summary data based on the three-dimensional model of the target building modeling object and the target building modeling control instruction in combination with data processing; the BIM building modeling operation execution unit performs BIM building modeling operations according to the BIM building modeling summary data in combination with BIM building modeling software.

[0014] (III) Beneficial effects The present invention provides a BIM intelligent building modeling method and system based on a large language model. It has the following beneficial effects: 1. Through the building modeling object feature information collection unit, the voice input module is used to quickly and accurately obtain the feature information of the target building modeling object, providing reliable data support for efficient BIM building modeling; the building modeling object type analysis unit, based on the building modeling object feature information, combines with the intelligent large language model and the scientifically preset feature information of different types of building modeling objects to conduct an efficient and intelligent analysis of the BIM building modeling object types, realizing the efficient and accurate identification of building component, main structure, and mechanical and electrical equipment object types in BIM building modeling based on the large language model; the building modeling object 3D model search unit, according to the building modeling object type analysis results, combines with the intelligent recognition algorithm and the 3D model information of different types of building modeling objects stored in the big data to conduct an intelligent search for the 3D model of the BIM building modeling object, realizing the scientific and efficient generation of the 3D structure of the BIM building modeling object based on the designer's natural language, and improving the efficiency of BIM building modeling.

[0015] 2. Through the mutual cooperation among the building modeling object 3D image information collection unit, the building modeling scene 3D model collection unit, the building modeling scene 3D image information collection unit, and the building modeling object and building modeling scene position, shape, and size 3D image information collection unit, the dynamic and scientific collection of the 3D image information of the target building modeling object, the 3D model information of the building modeling scene, the 3D image information of the building modeling scene, and the 3D information of the position, shape, and size relationship between the target building modeling object and the building modeling scene is realized, providing reliable data support for accurate and efficient BIM building modeling; the building modeling object and building modeling scene position, shape, and size modeling requirement text information generation unit, based on the intelligent large language model, conducts intelligent translation of the modeling requirement text information of the spatial position and size between the building modeling object and the building modeling scene and the shape and size of the building modeling object, realizing the scientific and standard generation of the building modeling object and building modeling scene position, shape, and size relationship modeling requirement text information based on the designer's design concept; the building modeling control instruction screening unit, based on the building modeling object and building modeling scene position, shape, and size relationship modeling requirement text data, combines with the intelligent search algorithm and the building modeling control instructions for different modeling requirements stored in the standard to conduct intelligent matching of the BIM building modeling control instructions, realizing the intelligent and precise selection of the BIM building modeling control instructions based on the designer's natural language and design concept, and realizing the intelligent BIM building modeling.

[0016] 3. Through the mutual cooperation of the BIM building modeling information construction unit and the BIM building modeling operation execution unit, based on the 3D model information of the target building modeling object and the target building modeling control instruction information, the BIM building modeling summary information is autonomously and efficiently constructed by combining data processing. At the same time, the BIM building modeling operation is dynamically and accurately executed in combination with the BIM building modeling software, realizing the dynamic and intelligent BIM building modeling based on the designer's natural language and design concept image information, and improving the accuracy and applicability of BIM building modeling. Brief Description of the Drawings

[0017] Figure 1 It is a schematic diagram of modules of a BIM intelligent building modeling system based on a large language model provided by the present invention; Figure 2 It is a flowchart of a BIM intelligent building modeling method based on a large language model provided by the present invention. Detailed Embodiments

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] The embodiments of a BIM intelligent building modeling method and system based on a large language model are as follows: Embodiment 1: Please refer to Figure 1 - Figure 2 , a BIM intelligent building modeling method based on a large language model, the method includes the following steps: S1. Collect characteristic text data of building modeling objects; S2. Perform BIM building modeling object type analysis and processing on the basis of the characteristic text data of building modeling objects and the characteristic text data of different types of building modeling objects to generate building modeling object type analysis data; S3. Perform three-dimensional model search processing on the BIM building modeling object according to the building modeling object type analysis data and the three-dimensional model data of different types of building modeling objects to generate target building modeling object three-dimensional model data; S4. Collect three-dimensional image data of building modeling objects, three-dimensional model data of building modeling scenes, three-dimensional image data of building modeling scenes, and three-dimensional image data of the position, shape and size relationship between building modeling objects and building modeling scenes; S5. Perform identification and processing of modeling requirement text information on the spatial position and size of the building modeling object and the building modeling scene and the shape and size of the building modeling object based on the three-dimensional image data of the building modeling object, the three-dimensional image data of the building modeling scene, and the three-dimensional image data of the position, shape and size relationship between the building modeling object and the building modeling scene to generate building modeling object and building modeling scene position, shape and size relationship modeling requirement text data; S6. Screen and search the BIM building modeling control instruction data based on the modeling requirement text data of the relationship between the building modeling object and the location, shape, and size of the building modeling scene, and generate the target building modeling control instruction data; S7. Construct the BIM building modeling summary data and execute the BIM building modeling operation.

[0020] Further, please refer to Figure 1 - Figure 2 , and the operation steps for collecting the characteristic text data of the building modeling object are as follows: S11. Online collect the characteristic description text information of the designer about the specific BIM building modeling object through the voice input module, and generate the characteristic text data of the building modeling object .

[0021] The operation steps for analyzing and processing the BIM building modeling object type based on the characteristic text data of the building modeling object and the characteristic text data of different types of building modeling objects to generate the building modeling object type analysis data are as follows: S21. Establish a set of characteristic text data of different types of building modeling objects , where represents the characteristic text data of different types of building modeling objects corresponding to the th type of BIM building modeling object type, represents the maximum value of the number of BIM building modeling object types; the BIM building modeling object types include walls, floors, doors and windows, roofs, stairs, railings, beams, columns, slabs, shear walls, pile foundations, pipes, valves, and pumps; the characteristic text data of different types of building modeling objects represents the standard building modeling object characteristic text information set for different types of BIM building modeling objects; S22. Use the intelligent large language model to match the characteristic text data of the building modeling object with the characteristic text data of different types of building modeling objects in the set of characteristic text data of different types of building modeling objects to perform building modeling feature character matching, search for the characteristic text data of different types of building modeling objects that matches the characteristic text data of the building modeling object , and construct the building modeling object type analysis data corresponding to the building modeling object type text information, and the intelligent large language model includes any one of DeepSeekV3, Alibaba Tongyi Qianwen Qwen2.5, and Tencent Hunyuan Turbo.

[0022] ​Perform 3D model search processing on BIM building modeling objects based on building modeling object type analysis data and 3D model data of different types of building modeling objects, and the operation steps for generating 3D model data of target building modeling objects are as follows: S31. Establish a set of 3D model data for different types of building modeling objects , where represents the 3D model data of different types of building modeling objects corresponding to the th type of BIM building modeling object type, and the characteristic text data of different types of building modeling objects represents the standard building modeling 3D entity data set for different types of BIM building modeling objects; S32. Perform keyword matching of building modeling object type for the building modeling object type analysis data and the 3D model data of different types of building modeling objects in the set of 3D model data of different types of building modeling objects to search for the 3D model data of different types of building modeling objects corresponding to the building modeling object type analysis data , and construct the 3D model data of the target building modeling object , and the specific operation steps for generating the 3D model data of the target building modeling object are as follows: S321. Initialize all the 3D model data of different types of building modeling objects as nectar sources in the search space of the set of 3D model data of different types of building modeling objects ; at the same time, initialize the number of nectar sources as , the current iteration number t, the maximum iteration number T, and the initial position of the nectar source . The calculation formula for the initial position is as follows: is randomly generated in the search space of the set of 3D model data of different types of building modeling objects , where represents a random number within , and and respectively represent the upper and lower bounds in the search space of the set of 3D model data of different types of building modeling objects ; and respectively represent the upper and lower bounds in the search space of the set of 3D model data of different types of building modeling objects ; S322. Assign employed bees to the initial position . At the beginning of the search, the employed bees are in the search space of the set of 3D model data of different types of building modeling objects and are composed of all the 3D model data of different types of building modeling objects The initial position formed Around, according to the formula , in the three-dimensional model data set of different types of building modeling objects Search in the search space to generate three-dimensional model data of different types of building modeling objects that match the building modeling object type analysis data ; among them New nectar source position ; where Indicates randomly selecting a nectar source that is not equal to i among nectar sources, Indicates the randomly selected nectar source Position in the search space of the three-dimensional model data set of different types of building modeling objects , Indicates a random number in the range [-1, 1]; S323. Based on the fitness value of the new nectar source position , determine the retained nectar sources according to the greedy selection method, and retain the three-dimensional model data of different types of building modeling objects that match the building modeling object type analysis data ; ; S324. The employed bees share the nectar source information, and the follower bees search for the three-dimensional model data of different types of building modeling objects in the search space of the three-dimensional model data set of different types of building modeling objects according to the probability value, and the follower bees search out the three-dimensional model data of different types of building modeling objects that match the building modeling object type analysis data based on the search results shared by the employed bees. Calculate the probability that the nectar source found by the employed bees is followed according to the probability formula. The probability calculation formula is as follows: , where Indicates the probability that the new nectar source position is found and followed by the employed bees, Indicates the fitness value of the new nectar source position shared by the follower bees according to the employed bees ; S325. The follower bees search in the same way as the employed bees, and search out the three-dimensional model data of different types of building modeling objects that match the building modeling object type analysis data in the search space of the three-dimensional model data set of different types of building modeling objects according to the greedy selection method ; Nectar source and retain it; S326. The scout bees are in the three-dimensional model data set of different types of building modeling objects During the random search process in the search space, if the nectar source location reaches the iteration threshold after t iterations of search and no better nectar source is found, the location of this nectar source will be abandoned, and the employed bee corresponding to it will become a scout bee; the formula for the scout bee to randomly generate a new nectar source location in the search space is as follows: , where represents generating a new nectar source after iterations at the position in the search space of the 3D model data set of different types of building modeling objects , represents generating a new nectar source after t iterations at the position in the search space of the 3D model data set of different types of building modeling objects ; S327. When the algorithm meets the maximum number of iterations T, output the 3D model data of different types of building modeling objects matching the building modeling object type analysis data , and generate the 3D model data of the target building modeling object through data identification .

[0023] Through the building modeling object feature information acquisition unit, use the voice input module to quickly and accurately obtain the feature information of the target building modeling object, providing reliable data support for efficient BIM building modeling; the building modeling object type analysis unit combines the feature information of the building modeling object with the intelligent large language model and the feature information of different types of building modeling objects preset scientifically to perform efficient and intelligent analysis of the BIM building modeling object type, realizing the efficient and accurate identification of building components, main structures, and electromechanical equipment object types in BIM building modeling based on the large language model; the building modeling object 3D model search unit combines the building modeling object type analysis results with the intelligent recognition algorithm and the 3D model information of different types of building modeling objects based on big data storage to perform intelligent search of the 3D model of the BIM building modeling object, realizing the scientific and efficient generation of the 3D structure of the BIM building modeling object based on the natural language of the designer and improving the efficiency of BIM building modeling.

[0024] Furthermore, please refer to Figure 1 - Figure 2 . The operation steps for collecting 3D image data of building modeling objects, 3D model data of building modeling scenes, 3D image data of building modeling scenes, and 3D image data of the position, shape, and size relationship between building modeling objects and building modeling scenes are as follows: S41. The 3D model data of the target building modeling object Run the BIM building modeling software and cooperate with the screenshot software to collect the axonometric drawing of the target building modeling object, and generate the three-dimensional image data of the building modeling object ; The BIM building modeling software includes any one of Autodesk Revit, ArchiCAD, and Bentley Systems, and the screenshot software includes any one of the screenshot tools built into Windows, Snipasteh, and Jinzhou Screenshot Software; Online obtain the three-dimensional entity information of the building modeling scene of the modeling design assembly object as the target building modeling object through the BIM building modeling software, and generate the three-dimensional model data of the building modeling scene ; Online obtain the three-dimensional model data of the building modeling scene through the screenshot software The corresponding axonometric drawing of the building modeling scene in the BIM building modeling software, and generate the three-dimensional image data of the building modeling scene ; Online synchronously collect the designer's positioning dimension design concept of the spatial position of the target building modeling object and the building modeling scene, as well as the shape dimension design concept of the target building modeling object through the brain-computer interface device, form the three-dimensional image information of the position dimension annotation and shape dimension annotation of the building modeling object, and generate the three-dimensional image data of the position and shape dimension relationship between the building modeling object and the building modeling scene , The brain-computer interface device includes any one of Neuralink brain-computer interface, North Brain No. 1 system, and mixed reality brain-computer interface system.

[0025] Based on the three-dimensional image data of the building modeling object, the three-dimensional image data of the building modeling scene, and the three-dimensional image data of the position and shape dimension relationship between the building modeling object and the building modeling scene, perform the identification and processing of the text information of the modeling requirements for the spatial position dimensions of the building modeling object and the building modeling scene and the shape dimensions of the building modeling object, and generate the operation steps of the text data of the modeling requirements for the position and shape dimension relationship between the building modeling object and the building modeling scene as follows: S51. Import the three-dimensional image data of the building modeling object , the three-dimensional image data of the building modeling scene , and the three-dimensional image data of the position and shape dimension relationship between the building modeling object and the building modeling scene into the intelligent large language model respectively. The intelligent large language model performs the identification and processing of the text information of the spatial position dimensions of the target building modeling object and the building modeling scene and the text information of the shape dimensions of the target building modeling object based on the image information, and generates the text data of the modeling requirements for the position and shape dimension relationship between the building modeling object and the building modeling scene , the text data representation of the modeling requirements for the location, shape, and size relationship between the building modeling object and the building modeling scene represents the positioning dimension text information of the target building modeling object and the building modeling scene, and the shape and size text information of the target building modeling object.

[0026] According to the text data of the modeling requirements for the location, shape, and size relationship between the building modeling object and the building modeling scene and the building modeling control instruction data for different modeling requirements, the operation steps for screening and searching the BIM building modeling control instructions to generate the target building modeling control instruction data are as follows: S61. Establish a set of building modeling control instruction data for different modeling requirements , ; where represents the building modeling control instruction data for different modeling requirements corresponding to the th type of building modeling requirement feature combination type. The building modeling requirement feature combination type represents an index data type formed by data combination mainly from the positioning dimension text information of the target building modeling object and the building modeling scene, and the shape and size text information of the target building modeling object, which is used to search for building modeling control instructions; the building modeling control instruction data for different modeling requirements represents the building modeling control instruction information of the BIM building modeling software set according to different building modeling requirement feature combination types; S62. Use the Rabin-Karp search algorithm to match the text data of the modeling requirements for the location, shape, and size relationship between the building modeling object and the building modeling scene with the building modeling control instruction data in the set of building modeling control instruction data for different modeling requirements to search for the building modeling control instruction data for different modeling requirements corresponding to the text data of the modeling requirements for the location, shape, and size relationship between the building modeling object and the building modeling scene , and identify the target building modeling control instruction data through data , and .

[0027] ​Through the mutual cooperation among the three-dimensional image information acquisition unit of the building modeling object, the three-dimensional model acquisition unit of the building modeling scene, the three-dimensional image information acquisition unit of the building modeling scene, and the three-dimensional image information acquisition unit of the position, shape, and size of the building modeling object and the building modeling scene, the dynamic scientific acquisition of the three-dimensional image information of the target building modeling object, the three-dimensional model information of the building modeling scene, the three-dimensional image information of the building modeling scene, and the three-dimensional information of the position, shape, and size relationship between the target building modeling object and the building modeling scene is realized, providing reliable data support for accurate and efficient BIM building modeling; the text information generation unit for the modeling requirements of the position, shape, and size of the building modeling object and the building modeling scene performs intelligent translation of the text information for the modeling requirements of the spatial position and size of the building modeling object and the building modeling scene and the shape and size of the building modeling object based on the intelligent large language model, realizing the scientific and standard generation of the text information for the modeling requirements of the position, shape, and size relationship between the building modeling object and the building modeling scene based on the design concept of the designer; the building modeling control instruction screening unit performs intelligent matching of the BIM building modeling control instructions based on the text data of the modeling requirements for the position, shape, and size relationship between the building modeling object and the building modeling scene, combined with the intelligent search algorithm and different modeling requirement building modeling control instructions stored in the standard, realizing the intelligent and accurate selection of the BIM building modeling control instructions based on the natural language of the designer and the design concept of the designer, and realizing the intelligent BIM building modeling.

[0028] Further, please refer to Figure 1 - Figure 2 to construct the BIM building modeling summary data and the operation steps for performing the BIM building modeling operation are as follows: S71. Combine the three-dimensional model data of the target building modeling object and the target building modeling control instruction data through data combination identification to construct the BIM building modeling summary data , where ; S72. Import the three-dimensional model data of the target building modeling object in the BIM building modeling summary data into the BIM building modeling software to run in the same building modeling space as the three-dimensional model data of the building modeling scene. The BIM building modeling software performs modeling operations on the spatial position of the target building modeling object and the building modeling scene and the spatial shape of the target building modeling object according to the target building modeling control instruction data .

[0029] Through the mutual cooperation of the BIM building modeling information construction unit and the BIM building modeling operation execution unit, based on the three-dimensional model information of the target building modeling object and the target building modeling control instruction information, combined with data processing, the BIM building modeling summary information is autonomously and efficiently constructed. At the same time, combined with the BIM building modeling software, the BIM building modeling operation is dynamically and accurately executed, realizing BIM building modeling based on the natural language of designers and the design concept image information, and improving the accuracy and applicability of BIM building modeling.

[0030] Embodiment 2: Please refer to Figure 1 - Figure 2 , a BIM intelligent building modeling system based on a large language model, used to implement a BIM intelligent building modeling method based on a large language model. The system includes a BIM building modeling object analysis module, a BIM building modeling instruction analysis module, and a BIM building modeling execution module; The BIM building modeling object analysis module includes a building modeling object feature information collection unit, a feature information storage unit for different types of building modeling objects, a building modeling object type analysis unit, a three-dimensional model storage unit for different types of building modeling objects, and a three-dimensional model search unit for building modeling objects; The building modeling object feature information collection unit collects the building modeling object feature text data through the voice input module; the feature information storage unit for different types of building modeling objects is used to store the feature text data of different types of building modeling objects; the building modeling object type analysis unit combines the building modeling object feature text data with the intelligent large language model and the feature text data of different types of building modeling objects to perform BIM building modeling object type analysis and processing, generating building modeling object type analysis data; the three-dimensional model storage unit for different types of building modeling objects is used to store the three-dimensional model data of different types of building modeling objects; the three-dimensional model search unit for building modeling objects performs three-dimensional model search processing of the BIM building modeling object according to the building modeling object type analysis data and the three-dimensional model data of different types of building modeling objects, generating the three-dimensional model data of the target building modeling object; The BIM building modeling instruction analysis module includes a three-dimensional image information collection unit for building modeling objects, a three-dimensional model collection unit for building modeling scenes, a three-dimensional image information collection unit for building modeling scenes, a three-dimensional image information collection unit for the position, shape, and size of building modeling objects and building modeling scenes, a text information generation unit for the position, shape, and size modeling requirements of building modeling objects and building modeling scenes, a storage unit for building modeling control instructions with different modeling requirements, and a building modeling control instruction screening unit; The three-dimensional image information acquisition unit of the building modeling object collects the three-dimensional image data of the building modeling object through screenshot software; the three-dimensional model acquisition unit of the building modeling scene collects the three-dimensional model data of the building modeling scene through BIM building modeling software; the three-dimensional image information acquisition unit of the building modeling scene collects the three-dimensional image data of the building modeling scene through screenshot software; the three-dimensional image information acquisition unit of the position, shape and size of the building modeling object and the building modeling scene collects the three-dimensional image data of the relationship between the position, shape and size of the building modeling object and the building modeling scene through a brain-computer interface device; the text information generation unit for the modeling requirements of the position, shape and size of the building modeling object and the building modeling scene performs identification and processing on the text information of the modeling requirements for the spatial position and size of the building modeling object and the building modeling scene and the shape and size of the building modeling object based on the three-dimensional image data of the building modeling object, the three-dimensional image data of the building modeling scene, and the three-dimensional image data of the relationship between the position, shape and size of the building modeling object and the building modeling scene in combination with an intelligent large language model, and generates the text data of the modeling requirements for the relationship between the position, shape and size of the building modeling object and the building modeling scene; the storage unit for building modeling control instructions with different modeling requirements is used to store the data of building modeling control instructions with different modeling requirements; the building modeling control instruction screening unit performs screening and search processing on the BIM building modeling control instructions according to the text data of the modeling requirements for the relationship between the position, shape and size of the building modeling object and the building modeling scene and the data of the building modeling control instructions with different modeling requirements, and generates the target building modeling control instruction data; The BIM building modeling execution module includes a BIM building modeling information construction unit and a BIM building modeling operation execution unit; The BIM building modeling information construction unit constructs the BIM building modeling summary data based on the three-dimensional model of the target building modeling object and the target building modeling control instruction in combination with data processing; the BIM building modeling operation execution unit performs the BIM building modeling operation according to the BIM building modeling summary data and in combination with the BIM building modeling software.

[0031] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A BIM intelligent building modeling method based on a large language model, characterized in that: The method comprises the following steps: S1, collecting characteristic text data of building modeling objects; S2. Perform BIM building modeling object type analysis and processing to generate building modeling object type analysis data; S3, performing a three-dimensional model search process on a BIM building modeling object to generate three-dimensional model data of a target building modeling object; S4, collecting three-dimensional image data of building modeling objects, three-dimensional model data of building modeling scenes, three-dimensional image data of building modeling scenes, and three-dimensional image data of the relationship between the position, shape and size of the building modeling objects and the building modeling scenes; S5, performing recognition processing on modeling requirement text information of the spatial position size of the architectural modeling object and the architectural modeling scene and the shape size of the architectural modeling object, and generating modeling requirement text data on the relationship between the position, shape and size of the architectural modeling object and the architectural modeling scene; S6, performing BIM building modeling control instruction screening and search processing to generate target building modeling control instruction data; S7. Construct BIM building modeling summary data and execute BIM building modeling tasks.

2. The BIM intelligent building modeling method based on a large language model according to claim 1, characterized in that: The S1 comprises the following steps: S11. Collect the designer's feature description text information about specific BIM building modeling objects online through the voice input module, and generate building modeling object feature text data .

3. The BIM intelligent building modeling method based on a large language model according to claim 2, characterized in that: The S2 comprises the following steps: S21. Establishing feature text data sets of different types of building modeling objects , ;in Indicates Different types of building modeling object feature text data corresponding to different types of BIM building modeling object types, Indicates the maximum number of BIM building modeling object types; S22, using an intelligent large language model to With the As stated in Perform architectural modeling feature character matching and search for Matching the Corresponding building modeling object type text information, and constructing building modeling object type analysis data .

4. The BIM intelligent building modeling method based on a large language model according to claim 3 is characterized in that: The S3 comprises the following steps: S31. Establishing 3D model data sets of different types of building modeling objects ,in Indicates 3D model data of different types of building modeling objects corresponding to the types of BIM building modeling objects; S32, the With the As stated in Perform keyword matching of building modeling object types and search for the The corresponding , and construct the 3D model data of the target building modeling object , execute to generate the three-dimensional model data of the target building modeling object The specific steps are as follows: S321, in the In the search space, all the Initialize as a honey source ; Initialize the honey source at the same time The quantity is , the current number of iterations t and the maximum number of iterations T and the honey source Initial position ; S322, initial position Assign employed bees. At the beginning of the search, the employed bees The search space consists of all the Initial position of the composition around, in the The search space of Matching the New honey source location ; S323, based on the location of the new nectar source The fitness value of the nectar source is determined by the greedy selection method, and the nectar source retained is the same as that Matching the ; S324, the employed bees share the nectar source information, and the follower bees follow the probability value in the In the search space of Search, follow the bee to search for the search results shared by the hired bee. Matching the , calculate the probability that the nectar source found by the employed bees is followed; S325, the follower bee uses the same method as the employed bee to search, according to the greedy selection method described in Search the search space for Matching the Honey Source and retain; S326, the scout bee In the random search process in the search space, if the location of the honey source After t iterations, the search reaches the iteration threshold No better nectar source was found. The location of the nectar source The bees will be abandoned, and the corresponding employed bees will become scout bees. The scout bees will randomly generate new nectar source locations in the search space. S327, when the algorithm meets the maximum number of iterations T, the output is the same as the Matching the , and generate the three-dimensional model data of the target building modeling object through data identification .

5. The BIM intelligent building modeling method based on a large language model according to claim 4 is characterized in that: The S4 comprises the following steps: S41, the Import BIM building modeling software and run it, and use screenshot software to collect the axonometric drawing of the target building modeling object, and generate 3D image data of the building modeling object ; The BIM architectural modeling software is used to obtain the three-dimensional entity information of the architectural modeling scene as the modeling design assembly object of the target architectural modeling object online, and generate the three-dimensional model data of the architectural modeling scene ; Get the screenshots online The corresponding axonometric drawing of the building modeling scene in the BIM building modeling software, and the generation of 3D image data of the building modeling scene ; The brain-computer interface device is used to synchronously collect the designer's positioning dimension design ideas about the spatial position of the target building modeling object and the building modeling scene, as well as the shape dimension design ideas of the target building modeling object, to form three-dimensional image information of the position dimension annotation and shape dimension annotation of the building modeling object, and to generate three-dimensional image data of the position, shape and size relationship between the building modeling object and the building modeling scene .

6. The BIM intelligent building modeling method based on a large language model according to claim 5, characterized in that: The S5 comprises the following steps: S51, the , , The intelligent large language model is respectively imported into the intelligent large language model, which recognizes and processes the spatial position and size text information of the target building modeling object and the building modeling scene, as well as the shape and size text information of the target building modeling object based on the image information, and generates the modeling requirements text data of the position, shape and size relationship between the building modeling object and the building modeling scene. .

7. The BIM intelligent building modeling method based on a large language model according to claim 6 is characterized in that: The S6 comprises the following steps: S61. Establishing building modeling control instruction data sets for different modeling requirements , ;in Indicates Architectural modeling control instruction data of different modeling requirements corresponding to the architectural modeling requirement feature combination types; S62, using the Rabin-Karp search algorithm to With the As stated in Match the building modeling requirement feature information and search for the The corresponding , and after data identification, the target building modeling control instruction data .

8. The BIM intelligent building modeling method based on a large language model according to claim 7, characterized in that: The S7 comprises the following steps: S71, the , Construct BIM building modeling summary data through data combination identification ; S72, the As stated in Import into BIM architectural modeling software Running in the same building modeling space, the BIM building modeling software follows the target building modeling control instruction data Modeling is performed on the spatial position of the target building modeling object and the building modeling scene, as well as the spatial shape of the target building modeling object.

9. A BIM intelligent building modeling system based on a large language model, used to implement a BIM intelligent building modeling method based on a large language model as claimed in any one of claims 1 to 8, characterized in that: The system includes a BIM building modeling object analysis module, a BIM building modeling instruction analysis module, and a BIM building modeling execution module.

Images