BIM family library intelligent management method based on generative large model

Abstract

The invention discloses a BIM family library intelligent management method based on a generative large model. The BIM family library intelligent management method comprises the steps of obtaining natural language data representing user requirements; based on a predefined cue word template, converting the natural language data into input data of a generative large model; inputting the input data into the generative large model to generate a corresponding semantic expression; based on different types of user demands, executing the following operations: if the user demand is component retrieval, retrieving an optimal family library component in the BIM family library according to a semantic retrieval formula; if the user demand is component generation, inputting a semantic generation form into a BIM creation system through an API, and generating a target component through a parameterization design module; if the user demand is component scheme design, inputting the semantic rule into the generative large model to obtain a component parameterization design scheme conforming to the rule; and if the user demand is component selection, inputting the semantic recommendation into the generative large model to obtain a family library component recommendation result conforming to the user preference.

Description

Technical Field

[0001] The present invention relates to the technical field of large models, and in particular to a BIM family library intelligent management method based on a generative large model. Background Art

[0002] With the rapid development of building information technology, BIM (Building Information Modeling) technology has been widely used in engineering design, construction management and operation and maintenance management. Through three-dimensional digital modeling, BIM can integrate relevant information throughout the life cycle of a building, providing efficient collaborative support and accurate data management for construction projects. As an important tool for storing standardized components and their attribute information, BIM family library plays a core role in component management, design reuse and project collaboration. However, traditional BIM family library management methods still have many problems and are difficult to meet the needs of modern construction projects for efficiency, intelligence and dynamic adaptability.

[0003] At present, the main problems faced by BIM family library management include:

[0004] 1. Low retrieval efficiency: Traditional family library management relies on keyword search or rule matching, which makes it difficult to understand users' complex semantic needs, resulting in low retrieval efficiency and a low match between recommended results and actual needs.

[0005] 2. Insufficient dynamic update capabilities: Traditional family libraries are difficult to dynamically adjust the properties and parameters of components based on real-time data feedback during construction, design changes, or operation and maintenance, resulting in a lack of flexibility and adaptability in the content of the family library.

[0006] 3. Poor collaboration: In multi-project or multi-team collaboration, family library resources are difficult to integrate efficiently. The creation and use of duplicate components significantly increases the design workload and reduces collaboration efficiency.

[0007] 4. Low level of intelligence: Existing management methods fail to fully utilize artificial intelligence technology, especially the powerful semantic understanding and generation capabilities of generative large language models. The family library still relies mainly on manual operations in component generation, optimization and recommendation.

[0008] The introduction of generative large models (such as LLAMA, GPT, etc.) provides new possibilities for the intelligent development of BIM family library management. With its powerful natural language understanding and generation capabilities, such models can extract complex semantics from natural language requirements input by users and generate components or optimization solutions that meet the requirements. In addition, generative large models can dynamically optimize based on historical data and real-time feedback through deep learning technology, thereby achieving efficient management and intelligent recommendation of family library resources.

[0009] Although generative big models have demonstrated excellent performance in natural language processing, data generation and other fields, their application in BIM family library management is still in the exploratory stage. Current research and practice mainly focus on text processing, simple task optimization and preliminary component recommendation, while there is no systematic solution for core needs such as dynamic update, collaborative optimization and full life cycle management of family library management. Therefore, the deep integration of generative big models with BIM technology to realize intelligent management, optimization and dynamic update of family libraries has become an important direction for solving the problem of building informatization. Summary of the invention

[0010] The present invention proposes a BIM family library intelligent management method based on a generative large model. Through technical means such as semantic understanding, component generation, dynamic optimization and collaborative management, it solves the problems of low retrieval efficiency, poor dynamic response capability and insufficient collaboration in traditional BIM family library management, and significantly improves the management efficiency and intelligence level of the entire life cycle of construction projects.

[0011] In order to achieve the above-mentioned invention object, the technical solution provided by the present invention includes:

[0012] The intelligent management method of BIM family library based on generative large model includes:

[0013] S1. Obtain natural language data representing user needs;

[0014] S2. Based on a predefined prompt word template, convert the natural language data into input data of a generative large model, wherein the prompt word template includes format requirements and examples for input content and output content;

[0015] S3, inputting the input data into the generative big model to generate the corresponding semantic expression;

[0016] S4. Based on different types of user needs, perform the following operations:

[0017] If the user requirement is component retrieval, the semantic expression is a semantic retrieval formula, and the optimal family library component is retrieved in the BIM family library according to the semantic retrieval formula;

[0018] If the user requirement is component generation, the semantic expression is a semantic generation formula, and the semantic generation formula is input into the BIM creation system through the API to generate the target component through the parametric design module;

[0019] If the user requirement is component solution design, the semantic expression is a semantic rule formula, and the semantic rule formula is input into the generative macro model to obtain a component parametric design solution that complies with the rule;

[0020] If the user's demand is component selection, the semantic expression is a semantic recommendation formula, and the semantic recommendation formula is input into the generative large model to obtain the family library component recommendation results that meet the user's preferences.

[0021] Preferably, the method further comprises the steps of:

[0022] S0. Fine-tune the generative big model based on domain knowledge. The fine-tuning is based on the semantic features and business needs of the construction field. By optimizing model parameters and adjusting semantic features, the generative big model's ability to understand and generate semantics related to the BIM family library is enhanced to ensure efficient adaptation of the generative big model to the BIM family library.

[0023] Preferably, the prompt word template is designed according to different task requirements, specifically including:

[0024] The prompt word template for component retrieval requirements includes semantic keywords describing user requirements and their context information;

[0025] The prompt word template for component generation requirements includes the geometric constraints, attribute requirements and target function description of the target component;

[0026] The prompt word template for component solution design requirements includes parametric expressions of design rules and specifications;

[0027] The prompt word template for component selection needs includes user preferences, historical selection data and recommendation conditions.

[0028] Preferably, the target component further includes a parameter tag, and the parameter tag includes the geometric characteristics, material properties, connection method and applicable scope of the target component.

[0029] Preferably, the following steps are also included to improve the real-time response performance in the management process:

[0030] Lightweight the large generative model, optimize model parameters and reduce complexity;

[0031] In the execution of the management method, edge computing architecture is used to share the computing load, and tasks requiring high computing power are deployed and executed in the cloud;

[0032] Use Web3D technology to directly render and interact with 3D models on the user side to ensure the smoothness and real-time performance of 3D visualization operations.

[0033] Preferably, the following steps are also included to ensure the security of the BIM family library and the flexible allocation of resources:

[0034] The BIM family library is stored in a distributed cloud database, and the family library resources are encrypted through the encryption interface of the generative large model;

[0035] In the process of executing the management method, set user operation restrictions based on the access rights control mechanism to ensure the privacy and integrity of the family library resources in a multi-user collaborative environment;

[0036] Dynamically allocate computing and storage resources according to project requirements, including: allocating cloud storage capacity on demand to adapt to family library management of projects of different sizes; adjusting computing resource allocation according to real-time load to ensure efficient use of resources and smooth execution of management tasks.

[0037] Preferably, it also includes:

[0038] Step S5: Obtain real-time feedback data throughout the project life cycle;

[0039] Step S6: Analyze the real-time feedback data using the generative big model to generate an optimization solution.

[0040] Preferably, the method of using a generative large model to analyze real-time feedback data and generate an optimization solution includes:

[0041] Compare the actual parameters of the components in the real-time feedback data with the corresponding theoretical parameters of the components in the BIM family library. If there is an error exceeding the preset threshold, the generative model is input according to the actual parameters of the components to generate an updated semantic generative formula, and the updated semantic generative formula is input into the BIM creation system through the API to generate an updated target component through the parametric design module.

[0042] Preferably, the following steps are also included to achieve the collaboration and knowledge transfer of family library resources among multiple projects:

[0043] Use generative big models to analyze and share cross-project data and identify common component requirements between different projects;

[0044] Based on the identification results, family components with high matching degree are recommended in the new project to meet the project requirements;

[0045] Summarize the component usage experience accumulated in different projects and incorporate the experience into the family library optimization strategy through the transfer learning function.

[0046] Beneficial Effects

[0047] 1. Improve the intelligent level of family database management

[0048] The present invention utilizes the powerful semantic parsing capability of the generative big model, which can deeply understand the natural language requirements input by users without relying on complex keyword matching rules, greatly improving the retrieval efficiency of family library resources and the accuracy of component recommendation. With the generation capability of the generative big model, users only need to provide a simple demand description to automatically generate components or optimization solutions that meet the design requirements, reducing manual intervention and improving the intelligence of the family library.

[0049] 2. Realize dynamic update and adaptive optimization of family library resources

[0050] The present invention dynamically adjusts the component parameters, geometric shapes and attribute information in the family library by acquiring construction feedback data, design change information or operation and maintenance requirements in real time.

[0051] 3. Enhance the coordination and resource integration capabilities of family database management

[0052] By utilizing the transfer learning function of the generative large model, we can realize the sharing and integration of family library resources among multiple projects, quickly identify the common component requirements among different projects, and give priority to recommending components with high matching degree. By summarizing and integrating cross-project experience, we can reduce duplication of work, improve team collaboration efficiency, and significantly reduce management costs.

[0053] 4. Improve system response performance and user experience

[0054] By lightweight processing of large generative models, and adopting edge computing architecture and Web3D technology, the present invention can complete tasks with high computing requirements in the cloud, while achieving real-time rendering and interaction on the user side, ensuring smooth operation in high-concurrency scenarios. The lightweight design significantly shortens the response time of retrieval and generation tasks, and improves the user's operating efficiency in family library management.

[0055] 5. Support intelligent management of the entire life cycle of construction projects

[0056] The present invention is not only applicable to the design stage, but also can run through the construction and operation and maintenance stages, realizing the full life cycle management of family library resources, and providing strong support for efficient collaboration and intelligent decision-making in building informatization. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A schematic diagram of a process flow of a BIM family library intelligent management method based on a generative large model provided in a preferred embodiment of the present invention;

[0058] Figure 2 A management system structure and information transmission schematic diagram based on the BIM family library intelligent management method provided by the present invention is provided in a preferred embodiment of the present invention;

[0059] Figure 3 A schematic diagram of component information with parameter tags as automatically generated descriptions in another preferred embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention is further described below in conjunction with the accompanying drawings. In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside", "outside" and the like indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings, which are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention.

[0061] Embodiment 1

[0062] like Figure 1 and Figure 2 As shown, this embodiment provides a BIM family library intelligent management method based on a generative large model, including:

[0063] S1. Obtain natural language data that represents user needs.

[0064] In the present invention, the user demand refers to the specific request of the user for the BIM family library management system regarding components, schemes or related information. These demands are usually input into the system in the form of natural language to guide the generative large model to complete the corresponding operations. User demands can be divided into the following types:

[0065] 1. Component retrieval requirements

[0066] Users hope to quickly find BIM family components that meet certain specific conditions through the system. For example:

[0067] "Find Structural Steel Beams for Residential Construction."

[0068] “Recommend a concrete column that meets earthquake resistance requirements.”

[0069] 2. Component generation requirements

[0070] Users want to generate new components through the system to meet specific geometric shapes, properties or design goals. For example:

[0071] "Generate a standard concrete wall that is 6 meters long and 3 meters high."

[0072] “Design an energy-saving glass window suitable for high-altitude cold regions.”

[0073] 3. Component design requirements

[0074] Users need the system to generate a complete component design solution based on specific industry specifications or design rules. For example:

[0075] “Designing a Building Staircase to Comply with Fire Codes.”

[0076] “Design of seismic steel columns for high-rise buildings.”

[0077] 4. Component selection requirements

[0078] Users need the system to recommend family components that suit their needs based on historical records or preferences. For example:

[0079] “Recommend exterior wall decoration materials suitable for high-rise residential buildings.”

[0080] “Pick a lightweight steel door that’s affordable.”

[0081] S2. Based on a predefined prompt word template, convert the natural language data into input data of a generative large model, wherein the prompt word template includes format requirements and examples for input content and output content.

[0082] A predefined prompt word template refers to a structured rule or framework used to guide a generative large model (such as an LLAMA model) to parse user input data and generate output results. The main function of the prompt word template is to standardize the format of input and output so that the large model can more accurately understand user needs and perform corresponding operations.

[0083] The user's requirements input in natural language are converted into a standardized data format that can be understood and directly used by the generative large model through certain processing steps or rules. The core goal of this process is to convert unstructured natural language requirements into input with a clear semantic structure through prompt word templates, so that the large model can perform tasks. The prompt word template is designed according to different task requirements. In some preferred embodiments, the construction method of the prompt word template specifically includes:

[0084] The prompt word template for component retrieval requirements includes semantic keywords describing user requirements and their context information;

[0085] The prompt word template for component generation requirements includes the geometric constraints, attribute requirements and target function description of the target component;

[0086] The prompt word template for component solution design requirements includes parametric expressions of design rules and specifications;

[0087] The prompt word template for component selection needs includes user preferences, historical selection data and recommendation conditions.

[0088] Those skilled in the art should understand that the semantic analysis and keyword extraction of the natural language data input by the user can be achieved through natural language processing (NLP) technology, which is not the focus of the present invention and will not be described in detail.

[0089] S3. Input the input data into the generative big model to generate the corresponding semantic expression. The semantic expression is the structured output generated by the generative big model after parsing the input data. It can accurately describe the semantic content of user needs and directly guide subsequent operations. The form and content of the semantic expression vary depending on the task requirements.

[0090] S4. Based on different types of user needs, perform the following operations:

[0091] If the user's requirement is to search for components, the semantic expression is a semantic search formula, and the optimal component of the BIM family library is searched according to the semantic search formula. By using the natural language understanding ability of the generative model, the family library requirements can be described in natural language without relying on specific keywords or naming rules, and accurate search results can be generated through semantic analysis. At the same time, based on context understanding and user historical operation habits, the optimal family library component can be dynamically recommended.

[0092] If the user requirement is component generation, the semantic expression is a semantic generation formula, which is input into the BIM creation system through an API to generate a target component through a parametric design module. There is no need to manually draw complex components, and only simple geometric constraints, attribute requirements or functional goals are required to generate a family component that meets the requirements. In some preferred embodiments, the target component also includes a parameter tag, which includes the geometric features, material properties, connection methods and applicable scope of the target component. Figure 3 As shown in the figure, the parameter label is an automatically generated description, which not only provides comprehensive component information for designers, but also provides data support for subsequent component management, query and optimization. Through the automatic description capability of the generative model, the detailed information of each component in the family library can be quickly generated and maintained, thereby greatly improving the management efficiency and accuracy of the family library.

[0093] If the user requirement is for component scheme design, the semantic expression is a semantic rule-based expression, and the semantic rule-based expression is input into the generative large model to obtain a component parametric design scheme that complies with the rules. In the component scheme design task, the present invention parses the design requirements input by the user through the generative large model and converts it into a semantic rule-based expression. The semantic rule-based expression clarifies the design goals and specification requirements, and serves as an input to guide the large model to generate a parametric design scheme that complies with the specifications in combination with the industry design rule library. The generated design scheme includes the geometric characteristics, material properties and design rule compliance of the component, and can be directly used for design verification, component generation or family library updates, which significantly improves the automation level and efficiency of component design, while ensuring that industry specifications and user needs are met. For example, for fire protection, earthquake resistance, energy saving and other specifications in the construction industry, the present invention can intelligently generate component design schemes that comply with the rules to help users avoid risks in the early stages of design.

[0094] If the user demand is component selection, the semantic expression is a semantic recommendation formula, and the semantic recommendation formula is input into the generative large model to obtain the family library component recommendation results that meet the user's preferences. In the component selection task, the present invention extracts commonly used components, parameter adjustment habits, and special demand preferences by analyzing the user's family library usage behavior in historical projects. In new projects, the present invention can generate intelligent suggestions based on these behavior patterns, and give priority to recommending family components that meet user habits, thereby improving the design experience and efficiency. The semantic recommendation formula is used as input to guide the large model to perform semantic matching in the family library resources, screen out family components that meet the needs, and generate a recommendation list sorted by matching degree. The recommendation results not only include the parameters, functions, and applicable scope of the component, but also optimize the matching conditions in combination with the user's historical data and current needs, significantly improving the intelligence level, efficiency, and user experience of component selection.

[0095] In some preferred embodiments, in order to improve the real-time response performance in the management process, the generative large model (such as the LLAMA model) is lightweighted, and the response time of the retrieval and generation tasks is effectively shortened by optimizing the model parameters and reducing the computational complexity, ensuring that an efficient user experience can still be maintained in high-concurrency scenarios. In order to further optimize the system performance, the present invention adopts an edge computing architecture to share the computing load between the cloud and the local device. During the execution of the management method, tasks with high computing power requirements are deployed in the cloud for complex data processing and model reasoning, while rendering and interactive tasks are assigned to edge devices, making full use of the computing power of edge devices to achieve rapid response and data display. In addition, this embodiment integrates Web3D technology, directly renders and interacts with three-dimensional models on the user side, effectively reduces the computing pressure of local devices, and ensures the smoothness and real-time performance of three-dimensional visualization operations. The combination of Web3D technology with lightweight models and edge computing architecture realizes the smooth operation and real-time response capability of the system on various devices by optimizing computing resource scheduling. This design improves the stability and user experience of the system in high-concurrency scenarios, and provides technical support for the efficiency and intelligence of BIM family library management.

[0096] In some other preferred embodiments, the following steps are also included to ensure the security of the BIM family library and the flexible allocation of resources:

[0097] The BIM family library is stored in a distributed cloud database, and the family library resources are encrypted through the encryption interface of the generative large model; during the execution of the management method, user operation restrictions based on the access permission control mechanism are set to ensure the privacy and integrity of family library resources in a multi-user collaborative environment; computing and storage resources are dynamically allocated according to project requirements, including: allocating cloud storage capacity on demand to adapt to family library management of projects of different sizes; adjusting computing resource allocation according to real-time load to ensure efficient use of resources and smooth execution of management tasks.

[0098] In order to dynamically manage the family library resources in the entire life cycle of the project, automatically update the family library according to the real-time feedback of the construction data, and ensure that the family library is always in the best state, the present invention also includes the steps of:

[0099] Step S5: Obtain real-time feedback data throughout the project life cycle;

[0100] Step S6: Analyze the real-time feedback data using the generative big model to generate an optimization solution, preferably:

[0101] Compare the actual parameters of the components in the real-time feedback data with the corresponding theoretical parameters of the components in the BIM family library. If there is an error exceeding the preset threshold, the generative model is input according to the actual parameters of the components to generate an updated semantic generative formula, and the updated semantic generative formula is input into the BIM creation system through the API to generate an updated target component through the parametric design module.

[0102] In order to achieve efficient collaboration and knowledge transfer of family library resources among multiple projects, quickly identify common component requirements among different projects, and give priority to recommending highly matching components in new projects. At the same time, the experience of different projects is summarized and integrated into the family library optimization strategy to reduce duplication of work and improve team collaboration efficiency. The present invention also includes the following steps:

[0103] Use generative big models to analyze and share cross-project data and identify common component requirements between different projects;

[0104] Based on the identification results, family components with high matching degree are recommended in the new project to meet the project requirements;

[0105] Summarize the component usage experience accumulated in different projects and incorporate the experience into the family library optimization strategy through the transfer learning function.

[0106] Those skilled in the art should know that the generative large model is a general model, and its semantic parsing and generation capabilities are usually based on pre-trained data in a wide range of fields. However, in specific fields (such as building information modeling, BIM), the general model may have the following problems:

[0107] 1. Insufficient domain understanding: The model lacks a deep understanding of specialized terms in the construction field (such as "parametric design of components", "fire-resistant steel beams", etc.), which may lead to misunderstanding of user needs or inaccurate results.

[0108] 2. Insufficient accuracy and applicability: The general model has not been fine-tuned with data from the construction field and cannot effectively adapt to the management tasks of the BIM family library, such as component retrieval, generation or recommendation.

[0109] 3. The generated results are not well context-relevant: In a specific context, the general model may not be able to effectively associate the specifications in the construction field with the actual application scenarios, and the generated semantic expressions may not meet the actual needs.

[0110] In order to solve the above problems and improve the applicability of the model in the BIM family library management task, some preferred embodiments also include the following steps:

[0111] S0. Fine-tune the generative big model based on domain knowledge. The fine-tuning is based on the semantic features and business needs of the construction field. By optimizing model parameters and adjusting semantic features, the generative big model's ability to understand and generate semantics related to the BIM family library is enhanced to ensure efficient adaptation of the generative big model to the BIM family library.

[0112] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A BIM family library intelligent management method based on a generative large model, characterized in that: include: S1. Obtain natural language data representing user needs; S2. Based on a predefined prompt word template, convert the natural language data into input data of a generative large model, wherein the prompt word template includes format requirements and examples for input content and output content; S3, inputting the input data into the generative big model to generate the corresponding semantic expression; S4. Based on different types of user needs, perform the following operations: If the user requirement is component retrieval, the semantic expression is a semantic retrieval formula, and the optimal family library component is retrieved in the BIM family library according to the semantic retrieval formula; If the user requirement is component generation, the semantic expression is a semantic generation formula, and the semantic generation formula is input into the BIM creation system through the API to generate the target component through the parametric design module; If the user requirement is component solution design, the semantic expression is a semantic rule formula, and the semantic rule formula is input into the generative macro model to obtain a component parametric design solution that complies with the rule; If the user's demand is component selection, the semantic expression is a semantic recommendation formula, and the semantic recommendation formula is input into the generative large model to obtain the family library component recommendation results that meet the user's preferences.

2. The BIM family library intelligent management method based on the generative large model according to claim 1 is characterized in that: Also includes the steps: S0. Fine-tune the generative big model based on domain knowledge. The fine-tuning is based on the semantic features and business needs of the construction field. By optimizing model parameters and adjusting semantic features, the generative big model's ability to understand and generate semantics related to the BIM family library is enhanced to ensure efficient adaptation of the generative big model to the BIM family library.

3. The BIM family library intelligent management method based on the generative large model according to claim 1, characterized in that: The prompt word template is designed according to different task requirements, specifically including: The prompt word template for component retrieval requirements includes semantic keywords describing user requirements and their context information; The prompt word template for component generation requirements includes the geometric constraints, attribute requirements and target function description of the target component; The prompt word template for component solution design requirements includes parametric expressions of design rules and specifications; The prompt word template for component selection needs includes user preferences, historical selection data and recommendation conditions.

4. The BIM family library intelligent management method based on the generative large model according to claim 1 is characterized in that: The target component also includes a parameter tag, which includes the geometric characteristics, material properties, connection methods and application scope of the target component.

5. The BIM family library intelligent management method based on the generative large model according to claim 1 is characterized in that: The following steps are also included to improve the real-time response performance during management: Lightweight the large generative model, optimize model parameters and reduce complexity; In the execution of the management method, edge computing architecture is used to share the computing load, and tasks requiring high computing power are deployed and executed in the cloud; Use Web3D technology to directly render and interact with 3D models on the user side to ensure the smoothness and real-time performance of 3D visualization operations.

6. The BIM family library intelligent management method based on the generative large model according to claim 1 is characterized in that: The following steps are also included to ensure the security of the BIM family library and flexible resource allocation: The BIM family library is stored in a distributed cloud database, and the family library resources are encrypted through the encryption interface of the generative large model; In the process of executing the management method, set user operation restrictions based on the access rights control mechanism to ensure the privacy and integrity of the family library resources in a multi-user collaborative environment; Dynamically allocate computing and storage resources according to project requirements, including: allocating cloud storage capacity on demand to adapt to family library management for projects of different sizes; Adjust computing resource allocation according to real-time load to ensure efficient use of resources and smooth execution of management tasks.

7. The BIM family library intelligent management method based on the generative large model according to claim 1 is characterized in that: Also includes: Step S5: Obtain real-time feedback data throughout the project life cycle; Step S6: Analyze the real-time feedback data using the generative big model to generate an optimization solution.

8. The BIM family library intelligent management method based on the generative large model according to claim 7 is characterized in that: Methods for using generative big models to analyze real-time feedback data and generate optimization solutions include: Compare the actual parameters of the components in the real-time feedback data with the corresponding theoretical parameters of the components in the BIM family library. If there is an error exceeding the preset threshold, the generative model is input according to the actual parameters of the components to generate an updated semantic generative formula, and the updated semantic generative formula is input into the BIM creation system through the API to generate an updated target component through the parametric design module.

9. The BIM family library intelligent management method based on the generative large model according to claim 1 is characterized in that: The following steps are also included to achieve the collaboration and knowledge transfer of family library resources among multiple projects: Use generative big models to analyze and share cross-project data and identify common component requirements between different projects; Based on the identification results, family components with high matching degree are recommended in the new project to meet the project requirements; Summarize the component usage experience accumulated in different projects and incorporate the experience into the family library optimization strategy through the transfer learning function.

Images