A lightweight BIM model processing method and system based on AI

Abstract

This invention relates to the field of Building Information Modeling (BIM) optimization, and provides an AI-based lightweight processing method and system for BIM models. By performing multi-dimensional analysis on the geometric, semantic, business, topological, access, and operational environment characteristics of the BIM model, it automatically determines lightweight strategies for different components or component sets, achieving differentiated processing of model geometric data, attribute data, and display levels. While ensuring the integrity of key business information, it significantly reduces model loading time, rendering pressure, and network transmission volume, improving the model's visualization efficiency and business support capabilities in different scenarios.

Description

Technical Field

[0001] This invention relates to the field of building information modeling optimization, and more specifically, to an AI-based method and system for lightweight processing of BIM models. Background Technology

[0002] With the widespread application of BIM technology in transportation engineering, building engineering, municipal engineering, electromechanical operation and maintenance, and digital twin platforms, BIM models have gradually extended from the design stage to the construction stage, operation and maintenance stage, and business visualization stage. BIM models typically contain a large amount of geometric information, attribute information, material texture information, and topological relationships between components. The models are large in size, deep in hierarchy, and complex in data type.

[0003] Existing BIM models commonly suffer from the following problems when displayed on the cloud, on the web, for mobile inspection, integrated with GIS platforms, and called by business platforms: 1. The original data volume of the model is too large, and loading it directly takes a long time, which affects the efficiency of the first screen display; 2. Different components have different levels of importance, but existing lightweight solutions usually adopt a uniform strategy of reducing surface area, compressing, and eliminating components, which cannot make differentiated optimizations for different business scenarios. 3. Lightweight processing often focuses only on geometric compression, ignoring the preservation of attributes, topological relationships, and business semantics. As a result, although the model can be displayed, it cannot support subsequent retrieval, statistics, linkage, and early warning analysis.

[0004] Therefore, there is an urgent need for a technical solution that can automatically generate differentiated lightweight strategies by combining BIM model characteristics, business scenarios, terminal performance, and historical usage behavior, in order to improve model loading efficiency, rendering efficiency, and business availability. Summary of the Invention

[0005] This invention addresses the technical problems existing in the prior art by providing an AI-based lightweight BIM model processing method and system to overcome the shortcomings of existing lightweight BIM model solutions.

[0006] According to a first aspect of the present invention, an AI-based lightweight processing method for BIM models is provided, comprising: Step S1: Obtain the BIM model data to be processed and the context constraint information related to model lightweighting; Step S2: Based on the BIM model data to be processed and the context constraint information, extract multi-dimensional features of the target components or sets of target components in the BIM model to be processed. The multi-dimensional features include at least geometric complexity features, semantic features, attribute value features, topological relationship features, access behavior features, and scene environment features. Step S3: Take the multi-dimensional features of the target component or set of target components in the BIM model to be processed as feature vectors, input them into the AI ​​strategy generation model, and output a lightweight solution for the target component or set of target components. Step S4: According to the lightweighting scheme, perform differentiated lightweighting processing on the target component or set of target components to generate a lightweight BIM model.

[0007] Based on the above technical solution, the present invention can also be improved as follows.

[0008] Optionally, the BIM model data to be processed includes at least one or more of the following: component geometric data, component attribute data, component classification information, component material and texture information, component hierarchical relationship, and component spatial location and topological relationship; The context constraint information includes one or more of the following: business scenario type, user role information, terminal device performance parameters, network bandwidth and latency information, historical access popularity, current view range, key focus area, and historical rendering stuttering records.

[0009] Optionally, step S3 involves taking the multidimensional features of the target component or set of target components in the BIM model to be processed as feature vectors, inputting them into the AI ​​strategy generation model, and outputting a lightweight strategy scheme for the target component or set of target components, including: The multidimensional features of the target component or set of target components are normalized to construct the feature vector corresponding to the target component or set of target components, and the feature vector is input into the pre-trained AI policy generation model. The lightweight strategy parameters for the target component or set of target components are output based on the AI ​​strategy generation model. The lightweight strategy parameters include one or more of the following: target LOD level, geometric simplification ratio, texture resolution compression level, attribute preservation level, whether to preserve internal detailed structure, whether to use instantiation reuse, whether to perform block loading, whether to perform streaming loading, loading priority, cache priority, whether to retain only bounding box or proxy body, and whether to retain the complete accuracy of key business components. Based on the strategic parameters, a lightweight strategy scheme for the target component or set of target components is generated.

[0010] Optionally, the geometric complexity features include the number of triangles, the number of vertices, the bounding box volume, the number of surfaces, the detail density, and the geometric repetition rate; the semantic features include component category, professional affiliation, functional purpose, business tag, and risk level; the attribute value features include the number of attribute fields, key attribute identifiers, the number of business-related attributes, and query frequency; the topological relationship features include hierarchical relationships, connection relationships, adjacency relationships, attachment relationships, and association centrality; the access behavior features include user click frequency, view dwell time, number of queries, and number of linked calls; and the scene environment features include: terminal GPU capability, memory capacity, network status, display resolution, and scene target frame rate.

[0011] Optionally, the AI ​​strategy generation model may employ one or more of the following: gradient boosting tree model, deep neural network model, graph neural network model, multi-task learning model, reinforcement learning model, or a hybrid model combining rule model and machine learning model; for BIM models with complex hierarchies and topological relationships, graph neural networks are used to extract component relationship features; for the continuous optimization process of lightweight strategies, reinforcement learning models are used to update strategy parameters based on operational feedback.

[0012] Optionally, the lightweight strategy parameters for the target component or set of target components output by the AI ​​strategy generation model include: Based on the AI ​​strategy, the model generates a business importance score and a lightweight tolerance score for the target component or set of target components. Based on the business importance score and the lightweight tolerance score, the value range of each lightweight parameter is determined, and the corresponding lightweight solution is generated.

[0013] Optionally, step S4, according to the lightweighting scheme, performs differentiated lightweighting processing on the target component or set of target components, including: Perform mesh simplification on components with low importance and high complexity; Perform instantiation replacement on highly repetitive components; Perform attribute pruning or lazy loading on attributes that are accessed infrequently; Key components should retain their original or higher precision. Perform hierarchical compression on texture resources; The model is spatially partitioned and hierarchically reorganized; Set different loading priorities for different blocks.

[0014] The key components include one or more of the following: bridge main beam, bridge pier, bearing, key section of tunnel lining, key components of slope protection, and core components of electromechanical equipment.

[0015] Optionally, step S4, which involves performing differentiated lightweight processing on the target component or set of target components according to the lightweight scheme to generate a lightweight BIM model, further includes: Based on the BIM model before lightweighting, perform a consistency check on the BIM model after lightweighting. If the lightweighting result does not meet the preset error threshold or business availability threshold, return to step S5 to readjust the lightweighting strategy. The consistency verification includes one or more of the following: geometric integrity verification, topological continuity verification, attribute mapping integrity verification, component identifier uniqueness verification, business function availability verification, and visual error threshold verification.

[0016] Optionally, step S4, which involves performing differentiated lightweight processing on the target component or set of target components according to the lightweighting scheme to generate a lightweight BIM model, further includes: During the operation of the lightweight BIM model, feedback data is collected in real time, including user interaction logs, frame rate, loading time, lag locations, hot spots, and query behavior. Based on the feedback data, the AI ​​policy generation model is incrementally trained or its policy parameters are updated in a lightweight manner to continuously optimize the AI ​​policy generation model.

[0017] According to a second aspect of the present invention, an AI-based lightweight BIM model processing system is provided, comprising: The acquisition module is used to acquire the BIM model data to be processed and the context constraint information related to model lightweighting; The extraction module is used to extract multi-dimensional features of the target components or sets of target components in the BIM model to be processed based on the BIM model data to be processed and the context constraint information. The multi-dimensional features include at least geometric complexity features, semantic features, attribute value features, topological relationship features, access behavior features and scene environment features. The generation module is used to take the multi-dimensional features of the target component or set of target components in the BIM model to be processed as feature vectors, input them into the AI ​​strategy generation model, and generate a lightweight solution for the target component or set of target components. The execution module is used to perform differentiated lightweight processing on the target component or set of target components according to the lightweight scheme, and generate a lightweight BIM model.

[0018] This invention provides an AI-based lightweight BIM model processing method and system. By performing multi-dimensional analysis on the geometric, semantic, business, topological, access, and operating environment characteristics of the BIM model, it automatically determines lightweight strategies for different components or component sets, achieving differentiated processing of model geometric data, attribute data, and display levels. While ensuring the integrity of key business information, it significantly reduces model loading time, rendering pressure, and network transmission volume, improving the model's visualization efficiency and business support capabilities in different scenarios. Attached Figure Description

[0019] Figure 1 A flowchart of an AI-based lightweight BIM model processing method provided in one embodiment of the present invention; Figure 2 This is a block diagram of an AI-based lightweight BIM model processing system provided in one embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined with each other to form feasible technical solutions. Such combinations are not constrained by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0021] Addressing the shortcomings of existing BIM model lightweighting methods, this invention proposes an AI-based BIM model lightweighting processing method. First, it acquires the BIM model to be processed and related business context information, and extracts multi-dimensional features from the components in the model. Then, it uses a trained artificial intelligence model to predict the importance, display requirements, and compression tolerance of each component or set of components, automatically generating corresponding lightweighting strategies. Next, based on these lightweighting strategies, it performs differentiated geometric simplification, attribute clipping, texture compression, hierarchical reorganization, and block encapsulation on the model. Finally, it dynamically adjusts the lightweighting strategies based on rendering feedback and user interaction behavior, achieving closed-loop optimization of model lightweighting.

[0022] Figure 1The following is a flowchart illustrating an AI-based lightweight BIM model processing method according to an embodiment of the present invention. Figure 1 As shown, the method includes the following steps: Step S1: Obtain the BIM model data to be processed and the context constraint information related to model lightweighting.

[0023] Understandably, the BIM model data to be lightweighted includes at least one or more of the following: component geometric data, component attribute data, component classification information, component material and texture information, component hierarchical relationship, and component spatial location and topological relationship.

[0024] BIM models can be derived from standard or non-standard model files exported from Revit, IFC, Bentley, Navisworks, or other BIM modeling / integration platforms.

[0025] This includes acquiring contextual constraint information related to model lightweighting. Contextual constraint information includes one or more of the following: business scenario type, user role information, terminal device performance parameters, network bandwidth and latency information, historical access popularity, current viewpoint range, key areas of focus, and historical rendering stuttering records. Business scenario types may include design review, construction demonstration, mobile inspection, maintenance management, asset management, digital twin overview, emergency command, etc.

[0026] Step S2: Based on the BIM model data to be processed and the context constraint information, extract multidimensional features of the target components or sets of target components in the BIM model to be processed.

[0027] Understandably, based on the BIM model data to be lightweighted and related contextual constraint information obtained in step S1, multidimensional features of the target components or sets of target components in the BIM model are extracted. When extracting component features, multidimensional features can be extracted for each component individually. Alternatively, components can be divided into multiple groups based on component category, spatial location, functional purpose, or business scenario; multidimensional features of components within the same group are extracted; subsequently, a unified or semi-unified lightweighting strategy can be generated for components within the same group to reduce strategy fragmentation and improve lightweighting execution efficiency.

[0028] The multidimensional features include at least geometric complexity features, semantic features, attribute value features, topological relationship features, access behavior features, and scene environment features. Geometric complexity features include the number of triangles, the number of vertices, the volume of the bounding box, the number of surfaces, the detail density, and the geometric repetition rate. Semantic features include component category, professional affiliation, functional purpose, business label, and risk level. Attribute value features include the number of attribute fields, key attribute identifiers, the number of business-related attributes, and query frequency. Topological relationship features include hierarchical relationships, connection relationships, adjacency relationships, attachment relationships, and association centrality. Access behavior features include user click frequency, view dwell time, number of queries, and number of linked calls. Scene environment features include: terminal GPU capabilities, memory capacity, network status, display resolution, and scene target frame rate.

[0029] Step S3: Take the multi-dimensional features of the target component or set of target components in the BIM model to be processed as feature vectors, input them into the AI ​​strategy generation model, and output a lightweight solution for the target component or set of target components.

[0030] Specifically, the multidimensional features of each extracted component are normalized to construct a feature vector corresponding to each component or set of components, which is then input into a pre-trained AI policy generation model.

[0031] AI policy generation models can employ one or more of the following: gradient boosting tree model, deep neural network model, graph neural network model, multi-task learning model, reinforcement learning model, and hybrid model combining rule-based model and machine learning model.

[0032] Preferably, for BIM models with complex hierarchies and topological relationships, graph neural networks are used to extract component relationship features; for the continuous optimization process of the lightweight strategy, a reinforcement learning model is used to update the lightweight strategy parameters based on operational feedback.

[0033] The AI ​​strategy generation model outputs lightweight strategy parameters for the target component based on the multidimensional features of the target component or set of target components. The lightweight strategy parameters include one or more of the following: target LOD level, geometric simplification ratio, texture resolution compression level, attribute preservation level, whether to retain internal detailed structure, whether to use instantiation reuse, whether to perform block loading, whether to perform streaming loading, loading priority, cache priority, whether to retain only the bounding box or proxy body, and whether to retain the complete accuracy of key business components.

[0034] The AI ​​strategy generation module outputs a "business importance score" and a "lightweight tolerance score" for each component based on its multidimensional characteristics. It then determines the range of values ​​for the lightweight strategy parameters of the target component based on these scores, thereby generating the corresponding lightweight solution.

[0035] During the generation of lightweight solutions, industry rules can be introduced to perform secondary corrections. For example: safety-critical components must not be reduced to the lowest LOD; components that need to participate in business statistics must retain core attributes; components that participate in early warning analysis must retain unique codes and location attributes. Furthermore, during the generation of lightweight strategies, the following objectives should be optimized simultaneously: minimizing model size, minimizing initial loading time, maximizing the retention of business attributes, minimizing visual errors, and maximizing target frame rate stability.

[0036] Step S4: According to the lightweighting scheme, perform differentiated lightweighting processing on the target component or set of target components to generate a lightweight BIM model.

[0037] Specifically, based on the generated lightweight scheme, differentiated lightweight processing is performed on the target components. This differentiated processing includes one or more of the following: performing mesh simplification on low-importance and high-complexity components; performing instantiation replacement on highly repetitive components; performing attribute pruning or deferred loading on low-access-frequency attributes; preserving the original or higher precision of key components; performing hierarchical compression on texture resources; performing spatial partitioning and hierarchical reorganization on the model; and setting different loading priorities for different partitions.

[0038] Key components may include bridge main beams, piers, bearings, key sections of tunnel lining, key components of slope protection, and core components of electromechanical equipment.

[0039] After lightweighting the target components of the BIM model based on the lightweighting scheme, a lightweight BIM model is generated. The lightweight BIM model is then compared with the original BIM model to perform a consistency check. This consistency check includes: geometric integrity check, topological continuity check, attribute mapping integrity check, component identifier uniqueness check, business function availability check, and visual error threshold check. If the lightweighting result does not meet the preset error threshold or business availability threshold, the process returns to step S3 to readjust the lightweighting strategy.

[0040] After consistency verification, the lightweight model file and its associated index data are output, mainly including the lightweight model file, attribute mapping table, block index file, LOD mapping file, view priority configuration, and dynamic loading strategy configuration file.

[0041] The output results can be directly loaded by web-based systems, GIS platforms, digital twin platforms, mobile inspection systems, or business management systems.

[0042] It should be noted that during the operation of the lightweight BIM model, feedback data such as user interaction logs, frame rate, loading time, lag locations, hot spots, and query behavior are collected in real time. Based on the feedback data, the AI ​​strategy generation model is incrementally trained or its parameters are updated to continuously optimize the subsequent lightweight model effect.

[0043] In practical BIM applications, the system can predict the key visible areas at the next moment based on the user's current perspective, camera path, historical roaming trajectory, and hotspot areas, and preload high-priority lightweight models for components that are about to enter the visible area.

[0044] Lightweight processing for digital twin platforms of highway bridges is implemented, taking a bridge BIM model as an example. The original model includes components such as piers, cap beams, main beams, supports, crash barriers, ancillary facilities, maintenance access, and electromechanical equipment. After acquiring the model, the system extracts the geometric complexity, number of attributes, business importance, and access frequency of each component.

[0045] Among them, the main beams, piers, and bearings of the bridge are marked as high-importance components, the crash barriers and ancillary facilities are marked as medium-importance components, and some small ancillary components with high repetition are marked as low-importance components.

[0046] Based on the business scenario of "maintenance management + comprehensive display", the AI ​​model outputs the following strategies: retain high-precision geometry and all key attributes for the main beam, pier, and bearing of the bridge; retain medium-precision geometry and basic attributes for the crash barrier; replace repetitive accessories with instantiated proxy models; adopt an on-demand loading strategy for internal detail components that do not participate in the current display task; and perform medium-level compression on textures.

[0047] After processing, the overall size of the bridge model is significantly reduced, improving the loading efficiency of the first screen, while the key defect analysis and asset location functions remain unaffected.

[0048] It should be noted that in this embodiment of the invention, the AI ​​policy generation model can be replaced with different types of supervised learning models or reinforcement learning models. The lightweight object can be a single component, a set of components, a floor partition, a spatial block, or a professional layer, and the component category and business rules can be replaced according to different scenarios such as transportation, architecture, municipal administration, water conservancy, and industrial parks. The format of the lightweight output result can be adapted to the target platform as glTF, 3D Tiles, OBJ, a proprietary lightweight format, or a combination thereof; and the consistency verification of the lightweight model can be performed using either a graphical error evaluation method or a business rule evaluation method.

[0049] Figure 2 This invention illustrates an AI-based lightweight BIM model processing system according to an embodiment of the present invention. The system includes: The acquisition module 201 is used to acquire the BIM model data to be processed and the context constraint information related to model lightweighting; Extraction module 202 is used to extract multi-dimensional features of target components or sets of target components in the BIM model to be processed based on the BIM model data to be processed and the context constraint information. The multi-dimensional features include at least geometric complexity features, semantic features, attribute value features, topological relationship features, access behavior features and scene environment features. The generation module 203 is used to take the multi-dimensional features of the target component or set of target components in the BIM model to be processed as feature vectors, input them into the AI ​​strategy generation model, and generate a lightweight solution for the target component or set of target components. The execution module 204 is used to perform differentiated lightweight processing on the target component or set of target components according to the lightweight scheme, and generate a lightweight BIM model.

[0050] It is understood that the AI-based lightweight BIM model processing system provided by the present invention corresponds to the AI-based lightweight BIM model processing method provided in the foregoing embodiments. The relevant technical features of the AI-based lightweight BIM model processing system can be referred to the relevant technical features of the AI-based lightweight BIM model processing method, and will not be repeated here.

[0051] The present invention provides an AI-based lightweight BIM model processing method and system, which has the following advantages compared with the prior art: (1) Achieve differentiated lightweighting: Unlike the traditional unified reduction and uniform trimming methods, this invention can automatically generate differentiated lightweighting strategies based on the importance of components, business value and access popularity, avoiding the problem of "over-compressing key components and retaining too many irrelevant components".

[0052] (2) Simultaneously consider geometry, attributes and business capabilities: not only optimize the geometric model, but also retain and index attribute data, topological relationships and business tags in a hierarchical manner, so that the lightweight model still has query, statistics, linkage and early warning capabilities.

[0053] (3) Enhance adaptability to different terminal and network environments: By introducing terminal performance parameters and network status as feature inputs, the present invention can automatically generate different lightweight solutions that are adapted to PC, mobile, weak network environment and large screen scenario.

[0054] (4) Support for continuous learning and closed-loop optimization: By introducing rendering feedback and user behavior data, this invention can continuously optimize the lightweight strategy and avoid the long-term failure of static solutions.

[0055] (5) Improve the overall performance of the platform: While ensuring the key business capabilities, it can significantly reduce the amount of model data transmission, shorten the first loading time of the model, and improve the page response speed and scene rendering smoothness.

[0056] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0057] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0058] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0059] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0060] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0061] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0062] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A lightweight BIM model processing method based on AI, characterized in that, include: Step S1: Obtain the BIM model data to be processed and the context constraint information related to model lightweighting; Step S2: Based on the BIM model data to be processed and the context constraint information, extract multi-dimensional features of the target components or sets of target components in the BIM model to be processed. The multi-dimensional features include at least geometric complexity features, semantic features, attribute value features, topological relationship features, access behavior features, and scene environment features. Step S3: Take the multi-dimensional features of the target component or set of target components in the BIM model to be processed as feature vectors, input them into the AI ​​strategy generation model, and output a lightweight solution for the target component or set of target components. Step S4: According to the lightweighting scheme, perform differentiated lightweighting processing on the target component or set of target components to generate a lightweight BIM model.

2. The AI-based lightweight BIM model processing method according to claim 1, characterized in that, The BIM model data to be processed includes at least one or more of the following: component geometric data, component attribute data, component classification information, component material and texture information, component hierarchical relationship, and component spatial location and topological relationship. The context constraint information includes one or more of the following: business scenario type, user role information, terminal device performance parameters, network bandwidth and latency information, historical access popularity, current view range, key focus area, and historical rendering stuttering records.

3. The AI-based lightweight BIM model processing method according to claim 1, characterized in that, Step S3 involves taking the multidimensional features of the target component or set of target components in the BIM model to be processed as feature vectors, inputting them into the AI ​​strategy generation model, and outputting a lightweight strategy scheme for the target component or set of target components, including: The multidimensional features of the target component or set of target components are normalized to construct the feature vector corresponding to the target component or set of target components, and the feature vector is input into the pre-trained AI policy generation model. The lightweight strategy parameters for the target component or set of target components are output based on the AI ​​strategy generation model. The lightweight strategy parameters include one or more of the following: target LOD level, geometric simplification ratio, texture resolution compression level, attribute preservation level, whether to preserve internal detailed structure, whether to use instantiation reuse, whether to perform block loading, whether to perform streaming loading, loading priority, cache priority, whether to retain only bounding box or proxy body, and whether to retain the complete accuracy of key business components. Based on the strategic parameters, a lightweight strategy scheme for the target component or set of target components is generated.

4. The AI-based lightweight BIM model processing method according to claim 1 or 3, characterized in that, The geometric complexity features include the number of triangles, the number of vertices, the volume of the bounding box, the number of surfaces, the detail density, and the geometric repetition rate; the semantic features include component category, professional affiliation, functional purpose, business label, and risk level. The attribute value characteristics include the number of attribute fields, key attribute identifiers, number of business-related attributes, and query frequency; The topological relationship features include hierarchical relationships, connection relationships, adjacency relationships, attachment relationships, and association centrality. The access behavior characteristics include user click frequency, viewing duration, number of queries, and number of linked calls; The scene environment features include: terminal GPU capabilities, memory capacity, network status, display resolution, and scene target frame rate.

5. The AI-based lightweight BIM model processing method according to claim 1, characterized in that, The AI ​​strategy generation model adopts one or more of the following: gradient boosting tree model, deep neural network model, graph neural network model, multi-task learning model, reinforcement learning model, and a hybrid model combining rule model and machine learning model; for BIM models with complex hierarchies and topological relationships, graph neural networks are used to extract component relationship features; for the continuous optimization process of lightweight strategy, reinforcement learning model is used to update strategy parameters based on operational feedback.

6. The AI-based lightweight BIM model processing method according to claim 3, characterized in that, The lightweight strategy parameters for the target component or set of target components are output by the AI ​​strategy generation model, including: Based on the AI ​​strategy, the model generates a business importance score and a lightweight tolerance score for the target component or set of target components. Based on the business importance score and the lightweight tolerance score, the value range of each lightweight parameter is determined, and the corresponding lightweight solution is generated.

7. The AI-based lightweight BIM model processing method according to claim 1, characterized in that, Step S4, according to the lightweighting scheme, performs differentiated lightweighting processing on the target component or set of target components, including: Perform mesh simplification on components with low importance and high complexity; Perform instantiation replacement on highly repetitive components; Perform attribute pruning or lazy loading on attributes that are accessed infrequently; Key components should retain their original or higher precision. Perform hierarchical compression on texture resources; The model is spatially partitioned and hierarchically reorganized; Set different loading priorities for different blocks; The key components include one or more of the following: bridge main beam, bridge pier, bearing, key section of tunnel lining, key components of slope protection, and core components of electromechanical equipment.

8. The AI-based lightweight BIM model processing method according to claim 1, characterized in that, Step S4 involves performing differentiated lightweight processing on the target component or set of target components according to the lightweight scheme to generate a lightweight BIM model, and then further includes: Based on the BIM model before lightweighting, perform a consistency check on the BIM model after lightweighting. If the lightweighting result does not meet the preset error threshold or business availability threshold, return to step S5 to readjust the lightweighting strategy. The consistency verification includes one or more of the following: geometric integrity verification, topological continuity verification, attribute mapping integrity verification, component identifier uniqueness verification, business function availability verification, and visual error threshold verification.

9. The AI-based lightweight BIM model processing method according to claim 1, wherein step S4, according to the lightweight scheme, performs differentiated lightweight processing on the target component or set of target components to generate a lightweight BIM model, further includes: During the operation of the lightweight BIM model, feedback data is collected in real time, including user interaction logs, frame rate, loading time, lag locations, hot spots, and query behavior. Based on the feedback data, the AI ​​policy generation model is incrementally trained or its policy parameters are updated in a lightweight manner to continuously optimize the AI ​​policy generation model.

10. A lightweight BIM model processing system based on AI, characterized in that, include: The acquisition module is used to acquire the BIM model data to be processed and the context constraint information related to model lightweighting; The extraction module is used to extract multi-dimensional features of the target components or sets of target components in the BIM model to be processed based on the BIM model data to be processed and the context constraint information. The multi-dimensional features include at least geometric complexity features, semantic features, attribute value features, topological relationship features, access behavior features and scene environment features. The generation module is used to take the multi-dimensional features of the target component or set of target components in the BIM model to be processed as feature vectors, input them into the AI ​​strategy generation model, and generate a lightweight solution for the target component or set of target components. The execution module is used to perform differentiated lightweight processing on the target component or set of target components according to the lightweight scheme, and generate a lightweight BIM model.

Images