Building engineering data processing system based on BIM

Abstract

The invention provides a BIM-based building engineering data processing system, and relates to the technical field of engineering data processing. According to the BIM building engineering data processing system, all modules in an existing BIM building engineering data processing system are optimized, and the performance and stability of the system can be remarkably improved on the premise that the model precision is guaranteed. Compared with the prior art, the method has the advantages that component priority adjustment is more accurate, actual use requirements are met, geometric simplification effects are more balanced, accuracy of key design elements can be reserved, material rendering is more stable and efficient, adaptability is high, multi-layer grid division is more dynamic and intelligent, user requirements can be met in real time, and interaction experience is improved.

Description

Technical Field

[0001] This invention belongs to the field of engineering data processing technology, and specifically relates to a BIM-based building engineering data processing system. Background Technology

[0002] With the widespread application of Building Information Modeling (BIM) technology in the field of building engineering, the complexity and data volume of BIM models are increasing day by day. Traditional data processing methods are no longer sufficient to meet the needs of efficiently processing large-scale BIM models. Existing technologies have the following problems: insufficient component identification accuracy in BIM models, making it difficult to accurately distinguish and identify the relationships between complex components, resulting in poor subsequent processing effects; difficulty in maintaining the accuracy of key structures during geometric simplification while effectively reducing the computational complexity of non-critical areas; lack of intelligent adjustment strategies in the material rendering process, failing to automatically optimize rendering parameters according to different hardware performance and system load; and insufficient flexibility in mesh generation methods, making it difficult to dynamically adjust mesh density based on user interaction behavior and viewpoint changes.

[0003] Therefore, there is an urgent need to develop a system that can efficiently process BIM model data to solve the above problems and improve the efficiency and user experience of BIM applications. Summary of the Invention

[0004] This invention provides a BIM-based building engineering data processing system, the technical solution of which includes the following aspects.

[0005] According to one aspect of the embodiments of this application, a BIM-based building engineering data processing system is provided. The system includes: a model import module, a geometry simplification module, an automatic material rendering module, and a multi-layer mesh generation module. The model import module accurately identifies various components in the BIM model by combining deep learning algorithms of convolutional neural networks and graph convolutional networks, and dynamically adjusts the priority of key components based on an adaptive scene prediction model. The adaptive scene prediction model can perform joint analysis based on real-time environmental data and historical data to ensure that the priority adjustment of key components better meets actual usage needs, thereby improving the accuracy of component identification and priority adjustment.

[0006] Furthermore, the deep learning algorithm of the model import module further includes using graph convolutional networks to model spatial relationships, accurately modeling the relationships between components of complex structures, so as to improve the accuracy of component identification and the rationality of priority adjustment.

[0007] Furthermore, the geometry simplification module introduces a multi-scale generative adversarial network structure, in which both the generator and discriminator adopt a multi-scale design, enabling geometric simplification of the BIM model at multiple scale levels. This optimized structure can effectively preserve the geometric accuracy of key design elements and reduce details in non-critical areas to lower computational complexity. Through an adaptive accuracy adjustment mechanism, different degrees of simplification are automatically selected based on the importance of components, user needs, and hardware performance, achieving a balance between model simplification and accuracy.

[0008] Furthermore, the multi-scale generative adversarial network of the geometry simplification module includes at least two scale levels, where the lower scale is used to process relatively coarse geometric features, while the higher scale retains key design elements with rich details. The accuracy of key structural features is ensured through layer-by-layer optimization, while non-critical regions are simplified to lower-precision models.

[0009] Furthermore, the geometry simplification module further includes a component importance assessment mechanism that prioritizes the retention of accuracy for key components, while further simplifying areas that do not affect the final architectural effect, thereby optimizing the overall geometry simplification process and reducing computational complexity.

[0010] Furthermore, the automatic material rendering module adopts a rendering strategy based on deep reinforcement learning to dynamically adjust the texture quality and rendering parameters of each component. The deep reinforcement learning algorithm performs real-time optimization based on the priority of different components, ambient lighting conditions, and rendering hardware performance to ensure the best balance between high-quality rendering effects and system performance. In addition, the algorithm can intelligently select a suitable rendering strategy to avoid system rendering bottlenecks when the performance load is high.

[0011] Furthermore, the deep reinforcement learning algorithm of the automatic material rendering module automatically selects the optimal texture quality and rendering parameters in different rendering environments by setting multiple rendering strategies to balance rendering quality and system performance; under low-performance devices or high load conditions, it automatically reduces texture quality and rendering accuracy to improve the stability of system operation.

[0012] Furthermore, the optimization strategy of the automatic material rendering module can adjust the material allocation of components according to the real-time changes of the rendering scene, and select an appropriate rendering algorithm according to the actual performance of the hardware device, so as to ensure good compatibility and performance on different hardware platforms.

[0013] Furthermore, the multi-layer mesh partitioning module introduces an intelligent mesh adjustment mechanism based on user interaction behavior and perspective changes. This mechanism dynamically adjusts the hierarchical structure and density of mesh details according to the user's current perspective, interaction frequency, hardware performance, and system load. The multi-layer mesh partitioning module uses a load-balancing-based mesh optimization algorithm to adjust the mesh density according to the real-time system load and the real-time nature of user interaction, reducing unnecessary computation and ensuring a smooth interactive experience even under high system load. It also dynamically adjusts mesh details through an adaptive algorithm to avoid performance bottlenecks caused by over-computation.

[0014] Compared with the prior art, this application has the following beneficial effects.

[0015] This invention optimizes various modules in existing BIM building engineering data processing systems, significantly improving system performance and stability while maintaining model accuracy. Specifically, this manifests in: 1) more accurate component priority adjustment, meeting actual usage needs; 2) more balanced geometric simplification, preserving the accuracy of key design elements; 3) more stable and efficient material rendering with strong adaptability; and 4) more dynamic and intelligent multi-layer mesh generation, adapting to user needs in real time and enhancing the interactive experience. Attached Figure Description

[0016] Figure 1 This application provides a structural diagram of a BIM-based building engineering data processing system as one embodiment. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0018] Please refer to Figure 1 This illustrates a BIM-based building engineering data processing system provided in one embodiment of this application. The technical solution of the present invention is: a BIM-based building engineering data processing system, including a model import module, a geometry simplification module, an automatic material rendering module, and a multi-layer mesh generation module.

[0019] The model import module accurately identifies various components in the BIM model by combining deep learning algorithms of convolutional neural networks and graph convolutional networks. Based on an adaptive scene prediction model, it dynamically adjusts the priority of key components. The adaptive scene prediction model performs joint analysis of real-time environmental data and historical data to ensure that the priority adjustment of key components better meets actual usage needs, improving the accuracy of component identification and priority adjustment. The deep learning algorithm in the model import module further includes spatial relationship modeling using graph convolutional networks to accurately model the relationships between components in complex structures, thereby improving the accuracy of component identification and the rationality of priority adjustment.

[0020] The geometry simplification module introduces a multi-scale generative adversarial network (GAN) structure, where both the generator and discriminator employ multi-scale design, enabling geometric simplification of the BIM model at multiple scale levels. This optimized structure effectively preserves the geometric accuracy of key design elements while reducing details in non-critical areas, thereby lowering computational complexity. Furthermore, an adaptive accuracy adjustment mechanism automatically selects different simplification levels based on component importance, user needs, and hardware performance, achieving a balance between model simplification and accuracy. The multi-scale GAN of the geometry simplification module comprises at least two scale levels. Lower scales handle coarser geometric features, while higher scales preserve detailed key design elements. Layer-by-layer optimization ensures the accuracy of key structural features, while non-critical areas are simplified to lower-precision models. The geometry simplification module further includes a component importance assessment mechanism that prioritizes the preservation of accuracy for key components, while further simplifying areas that do not affect the final architectural effect, thus optimizing the overall geometry simplification process and reducing computational complexity.

[0021] The automatic material rendering module employs a deep reinforcement learning-based rendering strategy to dynamically adjust the texture quality and rendering parameters of each component. The deep reinforcement learning algorithm optimizes in real-time based on the priority of different components, ambient lighting conditions, and rendering hardware performance to ensure an optimal balance between high-quality rendering and system performance. Furthermore, the algorithm intelligently selects suitable rendering strategies to avoid system rendering bottlenecks under high performance loads. The deep reinforcement learning algorithm in the automatic material rendering module automatically selects the optimal texture quality and rendering parameters in different rendering environments by setting multiple rendering strategies to balance rendering quality and system performance. Under low-performance devices or high loads, it automatically reduces texture quality and rendering precision to improve system stability. The optimization strategy of the automatic material rendering module can adjust the material allocation of components according to real-time changes in the rendering scene and select appropriate rendering algorithms based on the actual performance of the hardware, ensuring good compatibility and performance across different hardware platforms.

[0022] The multi-layer meshing module introduces an intelligent mesh adjustment mechanism based on user interaction behavior and perspective changes. This mechanism dynamically adjusts the hierarchical structure and density of mesh details according to the user's current perspective, interaction frequency, hardware performance, and system load. The multi-layer meshing module uses a load-balancing-based mesh optimization algorithm to adjust mesh density according to real-time system load and the real-time nature of user interaction, reducing unnecessary computation and ensuring a smooth interactive experience even under high system load. Furthermore, it dynamically adjusts mesh details through an adaptive algorithm to avoid performance bottlenecks caused by over-computation.

[0023] Example 1, such as Figure 1 As shown, the BIM-based building engineering data processing system provided by this invention includes a model import module, a geometry simplification module, an automatic material rendering module, and a multi-layer mesh generation module. The various modules of the system work collaboratively through data flow to jointly complete the BIM model processing task.

[0024] The model import module uses deep learning algorithms combining convolutional neural networks and graph convolutional networks to identify various components in the BIM model. The convolutional neural network is mainly responsible for extracting local features of the components, while the graph convolutional network is used to construct a spatial relationship model between the components.

[0025] The specific implementation process is as follows.

[0026] First, the BIM model is converted into a representation that can be processed by deep learning algorithms, including geometric features, topological relationships, and semantic information.

[0027] Convolutional neural networks extract features from each component to obtain local feature representations of the component.

[0028] Graph convolutional networks construct graph structures and model spatial relationships based on the topological relationships between components.

[0029] The adaptive scenario prediction model performs joint analysis of real-time environmental data and historical data to dynamically adjust the priority of key components.

[0030] Output the identification results and the component information after priority adjustment for use by subsequent modules.

[0031] The adaptive scenario prediction model employs a recurrent neural network structure, enabling it to process time-series data and capture historical usage patterns. By analyzing user interaction behavior, component usage frequency, and system performance data, the model predicts the importance of components in different scenarios, thereby dynamically adjusting component priorities.

[0032] The geometry simplification module introduces a multi-scale generative adversarial network structure to perform geometric simplification on the BIM model. This network consists of two parts: a generator and a discriminator, both of which employ a multi-scale design.

[0033] The generator is responsible for generating the simplified geometric model, the structure of which includes...

[0034] Encoder: Encodes the original geometric model into a latent representation.

[0035] Multi-scale feature extractor: Extracts geometric features at different scales.

[0036] Decoder: Generates a simplified geometric model based on the latent representation and multi-scale features.

[0037] The discriminator is responsible for evaluating the quality of the simplified model, and its structure includes...

[0038] Multi-scale feature comparator: compares the feature differences between the original model and the simplified model at different scales.

[0039] Geometric Fidelity Evaluator: Evaluates the extent to which a simplified model retains key geometric features.

[0040] Computational complexity estimator: Evaluates the reduction in computational complexity of the simplified model.

[0041] The multi-scale generative adversarial network for the geometry simplification module comprises at least two scale levels: lower scales handle coarse geometric features, while higher scales preserve detailed key design elements. Through layer-by-layer optimization, the accuracy of key structural features is ensured, while non-critical regions are simplified to lower-precision models.

[0042] In addition, the geometry simplification module also includes a component importance assessment mechanism, which evaluates component importance based on the following factors.

[0043] The functional role of components in building structures.

[0044] The visibility of the components and their impact on the appearance.

[0045] The frequency of component usage during the interaction process.

[0046] The degree of connection between the component and other key components.

[0047] Based on the evaluation results, the accuracy of key components is retained first, while areas that do not affect the final architectural effect are further simplified, thereby optimizing the effect of the entire geometric simplification process and reducing computational complexity.

[0048] The automatic material rendering module adopts a rendering strategy based on deep reinforcement learning to dynamically adjust the texture quality and rendering parameters of each component.

[0049] Deep reinforcement learning frameworks include...

[0050] State space: includes information such as component priority, ambient lighting conditions, rendering hardware performance, and system load.

[0051] Action space: includes operations such as texture quality level, rendering parameter adjustment, and rendering algorithm selection.

[0052] Reward function: comprehensively considers factors such as rendering quality, system resource consumption, and rendering performance.

[0053] Policy network: Selects the optimal rendering strategy based on the current state.

[0054] Deep reinforcement learning algorithms automatically select the optimal texture quality and rendering parameters under different rendering environments by setting multiple rendering strategies. On low-performance devices or under high load, they automatically reduce texture quality and rendering precision to improve system stability.

[0055] The automatic material rendering module's optimization strategy adjusts the material allocation of components based on real-time changes in the rendering scene. For example, higher-quality materials are assigned to components in the visual focal area, while simplified materials are used for areas far from the viewpoint or occluded. Simultaneously, appropriate rendering algorithms are selected based on the actual performance of the hardware devices; for instance, physically based rendering (PBR) is used on high-performance devices, while simplified lighting models are used on low-performance devices, ensuring good compatibility and performance across different hardware platforms.

[0056] The multi-layer meshing module introduces an intelligent mesh adjustment mechanism based on user interaction behavior and perspective changes. It dynamically adjusts the hierarchical structure and density of mesh details according to the user's current perspective, interaction frequency, hardware performance, and system load.

[0057] The multi-layer mesh generation algorithm includes the following steps.

[0058] Initial mesh generation: Generate an initial multi-level mesh structure based on the BIM model.

[0059] Interaction behavior analysis: Collect and analyze user interaction behavior data, including viewpoint movement, operation frequency, etc.

[0060] System load monitoring: Real-time monitoring of system hardware performance and load.

[0061] Dynamic mesh density adjustment: The level of detail in the mesh is dynamically adjusted based on the analysis results.

[0062] This module employs a load-balancing-based mesh optimization algorithm, adjusting mesh density based on real-time system load and the real-time nature of user interaction. Under high system load, mesh density is reduced in non-critical areas to minimize unnecessary computation; in areas with frequent user interaction, mesh density is increased to improve accuracy. An adaptive algorithm dynamically adjusts mesh details to avoid performance bottlenecks caused by over-computation, ensuring a smooth user experience under various load conditions.

[0063] Example 2, based on Example 1, further optimizes the implementation methods of each module.

[0064] The graph convolutional network in the model import module employs an improved attention mechanism that more accurately captures spatial dependencies between components. This mechanism improves the accuracy of spatial relationship modeling by calculating attention weights between components and focusing on neighboring components that are highly correlated with the current component.

[0065] The multi-scale generative adversarial network in the geometry simplification module adds a conditional control mechanism, allowing users to set constraints on the degree of simplification according to actual needs. Simultaneously, a geometric feature preservation loss function is introduced to ensure the preservation of key geometric features during the simplification process.

[0066] The automatic material rendering module employs a deep reinforcement learning algorithm based on meta-learning, enabling it to quickly adapt to new rendering environments and tasks. This algorithm accumulates experience and knowledge through pre-training in various rendering scenarios, allowing it to quickly find the optimal rendering strategy when faced with new rendering tasks.

[0067] The multi-layer mesh generation module introduces an adaptive mesh generation method based on user-interested regions. This method analyzes the areas of user focus through eye tracking or mouse movement trajectory analysis, and performs finer mesh generation on these areas, improving the rendering quality and accuracy of the user-focused regions.

Claims

1. A BIM-based building engineering data processing system, characterized in that, The system includes: a model import module, a geometry simplification module, an automatic material rendering module, and a multi-layer mesh generation module; The model import module accurately identifies various components in the BIM model by combining deep learning algorithms of convolutional neural networks and graph convolutional networks, and dynamically adjusts the priority of key components based on an adaptive scene prediction model. The adaptive scene prediction model can perform joint analysis based on real-time environmental data and historical data to ensure that the priority adjustment of key components is more in line with actual usage needs, thereby improving the accuracy of component identification and priority adjustment.

2. The BIM-based building engineering data processing system according to claim 1, characterized in that, The deep learning algorithm in the model import module further includes using graph convolutional networks to model spatial relationships, accurately modeling the relationships between components of complex structures, so as to improve the accuracy of component identification and the rationality of priority adjustment.

3. The BIM-based building engineering data processing system according to claim 1, characterized in that, The geometry simplification module introduces a multi-scale generative adversarial network structure, in which both the generator and discriminator adopt a multi-scale design, enabling geometric simplification of the BIM model at multiple scale levels. This optimized structure can effectively preserve the geometric accuracy of key design elements and reduce details in non-critical areas to reduce computational complexity. Furthermore, through an adaptive accuracy adjustment mechanism, different degrees of simplification are automatically selected based on the importance of components, user needs, and hardware performance, achieving a balance between model simplification and accuracy.

4. A BIM-based building engineering data processing system according to claim 3, characterized in that, The multi-scale generative adversarial network of the geometry simplification module includes at least two scale levels, where the lower scale is used to process relatively coarse geometric features, while the higher scale retains key design elements with rich details. The accuracy of key structural features is ensured through layer-by-layer optimization, while non-critical regions are simplified to a lower-precision model.

5. A BIM-based building engineering data processing system according to claim 3, characterized in that, The geometry simplification module further includes a component importance assessment mechanism that prioritizes the retention of accuracy for key components, while further simplification is applied to areas that do not affect the final architectural effect, thereby optimizing the overall geometry simplification process and reducing computational complexity.

6. A BIM-based building engineering data processing system according to claim 1, characterized in that, The automatic material rendering module adopts a rendering strategy based on deep reinforcement learning to dynamically adjust the texture quality and rendering parameters of each component. The deep reinforcement learning algorithm performs real-time optimization based on the priority of different components, ambient lighting conditions, and rendering hardware performance to ensure the best balance between high-quality rendering effects and system performance. Furthermore, the algorithm can intelligently select a suitable rendering strategy to avoid system rendering bottlenecks when the performance load is high.

7. A BIM-based building engineering data processing system according to claim 6, characterized in that, The deep reinforcement learning algorithm of the automatic material rendering module automatically selects the optimal texture quality and rendering parameters in different rendering environments by setting multiple rendering strategies to balance rendering quality and system performance; under low-performance devices or high load conditions, it automatically reduces texture quality and rendering accuracy to improve the stability of system operation.

8. A BIM-based building engineering data processing system according to claim 6, characterized in that, The optimization strategy of the automatic material rendering module can adjust the material allocation of components according to the real-time changes of the rendering scene, and select an appropriate rendering algorithm according to the actual performance of the hardware device, so as to ensure good compatibility and performance on different hardware platforms.

9. A BIM-based building engineering data processing system according to claim 1, characterized in that, The multi-layer mesh partitioning module introduces an intelligent mesh adjustment mechanism based on user interaction behavior and perspective changes. This mechanism dynamically adjusts the hierarchical structure and density of mesh details according to the user's current perspective, interaction frequency, hardware performance, and system load. The multi-layer mesh partitioning module uses a load-balancing-based mesh optimization algorithm to adjust the mesh density according to the real-time system load and the real-time nature of user interaction, reducing unnecessary computation and ensuring a smooth interactive experience even under high system load. It also dynamically adjusts mesh details through an adaptive algorithm to avoid performance bottlenecks caused by over-computation.

Images