Curtain wall space curved surface design method and system

Abstract

The invention relates to the field of building design and engineering, in particular to a curtain wall space curved surface design method and system, and the method comprises the steps: historical project data collection and feature extraction, application of a convolutional neural network to generate a preliminary curved surface, generative adversarial network optimization curved surface data, and performance evaluation through finite element analysis. And by introducing user feedback data, the design is continuously optimized, so that the finally confirmed curved surface design reaches high standards in aesthetic property and functionality. According to the method, the problems of low efficiency and inconsistent design of a traditional design method are effectively solved, and a scientific and flexible solution is provided for curtain wall design.

Description

Technical Field

[0001] The present invention relates to the field of architectural design and engineering, and in particular to a curtain wall space curved surface design method and system. Background Art

[0002] As a key component of the building facade, the curtain wall not only affects the aesthetics of the building, but also has a profound impact on structural safety, energy efficiency and environmental adaptability. However, the existing technology (Chinese invention patent, publication number: CN116049951A, name: Architectural curtain wall spatial surface optimization design method and system based on BIM technology) often relies on traditional two-dimensional design methods and manual parameter adjustment in curtain wall design, which leads to low design efficiency, unpredictable effects, and inability to fully utilize the advantages of computer-aided design. These traditional technologies usually include experience-based design methods and static geometric calculation methods, which make designers face challenges when dealing with complex and special-shaped surfaces. Due to the lack of flexible design tools, the final design often fails to achieve the expected aesthetics and functionality, resulting in material waste and construction difficulties. Summary of the invention

[0003] In view of the many problems existing in the above-mentioned prior art, the present invention provides a curtain wall space curved surface design method and system. The present invention uses a data-driven method to extract design features using a convolutional neural network and generates a high-quality curved surface model through a generative adversarial network. The performance evaluation data is analyzed using a machine learning regression model to achieve optimized design. The present invention not only improves the design efficiency, but also ensures the structural safety and aesthetics of the curved surface.

[0004] The curtain wall space surface design method includes the following steps:

[0005] Collect historical project data, including design parameters, material properties, and construction difficulty, generate cleaned project data through data sorting and cleaning, and use principal component analysis to extract features from the cleaned project data to generate design feature data;

[0006] Use convolutional neural networks to generate preliminary surface data based on design feature data, and use generative adversarial networks to generate adversarial surface data, and finally generate adversarial generated surface data; use shape optimization algorithms to perform real-time shape optimization on adversarial generated surface data to generate optimized surface data;

[0007] Inputting the optimized surface data into the finite element analysis software for performance evaluation, generating performance evaluation data, and analyzing the performance evaluation data to identify feedback parameter data; using a machine learning regression model to analyze the feedback parameter data to adjust the optimized surface data, generating adjusted surface data, and performing multi-objective optimization to generate final optimized design data;

[0008] Import the final optimized design data into the virtual reality platform, build virtual scene data, conduct user experience testing, collect user feedback data, and make final adjustments based on user feedback data to confirm the final design data.

[0009] Preferably, the design feature data includes key parameters that affect the surface design, specifically surface thickness, material properties, construction difficulty and design style, and the design feature data is generated by performing principal component analysis on the cleaned project data.

[0010] Preferably, the structure of the convolutional neural network includes multiple convolutional layers, activation layers and fully connected layers, each convolutional layer is used to extract spatial features in the design feature data, and the preliminary surface data output by the convolutional neural network after training contains preliminary generated surface model information.

[0011] Preferably, the generator of the generative adversarial network receives preliminary surface data and generates adversarial generated surface data, and the discriminator of the generative adversarial network is used to evaluate the quality of the generated adversarial generated surface data.

[0012] Preferably, the shape optimization algorithm performs real-time shape optimization on the generated surface data. The optimization process includes topological optimization of the generated surface to minimize the use of materials and improve the structural strength of the surface, and generate optimized surface data. The goal of the topological optimization is to minimize the material stress σ, and the calculation formula is:

[0013]

[0014] Among them, σ represents material stress; F represents the applied force; and A represents the cross-sectional area.

[0015] Preferably, the finite element analysis software is used to perform performance evaluation on the optimized surface data, and the performance evaluation data includes stress distribution, deformation and safety factor, and the safety factor calculation formula is:

[0016]

[0017] Where SF is the safety factor; σ ult Indicates the ultimate strength of the material; σ actual Represents the actual stress.

[0018] Preferably, the machine learning regression model is a random forest regression model, which uses feedback parameter data to analyze performance evaluation data to generate prediction results for adjusting and optimizing surface data, and the prediction results are prediction values ​​based on input features.

[0019] Preferably, the construction of the virtual scene data includes the setting of light sources, the configuration of camera viewing angles, and the addition of interactive elements to support the user's experience in a virtual reality environment and collect user feedback data to further optimize the design.

[0020] Preferably, the user feedback data includes the user's evaluation and suggestions on the final confirmed design data, and the user feedback data is used to guide the final adjustment.

[0021] A system for implementing the curtain wall space curved surface design method, comprising:

[0022] Data collection module, used to collect historical project data, including design parameters, material properties and construction difficulty, and generate cleaned project data through data collation and cleaning;

[0023] A feature extraction module is used to extract features from the cleaned project data using principal component analysis to generate design feature data;

[0024] The surface generation module uses a convolutional neural network to generate preliminary surface data based on the design feature data, and performs adversarial generation on the preliminary surface data through a generative adversarial network to finally generate adversarial generated surface data;

[0025] Shape optimization module, used to perform real-time shape optimization on adversarially generated surface data and generate optimized surface data;

[0026] A performance evaluation module, connected to the finite element analysis software, inputs the optimized surface data to generate performance evaluation data, and analyzes the performance evaluation data to identify feedback parameter data;

[0027] A machine learning module uses a machine learning regression model to analyze feedback parameter data to adjust and optimize surface data, generate adjusted surface data, and perform multi-objective optimization to generate final optimized design data;

[0028] The virtual reality module is used to import the final optimized design data into the virtual reality platform, build virtual scene data, conduct user experience testing, collect user feedback data, and make final adjustments based on the user feedback data to confirm the final design data.

[0029] Compared with the prior art, the advantages and beneficial effects of the present invention are:

[0030] By introducing advanced deep learning technologies, especially convolutional neural networks and generative adversarial networks, the present invention realizes intelligent optimization of curtain wall space surface design.

[0031] The present invention can not only efficiently generate diverse surface designs, but also evaluate performance in real time, greatly improving the accuracy and reliability of the design, thereby solving the problems of inefficiency and design inconsistency in traditional technologies.

[0032] Through the user feedback mechanism, the present invention can continuously optimize the design to make it more in line with actual needs and ensure the quality and applicability of the final product. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the process of the present invention;

[0034] Figure 2 This is a diagram of the data analysis results in the present invention;

[0035] Figure 3 It is a result display diagram of the present invention;

[0036] Figure 4 This is a system structure block diagram of the present invention. DETAILED DESCRIPTION

[0037] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0038] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0039] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0040] like Figure 1 As shown, the curtain wall space surface design method includes the following steps:

[0041] Collect historical project data, including design parameters, material properties, and construction difficulty, generate cleaned project data through data sorting and cleaning, and use principal component analysis to extract features from the cleaned project data to generate design feature data;

[0042] Preferably, the design feature data includes key parameters that affect the surface design, specifically surface thickness, material properties, construction difficulty and design style, and the design feature data is generated by performing principal component analysis on the cleaned project data.

[0043] Through data collation and cleaning, cleaned project data is generated. This process aims to remove duplicate, erroneous or incomplete information to ensure the accuracy of the data. Cleaned project data provides a reliable basis for further analysis.

[0044] When generating design feature data, principal component analysis (PCA) is applied to the cleaned project data to extract key parameters that affect surface design. PCA is a statistical technique that can identify the features that contribute most to data variance by reducing the data from a high-dimensional space to a low-dimensional space. Through this method, multiple variables in the original data set can be effectively converted into a few principal components, thereby reducing the data dimension and retaining as much information as possible. These principal components are called design feature data, which specifically include key parameters such as surface thickness, material properties, construction difficulty, and design style. These design feature data not only facilitate subsequent model training and surface generation, but also provide designers with important decision-making basis to ensure the optimization of design solutions in terms of aesthetics, structural safety, and construction feasibility.

[0045] For example, in practical applications, the thickness of the surface in the design feature data may directly affect the load-bearing capacity and visual effect of the curtain wall. If a design scheme shows that the thickness of the surface is a major influencing factor after principal component analysis, the designer can adjust the thickness parameters accordingly to optimize the structural performance and meet the aesthetic requirements. Through this data-driven design method, the design of the curtain wall space surface not only becomes more scientific and reasonable, but also improves the quality and construction efficiency of the overall project.

[0046] Use convolutional neural networks to generate preliminary surface data based on design feature data, and use generative adversarial networks to generate adversarial surface data, and finally generate adversarial generated surface data; use shape optimization algorithms to perform real-time shape optimization on adversarial generated surface data to generate optimized surface data;

[0047] Preferably, the structure of the convolutional neural network includes multiple convolutional layers, activation layers and fully connected layers, each convolutional layer is used to extract spatial features in the design feature data, and the preliminary surface data output by the convolutional neural network after training contains preliminary generated surface model information.

[0048] As a powerful deep learning tool, convolutional neural network (CNN) plays a key role. Its structure usually consists of multiple convolutional layers, activation layers and fully connected layers. Specifically, the convolutional layer is used to extract spatial features in the design feature data, which include the shape, texture and other geometric properties of the surface. By using convolution operations, the network can automatically identify local features in the input data to capture complex spatial relationships. For example, the convolutional layer can learn the edge features of different surface shapes, so that the model can better understand the design style.

[0049] Each convolutional layer is followed by an activation layer, which usually uses ReLU (Rectified Linear Unit) as the activation function. This process allows the network to introduce nonlinearity and further enhance the model's expressiveness. Through this structure, convolutional neural networks can extract increasingly complex features layer by layer, thereby achieving in-depth analysis of design feature data.

[0050] After training, the preliminary surface data output by the convolutional neural network contains preliminary generated surface model information. This output data not only reflects the structural characteristics of the input design feature data, but also provides an important basis for the subsequent generative adversarial network (GAN) and shape optimization algorithm. By inputting these preliminary surface data into the subsequent optimization and generation modules, more complex and accurate surface designs can be achieved. For example, if the preliminary surface data indicates that a specific shape meets the design requirements both visually and structurally, then the surface model can be used as the input of the generative adversarial network to generate more diverse and novel surface designs.

[0051] Preferably, the generator of the generative adversarial network receives preliminary surface data and generates adversarial generated surface data, and the discriminator of the generative adversarial network is used to evaluate the quality of the generated adversarial generated surface data.

[0052] Generative adversarial networks (GANs) play an important role as an innovative generative model. Specifically, GANs consist of two main components: a generator and a discriminator. The generator is responsible for receiving preliminary surface data and generating adversarial surface data based on it. By processing the preliminary surface data, the generator can create a surface model that is visually and structurally similar to the original data but novel. This process is particularly critical in the design stage because it can provide a variety of design solutions.

[0053] The generator works by generating new surface data through a series of complex calculations and parameter adjustments. During this process, the generator continuously extracts features from the preliminary surface data and uses these features to generate new surface designs. These newly generated surface models may contain different shapes, textures, and design styles, providing designers with a wealth of choices to meet different project requirements.

[0054] At the same time, the role of the discriminator is to evaluate the quality of the generated adversarial surface data. The discriminator learns how to distinguish between the generated surface data and the real design data by comparing them. It outputs a probability value indicating the possibility that a certain generated surface data is real data. Through such a feedback mechanism, the generator can continuously adjust its parameters and optimize the generation process to improve the quality and authenticity of the generated surface data.

[0055] For example, suppose the generator outputs a new surface design that visually mimics an existing high-quality design. The discriminator analyzes the design and finds that it has defects in curvature and material effects, and returns a low probability value. This feedback prompts the generator to readjust the generation algorithm and optimize the curvature parameters of the surface until the generated surface data obtains a higher probability of authenticity in the discriminator's evaluation.

[0056] In summary, the application of generative adversarial networks in curtain wall space surface design not only creates diverse adversarial generated surface data through the generator, but also ensures the high quality and reliability of these data through the discriminator.

[0057] Preferably, the shape optimization algorithm performs real-time shape optimization on the generated surface data. The optimization process includes topological optimization of the generated surface to minimize the use of materials and improve the structural strength of the surface, and generate optimized surface data. The goal of the topological optimization is to minimize the material stress σ, and the calculation formula is:

[0058]

[0059] Among them, σ represents material stress; F represents the applied force; and A represents the cross-sectional area.

[0060] The application of shape optimization algorithms is crucial to improving the structural performance and material utilization efficiency of designs. Specifically, the shape optimization algorithm performs real-time shape optimization on adversarial generated surface data and adopts topology optimization methods to minimize material usage and enhance the structural strength of the surface. Topology optimization can find the optimal material layout solution by analyzing the distribution of materials under specific loads, ensuring that the amount of material used is minimized while meeting functional requirements.

[0061] Through the above formula, the algorithm can quantify the stress distribution of the surface when the load is applied, and then optimize the shape of the surface, reduce the stress concentration area, and prevent potential material damage or failure. This method ensures the safety and reliability of the design, especially when facing complex external environments.

[0062] For example, in a specific implementation case, suppose that the surface data generated by the adversarial model shows obvious stress concentration after applying topology optimization, resulting in material stress in some areas being higher than the acceptable range. Through the shape optimization algorithm, the shape and material distribution of these high-stress areas can be adjusted in real time, such as increasing the thickness of the surface or adjusting the curvature to achieve uniform stress distribution, thereby effectively reducing material stress and achieving the design goal.

[0063] Through this optimization process, the optimized surface data finally generated not only meets the design requirements in terms of aesthetics, but also performs well in terms of structural strength and material economy.

[0064] Inputting the optimized surface data into the finite element analysis software for performance evaluation, generating performance evaluation data, and analyzing the performance evaluation data to identify feedback parameter data; using a machine learning regression model to analyze the feedback parameter data to adjust the optimized surface data, generating adjusted surface data, and performing multi-objective optimization to generate final optimized design data;

[0065] Preferably, Figure 2 As shown, the finite element analysis software is used to perform performance evaluation on the optimized surface data. The performance evaluation data includes stress distribution, deformation and safety factor. The safety factor calculation formula is:

[0066]

[0067] Where SF is the safety factor; σ ult Indicates the ultimate strength of the material; σ actual Represents the actual stress.

[0068] Finite element analysis software can be used to evaluate the performance of the optimized surface in multiple dimensions, including key parameters such as stress distribution, deformation, and safety factor. These performance evaluation data can provide designers with in-depth insights to ensure that the design has sufficient safety and stability in practical applications.

[0069] The application of the above formula can help designers evaluate the safety performance of the surface under a specific load. Specifically, when the calculated safety factor is greater than 1, it means that the surface design is safe under the applied load; when the safety factor is less than 1, it indicates that there may be a risk of material failure, and the design needs to be adjusted.

[0070] For example, suppose the finite element analysis results show that the actual stress of an optimized surface under a specific environment is 200MPa, while the ultimate strength of the material is 500MPa. The safety factor calculated by the formula is 2.5, indicating that the design has a good safety margin. In this case, the designer can confirm that the surface design is feasible in practical applications and can withstand higher loads without failure.

[0071] On the other hand, if the safety factor is found to be lower than 1 during the evaluation process, the designer needs to re-examine the design parameters of the surface, such as increasing the thickness of the surface or changing the material, to ensure that the safety meets the design standards. Through this systematic evaluation process, the finite element analysis software not only improves the reliability of the design, but also provides a scientific basis for optimizing the design, so that the curtain wall design achieves a good balance between aesthetics and safety.

[0072] Preferably, the machine learning regression model is a random forest regression model, which uses feedback parameter data to analyze performance evaluation data to generate prediction results for adjusting and optimizing surface data, and the prediction results are prediction values ​​based on input features.

[0073] As an integrated learning method, the random forest regression model uses multiple decision trees for training and prediction to generate prediction results for adjusting and optimizing surface data. The random forest regression model can handle complex nonlinear relationships and provide efficient feature selection and importance evaluation, which makes it perform well when processing performance evaluation data.

[0074] In specific applications, the random forest regression model uses feedback parameter data to analyze performance evaluation data. This process first involves the integration of historical design data and feedback parameter data. By comparing these data sets, the model can identify important factors that affect performance. For example, in the previous performance evaluation, if the deformation and stress level of a surface design under a specific load caused concern, the random forest regression model can generate predictions based on this data, indicating which design features (such as material selection, surface thickness, etc.) should be adjusted to optimize the overall performance of the surface.

[0075] The prediction result is not just a numerical output based on the input features, it also provides an intuitive guide that enables designers to make effective adjustments for specific design goals. For example, suppose the feedback parameter data shows that the actual stress of the optimized surface under a certain condition exceeds expectations, and the random forest model predicts that increasing the thickness of the surface will help reduce the stress level and increase the safety factor. Designers can make corresponding modifications based on this to ensure the feasibility and safety of the design in practical applications.

[0076] In addition, an important advantage of the random forest regression model is that it scores the importance of each feature during the model training process. Designers can use this to determine which design parameters play a key role in performance optimization, thereby performing more accurate optimization and resource allocation. This data-driven approach not only improves the scientificity and reliability of the design, but also makes the design process more efficient and flexible.

[0077] In summary, the application of random forest regression model in curtain wall space surface design provides practical prediction results through in-depth analysis of feedback parameter data and performance evaluation data, and effectively guides the adjustment of optimized surface data.

[0078] like Figure 3 As shown, the final optimized design data is imported into the virtual reality platform, virtual scene data is constructed, user experience testing is carried out, user feedback data is collected, and final adjustments are made based on the user feedback data to confirm the final design data.

[0079] Preferably, the construction of the virtual scene data includes the setting of light sources, the configuration of camera viewing angles, and the addition of interactive elements to support the user's experience in a virtual reality environment and collect user feedback data to further optimize the design.

[0080] The construction of virtual scene data is an important part of user experience testing, which aims to create an interactive virtual reality environment so that designers and users can experience the design effects more intuitively. The process of constructing virtual scene data includes the setting of light sources, the configuration of camera perspectives, and the addition of interactive elements. These factors together affect the user's immersion and interactive experience in the virtual environment.

[0081] The setting of light sources plays a vital role in virtual scenes. Appropriate lighting can enhance the visual effect of curved surface design, making it show real texture and details at different angles and conditions. For example, when designing curtain walls, the use of directional light sources can simulate the effect of sunlight, produce shadows and highlights, and better reflect the characteristics of materials and the complexity of design. Through the reasonable configuration of light sources, designers can observe the design performance under different lighting conditions and make timely adjustments to meet aesthetic and functional requirements.

[0082] The configuration of the camera perspective affects the user's observation angle and viewing experience. By setting different camera perspectives in a virtual reality environment, users can examine the curtain wall design from multiple angles and understand its visual effects and sense of space in the actual building. Designers can preset multiple perspectives so that users can easily switch and deeply experience every detail of the design. The camera's movement and rotation functions also allow users to observe the design from different heights and distances, which is very important for judging the applicability and aesthetics of the design.

[0083] The addition of interactive elements is a key step in improving user experience. By introducing interactive elements such as buttons, menus, and sliders in a VR environment, users can interact with the design directly. This interaction not only enables users to freely explore different aspects of the design, but also allows users to collect feedback data during the experience. For example, users can adjust the thickness of a surface through a slider and observe the design changes in real time, and designers can collect this feedback data to further optimize the design.

[0084] As mentioned above, the construction of virtual scene data provides users with an immersive design experience through reasonable light source settings, flexible camera perspective configuration, and the addition of interactive elements.

[0085] Preferably, the user feedback data includes the user's evaluation and suggestions on the final confirmed design data, and the user feedback data is used to guide the final adjustment.

[0086] User feedback data mainly includes users’ evaluation and suggestions on the final confirmed design data. This information not only reflects the users’ subjective feelings, but also provides specific directions for improvement, which can provide valuable guidance for designers in subsequent adjustments.

[0087] The process of collecting user feedback data is usually carried out through interactive functions in the virtual reality environment. Users can provide feedback through the evaluation system when experiencing the design. This system can be a simple scoring mechanism or an open-ended opinion collection, which ensures that designers can obtain comprehensive user experience information through a combination of quantitative and qualitative methods. For example, when experiencing a certain curved design, users may give an evaluation of the aesthetics, functionality, and overall experience through the scoring system, and submit specific suggestions for design improvements through a text box. Such a feedback mechanism allows the user's voice to be directly conveyed to the design team, helping them understand the pros and cons of the design.

[0088] In principle, the analysis of user feedback data usually uses statistics and data mining techniques to organize and classify the collected feedback information and identify the design issues and suggestions that users are generally concerned about. This process can reveal potential defects or deficiencies in the design and help designers to be more targeted when adjusting the design. For example, if multiple users give negative comments on the thickness of the surface in the feedback data, designers can give priority to adjusting this parameter to improve user satisfaction.

[0089] In practical applications, user feedback data is not only an important basis for design iteration, but also a key factor in improving design competitiveness. By continuously collecting and analyzing user feedback, the design team can achieve dynamic docking with market demand, so that the final confirmed design is more in line with user expectations and actual use needs. This design optimization method based on user feedback not only improves the scientificity and rationality of the design, but also enhances the user's sense of participation and satisfaction, thereby promoting the innovation and development of curtain wall space surface design.

[0090] In summary, the importance of user feedback data in curtain wall space surface design is self-evident. By effectively collecting and analyzing this data, designers can make final adjustments more accurately to ensure that the design is both beautiful and practical, ultimately achieving higher user satisfaction and project success rate.

[0091] like Figure 4 As shown, a system for implementing the curtain wall space curved surface design method comprises:

[0092] The data collection module is used to collect historical project data, including design parameters, material properties and construction difficulty, and generate cleaned project data through data collation and cleaning; The data collection module is responsible for collecting historical project data, including design parameters, material properties and construction difficulty. This process generates cleaned project data through data collation and cleaning to ensure the reliability and validity of the data. These data provide a basis for subsequent analysis, enabling designers to make more scientific decisions based on past experience.

[0093] The feature extraction module is used to extract features from the cleaned project data using principal component analysis to generate design feature data. The feature extraction module uses principal component analysis (PCA) technology to process the cleaned project data, extract key features from it, and generate design feature data. This process can effectively reveal the main factors affecting the design, such as surface thickness and material properties, by reducing data dimensions and retaining important information. These feature data provide the necessary parameters for subsequent surface generation.

[0094] The surface generation module uses a convolutional neural network to generate preliminary surface data based on the design feature data, and generates adversarial generation on the preliminary surface data through a generative adversarial network to finally generate adversarial generated surface data; The surface generation module uses a convolutional neural network (CNN) to generate preliminary surface data based on the design feature data. The network can generate innovative and diverse surface designs by learning the spatial features in historical data. The generative adversarial network (GAN) further generates adversarial generation on the preliminary surface data to ensure that the generated surface achieves the best effect in terms of aesthetics and structural rationality. This generation process not only improves the flexibility of the design, but also provides designers with a variety of design options.

[0095] The shape optimization module is used to perform real-time shape optimization on the surface data generated by the adversarial model and generate optimized surface data; the shape optimization module performs real-time shape optimization on the surface data generated by the adversarial model to improve the structural strength of the surface and reduce material waste. Through the topology optimization algorithm, the module can adjust the shape of the surface to minimize material stress, thereby achieving a more efficient design.

[0096] The performance evaluation module is connected to the finite element analysis software, inputs the optimized surface data to generate performance evaluation data, and analyzes the performance evaluation data to identify feedback parameter data; the performance evaluation module is connected to the finite element analysis software, performs performance evaluation on the optimized surface data, and generates performance evaluation data, including stress distribution and deformation, etc. These data enable designers to evaluate the safety and functionality of the surface and identify feedback parameters that need to be adjusted.

[0097] The machine learning module uses a machine learning regression model to analyze feedback parameter data to adjust and optimize surface data, generate adjusted surface data, and perform multi-objective optimization to generate the final optimized design data. The machine learning module uses a machine learning regression model to analyze feedback parameter data to optimize surface design. This model can predict the effect of design adjustments based on historical data, thereby generating adjusted surface data and performing multi-objective optimization to ensure that the final design meets the requirements in terms of aesthetics, structural safety, and construction feasibility.

[0098] The virtual reality module is used to import the final optimized design data into the virtual reality platform, build virtual scene data, conduct user experience testing, collect user feedback data, and make final adjustments based on the user feedback data to confirm the final design data. The virtual reality module imports the final optimized design data into the virtual reality platform and builds virtual scene data. This enables designers and users to interact in a virtual environment, conduct user experience testing and collect feedback data. This feedback further guides the final adjustment of the design to ensure that the design not only meets the technical requirements, but also meets the actual needs of users.

[0099] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0100] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. The method for designing a curtain wall space surface is characterized by: The following steps are involved: Collect historical project data, including design parameters, material properties, and construction difficulty, generate cleaned project data through data sorting and cleaning, and use principal component analysis to extract features from the cleaned project data to generate design feature data; Use convolutional neural networks to generate preliminary surface data based on design feature data, and use generative adversarial networks to generate adversarial surface data, and finally generate adversarial generated surface data; use shape optimization algorithms to perform real-time shape optimization on adversarial generated surface data to generate optimized surface data; inputting the optimized surface data into finite element analysis software for performance evaluation, generating performance evaluation data, and analyzing the performance evaluation data to identify feedback parameter data; Use machine learning regression models to analyze feedback parameter data to adjust and optimize surface data, generate adjusted surface data, and perform multi-objective optimization to generate final optimized design data; Import the final optimized design data into the virtual reality platform, build virtual scene data, conduct user experience testing, collect user feedback data, and make final adjustments based on user feedback data to confirm the final design data.

2. The curtain wall space curved surface design method according to claim 1, characterized in that: The design feature data include key parameters that affect the surface design, specifically, the surface thickness, material properties, construction difficulty and design style. The design feature data is generated by performing principal component analysis on the cleaned project data.

3. The curtain wall space curved surface design method according to claim 1, characterized in that: The structure of the convolutional neural network includes multiple convolutional layers, activation layers and fully connected layers, each convolutional layer is used to extract spatial features in the design feature data, and the preliminary surface data output by the convolutional neural network after training contains preliminary generated surface model information.

4. The curtain wall space curved surface design method according to claim 1, characterized in that: The generator of the generative adversarial network receives preliminary surface data and generates adversarial generated surface data, and the discriminator of the generative adversarial network is used to evaluate the quality of the generated adversarial generated surface data.

5. The curtain wall space curved surface design method according to claim 1, characterized in that: The shape optimization algorithm performs real-time shape optimization on the generated surface data. The optimization process includes topological optimization of the generated surface to minimize the use of materials and improve the structural strength of the surface, and generate optimized surface data. The goal of the topological optimization is to minimize the material stress σ, and the calculation formula is: Among them, σ represents material stress; F represents the applied force; and A represents the cross-sectional area.

6. The curtain wall space curved surface design method according to claim 1, characterized in that: The finite element analysis software is used to perform performance evaluation on the optimized surface data, and the performance evaluation data includes stress distribution, deformation and safety factor. The safety factor calculation formula is: Where SF is the safety factor; σ ult Indicates the ultimate strength of the material; σ actual Represents the actual stress.

7. The curtain wall space curved surface design method according to claim 1, characterized in that: The machine learning regression model is a random forest regression model, which uses feedback parameter data to analyze performance evaluation data to generate a prediction result for adjusting and optimizing surface data, and the prediction result is a prediction value based on input features.

8. The curtain wall space curved surface design method according to claim 1, characterized in that: The construction of the virtual scene data includes the setting of light sources, the configuration of camera perspectives, and the addition of interactive elements to support the user's experience in a virtual reality environment and collect user feedback data to further optimize the design.

9. The curtain wall space curved surface design method according to claim 1, characterized in that: The user feedback data includes the user's evaluation and suggestions on the final confirmed design data, and the user feedback data is used to guide the final adjustment.

10. A system for implementing the curtain wall space curved surface design method according to any one of claims 1 to 9, characterized in that: include: Data collection module, used to collect historical project data, including design parameters, material properties and construction difficulty, and generate cleaned project data through data collation and cleaning; A feature extraction module is used to extract features from the cleaned project data using principal component analysis to generate design feature data; The surface generation module uses a convolutional neural network to generate preliminary surface data based on the design feature data, and performs adversarial generation on the preliminary surface data through a generative adversarial network to finally generate adversarial generated surface data; Shape optimization module, used to perform real-time shape optimization on adversarially generated surface data and generate optimized surface data; A performance evaluation module, connected to the finite element analysis software, inputs the optimized surface data to generate performance evaluation data, and analyzes the performance evaluation data to identify feedback parameter data; A machine learning module uses a machine learning regression model to analyze feedback parameter data to adjust and optimize surface data, generate adjusted surface data, and perform multi-objective optimization to generate final optimized design data; The virtual reality module is used to import the final optimized design data into the virtual reality platform, build virtual scene data, conduct user experience testing, collect user feedback data, and make final adjustments based on the user feedback data to confirm the final design data.

Images