Intelligent auxiliary identification method used in hyperbolic curtain wall parametric modeling design process

Abstract

The invention provides an intelligent auxiliary identification method used in a hyperbolic curtain wall parameterized modeling design process. The method comprises the following steps: acquiring a plurality of historical hyperbolic curtain wall models and corresponding historical design data; inputting the hole location distribution information, the geometric feature information and the avoidance mode information into a preset first optimization module, and outputting prediction connection optimization information and prediction node design information; training a first optimization module according to the node design information, the connection optimization information, the predicted connection optimization information and the predicted node design information; inputting the geometric feature information into a preset second optimization module, outputting predicted grid division information, obtaining an aesthetic property coefficient corresponding to the predicted grid division information, and completing training of the second optimization module; and inputting the geometric feature information and the node design information into a preset avoidance detection module, outputting predicted avoidance conflict information, and completing training of the avoidance detection module according to the predicted avoidance conflict information and the avoidance mode information.

Description

Technical Field

[0001] The present application relates to the technical field of architectural modeling, and in particular to an intelligent auxiliary recognition method used in the parametric modeling design process of hyperbolic curtain walls. Background Art

[0002] Hyperbolic curtain walls are widely used in modern architecture for their complex geometric forms and unique aesthetic effects. However, their design process involves a large number of key links that rely on manual judgment, including node design and connection optimization, grid division, and avoidance conflict detection and optimization. In current technologies, these links usually require independent models to be established for optimization, resulting in a waste of computing resources, complex model maintenance, and insufficient synergy.

[0003] Therefore, a method is urgently needed to solve at least one of the above problems. Summary of the invention

[0004] The present application provides an intelligent auxiliary recognition method for the parametric modeling and design process of hyperbolic curtain walls, aiming to solve the problem that hyperbolic curtain walls are widely used in modern buildings due to their complex geometric forms and unique aesthetic effects. However, its design process involves a large number of key links that require manual judgment, including node design and connection optimization, grid division, and avoidance conflict detection and optimization. In current technology, these links usually require the establishment of independent models for optimization, resulting in waste of computing resources, complex model maintenance, and insufficient synergy.

[0005] In a first aspect, the present application provides an intelligent auxiliary recognition method for use in a parametric modeling design process of a hyperbolic curtain wall, comprising:

[0006] Acquire multiple historical hyperbolic curtain wall models and corresponding historical design data; the historical design data at least includes node design information, connection optimization information, geometric feature information, avoidance method information, hole distribution information and connection optimization information;

[0007] Input the hole position distribution information, geometric feature information and avoidance method information into a preset first optimization module, and the first optimization module outputs predicted connection optimization information and predicted node design information; complete the training of the first optimization module according to the node design information, connection optimization information, predicted connection optimization information and predicted node design information;

[0008] The geometric feature information is input into a preset second optimization module, the second optimization module outputs predicted grid division information, an aesthetic coefficient corresponding to the predicted grid division information is obtained, and the training of the second optimization module is completed according to the aesthetic coefficient;

[0009] The geometric feature information and the node design information are input into a preset avoidance detection module, the avoidance detection module outputs predicted avoidance conflict information, and the avoidance detection module is trained according to the predicted avoidance conflict information and the avoidance mode information;

[0010] An intelligent auxiliary recognition model is constructed based on the trained first optimization module, the second optimization module and the avoidance detection module, which is used to generate optimization suggestions for the hyperbolic curtain wall model under modeling during the hyperbolic curtain wall modeling process, wherein the optimization suggestions include one or more of connection optimization suggestions, node design suggestions, grid division suggestions and avoidance design suggestions.

[0011] In some embodiments, constructing an intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module includes: inputting the node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetics coefficient, predicted avoidance conflict information and avoidance method information into a preset multi-task learning model, the multi-task learning model calculates the joint loss function corresponding to the first optimization module, the second optimization module and the avoidance detection module; completing the training of the first optimization module, the second optimization module and the avoidance detection module according to the joint loss function to form an intelligent auxiliary recognition model.

[0012] Exemplarily, the node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetics coefficient, predicted avoidance conflict information and avoidance method information are input into a preset multi-task learning model, including: calculating a first loss function according to the node design information, connection optimization information, predicted connection optimization information and predicted node design information; calculating a second loss function according to the aesthetics coefficient; calculating a third loss function according to the predicted avoidance conflict information and avoidance method information; inputting the first loss function, the second loss function and the third loss function into the multi-task learning model to calculate the joint loss function according to the first loss function, the second loss function and the third loss function.

[0013] It should be noted that, in some embodiments, the calculation of the joint loss function based on the first loss function, the second loss function and the third loss function includes: setting the first dynamic weight corresponding to the first loss function; setting the second dynamic weight corresponding to the second loss function; setting the third dynamic weight corresponding to the third loss function; calculating the joint loss function based on the first dynamic weight, the first loss function, the second dynamic weight, the second loss function, the third dynamic weight and the third loss function.

[0014] It should be noted that, in some embodiments, the expression of the joint loss function includes:

[0015] ;in, is the joint loss function, is the first dynamic weight, is the first loss function, is the second dynamic weight, is the second loss function, is the third dynamic weight, is the third loss function.

[0016] In some embodiments, obtaining the aesthetic coefficient corresponding to the predicted grid division information includes: obtaining a symmetry index, a proportion coordination index, and a grid density uniformity index corresponding to the predicted grid division information; and calculating the aesthetic coefficient based on the symmetry index, the proportion coordination index, and the grid density uniformity index.

[0017] Exemplarily, the method of obtaining the symmetry index, proportion coordination index, and grid density uniformity index corresponding to the predicted grid division information includes: obtaining the grid similarity on both sides of the symmetry axis corresponding to the predicted grid division information, using the grid similarity as the symmetry index; obtaining the difference coefficient between the aspect ratio of each grid corresponding to the predicted grid division information and a preset ideal ratio, using the difference coefficient as the proportion coordination index; obtaining the standard deviation corresponding to the area of ​​each grid, taking the inverse of the standard deviation and normalizing it, and outputting the grid density uniformity index.

[0018] Exemplarily, the expression of the aesthetic coefficient includes:

[0019] ;

[0020] in, is the aesthetics coefficient, is the first weight, is the symmetry index, is the proportional coordination index, is the grid density uniformity index, is the second weight, The third weight.

[0021] In some embodiments, before the hole distribution information, geometric feature information and avoidance method information are input into a preset first optimization module, it also includes: obtaining the design environment information corresponding to each of the historical hyperbolic curtain wall models; generating a format conversion relationship corresponding to the historical design data according to the design environment information; and converting each of the historical design data into a preset target format according to the format conversion relationship.

[0022] In some embodiments, after constructing the intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module, it also includes: generating an optimization effect score based on the hyperbolic curtain wall model in modeling, the optimization suggestions and the optimized hyperbolic curtain wall model; and completing the optimization of the auxiliary recognition model based on the optimization effect score, the hyperbolic curtain wall model in modeling and the optimization suggestions.

[0023] In a second aspect, the present application provides an intelligent auxiliary recognition device for use in a hyperbolic curtain wall parametric modeling design process, comprising:

[0024] A data acquisition unit, used to acquire a plurality of historical hyperbolic curtain wall models and corresponding historical design data; the historical design data at least includes node design information, connection optimization information, geometric feature information, avoidance information, hole distribution information and connection optimization information;

[0025] A data input unit, used to input the hole position distribution information, geometric feature information and avoidance method information into a preset first optimization module, the first optimization module outputs predicted connection optimization information and predicted node design information; and completes the training of the first optimization module according to the node design information, connection optimization information, predicted connection optimization information and predicted node design information;

[0026] A feature input unit, used to input the geometric feature information into a preset second optimization module, the second optimization module outputs predicted grid division information, obtains an aesthetic coefficient corresponding to the predicted grid division information, and completes the training of the second optimization module according to the aesthetic coefficient;

[0027] A training completion unit, used for inputting the geometric feature information and the node design information into a preset avoidance detection module, the avoidance detection module outputs predicted avoidance conflict information, and completing the training of the avoidance detection module according to the predicted avoidance conflict information and avoidance mode information;

[0028] The optimization output unit is used to build an intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module, and is used to generate optimization suggestions for the hyperbolic curtain wall model being modeled during the hyperbolic curtain wall modeling process, wherein the optimization suggestions include one or more of connection optimization suggestions, node design suggestions, grid division suggestions and avoidance design suggestions.

[0029] In a third aspect, the present application provides a computer device comprising a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement a method provided in any embodiment of the present application when executing the computer program.

[0030] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer-readable instructions are executed by the processor, one or more processors execute a method as provided in any embodiment of the present application.

[0031] The embodiment of the present application provides an intelligent auxiliary recognition method for the parametric modeling design process of a hyperbolic curtain wall. The intelligent auxiliary recognition method is mainly used for the parametric modeling design of the hyperbolic curtain wall, and improves the design efficiency and quality through machine learning technology. The specific technical process is as follows: multiple hyperbolic curtain wall models and their corresponding design data are obtained from past projects. These data include at least node design information, connection optimization information, geometric feature information, avoidance method information, hole position distribution information, etc. The hole position distribution information, geometric feature information and avoidance method information are input into a preset first optimization module. The first optimization module outputs predicted connection optimization information and predicted node design information. According to the actual historical node design information and connection optimization information, the first optimization module is trained so that it can accurately predict connection optimization and node design. The geometric feature information is input into a preset second optimization module. The second optimization module outputs predicted grid division information. The aesthetic coefficient corresponding to the predicted grid division information is calculated. The second optimization module is trained according to the aesthetic coefficient so that it can generate an aesthetic and reasonable grid division scheme. The geometric feature information and node design information are input into a preset avoidance detection module. The avoidance detection module outputs predicted avoidance conflict information. The avoidance detection module is trained according to the actual avoidance method information and the predicted avoidance conflict information, so that it can accurately detect and optimize the avoidance conflict. The trained first optimization module, the second optimization module and the avoidance detection module are integrated into an intelligent auxiliary recognition model. In the process of hyperbolic curtain wall modeling, the model can generate optimization suggestions for the hyperbolic curtain wall model currently being modeled, including connection optimization suggestions, node design suggestions, grid division suggestions and avoidance design suggestions.

[0032] Suppose we have a new hyperbolic curtain wall project that requires parametric modeling and design. The following are the specific application steps of this method: Collect the hyperbolic curtain wall models and their design data of past projects, such as node design drawings, connection methods, geometric features, avoidance methods, hole distribution, etc. Input the hole distribution information, geometric feature information, and avoidance method information of the new project. The first optimization module outputs the predicted connection optimization information and node design information. According to the prediction results, the designer can adjust the connection method and node design to achieve the best effect. Input the geometric feature information of the new project. The second optimization module outputs the predicted grid division information and calculates the aesthetic coefficient. According to the aesthetic coefficient, the designer can choose the most appropriate grid division scheme. Input the geometric feature information and node design information of the new project. The avoidance detection module outputs the predicted avoidance conflict information. According to the prediction results, the designer can adjust the design to avoid conflicts. The intelligent auxiliary recognition model combines the outputs of the above modules to generate comprehensive optimization suggestions to help designers complete design tasks efficiently.

[0033] The provided method has at least the following beneficial effects: Improve design efficiency: Through automated prediction and optimization, the time for manual judgment is reduced and the design efficiency is improved. Improve design quality: Based on historical data and machine learning algorithms, the generated optimization suggestions are more scientific and reasonable, which improves the design quality. Reduce resource waste: Avoid the situation of establishing independent models for optimization, and reduce the waste of computing resources. Simplify model maintenance: Through the integrated intelligent auxiliary recognition model, the workload of model maintenance is simplified. Enhance synergy effect: The collaborative work between different modules makes the entire design process smoother and more efficient. Improve aesthetics: Through the evaluation of the aesthetics coefficient, ensure that the final design is not only functional but also has good visual effects.

[0034] In summary, this intelligent assisted recognition method significantly improves the efficiency and quality of parametric modeling design of hyperbolic curtain walls through machine learning technology, providing strong support for modern architectural design.

[0035] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1It is a schematic flow chart of the steps of an intelligent auxiliary identification method for a hyperbolic curtain wall parametric modeling design process provided by an embodiment of the present application;

[0038] Figure 2 It is a schematic block diagram of the structure of an intelligent auxiliary recognition device for use in a parametric modeling design process of a hyperbolic curtain wall provided by an embodiment of the present application;

[0039] Figure 3 It is a schematic block diagram of the structure of a computer device provided in one embodiment of the present application.

[0040] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0042] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0043] It should be understood that, in order to facilitate the clear description of the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, the words "first", "second", etc. are used to distinguish the same items or similar items with substantially the same functions and effects. Those skilled in the art can understand that the words "first", "second", etc. do not limit the quantity and execution order, and the words "first", "second", etc. do not necessarily limit the difference.

[0044] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0045] It should also be understood that the term “and / or” used in the specification and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0046] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0047] Hyperbolic curtain walls are widely used in modern architecture for their complex geometric forms and unique aesthetic effects. However, their design process involves a large number of key links that rely on manual judgment, including node design and connection optimization, grid division, and avoidance conflict detection and optimization. In current technologies, these links usually require independent models to be established for optimization, resulting in a waste of computing resources, complex model maintenance, and insufficient synergy.

[0048] Therefore, a method is urgently needed to solve at least one of the above problems.

[0049] To solve the above problems, please refer to Figure 1 ,like Figure 1 As shown, the provided intelligent auxiliary recognition method for the parametric modeling design process of the hyperbolic curtain wall includes steps S101 to S105. The intelligent auxiliary recognition method for the parametric modeling design process of the hyperbolic curtain wall is executed by a computer device, which may be a single server or a server cluster, or may be a handheld terminal, a laptop computer, a wearable device, or a robot.

[0050] like Figure 1 As shown, the details of steps S101-S105 are as follows:

[0051] Step S101. Acquire multiple historical hyperbolic curtain wall models and corresponding historical design data; the historical design data at least includes node design information, connection optimization information, geometric feature information, avoidance information, hole distribution information and connection optimization information.

[0052] Specifically, this step aims to build a comprehensive dataset for subsequent model training. The historical hyperbolic curtain wall model can come from actual projects or existing design cases, and the historical design data covers node design information, connection optimization information, geometric feature information (such as curvature, shape, etc.), avoidance information (how to avoid structural conflicts), hole distribution information (hole location and size) and connection optimization information (such as what type of connectors to use and how to arrange them).

[0053] For example, data collection: obtain relevant case materials from company databases, public repositories or partners. Ensure that the data sources are diverse, including projects of different regions, sizes and complexities. Data preprocessing: clean the raw data to remove invalid records and standardize the format for subsequent analysis. To handle missing values, interpolation or deletion can be used. Normalize or standardize the data to ensure that each feature is on the same scale. Feature extraction: extract the various design parameters mentioned above as feature vectors for each case. Use computer vision technology to extract geometric features such as curvature, surface smoothness, etc. Convert text descriptions into structured data formats.

[0054] By establishing a high-quality and diverse training set, a solid foundation is laid for the subsequent machine learning tasks; at the same time, by learning from existing successful experiences, it can help new projects quickly find the right solution starting point. In addition, a diverse data set helps improve the generalization ability of the model, so that it can provide accurate suggestions when facing new projects.

[0055] Step S102. Input the hole distribution information, geometric feature information and avoidance information into a preset first optimization module, which outputs predicted connection optimization information and predicted node design information; and complete the training of the first optimization module based on the node design information, connection optimization information, predicted connection optimization information and predicted node design information.

[0056] Specifically, this stage focuses on using geometric properties and other related information to automatically derive reasonable connection schemes and node layouts. By inputting specific hole arrangement, basic morphological descriptors and other conditions into the pre-set algorithm framework, it is expected to obtain preliminary suggestions on the optimal joint configuration and support point settings.

[0057] For example, choose a neural network architecture suitable for solving regression problems, such as convolutional neural networks (CNN) or recurrent neural networks (RNN). Define a loss function to measure the difference between the predicted value and the true value. Commonly used ones are mean square error (MSE) or mean absolute error (MAE). Use supervised learning mode to send the prepared samples into the network and iterate repeatedly until convergence. Use batch gradient descent or other optimization algorithms to adjust model parameters. Set appropriate hyperparameters, such as learning rate, batch size, number of iterations, etc. Test the generalization ability of the model through cross-validation and other means to ensure that it can perform well on unseen data. Calculate indicators such as precision, recall, F1 score, etc. to evaluate model performance.

[0058] By automatically providing a reliable initial design solution, it reduces the need for manual intervention and speeds up the progress of the entire project. In addition, it can also discover potential design rules and guide future work directions. Through automated prediction, designers can focus on higher-level design decisions instead of tedious basic work.

[0059] Step S103. Input the geometric feature information into a preset second optimization module, the second optimization module outputs predicted grid division information, obtains the aesthetic coefficient corresponding to the predicted grid division information, and completes the training of the second optimization module according to the aesthetic coefficient.

[0060] Specifically, considering that visual effects are crucial to the appearance of buildings, an evaluation mechanism specifically for panel segmentation strategies is introduced here. That is, given a certain form, grid layouts under different combinations are tried, and they are scored and sorted according to pre-defined aesthetic standards, and the one with the highest score is selected as the recommendation result.

[0061] Aesthetic index definition: Combine psychological principles and art theory to extract a quantitative system to judge the quality of various patterns. Factors to consider include symmetry, proportion, rhythm, color matching, etc. Segmentation algorithm development: Implement a series of program codes that can flexibly change the spacing size, angle inclination, etc. Use optimization algorithms such as genetic algorithms and simulated annealing algorithms to find the optimal solution. Scoring system integration: Combine the two to form a closed-loop feedback loop, and continuously iterate and improve until the requirements are met. Design a user interface so that designers can intuitively see different grid division schemes and their scores.

[0062] It not only ensures that functional requirements are met, but also takes into account aesthetic considerations, making the final product more in line with the public's aesthetic preferences and enhancing market competitiveness. Through automated grid division, the design efficiency can be greatly improved while ensuring the consistency and aesthetics of the design.

[0063] Step S104. Input the geometric feature information and node design information into a preset avoidance detection module, the avoidance detection module outputs predicted avoidance conflict information, and completes the training of the avoidance detection module according to the predicted avoidance conflict information and avoidance method information.

[0064] Specifically, when there are other components, how to cleverly arrange their own positions to avoid collisions becomes a major challenge. For this reason, a functional component specifically responsible for checking and eliminating such hidden dangers is specially set up.

[0065] Research on conflict detection algorithms: Learn from classic practices in the field of computer graphics, such as using space partitioning trees to speed up queries. Use collision detection algorithms such as OBB (oriented bounding box) or AABB (axis-aligned bounding box) to detect potential conflicts. Automatic correction function implementation: Once a problem is found, take immediate measures to correct it, which may involve fine-tuning the coordinate system offset or modifying local construction details. Provide a variety of correction solutions for designers to choose from to meet different design needs.

[0066] User interface design: To make it easier for designers to understand the current situation and modification plan, it is necessary to provide intuitive and easy-to-understand operation guides. Design visualization tools to display conflict areas and suggested correction paths.

[0067] It greatly reduces the probability of errors caused by negligence and ensures the quality of the project. At the same time, it also improves work efficiency and avoids the trouble caused by repeated rework. Through automated conflict detection and correction, designers can complete the design faster and ensure the feasibility of the design.

[0068] Step S105. Construct an intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module, so as to generate optimization suggestions for the hyperbolic curtain wall model being modeled during the hyperbolic curtain wall modeling process, wherein the optimization suggestions include one or more of connection optimization suggestions, node design suggestions, grid division suggestions and avoidance design suggestions.

[0069] Specifically, all the tools introduced above are integrated together to form a complete intelligent assistance system, which can automatically generate a relatively complete set of construction drawing drafts after the user inputs a small amount of basic information.

[0070] Through API interface development: write the necessary application programming interface to enable smooth communication between various parts. Ensure the stability and security of the interface to prevent data leakage or tampering. Testing and debugging: conduct comprehensive functional and performance testing to find and fix existing defects. Use a variety of testing methods such as unit testing and integration testing to ensure the reliability and stability of the system. Write detailed user manuals and technical documents to facilitate others to understand and use. Including installation guides, instructions for use, FAQs, etc. Provide training courses for designers and engineers to help them familiarize themselves with the operation and use of the system. Provide online support and community forums to solve problems encountered by users during use.

[0071] It has significantly improved the level of intelligence in the architectural design industry and promoted technological innovation and development. It has also brought convenience to practitioners, allowing them to devote more energy to more creative activities. By comprehensively using various subsystems, designers can quickly generate high-quality design solutions and improve overall design efficiency and quality.

[0072] Through the above five steps, we have built a complete intelligent auxiliary recognition system that can effectively improve the efficiency and quality of hyperbolic curtain wall design. Each step has been carefully designed and optimized to ensure that the system can provide accurate and reliable design suggestions. This method not only reduces the need for manual intervention, but also greatly improves the consistency and aesthetics of the design, bringing revolutionary changes to the architectural design industry.

[0073] The provided intelligent auxiliary recognition method is mainly used for parametric modeling design of hyperbolic curtain walls, and improves design efficiency and quality through machine learning technology. The specific technical process is as follows: multiple hyperbolic curtain wall models and their corresponding design data are obtained from past projects. These data include at least node design information, connection optimization information, geometric feature information, avoidance method information, hole position distribution information, etc. The hole position distribution information, geometric feature information and avoidance method information are input into a preset first optimization module. The first optimization module outputs predicted connection optimization information and predicted node design information. According to the actual historical node design information and connection optimization information, the first optimization module is trained so that it can accurately predict connection optimization and node design. The geometric feature information is input into a preset second optimization module. The second optimization module outputs predicted grid division information. The aesthetic coefficient corresponding to the predicted grid division information is calculated. The second optimization module is trained according to the aesthetic coefficient so that it can generate an aesthetic and reasonable grid division scheme. The geometric feature information and node design information are input into a preset avoidance detection module. The avoidance detection module outputs predicted avoidance conflict information. The avoidance detection module is trained according to the actual avoidance method information and the predicted avoidance conflict information, so that it can accurately detect and optimize the avoidance conflict. The trained first optimization module, the second optimization module and the avoidance detection module are integrated into an intelligent auxiliary recognition model. In the process of hyperbolic curtain wall modeling, the model can generate optimization suggestions for the hyperbolic curtain wall model currently being modeled, including connection optimization suggestions, node design suggestions, grid division suggestions and avoidance design suggestions.

[0074] Suppose we have a new hyperbolic curtain wall project that requires parametric modeling and design. The following are the specific application steps of this method: Collect the hyperbolic curtain wall models and their design data of past projects, such as node design drawings, connection methods, geometric features, avoidance methods, hole distribution, etc. Input the hole distribution information, geometric feature information, and avoidance method information of the new project. The first optimization module outputs the predicted connection optimization information and node design information. According to the prediction results, the designer can adjust the connection method and node design to achieve the best effect. Input the geometric feature information of the new project. The second optimization module outputs the predicted grid division information and calculates the aesthetic coefficient. According to the aesthetic coefficient, the designer can choose the most appropriate grid division scheme. Input the geometric feature information and node design information of the new project. The avoidance detection module outputs the predicted avoidance conflict information. According to the prediction results, the designer can adjust the design to avoid conflicts. The intelligent auxiliary recognition model combines the outputs of the above modules to generate comprehensive optimization suggestions to help designers complete design tasks efficiently.

[0075] The provided method has at least the following beneficial effects: Improve design efficiency: Through automated prediction and optimization, the time for manual judgment is reduced and the design efficiency is improved. Improve design quality: Based on historical data and machine learning algorithms, the generated optimization suggestions are more scientific and reasonable, which improves the design quality. Reduce resource waste: Avoid the situation of establishing independent models for optimization, and reduce the waste of computing resources. Simplify model maintenance: Through the integrated intelligent auxiliary recognition model, the workload of model maintenance is simplified. Enhance synergy effect: The collaborative work between different modules makes the entire design process smoother and more efficient. Improve aesthetics: Through the evaluation of the aesthetics coefficient, ensure that the final design is not only functional but also has good visual effects.

[0076] In summary, this intelligent assisted recognition method significantly improves the efficiency and quality of parametric modeling design of hyperbolic curtain walls through machine learning technology, providing strong support for modern architectural design.

[0077] In some embodiments, constructing an intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module includes: inputting the node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetics coefficient, predicted avoidance conflict information and avoidance method information into a preset multi-task learning model, the multi-task learning model calculates the joint loss function corresponding to the first optimization module, the second optimization module and the avoidance detection module; completing the training of the first optimization module, the second optimization module and the avoidance detection module according to the joint loss function to form an intelligent auxiliary recognition model.

[0078] By collecting and organizing historical hyperbolic curtain wall models and their corresponding design data. Extract node design information, connection optimization information, geometric feature information, avoidance information, hole distribution information, etc. Select a neural network architecture suitable for multi-task learning, such as a structure with shared layers and multiple output heads. Define the loss function for each task (first loss function, second loss function, third loss function). Calculate the joint loss function and use dynamic weights to adjust the importance of each task. Input the extracted data into the multi-task learning model. Update the model parameters through the back-propagation algorithm to minimize the joint loss function. Perform multiple iterations until the model converges or reaches the predetermined number of training rounds. Use methods such as cross-validation to evaluate model performance. Adjust the model structure and hyperparameters based on the evaluation results to further optimize the model.

[0079] Exemplarily, the node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetics coefficient, predicted avoidance conflict information and avoidance method information are input into a preset multi-task learning model, including: calculating a first loss function according to the node design information, connection optimization information, predicted connection optimization information and predicted node design information; calculating a second loss function according to the aesthetics coefficient; calculating a third loss function according to the predicted avoidance conflict information and avoidance method information; inputting the first loss function, the second loss function and the third loss function into the multi-task learning model to calculate the joint loss function according to the first loss function, the second loss function and the third loss function.

[0080] Use mean square error (MSE) or mean absolute error (MAE) as the first loss function. Calculate the difference between the actual node design information and the predicted node design information. Calculate the aesthetics coefficient using the aesthetics scoring standard. Calculate the difference between the actual aesthetics coefficient and the predicted aesthetics coefficient. Calculate the third loss function: Use binary cross entropy or other classification loss functions to calculate the avoidance conflict information. Calculate the difference between the actual avoidance conflict information and the predicted avoidance conflict information. Weighted sum of the above three loss functions to obtain a joint loss function. Adjust the importance of each task through dynamic weights.

[0081] It should be noted that, in some embodiments, the calculation of the joint loss function based on the first loss function, the second loss function and the third loss function includes: setting the first dynamic weight corresponding to the first loss function; setting the second dynamic weight corresponding to the second loss function; setting the third dynamic weight corresponding to the third loss function; calculating the joint loss function based on the first dynamic weight, the first loss function, the second dynamic weight, the second loss function, the third dynamic weight and the third loss function.

[0082] Dynamic weights can be adjusted based on the performance of each task during training. For example, if the loss of a task is large, its weight can be increased so that it accounts for a larger proportion of the joint loss function.

[0083] It should be noted that, in some embodiments, the expression of the joint loss function includes:

[0084] ;in, is the joint loss function, is the first dynamic weight, is the first loss function, is the second dynamic weight, is the second loss function, is the third dynamic weight, is the third loss function.

[0085] It can be adjusted according to the performance of each task during training. For example, if the loss of a task is large, its corresponding value, thereby increasing its weight in the joint loss function. Use the above formula to calculate the joint loss function. Adjust the model parameters through gradient descent or other optimization algorithms to minimize the joint loss function.

[0086] In some embodiments, obtaining the aesthetic coefficient corresponding to the predicted grid division information includes: obtaining a symmetry index, a proportion coordination index, and a grid density uniformity index corresponding to the predicted grid division information; and calculating the aesthetic coefficient based on the symmetry index, the proportion coordination index, and the grid density uniformity index.

[0087] Get symmetry index: Calculate the grid similarity on both sides of the symmetry axis corresponding to the grid division information. The higher the similarity, the better the symmetry.

[0088] Obtaining the proportion coordination index: Calculate the difference coefficient between the aspect ratio of each grid and the preset ideal ratio. The smaller the difference, the better the proportion coordination.

[0089] Obtain the grid density uniformity index: Calculate the standard deviation corresponding to the area of ​​each grid. Take the inverse of the standard deviation and normalize it to obtain the grid density uniformity index.

[0090] Exemplarily, the method of obtaining the symmetry index, proportion coordination index, and grid density uniformity index corresponding to the predicted grid division information includes: obtaining the grid similarity on both sides of the symmetry axis corresponding to the predicted grid division information, using the grid similarity as the symmetry index; obtaining the difference coefficient between the aspect ratio of each grid corresponding to the predicted grid division information and a preset ideal ratio, using the difference coefficient as the proportion coordination index; obtaining the standard deviation corresponding to the area of ​​each grid, taking the inverse of the standard deviation and normalizing it, and outputting the grid density uniformity index.

[0091] Calculate symmetry index: Compare the grid layouts on both sides of the symmetry axis and calculate the similarity. The higher the similarity, the better the symmetry.

[0092] Calculate the proportion coordination index: preset the ideal ratio (such as the golden ratio), and calculate the difference between the actual length-width ratio of each grid and the ideal ratio. The smaller the difference, the better the proportion coordination.

[0093] Calculate the grid density uniformity index: Calculate the area standard deviation of each grid. Take the inverse of the standard deviation and normalize it to get the grid density uniformity index.

[0094] Exemplarily, the expression of the aesthetic coefficient includes:

[0095] ;

[0096] in, is the aesthetics coefficient, is the first weight, is the symmetry index, is the proportional coordination index, is the grid density uniformity index, is the second weight, The third weight.

[0097] Determined by experience or experiment , , The weights can reflect the importance of different indicators in the aesthetic evaluation. Substitute the values ​​of each indicator to calculate the aesthetic coefficient. By adjusting the weights, the impact of different indicators on aesthetics can be changed.

[0098] In some embodiments, before the hole distribution information, geometric feature information and avoidance method information are input into a preset first optimization module, it also includes: obtaining the design environment information corresponding to each of the historical hyperbolic curtain wall models; generating a format conversion relationship corresponding to the historical design data according to the design environment information; and converting each of the historical design data into a preset target format according to the format conversion relationship.

[0099] Obtain design environment information: Collect design environment information for each historical hyperbolic curtain wall model, such as project location, climate conditions, material type, etc.

[0100] Generate format conversion relations: Define the conversion rules from the original data format to the target format based on the design environment information. For example, unify dimensions of different units into the same unit, convert text descriptions into numerical features, etc.

[0101] Data conversion: Use the generated format conversion relationship to convert each historical design data into the target format. Ensure that all data is processed in the same scale and format. Ensure data consistency and standardization, improve the accuracy and efficiency of model training, reduce errors caused by inconsistent data formats, and enhance the robustness of the system.

[0102] In some embodiments, after constructing the intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module, it also includes: generating an optimization effect score based on the hyperbolic curtain wall model in modeling, the optimization suggestions and the optimized hyperbolic curtain wall model; and completing the optimization of the auxiliary recognition model based on the optimization effect score, the hyperbolic curtain wall model in modeling and the optimization suggestions.

[0103] Generate optimization effect scores: Calculate the changes in various performance indicators based on the hyperbolic curtain wall model before and after optimization. For example, calculate the degree of improvement in aesthetics and structural stability. Generate optimization effect scores based on various indicators. Adjust the parameters of the intelligent auxiliary recognition model based on the optimization effect scores. For example, adjust the dynamic weights in the multi-task learning model, improve the definition of the loss function, etc. Retrain the model to improve its performance in practical applications. Continuously optimize the model through the feedback mechanism to improve the practicality and accuracy of the model. This enables the intelligent auxiliary recognition system to better adapt to different design needs and provide higher quality design suggestions.

[0104] See also Figure 2 As shown, Figure 2 1 is a schematic diagram of the structure of an intelligent auxiliary recognition device 200 for use in a hyperbolic curtain wall parametric modeling design process provided by an embodiment of the present application. The intelligent auxiliary recognition device 200 for use in a hyperbolic curtain wall parametric modeling design process is used to execute the steps of the intelligent auxiliary recognition method for use in a hyperbolic curtain wall parametric modeling design process shown in the above-mentioned embodiments. The intelligent auxiliary recognition device 200 for use in a hyperbolic curtain wall parametric modeling design process can be a single server or a server cluster, or the intelligent auxiliary recognition device 200 for use in a hyperbolic curtain wall parametric modeling design process can be a terminal, which can be a handheld terminal, a laptop computer, a wearable device, a robot, etc.

[0105] like Figure 2 As shown, the intelligent auxiliary recognition device 200 used in the parametric modeling design process of the hyperbolic curtain wall includes:

[0106] The data acquisition unit 201 is used to acquire a plurality of historical hyperbolic curtain wall models and corresponding historical design data; the historical design data at least includes node design information, connection optimization information, geometric feature information, avoidance method information, hole distribution information and connection optimization information;

[0107] The data input unit 202 is used to input the hole position distribution information, geometric feature information and avoidance method information into a preset first optimization module, and the first optimization module outputs predicted connection optimization information and predicted node design information; the first optimization module is trained according to the node design information, connection optimization information, predicted connection optimization information and predicted node design information;

[0108] The feature input unit 203 is used to input the geometric feature information into a preset second optimization module, the second optimization module outputs predicted grid division information, obtains the aesthetic coefficient corresponding to the predicted grid division information, and completes the training of the second optimization module according to the aesthetic coefficient;

[0109] A training completion unit 204 is used to input the geometric feature information and the node design information into a preset avoidance detection module, the avoidance detection module outputs predicted avoidance conflict information, and completes the training of the avoidance detection module according to the predicted avoidance conflict information and avoidance mode information;

[0110] The optimization output unit 205 is used to build an intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module, and is used to generate optimization suggestions for the hyperbolic curtain wall model being modeled during the hyperbolic curtain wall modeling process, wherein the optimization suggestions include one or more of connection optimization suggestions, node design suggestions, grid division suggestions and avoidance design suggestions.

[0111] In some embodiments, constructing an intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module includes: inputting the node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetics coefficient, predicted avoidance conflict information and avoidance method information into a preset multi-task learning model, the multi-task learning model calculates the joint loss function corresponding to the first optimization module, the second optimization module and the avoidance detection module; completing the training of the first optimization module, the second optimization module and the avoidance detection module according to the joint loss function to form an intelligent auxiliary recognition model.

[0112] Exemplarily, the node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetics coefficient, predicted avoidance conflict information and avoidance method information are input into a preset multi-task learning model, including: calculating a first loss function according to the node design information, connection optimization information, predicted connection optimization information and predicted node design information; calculating a second loss function according to the aesthetics coefficient; calculating a third loss function according to the predicted avoidance conflict information and avoidance method information; inputting the first loss function, the second loss function and the third loss function into the multi-task learning model to calculate the joint loss function according to the first loss function, the second loss function and the third loss function.

[0113] It should be noted that, in some embodiments, the calculation of the joint loss function based on the first loss function, the second loss function and the third loss function includes: setting the first dynamic weight corresponding to the first loss function; setting the second dynamic weight corresponding to the second loss function; setting the third dynamic weight corresponding to the third loss function; calculating the joint loss function based on the first dynamic weight, the first loss function, the second dynamic weight, the second loss function, the third dynamic weight and the third loss function.

[0114] It should be noted that, in some embodiments, the expression of the joint loss function includes:

[0115] ;in, is the joint loss function, is the first dynamic weight, is the first loss function, is the second dynamic weight, is the second loss function, is the third dynamic weight, is the third loss function.

[0116] In some embodiments, obtaining the aesthetic coefficient corresponding to the predicted grid division information includes: obtaining a symmetry index, a proportion coordination index, and a grid density uniformity index corresponding to the predicted grid division information; and calculating the aesthetic coefficient based on the symmetry index, the proportion coordination index, and the grid density uniformity index.

[0117] Exemplarily, the method of obtaining the symmetry index, proportion coordination index, and grid density uniformity index corresponding to the predicted grid division information includes: obtaining the grid similarity on both sides of the symmetry axis corresponding to the predicted grid division information, using the grid similarity as the symmetry index; obtaining the difference coefficient between the aspect ratio of each grid corresponding to the predicted grid division information and a preset ideal ratio, using the difference coefficient as the proportion coordination index; obtaining the standard deviation corresponding to the area of ​​each grid, taking the inverse of the standard deviation and normalizing it, and outputting the grid density uniformity index.

[0118] Exemplarily, the expression of the aesthetic coefficient includes:

[0119] ;

[0120] in, is the aesthetics coefficient, is the first weight, is the symmetry index, is the proportional coordination index, is the grid density uniformity index, is the second weight, The third weight.

[0121] In some embodiments, before the hole distribution information, geometric feature information and avoidance method information are input into a preset first optimization module, it also includes: obtaining the design environment information corresponding to each of the historical hyperbolic curtain wall models; generating a format conversion relationship corresponding to the historical design data according to the design environment information; and converting each of the historical design data into a preset target format according to the format conversion relationship.

[0122] In some embodiments, after constructing the intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module, it also includes: generating an optimization effect score based on the hyperbolic curtain wall model in modeling, the optimization suggestions and the optimized hyperbolic curtain wall model; and completing the optimization of the auxiliary recognition model based on the optimization effect score, the hyperbolic curtain wall model in modeling and the optimization suggestions.

[0123] It should be noted that, those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working processes of the intelligent auxiliary identification device and each module used in the parametric modeling and design process of the hyperbolic curtain wall described above can refer to the corresponding processes in the embodiments of the intelligent auxiliary identification method used in the parametric modeling and design process of the hyperbolic curtain wall described in the above embodiments, and will not be repeated here.

[0124] The above-mentioned intelligent auxiliary recognition method for the parametric modeling design process of the hyperbolic curtain wall can be implemented in the form of a computer program. The computer program can be used in Figure 2 Run on the device shown.

[0125] See also Figure 3 , Figure 3 1 is a schematic block diagram of the structure of a computer device provided in an embodiment of the present application. The computer device includes a processor, a memory and a network interface connected via a device bus, wherein the memory may include a storage medium and an internal memory.

[0126] The storage medium can store an operating device and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any intelligent auxiliary recognition method used in the parametric modeling design process of the hyperbolic curtain wall.

[0127] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0128] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any intelligent auxiliary recognition method used in the parametric modeling design process of the hyperbolic curtain wall.

[0129] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the terminal to which the scheme of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0130] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0131] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0132] Acquire multiple historical hyperbolic curtain wall models and corresponding historical design data; the historical design data at least includes node design information, connection optimization information, geometric feature information, avoidance method information, hole distribution information and connection optimization information;

[0133] Input the hole position distribution information, geometric feature information and avoidance method information into a preset first optimization module, and the first optimization module outputs predicted connection optimization information and predicted node design information; complete the training of the first optimization module according to the node design information, connection optimization information, predicted connection optimization information and predicted node design information;

[0134] The geometric feature information is input into a preset second optimization module, the second optimization module outputs predicted grid division information, an aesthetic coefficient corresponding to the predicted grid division information is obtained, and the training of the second optimization module is completed according to the aesthetic coefficient;

[0135] The geometric feature information and the node design information are input into a preset avoidance detection module, the avoidance detection module outputs predicted avoidance conflict information, and the avoidance detection module is trained according to the predicted avoidance conflict information and the avoidance mode information;

[0136] An intelligent auxiliary recognition model is constructed based on the trained first optimization module, the second optimization module and the avoidance detection module, which is used to generate optimization suggestions for the hyperbolic curtain wall model under modeling during the hyperbolic curtain wall modeling process, wherein the optimization suggestions include one or more of connection optimization suggestions, node design suggestions, grid division suggestions and avoidance design suggestions.

[0137] In some embodiments, constructing an intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module includes: inputting the node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetics coefficient, predicted avoidance conflict information and avoidance method information into a preset multi-task learning model, the multi-task learning model calculates the joint loss function corresponding to the first optimization module, the second optimization module and the avoidance detection module; completing the training of the first optimization module, the second optimization module and the avoidance detection module according to the joint loss function to form an intelligent auxiliary recognition model.

[0138] Exemplarily, the node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetics coefficient, predicted avoidance conflict information and avoidance method information are input into a preset multi-task learning model, including: calculating a first loss function according to the node design information, connection optimization information, predicted connection optimization information and predicted node design information; calculating a second loss function according to the aesthetics coefficient; calculating a third loss function according to the predicted avoidance conflict information and avoidance method information; inputting the first loss function, the second loss function and the third loss function into the multi-task learning model to calculate the joint loss function according to the first loss function, the second loss function and the third loss function.

[0139] It should be noted that, in some embodiments, the calculation of the joint loss function based on the first loss function, the second loss function and the third loss function includes: setting the first dynamic weight corresponding to the first loss function; setting the second dynamic weight corresponding to the second loss function; setting the third dynamic weight corresponding to the third loss function; calculating the joint loss function based on the first dynamic weight, the first loss function, the second dynamic weight, the second loss function, the third dynamic weight and the third loss function.

[0140] It should be noted that, in some embodiments, the expression of the joint loss function includes:

[0141] ;in, is the joint loss function, is the first dynamic weight, is the first loss function, is the second dynamic weight, is the second loss function, is the third dynamic weight, is the third loss function.

[0142] In some embodiments, obtaining the aesthetic coefficient corresponding to the predicted grid division information includes: obtaining a symmetry index, a proportion coordination index, and a grid density uniformity index corresponding to the predicted grid division information; and calculating the aesthetic coefficient based on the symmetry index, the proportion coordination index, and the grid density uniformity index.

[0143] Exemplarily, the method of obtaining the symmetry index, proportion coordination index, and grid density uniformity index corresponding to the predicted grid division information includes: obtaining the grid similarity on both sides of the symmetry axis corresponding to the predicted grid division information, using the grid similarity as the symmetry index; obtaining the difference coefficient between the aspect ratio of each grid corresponding to the predicted grid division information and a preset ideal ratio, using the difference coefficient as the proportion coordination index; obtaining the standard deviation corresponding to the area of ​​each grid, taking the inverse of the standard deviation and normalizing it, and outputting the grid density uniformity index.

[0144] Exemplarily, the expression of the aesthetic coefficient includes:

[0145] ;

[0146] in, is the aesthetics coefficient, is the first weight, is the symmetry index, is the proportional coordination index, is the grid density uniformity index, is the second weight, The third weight.

[0147] In some embodiments, before the hole distribution information, geometric feature information and avoidance method information are input into a preset first optimization module, it also includes: obtaining the design environment information corresponding to each of the historical hyperbolic curtain wall models; generating a format conversion relationship corresponding to the historical design data according to the design environment information; and converting each of the historical design data into a preset target format according to the format conversion relationship.

[0148] In some embodiments, after constructing the intelligent auxiliary recognition model based on the trained first optimization module, the second optimization module and the avoidance detection module, it also includes: generating an optimization effect score based on the hyperbolic curtain wall model in modeling, the optimization suggestions and the optimized hyperbolic curtain wall model; and completing the optimization of the auxiliary recognition model based on the optimization effect score, the hyperbolic curtain wall model in modeling and the optimization suggestions.

[0149] It should be noted that technical personnel in the relevant field can clearly understand that, for the convenience and conciseness of description, the specific working process of the computer equipment and each module described above can refer to the corresponding process in the embodiment of the intelligent auxiliary identification method for the parametric modeling design process of the hyperbolic curtain wall described in the above embodiments, and will not be repeated here.

[0150] The present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the processor implements the steps of the intelligent auxiliary identification method for the parametric modeling design process of hyperbolic curtain walls as provided in any embodiment of the present application.

[0151] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc., equipped on the computer device.

[0152] Exemplarily, the medium is used to implement the following steps:

[0153] Acquire multiple historical hyperbolic curtain wall models and corresponding historical design data; the historical design data at least includes node design information, connection optimization information, geometric feature information, avoidance method information, hole distribution information and connection optimization information;

[0154] Input the hole position distribution information, geometric feature information and avoidance method information into a preset first optimization module, and the first optimization module outputs predicted connection optimization information and predicted node design information; complete the training of the first optimization module according to the node design information, connection optimization information, predicted connection optimization information and predicted node design information;

[0155] The geometric feature information is input into a preset second optimization module, the second optimization module outputs predicted grid division information, an aesthetic coefficient corresponding to the predicted grid division information is obtained, and the training of the second optimization module is completed according to the aesthetic coefficient;

[0156] The geometric feature information and the node design information are input into a preset avoidance detection module, the avoidance detection module outputs predicted avoidance conflict information, and the avoidance detection module is trained according to the predicted avoidance conflict information and the avoidance mode information;

[0157] An intelligent auxiliary recognition model is constructed based on the trained first optimization module, the second optimization module and the avoidance detection module, which is used to generate optimization suggestions for the hyperbolic curtain wall model under modeling during the hyperbolic curtain wall modeling process, wherein the optimization suggestions include one or more of connection optimization suggestions, node design suggestions, grid division suggestions and avoidance design suggestions.

[0158] It should be noted that technical personnel in the relevant field can clearly understand that, for the convenience and conciseness of description, the specific working process of the storage medium and each module described above can refer to the corresponding process in the embodiment of the intelligent auxiliary identification method for the parametric modeling design process of the hyperbolic curtain wall described in the above embodiments, and will not be repeated here.

[0159] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. An intelligent auxiliary recognition method for parametric modeling design of hyperbolic curtain walls, characterized in that: include: Obtain multiple historical hyperbolic curtain wall models and corresponding historical design data; The historical design data at least includes node design information, connection optimization information, geometric feature information, avoidance information, hole position distribution information and connection optimization information; The hole position distribution information, geometric feature information and avoidance method information are input into a preset first optimization module, and the first optimization module outputs predicted connection optimization information and predicted node design information; completing training of the first optimization module according to the node design information, the connection optimization information, the predicted connection optimization information, and the predicted node design information; The geometric feature information is input into a preset second optimization module, the second optimization module outputs predicted grid division information, an aesthetic coefficient corresponding to the predicted grid division information is obtained, and the training of the second optimization module is completed according to the aesthetic coefficient; The geometric feature information and the node design information are input into a preset avoidance detection module, the avoidance detection module outputs predicted avoidance conflict information, and the avoidance detection module is trained according to the predicted avoidance conflict information and the avoidance mode information; An intelligent auxiliary recognition model is constructed based on the trained first optimization module, the second optimization module and the avoidance detection module, which is used to generate optimization suggestions for the hyperbolic curtain wall model under modeling during the hyperbolic curtain wall modeling process, wherein the optimization suggestions include one or more of connection optimization suggestions, node design suggestions, grid division suggestions and avoidance design suggestions.

2. The method according to claim 1, characterized in that The intelligent auxiliary recognition model is constructed according to the trained first optimization module, the second optimization module and the avoidance detection module, including: The node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetic coefficient, predicted avoidance conflict information and avoidance method information are input into a preset multi-task learning model, and the multi-task learning model calculates the joint loss function corresponding to the first optimization module, the second optimization module and the avoidance detection module; The training of the first optimization module, the second optimization module and the avoidance detection module is completed according to the joint loss function to form an intelligent auxiliary recognition model.

3. The method according to claim 2, characterized in that The node design information, connection optimization information, predicted connection optimization information, predicted node design information, aesthetic coefficient, predicted avoidance conflict information and avoidance method information are input into a preset multi-task learning model, including: Calculate a first loss function based on the node design information, the connection optimization information, the predicted connection optimization information, and the predicted node design information; Calculate a second loss function according to the aesthetic coefficient; Calculating a third loss function according to the predicted avoidance conflict information and the avoidance mode information; The first loss function, the second loss function and the third loss function are input into the multi-task learning model to calculate the joint loss function according to the first loss function, the second loss function and the third loss function.

4. The method according to claim 3, characterized in that The calculating the joint loss function according to the first loss function, the second loss function and the third loss function includes: Setting a first dynamic weight corresponding to the first loss function; Setting a second dynamic weight corresponding to the second loss function; Setting a third dynamic weight corresponding to the third loss function; The joint loss function is calculated based on the first dynamic weight, the first loss function, the second dynamic weight, the second loss function, the third dynamic weight and the third loss function.

5. The method according to claim 4, characterized in that The expression of the joint loss function includes: ； in, is the joint loss function, is the first dynamic weight, is the first loss function, is the second dynamic weight, is the second loss function, is the third dynamic weight, is the third loss function.

6. The method according to claim 1, characterized in that The obtaining of the aesthetic coefficient corresponding to the predicted grid division information includes: Obtaining a symmetry index, a proportion coordination index, and a grid density uniformity index corresponding to the predicted grid division information; The aesthetic coefficient is calculated based on the symmetry index, the proportion coordination index, and the grid density uniformity index.

7. The method according to claim 6, characterized in that The step of obtaining the symmetry index, the proportion coordination index, and the grid density uniformity index corresponding to the predicted grid division information includes: Obtaining the grid similarity on both sides of the symmetry axis corresponding to the predicted grid division information, the grid similarity being used as the symmetry index; Obtaining a difference coefficient between the aspect ratio of each grid corresponding to the predicted grid division information and a preset ideal ratio, the difference coefficient being used as the ratio coordination index; The standard deviation corresponding to the area of ​​each grid is obtained, the inverse of the standard deviation is taken and normalized, and the grid density uniformity index is output.

8. The method according to claim 6, characterized in that The expression of the aesthetic coefficient includes: ； in, is the aesthetics coefficient, is the first weight, is the symmetry index, is the proportional coordination index, is the grid density uniformity index, is the second weight, The third weight.

9. The method according to claim 1, characterized in that: Before inputting the hole position distribution information, geometric feature information and avoidance method information into a preset first optimization module, the method further includes: Obtaining design environment information corresponding to each of the historical hyperbolic curtain wall models; generating a format conversion relationship corresponding to the historical design data according to the design environment information; Each of the historical design data is converted into a preset target format according to the format conversion relationship.

10. The method according to claim 1, characterized in that After the intelligent auxiliary recognition model is constructed according to the trained first optimization module, the second optimization module and the avoidance detection module, the method further includes: Generate optimization effect scores based on the hyperbolic curtain wall model in modeling, optimization suggestions and the optimized hyperbolic curtain wall model; The optimization of the auxiliary identification model is completed according to the optimization effect score, the hyperbolic curtain wall model in the modeling and the optimization suggestion.

Images