Building design management method and system based on Internet big data and AI technology

Abstract

The invention relates to the technical field of building design, in particular to a building design management method and system based on Internet big data and an AI technology. The automatic generation and intelligent optimization of the BIM model are realized through the AI technology, the modeling period is greatly shortened, the AI model is trained and optimized through the deep learning algorithm, the accuracy, integrity and flexibility of the AI recognition model generated after training are improved, the BIM model is output through the feature extraction of the AI recognition model on the building image, and the accuracy, integrity and flexibility of the AI recognition model are improved. The BIM model is optimized through an optimization algorithm and input parameters, diversified modeling requirements of different users are met, the accuracy of the BIM model is improved again, dynamic data of a public website and the BIM model are combined to generate a building design scheme, and an external platform is connected based on a Web technology and a real-time communication technology, so that the construction efficiency is improved. The information barrier between participants of the project is broken, and real-time sharing and cooperative work of data are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of architectural design, and specifically to an architectural design management method and system based on Internet big data and AI technology. Background Art

[0002] With the increasing scale and complexity of construction projects, traditional architectural design and construction management methods can no longer meet the needs of modern construction projects. After more than a decade of development, BIM technology has become relatively complete in its related functions and outputs, with a mature application and management system. However, its source: BIM model construction, is still in the field of manual labor, requiring manual modeling, and the accuracy and integrity of the model are difficult to guarantee. Existing modeling software lacks effective use of big data and in-depth application of AI technology, making it difficult to achieve refined management and innovative design of construction projects. Summary of the invention

[0003] The technical problem to be solved by the present invention is that the existing modeling software relies on manual modeling and lacks effective use of big data and in-depth application of AI technology, making it difficult to achieve refined management and innovative design of construction projects.

[0004] In order to solve the above problems, the present invention provides a solution that can automatically build models based on AI technology and can achieve refined management and innovative design of construction projects; The first technical solution of the present invention is: a method for architectural design management based on Internet big data and AI technology, comprising the following steps: Step 1: Obtain building sample images, build an AI model based on a deep learning framework, train and optimize the AI ​​model using the building sample images through the deep learning framework, and generate an AI recognition model; Step 2: Input the building image to be modeled into the AI ​​recognition model, which extracts features from the building image and constructs a BIM model based on the extracted features; Step 3: Optimize the BIM model using optimization algorithms and input parameters, including room size, door and window size, door and window location, and partition wall location parameters; Step 4: Obtain and store dynamic data on architectural design specifications, architectural design standards, building material prices, and the market from public websites; generate architectural design plans based on dynamic data and BIM models; Step 5: Send the architectural design plan to the external platform based on Web technology and real-time communication technology.

[0005] Preferably, a further technical solution of the present invention is: The model optimization steps of step two are as follows: obtain sample building images, mark the building component categories, building component locations, and building component sizes in the images, divide the marked images into training sets and validation sets according to the proportions, input the training set into the built AI model for model training, and generate prediction results. The validation set is used to detect the performance of the AI ​​model during the training process, and the verified AI model is output as an AI recognition model.

[0006] The optimization algorithm of step four includes finite element analysis, genetic algorithm and simulated annealing algorithm, and the BIM model is optimized through finite element analysis, genetic algorithm and simulated annealing algorithm.

[0007] The optimization of the BIM model through finite element analysis is specifically as follows: assigning properties, boundary conditions and load condition information of the building structure in the BIM model, discretizing the building structure into a finite number of units, solving the equilibrium equations of each unit, calculating the stress and strain distribution of the building structure under different conditions, and adjusting the parameters of the building structure in the BIM model according to the stress and strain distribution, thereby improving the structural stability.

[0008] The optimization of the BIM model by genetic algorithm is specifically as follows: the building structure in the BIM model is divided into an initial population according to the spatial layout, each individual in the initial population represents a possible spatial layout scheme, the fitness value of each individual is calculated according to the space utilization rate and the comfort level, the individuals in the initial population are selected, crossed and mutated, the selection operation tends to retain individuals with high fitness values, and after multiple rounds of selection, crossover and mutation operations, the spatial layout of the initial population gradually evolves to a better spatial layout scheme.

[0009] The optimization of the BIM model by the simulated annealing algorithm is specifically as follows: taking the combination of building envelope structure and equipment parameters in the BIM model as the initial solution, calculating the building energy consumption value of the initial solution, randomly perturbing the initial solution, generating a new combination solution, and accepting the new combination solution if the energy consumption value of the new combination solution is lower than the energy consumption value of the initial solution; if the new combination solution is higher than the energy consumption value of the current solution, the new combination solution is accepted with probability, and the probability gradually decreases with the iteration of the algorithm, and the new combination solution and the current solution are iteratively converged until they converge to an approximately optimal building envelope structure and equipment parameter combination solution.

[0010] The second technical solution of the present invention is: an architectural design management system based on Internet big data and AI technology, applied to the above-mentioned architectural design management method, including an architectural design management platform, a data acquisition module, a data storage module, an AI module and a collaborative management unit; The data acquisition module, data storage module, AI module and collaborative management unit are built on the architectural design management platform; The data acquisition module is used to obtain building sample images, building images and dynamic data, and send them to the data storage module for storage; The AI ​​module is used to train and optimize the AI ​​model based on the deep learning algorithm, generate the AI ​​recognition model, optimize the BIM model through the optimization algorithm and input parameters, and generate the building design plan based on the dynamic data and BIM model; The collaborative management unit is equipped with a Web technology stack and a communication unit, and is connected to an external platform through the Web technology stack and the communication unit.

[0011] Compared with the prior art, the present invention using the above technical solution has the following beneficial effects: The present invention combines AI technology with BIM modeling, and realizes the automatic generation and intelligent optimization of BIM models through AI technology, which greatly shortens the modeling cycle, reduces the errors and repetitive labor of manual modeling, and trains and optimizes the AI ​​model through a deep learning algorithm, thereby improving the accuracy, completeness and flexibility of the AI ​​recognition model generated after training. The BIM model is constructed by extracting features of building images through the AI ​​recognition model, and the BIM model is optimized through optimization algorithms and input parameters, which meets the diverse modeling needs of different users and once again improves the accuracy of the BIM model. The dynamic data of the public website is combined with the BIM model to generate a building design plan, and the external platform is connected based on Web technology and real-time communication technology, which breaks the information barriers between the project participants and realizes real-time data sharing and collaborative work. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a flow chart of Embodiment 1 of the present invention; Figure 2 This is an AI modeling framework diagram of the first embodiment of the present invention. DETAILED DESCRIPTION

[0013] The present invention will be further described below in conjunction with embodiments, the purpose of which is only to provide a better understanding of the content of the present invention. Therefore, the examples given do not limit the protection scope of the present invention.

[0014] See attached Figure 1-2 Embodiment 1 of the present invention discloses a method for managing architectural design based on Internet big data and AI technology, comprising the following steps: Step 1: Obtain building sample images, build an AI model based on a deep learning framework, train and optimize the AI ​​model using the building sample images through the deep learning framework, and generate an AI recognition model; Step 2: Input the building image to be modeled into the AI ​​recognition model, which extracts features from the building image and constructs a BIM model based on the extracted features; Step 3: Use optimization algorithms and input parameters to optimize and verify the BIM model. The input parameters include room size, door and window size, door and window location, and partition wall location parameters. Step 4: Obtain and store dynamic data on architectural design specifications, architectural design standards, building material prices, and the market from public websites; generate architectural design plans based on dynamic data and BIM models; Step 5: Send the architectural design plan to the external platform based on Web technology and real-time communication technology.

[0015] In this embodiment, the model optimization step of step one is specifically as follows: collect a large amount of sample data such as architectural drawings and construction site images. As architectural sample images, the diversity and quantity of sample data are crucial to the accuracy of the trained AI recognition model. The sample data needs to cover architectural drawings of various types, styles, and sizes and construction site images of different scenes and angles, and cover the data situations that may be encountered in actual applications as comprehensively as possible. The category, location, size and other information of the architectural component in the architectural sample images are annotated. The more accurate and detailed the annotation, the better the model learning effect and the higher the subsequent recognition accuracy; select a deep learning framework , such as TensorFlow or PyTorch, build AI models based on deep learning frameworks, and use their rich algorithm libraries and tools to implement image recognition functions. Within the selected deep learning framework, build AI models according to task requirements and data characteristics. The AI ​​model is a convolutional neural network model. The convolutional neural network model contains components such as convolutional layers, pooling layers, and fully connected layers. The convolutional layer is used to extract image features and slide convolutions on the image through different convolution kernels to capture local features in the image; the pooling layer downsamples the output of the convolutional layer to reduce the amount of data and the amount of model calculation while retaining the main features; the fully connected layer converts the pooling layer input The feature vectors are classified or regressed to output the final prediction results; the parameters required for AI model training are set, such as learning rate, number of iterations, batch size, etc. The learning rate determines the step size of the parameter update of the AI ​​model during the training process. A suitable learning rate can make the AI ​​model converge to the optimal solution faster. Too large a learning rate may cause the AI ​​model training to be unstable, while too small a learning rate will make the training speed too slow; the number of iterations is the number of rounds that the AI ​​model trains the entire training data set, which needs to be adjusted according to the actual situation to ensure that the AI ​​model fully learns the data features; the batch size refers to the number of samples input into the AI ​​model during each training, which affects the AI ​​model training Stability and efficiency; divide the labeled building sample images into training sets and validation sets according to the proportion, input the training set into the built AI model for model training. During the training process, the AI ​​model will generate prediction results based on forward propagation, and compare them with the labeled real results. The gradient of the loss function is calculated through the back-propagation algorithm, and then the parameters of the model are updated, and the weights and biases of the model are continuously adjusted to make the prediction results of the model gradually approach the real results; during the training process, the AI ​​model is regularly evaluated using the validation set, and the accuracy, recall rate, F1 value and other indicators of the AI ​​model on the validation set are calculated to evaluate the performance of the AI ​​model. If the performance of the AI ​​model does not meet expectations, the AI ​​model structure can be adjusted, such as adding or reducing convolutional layers, adjusting the size of the convolution kernel, etc.; the amount of training data can be increased, and more building sample images can be collected for annotation and training; the training parameters can be optimized, such as adjusting the learning rate, number of iterations, etc. to optimize the model, and the AI ​​model's recognition accuracy for different types of building images can be continuously improved.

[0016] In this embodiment, the feature extraction of step 2 is specifically as follows: adopt multi-source data-driven modeling, input the building image to be modeled into the AI ​​recognition model, such as CAD drawings, hand-drawn sketches, etc., and use the image recognition technology and natural language processing technology of the AI ​​recognition model to automatically extract building information, including building outlines, floor structures, room layouts, component parameters, etc., and then generate a BIM model. For example, through layer analysis and graphic recognition of CAD drawings, the position and size of components such as walls, doors and windows, beams and columns can be accurately identified; the text description in the sketch is semantically understood to supplement the relevant information of the building design drawing.

[0017] In this embodiment, the BIM model is optimized and verified by using input parameters as follows: a rich library of building components and parametric modeling tools are provided, and users can select appropriate components and templates according to project requirements, and quickly generate personalized BIM models by adjusting parameters. At the same time, it supports user-defined components and templates to meet special architectural design requirements. For example, when designing a residential project, users can select standard apartment modules from the component library, and quickly generate BIM models of different apartment types by modifying parameters such as room size, door and window positions; for complex and special-shaped buildings, users can customize component shapes and parameters to achieve innovative designs.

[0018] In this embodiment, the optimization algorithm of step three includes finite element analysis, genetic algorithm and simulated annealing algorithm, and the BIM model is optimized and verified by finite element analysis, genetic algorithm and simulated annealing algorithm.

[0019] In this embodiment, the optimization of the BIM model through finite element analysis is specifically as follows: importing the BIM model into the finite element analysis software, giving the BIM model the properties, boundary conditions, load conditions and other information of the building structure, the finite element analysis software discretizes the building structure into a finite number of units, solves the equilibrium equations of each unit, and calculates the stress and strain distribution of the building structure under different conditions. For example, when simulating earthquake effects, the stress state of each part of the structure can be clearly presented; according to the finite element analysis calculation results, the weak parts of the building structure in the BIM model are found, such as stress concentration areas or parts with excessive deformation. For these weak parts, the cross-sectional size of the components, material strength and other parameters are adjusted, for example, the cross-sectional area of ​​the beams and columns is increased, or building materials with higher strength grades are selected to improve the structural performance.

[0020] In this embodiment, the optimization of the BIM model by genetic algorithm is specifically as follows: the building structure in the BIM model is encoded according to the spatial layout to form an initial population, each individual in the initial population represents a possible spatial layout scheme, and the encoding method can be based on the digital representation of the room position, size and connection relationship, and the fitness function is set according to the space utilization rate, comfort, etc., and the fitness value of each individual is calculated. For example, the layout scheme with high space utilization rate, reasonable functional division and ergonomic principles has a higher fitness value; the initial population is subjected to genetic operations such as selection, crossover and mutation, and the selection operation tends to retain individuals with high fitness values; the crossover operation generates a new layout scheme by exchanging some genes of two individuals; the mutation operation randomly changes certain genes of the individual with a certain probability to introduce new layout possibilities. After multiple rounds of selection, crossover and mutation, the population gradually evolves to a better spatial layout scheme.

[0021] In this embodiment, the optimization of the BIM model by the simulated annealing algorithm is specifically as follows: taking the combination of building envelope structure and equipment parameters in the BIM model as the initial solution, such as the type of exterior wall insulation material, the heat transfer coefficient of the window, the operation mode of the air-conditioning system, etc., and calculating the building energy consumption value of the initial solution, randomly perturbing the initial solution to generate a new combination solution, if the energy consumption value of the new combination solution is lower than the energy consumption value of the initial solution, then accept the new combination solution; if the new combination solution is higher than the energy consumption value of the current solution, then accept the new combination solution with probability, and the probability gradually decreases with the iteration of the algorithm. In this way, the algorithm can jump out of the local optimal solution to a certain extent, and is more likely to find the global optimal energy consumption optimization solution; iteratively converge the new combination solution with the previous combination solution, and continuously repeat the process of random perturbation and acceptance criteria. As the temperature parameter gradually decreases, the search range of the genetic algorithm gradually narrows until it converges to an approximately optimal building envelope structure and equipment parameter combination solution, thereby achieving the purpose of reducing building energy consumption.

[0022] Embodiment 2 of the present invention discloses an architectural design management system based on Internet big data and AI technology, which is applied to the architectural design management method of embodiment 1, including an architectural design management platform, a data acquisition module, a data storage module, an AI module and a collaborative management unit; The data acquisition module, data storage module, AI module and collaborative management unit are built on the architectural design management platform; The data acquisition module is used to obtain building sample images, building images and dynamic data, and send them to the data storage module for storage; The data storage module is built with distributed storage technology, such as Hadoop distributed file system, to ensure high reliability and scalability of data, providing a solid foundation for subsequent data analysis and application; A deep learning framework is built in the AI ​​module, and the AI ​​module is trained and optimized based on the deep learning framework to generate an AI recognition model. The BIM model is optimized through optimization algorithms and input parameters, and is used to generate architectural design plans based on dynamic data and BIM models. The collaborative management unit is equipped with a Web technology stack and a communication unit, and provides a programming modeling interface API and a rich API function library, including model creation, editing, query, analysis and other functional functions. Through the Web technology stack and the communication unit, it is convenient for third-party applications to integrate with external platforms, supports multiple users to edit BIM models online at the same time, and realizes real-time sharing and synchronous updating of models; including data query interface, model operation interface, project management interface, etc., supports seamless connection with architectural design software, construction management system, enterprise resource planning (ERP) system, etc., realizes data interconnection and business process collaboration, develops project creation, editing, deletion and other functions on the collaborative management unit, supports the entry and management of basic project information (such as project name, location, scale, owner information, etc.), establishes project plan formulation functions, including work breakdown structure (WBS) creation, task allocation, duration estimation, resource allocation and other sub-functions, and displays project schedule plans through Gantt charts, network diagrams and other visual methods.

[0023] The architectural design management platform establishes a building component library and parametric modeling templates based on architectural design specifications and standards, and develops parametric modeling tools using programming languages ​​(such as Python, C#) and graphics processing libraries (such as OpenCASCADE), enabling users to quickly generate BIM models by adjusting parameters. At the same time, it supports users to customize components and templates, and realize personalized modeling needs by writing scripts or using visual programming tools.

[0024] In this embodiment, the data collection module uses Internet data collection technology to obtain the latest building specifications, standards, material prices, market trends and other data from the construction industry website, government department website, standard specification agency website, etc. through web crawler technology. The collected unstructured data is converted into structured data using data cleaning and conversion technology and stored in the data storage module.

[0025] The present invention combines AI technology with BIM modeling, and realizes the automatic generation and intelligent optimization of BIM models through AI technology, which greatly shortens the modeling cycle, reduces the errors and repetitive labor of manual modeling, and trains and optimizes the AI ​​model through a deep learning algorithm, thereby improving the accuracy, completeness and flexibility of the AI ​​recognition model generated after training. The AI ​​recognition model extracts the features of the building image and outputs the BIM model, and optimizes the BIM model through optimization algorithms and input parameters, thereby meeting the diverse modeling needs of different users and once again improving the accuracy of the BIM model. The dynamic data of the public website is combined with the BIM model to generate a building design plan, and is connected to the external platform based on Web technology and real-time communication technology, thereby breaking the information barriers between the project participants and realizing real-time data sharing and collaborative work.

[0026] The above description is only a preferred feasible embodiment of the present invention, and does not limit the scope of rights of the present invention. All equivalent changes made using the contents of the present specification and its drawings are included in the scope of rights of the present invention.

Claims

1. A building design management method based on Internet big data and AI technology, characterized in that: The following steps are involved: Step 1: Obtain building sample images, build an AI model based on a deep learning framework, train and optimize the AI ​​model using the building sample images through the deep learning framework, and generate an AI recognition model; Step 2: Input the building image to be modeled into the AI ​​recognition model, which extracts features from the building image and constructs a BIM model based on the extracted features; Step 3: Optimize the BIM model using optimization algorithms and input parameters, including room size, door and window size, door and window location, and partition wall location parameters; Step 4: Obtain and store dynamic data on architectural design specifications, architectural design standards, building material prices, and the market from public websites; generate architectural design plans based on dynamic data and BIM models; Step 5: Send the architectural design plan to the external platform based on Web technology and real-time communication technology.

2. The architectural design management method based on Internet big data and AI technology according to claim 1 is characterized in that: The model optimization steps in step one are as follows: obtain sample building images, mark the building component categories, building component locations, and building component sizes in the images, divide the marked images into training sets and validation sets according to the proportions, input the training set into the built AI model for model training, and generate prediction results. The validation set is used to detect the performance of the AI ​​model during the training process, and the verified AI model is output as an AI recognition model.

3. The architectural design management method based on Internet big data and AI technology according to claim 1 is characterized by: The optimization algorithm of step three includes finite element analysis, genetic algorithm and simulated annealing algorithm, and the BIM model is optimized through finite element analysis, genetic algorithm and simulated annealing algorithm.

4. The architectural design management method based on Internet big data and AI technology according to claim 3 is characterized in that: The optimization of the BIM model through finite element analysis is specifically as follows: assigning properties, boundary conditions and load condition information of the building structure in the BIM model, discretizing the building structure into a finite number of units, solving the equilibrium equations of each unit, calculating the stress and strain distribution of the building structure under different conditions, and adjusting the parameters of the building structure in the BIM model according to the stress and strain distribution, thereby improving the structural stability.

5. The architectural design management method based on Internet big data and AI technology according to claim 3 is characterized in that: The optimization of the BIM model by genetic algorithm is specifically as follows: the building structure in the BIM model is divided into an initial population according to the spatial layout, each individual in the initial population represents a possible spatial layout scheme, the fitness value of each individual is calculated according to the space utilization rate and the comfort level, the individuals in the initial population are selected, crossed and mutated, the selection operation tends to retain individuals with high fitness values, and after multiple rounds of selection, crossover and mutation operations, the spatial layout of the initial population gradually evolves to a better spatial layout scheme.

6. The architectural design management method based on Internet big data and AI technology according to claim 3 is characterized in that: The optimization of the BIM model by the simulated annealing algorithm is specifically as follows: taking the combination of building envelope structure and equipment parameters in the BIM model as the initial solution, calculating the building energy consumption value of the initial solution, randomly perturbing the initial solution, generating a new combination solution, and accepting the new combination solution if the energy consumption value of the new combination solution is lower than the energy consumption value of the current solution; if the new combination solution is higher than the energy consumption value of the initial solution, the new combination solution is accepted with probability, and the probability gradually decreases with the iteration of the algorithm, and the new combination solution and the current solution are iteratively converged until they converge to an approximately optimal building envelope structure and equipment parameter combination solution.

7. An architectural design management system based on Internet big data and AI technology, applied to the architectural design management method according to any one of claims 1 to 6, characterized in that: It includes architectural design management platform, data acquisition module, data storage module, AI module and collaborative management unit; The data acquisition module, data storage module, AI module and collaborative management unit are built on the architectural design management platform; The data acquisition module is used to obtain building sample images, building images and dynamic data, and send them to the data storage module for storage; The AI ​​module is used to train and optimize the AI ​​model based on the deep learning algorithm, generate the AI ​​recognition model, optimize the BIM model through the optimization algorithm and input parameters, and generate the building design plan based on the dynamic data and BIM model; The collaborative management unit is equipped with a Web technology stack and a communication unit, and is connected to an external platform through the Web technology stack and the communication unit.

Images