A computational decision-making aid method and system for weather-adaptive building cavity design

Abstract

The present invention relates to the technical field of building natural ventilation, and in particular to a computational auxiliary decision-making method and system for weather-adaptive building cavity design, the method comprising: generating an initial grid using a Delaunay triangulation method, adaptively refining the initial grid by a Laplace operator-based grid refinement method, obtaining an adaptive grid system, the adaptive grid system being used to dynamically adjust the grid density of a conditional PINNs network; constructing a multi-task learning framework in the conditional PINNs network, the multi-task learning framework being used to collaboratively predict multiple physical field variables in the conditional PINNs network; iteratively training the conditional PINNs network, and simultaneously using a gradient descent algorithm to minimize a loss function until a predetermined number of training iterations or loss convergence is reached, thereby generating a prediction model for key physical field variables in building cavity design. The present invention can efficiently and accurately predict key physical quantities in building cavity design, and can adapt to different building layouts and functional space characteristics.

Description

Technical Field

[0001] The present invention relates to the technical field of building natural ventilation, and in particular to a computational auxiliary decision-making method and system for weather-adaptive building cavity design. Background Art

[0002] In the vast field of modern architectural design, cavity design, as an innovative passive green technology, is gradually becoming a key strategy to improve the environmental performance and energy efficiency of public buildings. This technology cleverly uses the physical principles of nature, especially the heat pressure ventilation mechanism, and forms a natural air flow channel through the reasonable layout of the internal and external spaces of the building. It can achieve effective circulation and renewal of indoor air without relying on additional energy consumption, thereby significantly improving indoor air quality, reducing the load of the air conditioning system, and ultimately achieving the goal of energy conservation and emission reduction.

[0003] Although cavity design has shown great potential in green buildings, the complexity and precision requirements of its design process also bring many challenges. Traditional cavity design methods mainly rely on empirical formulas and computational fluid dynamics (CFD) simulations. Although these two methods have their value, they have exposed many limitations in practical applications:

[0004] 1. Low computational efficiency: Traditional CFD simulations often use a fixed grid system, which is difficult to flexibly adapt to the complex and changeable geometric shapes and flow characteristics inside the building cavity, resulting in huge calculations and long time consumption, especially when dealing with large-scale or high-precision building models. This problem is particularly prominent.

[0005] 2. Long design cycle: Due to low computing efficiency, the design iteration cycle is prolonged, making it difficult to quickly respond to market demands or adjust design solutions. This not only increases design costs, but may also delay project progress.

[0006] 3. Sensitive to initial parameters: The accuracy of CFD simulation results is highly dependent on the setting of initial conditions and boundary conditions, and slight changes in these parameters may have a significant impact on the simulation results, increasing the uncertainty and debugging difficulty in the design process.

[0007] 4. Separate simulation of physical quantities: Traditional methods often require separate simulation of different physical quantities such as wind speed and temperature, ignoring the interactions and influences between them, making it difficult to fully and accurately reflect the real fluid dynamics behavior in the building cavity.

[0008] 5. Insufficient capture of complex flow details: The limitation of fixed grids is also reflected in the difficulty in accurately capturing complex flow phenomena such as eddy currents and backflows inside the building cavity, which are of great significance for optimizing ventilation effects and reducing energy consumption. Summary of the invention

[0009] The object of the present invention is to provide a computational decision-making aid method and system for weather-adaptive building cavity design, which can efficiently and accurately predict key physical quantities in building cavity design, including wind speed, temperature, pressure distribution and PMV value, and can adapt to different building layouts and functional space characteristics to solve at least one of the above-mentioned prior art problems.

[0010] In a first aspect, the present invention provides a computational decision-making aid method for weather-adaptive building cavity design, the method specifically comprising:

[0011] Acquire building cavity design parameters, and construct a conditional PINNs network according to the building cavity design parameters, physical field variables and physical laws;

[0012] An initial grid is generated by using a Delaunay triangulation method, and the initial grid is adaptively refined by a Laplace operator-based grid refinement method to obtain an adaptive grid system, wherein the adaptive grid system is used to dynamically adjust the grid density of the conditional PINNs network;

[0013] Constructing a multi-task learning framework in the conditional PINNs network, wherein the multi-task learning framework is used to collaboratively predict multiple physical field variables in the conditional PINNs network;

[0014] Setting a loss function, wherein the loss function includes a data error loss function and a physical law residual loss function, wherein the data error loss function is used to measure the difference between the model prediction value and the actual observation value, and the physical law residual loss function is used to measure the difference between the model prediction value and the physical law;

[0015] Iteratively training the conditional PINNs network, while minimizing the loss function using a gradient descent algorithm, until a predetermined number of training iterations is reached or the loss converges, thereby generating a prediction model for key physical field variables in building cavity design;

[0016] A building structure diagram is obtained, and a building cavity thermal map is generated according to the prediction results of the key physical field variable prediction model in the building cavity design and the building structure diagram.

[0017] In a second aspect, the present invention provides a computational decision-making support system for weather-adaptive building cavity design, the system specifically comprising:

[0018] A first decision module is used to obtain building cavity design parameters and construct a conditional PINNs network according to the building cavity design parameters, physical field variables and physical laws;

[0019] A second decision module is used to generate an initial grid using a Delaunay triangulation method, and to adaptively refine the initial grid using a Laplace operator-based grid refinement method to obtain an adaptive grid system, wherein the adaptive grid system is used to dynamically adjust the grid density of the conditional PINNs network;

[0020] A third decision module is used to construct a multi-task learning framework in the conditional PINNs network, and the multi-task learning framework is used to collaboratively predict multiple physical field variables in the conditional PINNs network;

[0021] a fourth decision module, used to set a loss function, wherein the loss function includes a data error loss function and a physical law residual loss function, wherein the data error loss function is used to measure the difference between the model prediction value and the actual observation value, and the physical law residual loss function is used to measure the difference between the model prediction value and the physical law;

[0022] A fifth decision module is used to iteratively train the conditional PINNs network and simultaneously use a gradient descent algorithm to minimize the loss function until a predetermined number of training iterations is reached or the loss converges, thereby generating a prediction model for key physical field variables in building cavity design;

[0023] The sixth decision module is used to obtain a building structure diagram, and generate a building cavity thermal map according to the prediction results of the key physical field variable prediction model in the building cavity design and the building structure diagram.

[0024] In a third aspect, the present invention provides a computer device comprising: a memory and a processor and a computer program stored in the memory, and when the computer program is executed on the processor, it implements a computational auxiliary decision-making method for weather-adaptive building cavity design as described in any of the above methods.

[0025] In a fourth aspect, the present invention provides a computer-readable storage medium, characterized in that a computer program is stored thereon, and when the computer program is executed by a processor, it implements a computational auxiliary decision-making method for weather-adaptive building cavity design as described in any of the above methods.

[0026] Compared with the prior art, the present invention has at least one of the following technical effects:

[0027] 1. By combining adaptive grid refinement, a multi-task learning framework, and conditional PINNs, the present invention aims to improve the prediction accuracy and generalization ability of the model, thereby achieving the optimal design of building cavity performance adapted to climate characteristics.

[0028] 2. The present invention uses adaptive grid refinement technology to refine the grid only in key areas, reducing unnecessary calculations and improving simulation accuracy.

[0029] 3. The multi-task learning framework of the present invention can predict all key physical quantities simultaneously, providing more comprehensive analysis results and reducing the number of design iterations.

[0030] 4. The real-time feedback and multi-task prediction capabilities of the present invention enable designers to quickly evaluate different design solutions, thus accelerating the design decision-making process.

[0031] 5. The present invention reduces the reliance on traditional CFD simulation and shortens the design cycle through the automated PINN model.

[0032] 6. The present invention uses the PINN model to predict the natural ventilation effect of the building cavity, including key parameters such as wind speed and temperature distribution, and combines the predicted ventilation effect parameters to evaluate and optimize the indoor thermal comfort (PMV) to achieve a suitable living and working environment.

[0033] 7. The present invention improves the efficiency of natural ventilation and reduces the dependence on air conditioning and mechanical ventilation systems by optimizing the building cavity design, thereby reducing building energy consumption.

[0034] 8. The PINN model trained by the present invention has good generalization and can adapt to different architectural design scenarios, providing architects with a fast and accurate design evaluation tool. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 It is a flowchart of a computational auxiliary decision-making method for weather-adaptive building cavity design provided by an embodiment of the present invention;

[0037] Figure 2 It is a schematic diagram of the structure of a computational decision-making support system for weather-adaptive building cavity design provided by one embodiment of the present invention;

[0038] Figure 3 It is a structural schematic diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0040] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0041] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0042] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0043] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0044] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0045] In the embodiment of the present application, the execution subject of the process includes a terminal device, which includes but is not limited to: a server, a computer, a smart phone, a tablet computer, and other devices capable of executing the method disclosed in the present application. Figure 1 A flowchart of a computational auxiliary decision-making method for weather-adaptive building cavity design disclosed in the first embodiment of the present invention is shown, and the details are as follows:

[0046] S101, obtaining building cavity design parameters, and constructing a conditional PINNs network according to the building cavity design parameters, physical field variables and physical laws.

[0047] In this embodiment, in order to more accurately simulate and optimize the ventilation effect in the building cavity design, it is necessary to comprehensively consider the geometric parameters, physical field variables and physical laws of the building cavity. Traditional methods often have problems such as large amount of calculation and insufficient accuracy when dealing with these complex factors. Therefore, this embodiment proposes to use the conditional PINNs network to optimize the design parameters of the building cavity.

[0048] The building cavity design parameters include building geometry information, environmental information, other information and predicted values. The building geometry information includes building plane dimensions, number of building floors, building floor height, cavity cross-sectional dimensions, cavity vent dimensions of each floor, cavity top opening dimensions, and number of opening floors. The environmental information includes regional longitude and latitude, regional average temperature during the transition season, and regional average humidity during the transition season. Other information includes air density, aerodynamic viscosity, gravity, air specific heat capacity, air thermal conductivity, source term, metabolic rate, external power, average radiation temperature, clothing thermal conductivity coefficient, clothing surface temperature, convection heat transfer coefficient and other information. The predicted values ​​include specific values, actual measured values, and traditional CFD simulation values.

[0049] In this embodiment, compared with the traditional method, the conditional PINNs network can complete the optimization process of design parameters more quickly, reduce the number of design iterations and calculation time, and shorten the overall design cycle. By introducing physical laws as constraints, the conditional PINNs network can more accurately simulate and predict the fluid flow behavior in the building cavity, thereby improving the accuracy and reliability of the design parameters.

[0050] In some embodiments, in the above step S101, the building cavity design parameters include building layout information, material property information and environmental condition information; and constructing a conditional PINNs network according to the building cavity design parameters, physical field variables and physical laws specifically includes:

[0051] Converting the building layout information, the material property information and the environmental condition information into a first eigenvector, a second eigenvector and a third eigenvector respectively, and concatenating the first eigenvector, the second eigenvector and the third eigenvector to form a comprehensive eigenvector;

[0052] Design a conditional PINNs network, which includes an input layer, a hidden layer and an output layer. The input layer is used to receive the comprehensive feature vector and the physical field variable, and the output of the hidden layer is , represents the ReLU activation function, and They are The weight matrix and bias vector of the hidden layer, and Respectively represent the Layer and -1 layer activation value, the output layer is used to output the predicted value of the physical field variable;

[0053] Each physical law is embedded into the conditional PINNs network, and the physical law residual is calculated through the automatic differentiation mechanism.

[0054] In this embodiment, a conditional PINNs network is constructed, and the feature vector output by the conditional encoder is input. and physical field variables (wind pressure ,temperature , wind speed and ). The network structure can be divided into the following modules:

[0055] Input layer: accepts physical field variables and conditional eigenvectors Assume that the physical field variables are , the input layer input is .

[0056] Hidden layer: Several hidden layers, using a fully connected network (FCN), each layer uses a ReLU activation function; assuming that there are a total of The output of each hidden layer is:

[0057]

[0058] Output layer: output the predicted values ​​of physical field variables, wind pressure ,temperature , wind speed and .

[0059] The conditional PINNs model can automatically adjust the network structure and parameters according to different building layouts and functional space characteristics.

[0060] Furthermore, the comprehensive feature vector is ,in, represents the first eigenvector and , represents the second eigenvector and , represents the third eigenvector and , R represents the weight matrix in the neural network, n represents the number of rooms in the building layout information, m represents the number of vents in the building layout information, k represents the number of material properties, and p represents the number of environmental conditions.

[0061] In this embodiment, when designing a conditional PINNs network, it is first necessary to encode the input parameter information into a feature vector through a conditional encoder, using the following method:

[0062] Building layout encoding: Encode building layout information (such as room location, size, and vent location) into a multidimensional vector. Assume that the building has Rooms and The location and size of each room and the location of each vent can be used as features to form a feature vector .

[0063] Material property encoding: Encode material properties (such as thermal conductivity of wall materials and insulation materials) into a feature vector ,in is the number of material properties.

[0064] Environmental condition encoding: Encode environmental conditions (such as external temperature, etc.) into a feature vector ,in is the number of environmental conditions.

[0065] Finally, the above feature vectors are concatenated to form a comprehensive feature vector .

[0066] By introducing a conditional encoder, conditional information such as building layout and functional space characteristics is encoded into a feature vector and fused with the output of the shared encoder to adapt to different design conditions.

[0067] Furthermore, the physical laws include wind pressure formula, temperature formula, wind speed formula and thermal comfort formula;

[0068] The wind pressure formula is: , Indicates the air density (kg / m³), represents the velocity field (m / s);

[0069] The temperature formula is , represents temperature (K), Indicates the air density (kg / m³), represents the specific heat capacity of air (J / (kg·K)), represents the velocity field (m / s), represents the thermal conductivity of air (W / (m·K)), represents the source term, including internal heat source, radiation, etc. (W / m³), ∇ represents the Laplace operator (the Laplace operator is a second-order differential operator used to calculate the divergence of the gradient of a function in space, that is, the local rate of change of the function), It represents the rate of change of the temperature field T in space, that is, the divergence of the temperature gradient;

[0070] The wind speed formula is , represents the velocity field (m / s), Indicates the air density (kg / m³), Indicates pressure (Pa), represents the dynamic viscosity of air (Pa·s), represents external body force (such as gravity) (N / m³);

[0071] The thermal comfort formula is , represents the metabolic rate (W / m²), Indicates external power (W / m²), represents the water vapor partial pressure (Pa), represents the air temperature (°C), represents the mean radiant temperature (°C), represents the thermal conductivity of the clothing, Indicates the surface temperature of the garment (°C).

[0072] In this embodiment, the physical law formulas of wind pressure, temperature, wind speed and PMV are embedded in the network, and the physical law residuals are calculated by the automatic differentiation mechanism. Specifically, the derivative of the velocity field is calculated using the automatic differentiation mechanism, and the wind pressure formula and wind speed formula are embedded in the velocity field predicted by the network to calculate the wind pressure and wind speed respectively; the time derivative and space derivative of the temperature field are calculated using automatic differentiation, and the temperature formula is embedded in the temperature field predicted by the network to calculate the temperature; the PMV value is calculated based on the temperature, radiation temperature, water vapor partial pressure, etc. predicted by the network.

[0073] Through the above design and implementation, it can be ensured that the conditional PINNs network can effectively combine the building condition information and physical laws to improve the prediction accuracy and adaptability of the model.

[0074] S102, using a Delaunay triangulation method to generate an initial grid, and adaptively refining the initial grid by a Laplace operator-based grid refinement method to obtain an adaptive grid system, wherein the adaptive grid system is used to dynamically adjust the grid density of the conditional PINNs network.

[0075] In this embodiment, when constructing a conditional PINNs (physical information neural network) network, the generation of the initial grid and its subsequent adaptive refinement are crucial to improving the accuracy and efficiency of network prediction. The Delaunay triangulation method is widely used in the field of grid generation because it can generate triangular grids that meet certain geometric criteria. The Laplace operator-based grid refinement method can effectively adjust the grid density according to local features and achieve adaptive refinement of the grid. This embodiment will combine these two methods to construct an adaptive grid system for dynamically adjusting the grid density of the conditional PINNs network.

[0076] The initial mesh is generated using the Delaunay triangulation method. First, a series of discrete points (which can be uniformly distributed or distributed according to actual needs) are obtained according to the geometry and boundary conditions of the area to be analyzed. By calling an existing numerical computing library (such as the Delaunay function in SciPy), the discrete points are triangulated to generate an initial triangular mesh. Delaunay triangulation ensures that the generated triangular mesh satisfies the empty circle property, that is, the circumscribed circle of any triangle does not contain other vertices.

[0077] Define the criteria for mesh refinement. This is usually based on local error estimates or gradient changes of physical variables. For example, you can specify that when the gradient of a physical variable in a certain region exceeds a certain threshold, the mesh needs to be refined in that region.

[0078] Iterate the initial mesh to refine. In each round of refinement, calculate the Laplacian operator of each vertex (i.e., the weighted average of the position difference between the vertex and its adjacent vertices), and adjust the vertex position according to the value of the Laplacian operator, so that the mesh is automatically encrypted in the area that needs to be refined. During the refinement process, you can control the degree and effect of refinement by setting different refinement parameters (such as the number of iterations, the weight of the Laplacian operator, etc.).

[0079] The initial mesh generated by Delaunay triangulation is combined with the mesh refinement method based on the Laplace operator to construct an adaptive mesh system, which can automatically adjust the mesh density according to the changes in physical field variables and improve the mesh resolution in key areas.

[0080] When constructing a conditional PINNs network, the adaptive grid system is used to dynamically adjust the grid density of the network, so that the network can simulate and predict the physical processes in key areas more precisely while maintaining overall computational efficiency.

[0081] In this embodiment, an adaptive grid generation and refinement module is introduced in the input layer of the model, and local grid refinement is implemented in key areas (such as shaft entrances, corners, top openings, etc.) according to the gradient or error distribution of the physical field to improve the simulation accuracy of these areas while reducing the amount of calculation. The adaptive grid system can automatically adjust the grid density according to the changes in the physical field variables, so that the grid resolution is higher in areas where the physical process changes dramatically, thereby improving the precision and accuracy of the simulation. Compared with the global uniform grid, the adaptive grid system reduces unnecessary calculations and improves the overall computing efficiency by encrypting the grid only in the areas that need to be refined. This method has strong flexibility and can customize the standards and parameters of grid refinement according to different physical field characteristics and simulation requirements. It is suitable for grid generation and optimization in a variety of complex scenarios.

[0082] In some embodiments, in the above step S102, the initial grid is generated by using the Delaunay triangulation method, and the initial grid is adaptively refined by a Laplace operator-based grid refinement method to obtain an adaptive grid system, specifically including:

[0083] An initial mesh is generated by using a Delaunay triangulation method, and a preliminary simulation is performed on the initial mesh by using a finite element method to calculate the gradients of the velocity field and the temperature field;

[0084] Setting a mesh refinement standard, and determining a region to be refined on the initial mesh using a Laplace operator-based refinement algorithm according to the mesh refinement standard;

[0085] Refining the grid of the area to be refined according to the red-green refinement method to obtain an adaptive grid system;

[0086] The grid cell angles of the adaptive grid system are checked, and the grid node positions of the adaptive grid system are adjusted using a Laplacian smoothing algorithm.

[0087] In this embodiment, an adaptive mesh generation algorithm based on error estimation is used to generate an initial mesh based on the Delaunay triangulation method, and adaptive mesh refinement is performed in combination with a mesh refinement method based on the Laplace operator. The specific implementation steps are as follows:

[0088] Initial mesh generation: Generate the initial mesh using the Delaunay triangulation method to ensure the uniformity and good geometry of the mesh. This is achieved by calling the existing numerical calculation library (Delaunay function in SciPy)

[0089] Error estimation: Perform a preliminary simulation on the initial grid to calculate the gradients of the velocity field and temperature field. Use the finite element method to perform a preliminary simulation, and call the existing finite element software (COMSOL) for simulation. The calculation formula is:

[0090]

[0091] in, is a physical field (such as velocity or temperature), is a unit vector.

[0092] Mesh refinement criteria: Set refinement criteria, such as performing local mesh refinement when the velocity field gradient exceeds 0.1m / s² or the temperature change rate exceeds 1K / m.

[0093] Refinement algorithm: A refinement algorithm based on the Laplace operator is used to determine the area that needs to be refined by calculating the Laplace operator value at the grid node. The Laplace operator calculation formula is:

[0094]

[0095] in, It is a physical field. is the number of neighboring nodes of node 0.

[0096] Mesh refinement: In the areas that need to be refined, add mesh nodes to generate finer mesh elements. Use the red-green refinement method to refine the mesh to ensure that the mesh still maintains good geometry after refinement.

[0097] Mesh quality control: Ensure that the refined mesh system has good quality, such as orthogonality, uniformity, and minimum angle limit. Specific measures include:

[0098] Check the angles of the grid cells to make sure all angles are within a reasonable range (30° to 150°).

[0099] Adjust the mesh node positions to improve the orthogonality and uniformity of the mesh. Use the Laplacian smoothing algorithm:

[0100]

[0101] in, Is a node The new location, Is a node The neighboring node positions, is the number of neighboring nodes.

[0102] S103, constructing a multi-task learning framework in the conditional PINNs network, wherein the multi-task learning framework is used to collaboratively predict multiple physical field variables in the conditional PINNs network.

[0103] In this embodiment, a multi-task learning framework is constructed in a conditional PINNs (Physical Information Neural Networks) network, aiming to achieve collaborative optimization and mutual enhancement among these variables by simultaneously learning and predicting multiple physical field variables.

[0104] Specifically, consider a building cavity design problem, where multiple physical field variables need to be predicted simultaneously, such as wind speed, temperature, pressure, etc. There are certain interdependencies between these physical field variables. For example, changes in wind speed will affect the distribution of temperature, and changes in temperature will affect changes in pressure.

[0105] In the conditional PINNs network, a shared feature extractor is first designed. This extractor is responsible for extracting common features from the input data that are helpful for predicting multiple physical field variables. These features may include the geometry of the building cavity, boundary conditions, initial conditions, etc.

[0106] After the shared feature extractor, multiple parallel task-specific layers (also called "towers" or "heads") are designed, each of which is responsible for predicting a specific physical field variable. These task-specific layers can share some parameters or have independent parameters, depending on the correlation between tasks and the complexity of network design.

[0107] In order to optimize the prediction performance of multiple tasks at the same time, a multi-task loss function is designed. This function is usually a weighted sum of the loss functions of each task, and the weights can be adjusted according to the importance of the task or the characteristics of the data set.

[0108] The conditional PINNs network is trained using the preprocessed data. During the training process, the network optimizes its parameters by minimizing the multi-task loss function. Since multiple tasks share some feature extractors, the network is able to learn more generalized feature representations, thereby improving the overall prediction performance.

[0109] In this embodiment, through the multi-task learning framework, the conditional PINNs network can simultaneously learn and predict multiple physical field variables, and use the interdependence between them to improve the prediction accuracy. Compared with the single-task learning model, the multi-task learning model can learn a more generalized and robust feature representation. The multi-task learning framework reduces the amount of computation by sharing feature extractors, avoiding the overhead of training a model separately for each task. At the same time, since multiple tasks share some parameters, the network can converge to the optimal solution faster during the training process. By learning multiple related tasks at the same time, the conditional PINNs network can learn a more general feature representation, thereby improving the generalization ability of the model under different working conditions and conditions. This is of great significance for dealing with complex and changeable building cavity design problems.

[0110] Specifically, the multi-task learning framework includes a shared encoder structure and a task-specific decoder structure;

[0111] The shared encoder is used to encode the input physical field variables and the comprehensive feature vector into a shared latent representation;

[0112] The task-specific decoder structure is used to design a separate decoder for each physical field variable, and the decoder is used to take the output of the shared encoder as output and output the predicted value of the corresponding physical field variable.

[0113] In this embodiment, a deep neural network (DNN) is used as a shared encoder to encode the input physical field variables and conditional feature vectors into a shared latent representation. Assuming the input is, the output of the shared encoder is

[0114] The shared encoder consists of several hidden layers. The hidden layer structure is the hidden layer structure described in step 3. It adopts the FCN structure and uses the ReLU activation function in each layer. Assume that the shared encoder has The output of each hidden layer is:

[0115] in, is the ReLU activation function, and They are The weight matrix and bias vector of the layer. The output of the last layer This is the output of the shared encoder .

[0116] For each physical quantity (wind pressure ,temperature , wind speed and ) Design an independent decoder. Each decoder takes the output of the shared encoder as input and outputs the predicted value of the corresponding physical quantity.

[0117] The task-specific decoder also consists of several hidden layers, with the same structure as above. Assume that the task-specific decoder has hidden layer, the output of the last layer That is the predicted value of the corresponding physical quantity.

[0118] S104, setting a loss function, wherein the loss function includes a data error loss function and a physical law residual loss function, wherein the data error loss function is used to measure the difference between the model prediction value and the actual observation value, and the physical law residual loss function is used to measure the difference between the model prediction value and the physical law.

[0119] In this embodiment, when dealing with the problem of multi-physics field variable prediction in building cavity design, setting a comprehensive loss function including a data error loss function and a physical law residual loss function is an effective method, which can improve the prediction accuracy and physical consistency of the model. Specifically, the data error loss function usually selects mean square error (MSE) or mean absolute error (MAE) as a measurement standard to calculate the difference between the physical field variable values ​​(such as wind speed, temperature, pressure, etc.) predicted by the model and the actual observed values. The physical law residual loss function is used to evaluate whether the physical field variables predicted by the model satisfy the known physical laws, which is usually achieved by constructing residual terms based on physical laws.

[0120] Furthermore, the data error loss function is , , where K represents the number of physical field variables, is the weight coefficient of each physical quantity, is the data error loss of the kth physical quantity, N is the number of samples, and are the actual observed value and predicted value of the i-th sample respectively;

[0121] The physical law residual loss function is , ,in, is the weight coefficient of each physical law, is the residual loss of the mth physical law, N is the number of samples, is the physical law residual of the i-th sample;

[0122] The loss function is ,in, and It is the weight coefficient of data error loss and physical law residual loss, which is used to balance the importance of the two.

[0123] In this embodiment, the data error loss is used to measure the difference between the model prediction value and the actual observation value. In this embodiment, considering the existence of multiple physical quantities (wind pressure ,temperature , wind speed and PMV), so the data error loss can be defined as the weighted sum of the errors of each physical quantity.

[0124] for physical quantities, and the predicted value of each physical quantity is The actual observed value is , then the data error loss It can be defined as:

[0125]

[0126] in, is the weight coefficient of each physical quantity, It is The data error loss of physical quantities in this embodiment Value is , specifically, the mean square error (MSE) is used as the data error loss for each physical quantity:

[0127]

[0128] in, is the sample size, and They are The actual observed and predicted values ​​of samples.

[0129] The physical law residual loss is used to measure the difference between the model prediction value and the physical law. In this scenario, the physical law formulas of wind pressure, temperature, wind speed and PMV are considered, so the physical law residual loss can be defined as the weighted sum of the residuals of each physical law. The laws of physics ( ), the residual of each physical law is , then the physical law residual loss It can be defined as:

[0130]

[0131] in, is the weight coefficient of each physical law, It is Specifically, the mean square error (MSE) is used as the residual loss for each physical law:

[0132]

[0133] in, is the sample size, It is The residuals of the laws of physics for samples.

[0134] Total loss function It is the weighted sum of data error loss and physical law residual loss:

[0135]

[0136] in, and It is the weight coefficient of data error loss and physical law residual loss, which is used to balance the importance of the two.

[0137] S105, iteratively training the conditional PINNs network, while minimizing the loss function using a gradient descent algorithm, until a predetermined number of training iterations is reached or the loss converges, thereby generating a prediction model for key physical field variables in building cavity design.

[0138] In this embodiment, in each iteration, the automatic differentiation technique is used to calculate the gradient of the comprehensive loss function with respect to the network parameters, including the partial derivatives of the data error loss function and the physical law residual loss function with respect to the parameters.

[0139] The network weights and biases are updated using the calculated gradients according to a gradient descent algorithm (e.g. batch gradient descent, stochastic gradient descent, or mini-batch gradient descent). The update rule usually iterates with a certain step size in the opposite direction of the gradient.

[0140] Set a predetermined number of training iterations or loss convergence conditions as the criteria for stopping training. Check after each iteration whether the stopping condition is met. If so, stop training; otherwise, continue to the next iteration.

[0141] Through the iterative training process, the conditional PINNs network can gradually learn the complex mapping relationship of key physical field variables in building cavity design, thereby improving the accuracy of prediction.

[0142] In some embodiments, in the above step S105, the iterative training of the conditional PINNs network and the use of a gradient descent algorithm to minimize the loss function specifically include:

[0143] Initialize the network parameters of the conditional PINNs network according to the Xavier initialization method;

[0144] Iteratively training the conditional PINNs network, and in each iterative training, using forward propagation to calculate the predicted value of each physical field variable;

[0145] Calculate the loss of network parameters of the conditional PINNs network by the loss function;

[0146] Set the Adam optimizer and use the Adam optimizer to minimize the loss function. The Adam optimizer is , represents the learning rate, represents the exponential decay rate of the first-order moment estimate, represents the exponential decay rate of the second-order moment estimate.

[0147] In this embodiment, the Xavier initialization method is used to initialize the network parameters to ensure that the network parameters maintain an appropriate scale in the early stage of training to avoid gradient disappearance or explosion.

[0148]

[0149] in, represents uniform distribution, and are the number of input and output units of the current layer respectively. For the bias vector , initialized to a zero vector ( ).

[0150] The Adam optimizer is selected as the gradient descent algorithm, which combines the advantages of momentum and adaptive learning rate and can converge more stably. The parameters of the Adam optimizer include the learning rate , the exponential decay rate of the first-order moment estimate and the exponential decay rate of the second-order moment estimate .

[0151]

[0152] The default parameters are set to , , .

[0153] In each training iteration, forward propagation is first performed to calculate the predicted value of each physical quantity.

[0154] Shared encoder: Input Obtained through a shared encoder .

[0155] Task-specific decoders: Input into each task-specific decoder to obtain the predicted value of each physical quantity .

[0156] Calculate data error loss based on predicted values ​​and actual observed values , and calculate the physical law residual loss based on the predicted value and the physical law formula Finally, the total loss is calculated according to the formula .

[0157] The gradient of the total loss with respect to all parameters is calculated and the parameters are updated using the Adam optimizer.

[0158] Compute the gradient:

[0159]

[0160] Update parameters:

[0161]

[0162]

[0163] Repeat the above process of forward propagation, loss calculation, and back propagation until the predetermined number of training iterations is reached or the loss converges.

[0164] S106, obtaining a building structure diagram, and generating a building cavity thermal map according to the prediction results of the key physical field variable prediction model in the building cavity design and the building structure diagram.

[0165] In this embodiment, the building cavity thermal map can intuitively display the distribution of various physical field variables inside the building, providing an intuitive reference for architects, engineers, etc. By combining the prediction model and the thermal map, the distribution of various physical field variables in the building cavity can be predicted in the design stage, so as to discover and solve potential problems in advance and improve design efficiency and accuracy.

[0166] In some embodiments, in the above step S106, generating a building cavity thermal map according to the prediction results of the key physical field variable prediction model in the building cavity design and the building structure diagram specifically includes:

[0167] Obtain physical quantity data from the prediction results of the key physical field variable prediction model in the building cavity design, and the physical quantity data includes , , and , where each Z represents a two-dimensional array, , , and They represent the distribution of wind pressure, temperature, wind speed and thermal comfort on the building plane respectively;

[0168] Will , , and They are mapped to the color space respectively to form arrays C, and the initial thermal maps corresponding to the physical field variables are formed according to each array C;

[0169] Each initial thermal map is superimposed on the building structure map to form a target thermal map.

[0170] In this embodiment, the physical quantity data to be displayed are extracted from the model prediction results, assuming that they are , , and , where each Z is a two-dimensional array representing the distribution of the corresponding physical quantity on the building plane.

[0171] Select a suitable color mapping scheme (Jet) to map the value of each physical quantity to the color space. The color mapping function C can be defined as:

[0172]

[0173] Where C is a two-dimensional array of the same size as Z, and each element corresponds to a color value.

[0174] Heatmap generation: Generate a heatmap using the color-mapped array C. Each physical quantity corresponds to a heatmap, for a total of four. Each image is superimposed with the building plan structure and cavity position annotation map to form a heatmap of the corresponding physical quantity. The four images are displayed on the same page through the sub-image layout (2x2 grid) for easy comparison and analysis.

[0175] In order to interact and optimize the design, the visualization module should support user input and real-time update of the thermal map. Design an interactive interface to allow users to input different design parameters (building layout, cavity location, etc.), and re-run the model prediction based on the design parameters entered by the user to obtain new physical quantity data. , , and Generate and update the thermal map using the new physical quantity data, and update the display of the building plan structure and cavity position. By observing the changes in the thermal map, users can intuitively evaluate the impact of different design parameters on the distribution of physical quantities, and thus optimize the design. Based on the results of the observation and analysis, users adjust the design parameters, re-run the model prediction and thermal map update.

[0176] Reference Figure 2 An embodiment of the present invention provides a computational decision-making support system 2 for weather-adaptive building cavity design, and the system 2 specifically includes:

[0177] The first decision module 201 is used to obtain building cavity design parameters and construct a conditional PINNs network according to the building cavity design parameters, physical field variables and physical laws;

[0178] A second decision module 202 is used to generate an initial grid using a Delaunay triangulation method, and to adaptively refine the initial grid using a Laplace operator-based grid refinement method to obtain an adaptive grid system, wherein the adaptive grid system is used to dynamically adjust the grid density of the conditional PINNs network;

[0179] A third decision module 203 is used to construct a multi-task learning framework in the conditional PINNs network, and the multi-task learning framework is used to collaboratively predict multiple physical field variables in the conditional PINNs network;

[0180] A fourth decision module 204 is used to set a loss function, wherein the loss function includes a data error loss function and a physical law residual loss function, wherein the data error loss function is used to measure the difference between the model prediction value and the actual observation value, and the physical law residual loss function is used to measure the difference between the model prediction value and the physical law;

[0181] The fifth decision module 205 is used to iteratively train the conditional PINNs network and use a gradient descent algorithm to minimize the loss function until a predetermined number of training iterations is reached or the loss converges, thereby generating a prediction model for key physical field variables in building cavity design;

[0182] The sixth decision module 206 is used to obtain a building structure diagram, and generate a building cavity thermal map according to the prediction results of the key physical field variable prediction model in the building cavity design and the building structure diagram.

[0183] It is understandable that if Figure 1 The contents of the computational auxiliary decision-making method embodiment for weather-adaptive building cavity design shown in the embodiment are all applicable to the computational auxiliary decision-making system embodiment for weather-adaptive building cavity design. The functions specifically implemented by the computational auxiliary decision-making system embodiment for weather-adaptive building cavity design are similar to those of the embodiment shown in the embodiment. Figure 1 The computational auxiliary decision-making method embodiment for weather-adaptive building cavity design shown in FIG. 1 is the same as that shown in FIG. 1 , and the beneficial effects achieved are the same as those of FIG. Figure 1 The beneficial effects achieved by the illustrated embodiment of the computational auxiliary decision-making method for weather-adaptive building cavity design are also the same.

[0184] It should be noted that the information interaction, execution process and other contents between the above-mentioned systems are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0185] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0186] Reference Figure 3 The embodiment of the present invention further provides a computer device 3, comprising: a memory 302 and a processor 301 and a computer program 303 stored in the memory 302. When the computer program 303 is executed on the processor 301, a computational auxiliary decision-making method for weather-adaptive building cavity design as described in any one of the above methods is implemented.

[0187] The computer device 3 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The computer device 3 may include, but is not limited to, a processor 301 and a memory 302. Those skilled in the art will appreciate that Figure 3 It is only an example of computer device 3 and does not constitute a limitation on computer device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components, for example, it may also include input and output devices, network access devices, etc.

[0188] The processor 301 may be a central processing unit (CPU), or 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. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0189] In some embodiments, the memory 302 may be an internal storage unit of the computer device 3, such as a hard disk or memory of the computer device 3. In other embodiments, the memory 302 may also be an external storage device of the computer device 3, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device 3. Further, the memory 302 may also include both an internal storage unit of the computer device 3 and an external storage device. The memory 302 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program, etc. The memory 302 may also be used to temporarily store data that has been output or is to be output.

[0190] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the computer program implements a computational auxiliary decision-making method for weather-adaptive building cavity design as described in any one of the above methods.

[0191] In this embodiment, if the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, USB flash drive, mobile hard disk, disk or optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0192] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0193] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0194] In the embodiments disclosed in the present application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0195] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

Claims

1. A computational decision-making aid method for weather-adaptive building cavity design, characterized in that: The method specifically comprises: Acquire building cavity design parameters, and construct a conditional PINNs network according to the building cavity design parameters, physical field variables and physical laws; An initial grid is generated by using a Delaunay triangulation method, and the initial grid is adaptively refined by a Laplace operator-based grid refinement method to obtain an adaptive grid system, wherein the adaptive grid system is used to dynamically adjust the grid density of the conditional PINNs network; Constructing a multi-task learning framework in the conditional PINNs network, wherein the multi-task learning framework is used to collaboratively predict multiple physical field variables in the conditional PINNs network; Setting a loss function, wherein the loss function includes a data error loss function and a physical law residual loss function, wherein the data error loss function is used to measure the difference between the model prediction value and the actual observation value, and the physical law residual loss function is used to measure the difference between the model prediction value and the physical law; Iteratively training the conditional PINNs network, while minimizing the loss function using a gradient descent algorithm, until a predetermined number of training iterations is reached or the loss converges, thereby generating a prediction model for key physical field variables in building cavity design; A building structure diagram is obtained, and a building cavity thermal map is generated according to the prediction results of the key physical field variable prediction model in the building cavity design and the building structure diagram.

2. The method according to claim 1, characterized in that The building cavity design parameters include building layout information, material property information and environmental condition information; the conditional PINNs network is constructed according to the building cavity design parameters, physical field variables and physical laws, specifically including: Converting the building layout information, the material property information and the environmental condition information into a first eigenvector, a second eigenvector and a third eigenvector respectively, and concatenating the first eigenvector, the second eigenvector and the third eigenvector to form a comprehensive eigenvector; Design a conditional PINNs network, which includes an input layer, a hidden layer and an output layer. The input layer is used to receive the comprehensive feature vector and the physical field variable, and the output of the hidden layer is , represents the ReLU activation function, and They are The weight matrix and bias vector of the hidden layer, and Respectively represent the Layer and -1 layer activation value, the output layer is used to output the predicted value of the physical field variable; Each physical law is embedded into the conditional PINNs network, and the physical law residual is calculated through the automatic differentiation mechanism.

3. The method according to claim 2, characterized in that The comprehensive feature vector is ,in, represents the first eigenvector and , represents the second eigenvector and , represents the third eigenvector and , R represents the weight matrix in the neural network, n represents the number of rooms in the building layout information, m represents the number of vents in the building layout information, k represents the number of material properties, and p represents the number of environmental conditions. represents the weight matrix containing the number of rooms, the number of vents, the number of material properties, and the number of environmental conditions, represents the weight matrix containing the feature information of the number of rooms and the number of vents, represents the weight matrix containing the quantities of the material properties, A weight matrix representing the quantities that comprise the environmental conditions.

4. The method according to claim 2, characterized in that: Said physical laws include the wind pressure formula; The wind pressure formula is: , represents the air density, Represents the velocity field.

5. The method according to claim 1, characterized in that The method of using the Delaunay triangulation method to generate an initial grid and adaptively refining the initial grid by a Laplace operator-based grid refinement method to obtain an adaptive grid system specifically includes: An initial mesh is generated by using a Delaunay triangulation method, and a preliminary simulation is performed on the initial mesh by using a finite element method to calculate the gradients of the velocity field and the temperature field; Setting a mesh refinement standard, and determining a region to be refined on the initial mesh using a Laplace operator-based refinement algorithm according to the mesh refinement standard; Refining the grid of the area to be refined according to the red-green refinement method to obtain an adaptive grid system; The grid cell angles of the adaptive grid system are checked, and the grid node positions of the adaptive grid system are adjusted using a Laplacian smoothing algorithm.

6. The method according to claim 2, characterized in that The multi-task learning framework includes a shared encoder structure and a task-specific decoder structure; The shared encoder is used to encode the input physical field variables and the comprehensive feature vector into a shared latent representation; The task-specific decoder structure is used to design a separate decoder for each physical field variable, and the decoder is used to take the output of the shared encoder as output and output the predicted value of the corresponding physical field variable.

7. The method according to claim 1, characterized in that The data error loss function is , , where K represents the number of physical field variables, is the weight coefficient of each physical quantity, is the data error loss of the kth physical quantity, N is the number of samples, and are the actual observed value and predicted value of the i-th sample respectively; The physical law residual loss function is , ,in, is the weight coefficient of each physical law, is the residual loss of the mth physical law, N is the number of samples, is the physical law residual of the i-th sample, M represents the number of physical laws; The loss function is ,in, and It is the weight coefficient of data error loss and physical law residual loss, which is used to balance the importance of the two.

8. The method according to claim 1, characterized in that: The iterative training of the conditional PINNs network and the use of a gradient descent algorithm to minimize the loss function specifically include: Initialize the network parameters of the conditional PINNs network according to the Xavier initialization method; Iteratively training the conditional PINNs network, and in each iterative training, using forward propagation to calculate the predicted value of each physical field variable; Calculate the loss of network parameters of the conditional PINNs network by the loss function; Set the Adam optimizer and use the Adam optimizer to minimize the loss function. The Adam optimizer is , represents the learning rate, represents the exponential decay rate of the first-order moment estimate, represents the exponential decay rate of the second-order moment estimate.

9. The method according to any one of claims 1 to 8, characterized in that The generating of the building cavity thermal map according to the prediction results of the key physical field variable prediction model in the building cavity design and the building structure diagram specifically includes: Obtain physical quantity data from the prediction results of the key physical field variable prediction model in the building cavity design, and the physical quantity data includes , , and , where each Z represents a two-dimensional array, , , and They represent the distribution of wind pressure, temperature, wind speed and thermal comfort on the building plane respectively; Will , , and They are mapped to the color space respectively to form arrays C, and the initial thermal maps corresponding to the physical field variables are formed according to each array C; Each initial thermal map is superimposed on the building structure map to form a target thermal map.

10. A computational decision-making support system for weather-adaptive building cavity design, characterized in that: The system specifically comprises: A first decision module is used to obtain building cavity design parameters and construct a conditional PINNs network according to the building cavity design parameters, physical field variables and physical laws; A second decision module is used to generate an initial grid using a Delaunay triangulation method, and to adaptively refine the initial grid using a Laplace operator-based grid refinement method to obtain an adaptive grid system, wherein the adaptive grid system is used to dynamically adjust the grid density of the conditional PINNs network; A third decision module is used to construct a multi-task learning framework in the conditional PINNs network, and the multi-task learning framework is used to collaboratively predict multiple physical field variables in the conditional PINNs network; a fourth decision module, used to set a loss function, wherein the loss function includes a data error loss function and a physical law residual loss function, wherein the data error loss function is used to measure the difference between the model prediction value and the actual observation value, and the physical law residual loss function is used to measure the difference between the model prediction value and the physical law; A fifth decision module is used to iteratively train the conditional PINNs network and simultaneously use a gradient descent algorithm to minimize the loss function until a predetermined number of training iterations is reached or the loss converges, thereby generating a prediction model for key physical field variables in building cavity design; The sixth decision module is used to obtain a building structure diagram, and generate a building cavity thermal map according to the prediction results of the key physical field variable prediction model in the building cavity design and the building structure diagram.

Images