A microclimate prediction method and device based on machine learning, equipment and medium

Abstract

This invention relates to the field of architectural design technology, specifically to a microclimate prediction method, apparatus, device, and medium based on machine learning. The invention first sets design values ​​for the architectural design variables of a single building, and then applies a machine learning model to these design values ​​to predict the microclimate changes that the single building, designed based on these values, will bring to the surrounding environment. Because this invention is based on the design values ​​of a single building, it can predict the impact of a single building on the microclimate of its surrounding environment. Furthermore, using a machine learning model to predict microclimate does not require spending a significant amount of time building a building model; it can directly predict microclimate, thereby improving prediction efficiency.

Description

Technical Field

[0001] This invention relates to the field of architectural design technology, specifically to a microclimate prediction method, device, equipment, and medium based on machine learning. Background Technology

[0002] Different architectural designs can influence the microclimate of the building's environment. For example, the building's length, width, height, and the properties of the materials used all affect the microclimate. Current technology applies CFD models to parameters such as building density and average building height to predict the impact of buildings on the microclimate. However, since parameters such as building density are characteristics of all buildings rather than individual buildings, CFD models can only predict the microclimate of the city or region where a new or renovated building is located, and cannot predict the microclimate around a single building.

[0003] Therefore, existing technologies still need to be improved and enhanced. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a microclimate prediction method, device, equipment, and medium based on machine learning, which solves the problem that existing technologies cannot predict the microclimate around a single building.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] In a first aspect, the present invention provides a microclimate prediction method based on machine learning, comprising:

[0007] Determine the architectural design variables for a single building;

[0008] Set the design values ​​for the architectural design variables;

[0009] A machine learning model is applied to the design values ​​to predict the microclimate changes that the single building designed based on the design values ​​will bring to the surrounding environment.

[0010] In one implementation, determining the architectural design variables for a single building includes:

[0011] Obtain several original architectural design variables and set simulated values ​​for several of the original architectural design variables;

[0012] Based on the simulated values ​​of several original architectural design variables, an architectural design scene is generated;

[0013] Based on the architectural design scenario, generate a simulated geometric model of the building;

[0014] The simulation analysis examines the microclimate changes caused by incorporating the building's simulated geometric model.

[0015] Based on the microclimate simulation change data, sensitive variables are selected from several original building design variables;

[0016] Based on the aforementioned sensitive variables, the architectural design variables for a single building are obtained.

[0017] In one implementation, based on the microclimate simulation change data, sensitive variables are selected from several original building design variables, including:

[0018] A global sensitivity analysis algorithm is applied to the microclimate simulation change data to obtain the sensitivity of the microclimate to each of the original building design variables;

[0019] Based on the sensitivity of each of the original architectural design variables, sensitive variables are selected from several of the original architectural design variables.

[0020] In one implementation, setting the design value of the architectural design variable includes:

[0021] Determine the architectural form design variables and architectural thermal characteristic design variables among the architectural design variables;

[0022] Within the preset range of values ​​for the architectural form design variables, the design values ​​for the architectural form design variables are set.

[0023] Obtain the calculation formula corresponding to the building thermal characteristic design variable, and determine the design value of the building thermal characteristic design variable based on the calculation formula.

[0024] In one implementation, the building form design variables include building height, building orientation, and building aspect ratio; the building thermal characteristic design variables include the overall heat transfer coefficient of the building envelope, wall emissivity, and heat dissipation per unit building area; a machine learning model is applied to the design values ​​to predict microclimate changes in the surrounding environment caused by the single building designed based on the design values, including:

[0025] A temperature change prediction model based on SVR is applied to the design values ​​of the building height, the building orientation, the building length-to-width ratio, the overall heat transfer coefficient of the building envelope, the wall emissivity, and the heat dissipation per unit building area to predict the temperature change data of the single building designed based on the design values ​​to the surrounding environment.

[0026] The design values ​​of the building height, building orientation, building aspect ratio, overall heat transfer coefficient of the building envelope, wall emissivity, and heat dissipation per unit building area are applied to a wind speed change prediction model based on LightGBM to predict the wind speed change data of the single building designed based on the design values ​​to the surrounding environment, and the wind speed change data and the temperature change data are used as microclimate change data.

[0027] In one implementation, the regularization parameter of the SVR is 1, and the kernel function coefficient of the SVR is 0.001.

[0028] In one implementation, the LightGBM has 1000 trees, a maximum tree depth of 2, a maximum number of leaves of 3, a learning rate of 0.05, and a subsample ratio of 1 for training the LightGBM.

[0029] Secondly, embodiments of the present invention also provide a microclimate prediction device based on machine learning, wherein the device comprises the following components:

[0030] The design variable filtering module is used to determine the architectural design variables for a single building.

[0031] The design value design module is used to set the design values ​​of the architectural design variables;

[0032] The prediction module is used to apply a machine learning model to the design values ​​to predict the microclimate changes that the single building designed based on the design values ​​will bring to the surrounding environment.

[0033] Thirdly, embodiments of the present invention also provide a terminal device, wherein the terminal device includes a memory, a processor, and a machine learning-based microclimate prediction program stored in the memory and executable on the processor, wherein when the processor executes the machine learning-based microclimate prediction program, it implements the steps of the machine learning-based microclimate prediction method described above.

[0034] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a machine learning-based microclimate prediction program, wherein when the machine learning-based microclimate prediction program is executed by a processor, it implements the steps of the machine learning-based microclimate prediction method described above.

[0035] Beneficial Effects: This invention first sets design values ​​for the architectural design variables of a single building, and then applies a machine learning model to these values ​​to predict the microclimate changes brought about by the single building based on these design values ​​to the surrounding environment. Because this invention is based on the design values ​​of a single building, it can predict the impact of a single building on the microclimate of its surrounding environment. Furthermore, using a machine learning model to predict microclimate does not require spending a significant amount of time building a building model; it can directly predict microclimate, thereby improving prediction efficiency. Attached Figure Description

[0036] Figure 1 This is an overall flowchart of the present invention;

[0037] Figure 2 A structural diagram of the microclimate prediction device based on machine learning provided by the present invention;

[0038] Figure 3 This is a block diagram illustrating the internal structure of a terminal device provided in an embodiment of the present invention. Detailed Implementation

[0039] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0040] Research has found that different architectural designs can impact the microclimate of the building's environment. For example, the building's length, width, height, and the properties of the materials used all affect the microclimate. Current technology applies CFD models to parameters such as building density and average building height to predict the impact of buildings on microclimate. However, because parameters like building density represent the characteristics of all buildings, rather than individual buildings, CFD models can only predict the microclimate of the city or region where a new or renovated building is located, and cannot predict the microclimate around a single building.

[0041] To address the aforementioned technical problems, this invention provides a microclimate prediction method, device, equipment, and medium based on machine learning, which solves the problem that existing technologies cannot predict the microclimate around a single building.

[0042] The machine learning-based microclimate prediction method of this embodiment can be applied to terminal devices, which can be terminal products with data processing capabilities, such as computers. In this embodiment, as... Figure 1 As shown, the microclimate prediction method based on machine learning specifically includes the following steps:

[0043] S100, determines the architectural design variables for a single building;

[0044] S200, Set the design values ​​of the architectural design variables;

[0045] S300, apply a machine learning model to the design values ​​to predict microclimate change data of the surrounding environment caused by the single building designed based on the design values.

[0046] The application scenarios based on steps S100, S200, and S300 are as follows:

[0047] Before constructing or renovating an existing building complex, designers need to predict the potential microclimate changes the new or renovated building might cause to its surrounding environment, based on its design values. If the microclimate changes are significant, it indicates a substantial impact on the surrounding microclimate. Therefore, the design values ​​for the building need to be redesigned to match the smaller microclimate changes, thus optimizing the design and preventing the selection of a design scheme with adverse microclimate effects.

[0048] In this embodiment, step S100 includes the following specific steps S101 to S107:

[0049] S101, obtain several original architectural design variables and set simulated values ​​for several of the original architectural design variables.

[0050] S102, Based on the simulated values ​​of several original architectural design variables, generate an architectural design scene.

[0051] For each original architectural design variable, a corresponding simulated value is assigned. The simulated value represents the possible values ​​of that original architectural design variable. Then, the Latin hypercube sampling method is applied to the 200 sets of simulated values ​​of all original architectural design variables to generate 200 architectural design scenarios. Each set of simulated values ​​contains the simulated values ​​of all original architectural design variables.

[0052] S103, Generate a building simulation geometric model based on the architectural design scenario.

[0053] Architectural design scenarios are discrete data, meaning the form of an architectural design scenario is not a building. The GIS algorithm is used to generate 200 3D architectural simulation geometric models for the 200 architectural design scenarios mentioned above, in order to transform the architectural design scenarios in data form into visual architectural simulation geometric models.

[0054] S104, Simulate and analyze the microclimate simulation changes caused by the addition of the building simulation geometry model.

[0055] The Fluent software was used to simulate the microclimate of a region before and after the addition of the architectural simulation geometry model. The difference between the microclimate before and after the addition of the model is the microclimate simulation change data. This microclimate simulation change data includes temperature and wind speed changes. Since the number of 3D architectural simulation geometry models is 200, the number of microclimate simulation change data is also 200.

[0056] S105, apply a global sensitivity analysis algorithm to the microclimate simulation change data to obtain the sensitivity of the microclimate to each of the original variables of the building design.

[0057] In other words, the Morris global sensitivity analysis method was applied to 200 sets of microclimate simulation change data to calculate the significance of the simulated value of each building design original variable on temperature and wind speed changes. This means analyzing the sensitivity of temperature to each building design original variable and the sensitivity of temperature to each building design original variable.

[0058] S106, Based on the sensitivity of each of the original architectural design variables, sensitive variables are selected from the original architectural design variables.

[0059] The sensitive variables are selected from the original architectural design variables. As shown in Table 1, the sensitive variables include building height, building orientation, building width, building length, window-to-wall ratio, external wall heat transfer coefficient, external window heat transfer coefficient, wall emissivity, COP, and cooling load per unit cooling area. The value ranges of these ten sensitive variables are shown in Table 1. Designers need to focus on the impact of these six architectural design variables on the microclimate when designing buildings.

[0060] Table 1

[0061] Design variables scope unit Building height (h) 6～200 m Building orientation (o) 0～180 ° Building width (w) Fill in according to the actual site requirements. m Building length (l) Fill in according to the actual site requirements. m Window-to-wall ratio (r) 0～1 - <![CDATA[Exterior wall heat transfer coefficient (k1)]]> Fill in according to the specifications. <![CDATA[W / (m 2 ·K)]]> <![CDATA[Heat transfer coefficient of the outer window (k2)]]> Fill in according to the specifications. <![CDATA[W / (m 2 ·K)]]> Wall emissivity (e) 0～1 - COP 4～6 - Cooling load per unit cooling area (c) 80～150 <![CDATA[W / m 2 ]]>

[0062] S107. Based on the aforementioned sensitive variables, obtain the architectural design variables for a single building.

[0063] As shown in Table 2, the building design variables include building height, building orientation, building length-to-width ratio, overall heat transfer coefficient of building envelope, wall emissivity, and heat dissipation per unit building area.

[0064] Building height, building orientation, and building length-to-width ratio are building form design variables. The building length-to-width ratio is the ratio of building length to building width (both building length and building width are sensitive variables). The overall heat transfer coefficient of the building envelope, wall emissivity, and heat dissipation per unit building area are building thermal characteristic design variables. As shown in Table 2, the overall heat transfer coefficient of the building envelope x4: x4 = r*k2 + (1-r)*k1, where r is the window-to-wall ratio, k1 is the external wall heat transfer coefficient, and k2 is the external window heat transfer coefficient. The window-to-wall ratio, external wall heat transfer coefficient, and external window heat transfer coefficient are all sensitive variables. Wall emissivity x5: x5 = e, wall emissivity is also sensitive. Heat dissipation per unit building area x6: x6 = (1 + 1 / COP)*c*0.8, where c is the sensitive variable (cooling load per unit cooling area), and COP is also a sensitive variable.

[0065] Table 2

[0066]

[0067] In this embodiment, step S200 includes the following specific steps: determining the building form design variable and the building thermal characteristic design variable among the building design variables; setting the design value of the building form design variable within the preset value range of the building form design variable; obtaining the calculation formula corresponding to the building thermal characteristic design variable; and determining the design value of the building thermal characteristic design variable based on the calculation formula.

[0068] The architectural form design variables include building height, building orientation, and building aspect ratio. The design values ​​for these three architectural form design variables are set within the range specified in Table 1. The building thermal characteristic design variables include the overall heat transfer coefficient of the building envelope, wall emissivity, and heat dissipation per unit building area. The overall heat transfer coefficient of the building envelope is obtained using the formula x4 = r * k2 + (1 - r) * k1 in Table 2; the heat dissipation per unit building area is obtained using the formula x6 = (1 + 1 / COP) * c * 0.8 in Table 2.

[0069] In this embodiment, step S300 includes the following specific steps S301 and S302:

[0070] S301, apply an SVR-based temperature change prediction model to the design values ​​of the building height, building orientation, building aspect ratio, overall heat transfer coefficient of the building envelope, wall emissivity, and heat dissipation per unit building area to predict the temperature change data of the single building designed based on the design values ​​to the surrounding environment.

[0071] S302, apply a LightGBM-based wind speed change prediction model to the design values ​​of the building height, building orientation, building aspect ratio, overall heat transfer coefficient of the building envelope, wall emissivity, and heat dissipation per unit building area to predict the wind speed change data of the single building designed based on the design values ​​to the surrounding environment, and use the wind speed change data and the temperature change data as microclimate change data.

[0072] The Python program will automatically call the pre-trained and packaged proxy models, which include a temperature change prediction model based on SVR (Support Vector Regression) and a wind speed change prediction model based on LightGBM (Gradient Boosting Tree).

[0073] The parameter optimization process for the temperature change prediction model and the wind speed change prediction model is as follows:

[0074] Using 200 sets of CFD simulation data corresponding to different architectural designs, and optimizing the model parameters using the ten-fold cross-validation method, the parameters of the two prediction models were finally determined as shown in Table 3.

[0075] Table 3

[0076]

[0077]

[0078] As shown in Table 3, the temperature change prediction model uses the radial basis function (RBF) as the kernel function, with a regularization parameter of 1 and a kernel function coefficient of 0.001; the wind speed change prediction model has a mean absolute error (MAE) of 0.352, uses 1000 trees, has a maximum tree depth of 2, a maximum number of leaves of 3, a learning rate of 0.05, and a subsample ratio of 1.

[0079] Example 2, based on Example 1, applies the microclimate prediction method of Example 1 to optimize building design variables, including the following specific steps: obtaining pedestrian thermal discomfort in the surrounding environment of the building based on microclimate change data; obtaining building design values ​​related to energy consumption; obtaining total building energy consumption based on microclimate, building design values, and initial values ​​of building design variables; optimizing the initial values ​​of building design variables to synergistically optimize pedestrian thermal discomfort and total building energy consumption until both optimized pedestrian thermal discomfort and optimized total building energy consumption meet the optimization objectives, thus obtaining the optimization target value of the building design variables.

[0080] Building design values ​​include occupancy rate, light usage rate, electrical equipment usage rate and air conditioning equipment usage rate, window-to-wall ratio, wall specific heat, wall thermal absorption rate, wall solar energy absorption rate, wall visible light absorption rate, roof specific heat, roof thermal absorption rate, roof solar energy absorption rate, roof visible light absorption rate, floor specific heat, ground thermal absorption rate, window solar thermal gain coefficient, window visible light transmittance, and infiltration air mass flow coefficient, etc.

[0081] Minimize the building volume V and d min ≤d≤d max Under these two constraints, the optimization objective is pedestrian thermal discomfort D. dis To obtain the minimum value and total building energy consumption E tot The minimum value is obtained, and the optimization target value is the value of the architectural design variables that satisfies the optimization objective, where V = h * x * d 2 The value of V is 127500m 3 d max The value is 125m, d min The value is 15m, d is the building width, h is the building height, and x is the building aspect ratio. E tot =E CE +E LE +E EE E CE For cooling power consumption, E LE For lighting power consumption, E EE Electricity consumption for electrical equipment.

[0082] In summary, the mean absolute error (MAE) of the temperature change prediction model based on SVR (Support Vector Regression) in this invention is 0.194, and the mean absolute error (MAE) of the wind speed change prediction model based on LightGBM (Gradient Boosting Tree) is 0.352. Both prediction models exhibit very low absolute errors, ensuring the accuracy of the prediction results. Furthermore, the computation time for a single prediction has been reduced from over 5 hours to milliseconds. Therefore, this invention can quickly, comprehensively, and accurately predict the impact on local microclimate in the early stages of new building design or existing building renovation, helping designers achieve ideal microclimate effects during the design process or avoid unacceptable microclimate changes. In addition, this invention only requires inputting the values ​​of six design variables related to the building design into the model to obtain the prediction results immediately, thereby improving operational simplicity.

[0083] This embodiment also provides a microclimate prediction device based on machine learning, such as... Figure 2 As shown, the device comprises the following components:

[0084] Design variable filtering module 01 is used to determine the architectural design variables of a single building;

[0085] Design value design module 02 is used to set the design values ​​of the architectural design variables;

[0086] Prediction module 03 is used to apply a machine learning model to the design values ​​to predict the microclimate change data of the surrounding environment caused by the single building designed based on the design values.

[0087] Based on the above embodiments, the present invention also provides a terminal device, the principle block diagram of which can be as follows: Figure 3 As shown, the terminal device includes a processor, memory, network interface, and display screen connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a machine learning-based microclimate prediction method. The display screen can be an LCD screen or an e-ink screen.

[0088] Those skilled in the art will understand that Figure 3 The schematic diagram shown is only a partial structural diagram related to the present invention and does not constitute a limitation on the terminal device to which the present invention is applied. The specific terminal device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0089] In one embodiment, a terminal device is provided, the terminal device including a memory, a processor, and a machine learning-based microclimate prediction program stored in the memory and executable on the processor. When the processor executes the machine learning-based microclimate prediction program, it implements the following operation instructions:

[0090] Determine the architectural design variables for a single building;

[0091] Set the design values ​​for the architectural design variables;

[0092] A machine learning model is applied to the design values ​​to predict the microclimate changes that the single building designed based on the design values ​​will bring to the surrounding environment.

[0093] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A microclimate prediction method based on machine learning, characterized by, include: Determine the architectural design variables for a single building; Set the design values ​​for the architectural design variables; A machine learning model is applied to the design values ​​to predict the microclimate changes that the single building designed based on the design values ​​will bring to the surrounding environment.

2. The machine learning based microclimate prediction method of claim 1, wherein, The determination of architectural design variables for a single building includes: Obtain several original architectural design variables and set simulated values ​​for several of the original architectural design variables; Based on the simulated values ​​of several original architectural design variables, an architectural design scene is generated; Based on the architectural design scenario, generate a simulated geometric model of the building; The simulation analysis examines the microclimate changes caused by incorporating the building's simulated geometric model. Based on the microclimate simulation change data, sensitive variables are selected from several original building design variables; Based on the aforementioned sensitive variables, the architectural design variables for a single building are obtained.

3. The machine learning based microclimate prediction method of claim 2, wherein, Based on the microclimate simulation change data, sensitive variables were selected from several original building design variables, including: A global sensitivity analysis algorithm is applied to the microclimate simulation change data to obtain the sensitivity of the microclimate to each of the original building design variables; Based on the sensitivity of each of the original architectural design variables, sensitive variables are selected from several of the original architectural design variables.

4. The machine learning based microclimate prediction method of claim 1, wherein, Setting the design values ​​for the architectural design variables includes: Determine the architectural form design variables and architectural thermal characteristic design variables among the architectural design variables; Within the preset range of values ​​for the architectural form design variables, the design values ​​for the architectural form design variables are set. Obtain the calculation formula corresponding to the building thermal characteristic design variable, and determine the design value of the building thermal characteristic design variable based on the calculation formula.

5. The machine learning based microclimate prediction method of claim 4, wherein, The building form design variables include building height, building orientation and building length-to-width ratio; the building thermal characteristic design variables include the overall heat transfer coefficient of the building envelope, wall emissivity and heat dissipation per unit building area. Applying a machine learning model to the design values, predicting microclimate changes in the surrounding environment caused by the single building designed based on the design values, including: A temperature change prediction model based on SVR is applied to the design values ​​of the building height, the building orientation, the building length-to-width ratio, the overall heat transfer coefficient of the building envelope, the wall emissivity, and the heat dissipation per unit building area to predict the temperature change data of the single building designed based on the design values ​​to the surrounding environment. The design values ​​of the building height, building orientation, building aspect ratio, overall heat transfer coefficient of the building envelope, wall emissivity, and heat dissipation per unit building area are applied to a wind speed change prediction model based on LightGBM to predict the wind speed change data of the single building designed based on the design values ​​to the surrounding environment, and the wind speed change data and the temperature change data are used as microclimate change data.

6. The microclimate prediction method based on machine learning as described in claim 5, characterized in that, The regularization parameter of the SVR is 1, and the kernel function coefficient of the SVR is 0.

001.

7. The microclimate prediction method based on machine learning as described in claim 5, characterized in that, The LightGBM tree has 1000 trees, a maximum tree depth of 2, a maximum number of leaves of 3, a learning rate of 0.05, and a subsample ratio of 1 for training.

8. A microclimate prediction device based on machine learning, characterized in that, The device comprises the following components: The design variable filtering module is used to determine the architectural design variables for a single building. The design value design module is used to set the design values ​​of the architectural design variables; The prediction module is used to apply a machine learning model to the design values ​​to predict the microclimate changes that the single building designed based on the design values ​​will bring to the surrounding environment.

9. A terminal device, characterized in that, The terminal device includes a memory, a processor, and a machine learning-based microclimate prediction program stored in the memory and executable on the processor. When the processor executes the machine learning-based microclimate prediction program, it implements the steps of the machine learning-based microclimate prediction method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a machine learning-based microclimate prediction program, which, when executed by a processor, implements the steps of the machine learning-based microclimate prediction method as described in any one of claims 1-7.

Images