Data-driven traditional courtyard morphological characteristic comfort influence evaluation method

Abstract

The invention discloses a data-driven traditional courtyard morphological characteristic comfort influence evaluation method, which comprises the following steps: S1, data collection and arrangement: collecting morphological characteristic data of a traditional courtyard and thermal comfort data related to the morphological characteristic data, and constructing a morphological index evaluation system; s2, numerical simulation: constructing an ENVI-met model in combination with the data collected in S1, simulating microclimate of typical days in winter and summer to obtain a UTCI thermal comfort index, and calculating an average value of a simulation domain and time to represent the thermal comfort level of each courtyard; s3, building and analyzing a machine learning model: performing regression analysis on the relationship between the courtyard morphological characteristics and the thermal comfort index by using an XGBoost model; and S4: interpretability analysis: by means of an SHAP interpretation method, checking the importance and local dependency relationship of the morphological characteristics, visualizing a nonlinear correlation mode of a single variable, and understanding the direct influence degree of different morphological characteristics on thermal comfort and the cooperation or inhibition effect between the different morphological characteristics through an SHAP value.

Description

Technical Field

[0001] The present invention relates to the technical field of urban renewal, and particularly relates to a data-driven evaluation method for the impact of the morphological characteristics of traditional courtyards on comfort. Background Art

[0002] As an important part of cultural heritage, traditional courtyards not only carry rich historical and cultural information, but also embody unique architectural aesthetics and ecological wisdom. Their enclosed spatial layout, delicate structural details, and harmonious coexistence with the natural environment provide us with valuable experience and inspiration. Traditional courtyard buildings, with their unique enclosed spatial form, demonstrate excellent climate regulation functions. These buildings not only contribute to alleviating global warming and the urban heat island effect, but also are of great significance to promoting urban sustainable development.

[0003] However, in current research, the comprehensiveness of climate zones, the systematicness of morphological elements, and the non-linear mechanism have not been deeply explored. Especially in cities like Beijing, traditional courtyards have been affected by rapid urbanization, forming complex and highly heterogeneous spatial forms, which not only damage the traditional urban landscape, but also pose a threat to the thermal comfort of the courtyards. Traditional thermal comfort research methods are often based on a small number of morphological characteristics and are difficult to accurately capture the complex relationship between multi-dimensional environmental elements and thermal comfort. Therefore, a data-driven evaluation method for the impact of the morphological characteristics of traditional courtyards on comfort is proposed. Summary of the Invention

[0004] The purpose of the present invention is to solve the problem existing in the prior art that traditional thermal comfort research methods are often based on a small number of morphological characteristics and are difficult to accurately capture the complex relationship between multi-dimensional environmental elements and thermal comfort, and to propose a data-driven evaluation method for the impact of the morphological characteristics of traditional courtyards on comfort.

[0005] In order to achieve the above purpose, the present invention adopts the following technical scheme: A data-driven evaluation method for the impact of the morphological characteristics of traditional courtyards on comfort, comprising the following steps: S1: Data collection and collation: Through means such as on-site investigation and remote sensing image interpretation, collect the morphological characteristic data of traditional courtyards and the related thermal comfort data, and comprehensively consider the built environment and natural environment to construct a morphological index evaluation system; S2: Numerical simulation: Combine the data collected in S1 to construct an ENVI-met model, simulate the microclimate on typical days in winter and summer, obtain the UTCI (Universal Thermal Climate Index) thermal comfort index, and take the average value of the simulation domain and time to represent the thermal comfort level of each courtyard; S3: Machine learning model construction and analysis: Use the XGBoost model to perform regression analysis on the relationship between the morphological characteristics of the courtyard and the thermal comfort index; S4: Interpretability analysis: By means of the SHAP (Shapley Additive exPlanations) interpretation method, examine the importance of morphological features, local dependencies, and visualize the non-linear correlation patterns of individual variables. Understand the direct influence degree of different morphological features on thermal comfort through SHAP values, as well as their synergistic or inhibitory effects; S5: Revealing the influence mechanism and proposing optimization strategies: Comprehensively analyze the regression results of the machine learning model and the output results of the SHAP interpretation method, reveal the non-linear influence mechanism of courtyard morphological features on thermal comfort in winter and summer, and according to the revealed influence mechanism, propose optimization strategies for traditional courtyard morphology to achieve the improvement of thermal comfort.

[0006] The above further includes: Further, in S1, the morphological index evaluation system includes comprehensive indicators, built environment indicators, and natural environment indicators. The comprehensive indicators include the sky view factor, the built environment indicators include building coverage rate, shape index, aspect ratio, courtyard area, courtyard boundary ratio, and street boundary ratio, and the natural environment indicators include vegetation coverage rate.

[0007] Further, in S2, simulate the microclimate on typical days in winter and summer, obtain the UTCI (Universal Thermal Climate Index) thermal comfort index, and calculate the average value for the simulation domain and time. The specific steps are as follows: ENVI-met model construction: According to the data collected in S1, construct a three-dimensional model of each traditional courtyard in ENVI-met software, and set model parameters, including meteorological conditions (such as temperature, humidity, wind speed, wind direction, etc.), surface type, vegetation coverage, etc.; Microclimate simulation: Select the meteorological conditions on typical days in winter and summer as inputs, and run the ENVI-met model for microclimate simulation. During the simulation process, the model will calculate microclimate parameters such as air temperature, humidity, wind speed, and radiation at each time step; UTCI thermal comfort index calculation: Use the simulated microclimate parameters to calculate the UTCI thermal comfort index; Calculation of the average value for the simulation domain and time: Perform spatial averaging on the UTCI values within the simulation domain to represent the thermal comfort level of the entire courtyard, and perform temporal averaging on the UTCI values within the simulation time period to obtain the average thermal comfort level for the entire typical day.

[0008] Further, in the calculation of the UTCI thermal comfort index, the formula for calculating the UTCI thermal comfort index is expressed as ; ; Among them, Represents the thermal comfort of the th courtyard in the hour per unit time; Represents the number of simulation units in the th courtyard; Represents the temperature; Represents the temperature deviation caused by environmental or other factors; Represents the relative humidity; Represents the temperature adjustment time; Represents the wind speed; th unit in the hour of the value, and the calculation formula of this value is expressed as; ; Among them, Represents the thermal comfort index of the th courtyard; Represents 24 hours of a typical day in winter or summer.

[0009] Furthermore, in S3, the XGBoost model is used to perform a regression analysis on the relationship between the courtyard form characteristics and the thermal comfort index, including the following steps: Data preparation: Combine the courtyard form characteristics and the UTCI value into a data set, where the form characteristics are used as independent variables (features) and the UTCI value is used as the dependent variable (target variable); Data preprocessing: Clean the data, remove missing values or outliers, and perform standardization or normalization on the data to improve the convergence speed and performance of the model; XGBoost model construction: Set the parameters of the model, such as the learning rate, maximum depth, subsample ratio, etc., and these parameters can be optimized through methods such as cross-validation or grid search; Model training: Divide the data set into a training set and a test set, use the training set data to train the XGBoost model, and optimize the objective function by continuously adding new decision trees. The formula of the objective function is expressed as , where is the loss function, is the actual value, is the predicted value, is the regularization term of the kth decision tree. During the training process, monitor the changes in the loss function and evaluation metrics (such as the R² value) of the model; Model evaluation: Use the test set data to evaluate the performance of the model, and calculate the prediction accuracy (such as the R² value) and the mean square error (MSE) of the model; Regression analysis: Use the trained XGBoost model to conduct a regression analysis on the relationship between the courtyard morphological characteristics and the UTCI value, and analyze the feature importance and local dependence relationship output by the model to reveal the influence mechanism of the courtyard morphology on thermal comfort.

[0010] Furthermore, in S4, understand the direct influence degree of different morphological characteristics on thermal comfort through SHAP values. The calculation method of SHAP values is expressed as: Among them, represents the SHAP value of the th feature; represents the set of features participating in the prediction; represents the set of features; represents the number of features; represents the machine learning model used for interpretation; is the model output value under the feature combination S; represents the subset with the added feature ; represents the set ;

[0011] Furthermore, the SHAP method calculates a baseline value through the features to be explained by the model and the data distribution, and decomposes the variable to be explained by the model into SHAP values and the baseline value. The SHAP method assigns SHAP values to the indicators according to the contribution of each courtyard morphological characteristic to thermal comfort, expressed as; Where: ∈ represents whether the th morphological feature participates in the model prediction; represents the number of feature categories in the courtyard morphology; represents the th SHAP value of the morphological feature.

[0012] Furthermore, in S5, reveal the non-linear influence mechanism of the courtyard morphological characteristics on winter and summer thermal comfort, including analyzing the contribution degree, influence mode of each feature on thermal comfort, and their interaction relationship.

[0013] The present invention has the following beneficial effects: 8. In the present invention, an index system covering a variety of morphological characteristics is constructed, which can comprehensively and systematically explore the influence of traditional courtyard morphology on thermal comfort. By introducing the XGBoost model, the courtyard morphological parameters and thermal comfort data can be accurately fitted to reveal their non-linear relationship.

[0014] 9. In the present invention, by means of the SHAP interpretation method, the non-linear relationship between environmental factors and thermal comfort is clearly expressed, improving the transparency and credibility of the research results. It not only reveals the mathematical relationship between morphological characteristics and thermal comfort, but also provides a scientific basis and practical guidance for the optimization of traditional courtyard forms, the improvement of thermal comfort, and sustainable development in the future. Description of the Drawings

[0015] Figure 1 It is a step diagram of a data-driven method for evaluating the impact of traditional courtyard morphological characteristics on comfort proposed by the present invention. Detailed Embodiment

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0017] Please refer to Figure 1 As shown, the present invention is a data-driven method for evaluating the impact of traditional courtyard morphological characteristics on comfort, including the following steps: S1: Data collection and collation: Through means such as field investigation and remote sensing image interpretation, collect the morphological characteristic data of traditional courtyards and the related thermal comfort data, and comprehensively construct a morphological index evaluation system for the built environment and natural environment; S2: Numerical simulation: Combine the data collected in S1 to construct an ENVI-met model, simulate the microclimate on typical days in winter and summer, obtain the UTCI (Universal Thermal Climate Index) thermal comfort index, and take the average value of the simulation domain and time to represent the thermal comfort level of each courtyard; S3: Machine learning model construction and analysis: Use the XGBoost model to perform regression analysis on the relationship between courtyard morphological characteristics and thermal comfort index; S4: Interpretability analysis: By means of the SHAP (Shapley Additive exPlanations) interpretation method, test the importance of morphological characteristics, local dependence relationships, and visualize the non-linear correlation patterns of individual variables. Understand the direct impact degree of different morphological characteristics on thermal comfort through SHAP values, as well as their synergistic or inhibitory effects; S5: Revealing the Influence Mechanism and Proposing Optimization Strategies: Comprehensively analyze the regression results of the machine learning model and the output results of the SHAP interpretation method to reveal the non-linear influence mechanism of courtyard morphological characteristics on thermal comfort in winter and summer. According to the revealed influence mechanism, propose optimization strategies for traditional courtyard morphology to achieve the improvement of thermal comfort.

[0018] In one embodiment, for the above S1, in S1, the morphological index evaluation system includes comprehensive indicators, built environment indicators, and natural environment indicators. The comprehensive indicators include the sky view angle, and the built environment indicators include building coverage rate, shape index, aspect ratio, courtyard area, courtyard boundary ratio, and street boundary ratio. The natural environment indicators include vegetation coverage rate.

[0019] Table 1 Courtyard Morphology Design Parameter System Among them, represents the elevation angle of the i-th obstacle (such as a building or a tree), reflecting the impact of the height of the obstacle on the visible sky range. represents the azimuth angle span of the i-th obstacle (expressed in degrees), describing the horizontal coverage range of the obstacle. represents the area of the i-th functional area in the courtyard, used to quantify the total scale of the open space in the courtyard, affecting the heat capacity and ventilation potential of the thermal environment. The base area of the i-th building (m²). represents the total site area (m²), reflecting the building density. A high coverage rate may lead to an exacerbation of the heat island effect. P represents the total perimeter of the courtyard boundary (m), CA represents the total courtyard area (m²), and in the aspect ratio represents the length of the i-th area in the courtyard (m). represents the width of the i-th area, and in the courtyard boundary ratio represents the length of the i-th natural boundary (such as a hedge or a water body) in the courtyard (m), P represents the total perimeter of the courtyard, represents the length of the i-th boundary adjacent to the street in the courtyard (m). represents the area of the ii-th vegetation-covered area (m²).

[0020] In one embodiment, for the above S2, in S2, simulate the microclimate on typical days in winter and summer to obtain the UTCI (Universal Thermal Climate Index) thermal comfort index, and take the average value of the simulation domain and time. The specific steps are as follows: ENVI-met model construction: Based on the data collected in S1, a three-dimensional model of each traditional courtyard is constructed in ENVI-met software, and model parameters are set, including meteorological conditions (such as temperature, humidity, wind speed, wind direction, etc.), surface types, vegetation coverage, etc.; Microclimate simulation: Select the meteorological conditions of typical days in winter and summer as inputs, and run the ENVI-met model for microclimate simulation. During the simulation process, the model will calculate microclimate parameters such as air temperature, humidity, wind speed, and radiation at each time step; UTCI thermal comfort index calculation: Use the simulated microclimate parameters to calculate the UTCI thermal comfort index; Calculation of spatial and temporal averages in the simulation domain: Perform spatial averaging on the UTCI values within the simulation domain to represent the thermal comfort level of the entire courtyard, and perform temporal averaging on the UTCI values within the simulation time period to obtain the average thermal comfort level for the entire typical day.

[0021] In one embodiment, for the above UTCI thermal comfort index calculation, in the UTCI thermal comfort index calculation, the formula for calculating the UTCI thermal comfort index is expressed as ; ; Among them, represents the thermal comfort level of the th courtyard at the th hour per unit time; represents the number of simulation units in the th courtyard; represents temperature; represents the temperature deviation caused by the environment or other factors; represents relative humidity; represents the temperature adjustment time; represents wind speed; represents the th unit in the courtyard at the th hour, and the formula for calculating this value is expressed as; ; ; Among them, represents the thermal comfort index of the th courtyard; represents 24 hours of a typical day in winter or summer.

[0022] In one embodiment, for S3, in S3, a regression analysis of the relationship between courtyard morphological characteristics and thermal comfort index is performed using the XGBoost model, including the following steps: Data Preparation: Combine the courtyard morphological features and UTCI values into a dataset, where the morphological features are used as independent variables (features) and the UTCI values are used as the dependent variable (target variable). Data Preprocessing: Clean the data to remove missing values or outliers, and perform standardization or normalization on the data to improve the convergence speed and performance of the model. XGBoost Model Construction: Set the parameters of the model, such as the learning rate, maximum depth, subsample ratio, etc. These parameters can be optimized through methods such as cross-validation or grid search. Model Training: Divide the dataset into a training set and a test set. Use the training set data to train the XGBoost model, and optimize the objective function by continuously adding new decision trees. The formula of the objective function is , where is the loss function,[[]] is the actual value,[[]] is the predicted value,[[]] is the regularization term of the k-th decision tree. During the training process, monitor the changes in the loss function and evaluation metrics (such as the R² value) of the model. Model Evaluation: Use the test set data to evaluate the performance of the model, and calculate the prediction accuracy (such as the R² value) and mean squared error (MSE) of the model. Regression Analysis: Use the trained XGBoost model to perform regression analysis on the relationship between the courtyard morphological features and UTCI values, and analyze the feature importance and local dependence relationship output by the model to reveal the influence mechanism of the courtyard morphology on thermal comfort.

[0023] First, perform a visual analysis on the courtyard morphological features and UTCI values to initially understand their relationship. For example, plot a scatter plot of the sky view factor (SVF) and UTCI values to observe whether there is a correlation between them.

[0024] Then, use the training set data to train the XGBoost model and use the test set data to evaluate the performance of the model. Assume that the R² value of the trained model on the test set reaches 0.85, indicating that the model has good prediction ability.

[0025] Next, analyze the feature importance output by the model. Assume that the model shows that the sky view factor (SVF) is the most important feature, and its contribution degree is much higher than other features. This indicates that SVF has the most significant impact on the courtyard thermal comfort.

[0026] Finally, analyze the local dependence relationship output by the model. For example, plot the local dependence graph between SVF and UTCI values to observe the change trend of UTCI values under different SVF values. This helps to more deeply understand the influence mechanism of SVF on the courtyard thermal comfort.

[0027] In one embodiment, for S4 above, in S4, the direct influence degree of different morphological features on thermal comfort is understood through SHAP values, and the calculation method of SHAP values is expressed as: Wherein, represents the SHAP value of the -th feature; represents the set of features participating in the prediction; represents the set of features; represents the number of features; represents the machine learning model for interpretation; is the model output value under the feature combination S; represents the subset after adding the feature ; represents the set ; the number of elements in it.

[0028] In one embodiment, for SHAP above, the SHAP method calculates a baseline value through the features to be explained by the model and the data distribution , and decomposes the variable to be explained by the model into SHAP values and the baseline value. The SHAP method assigns SHAP values to the indicators according to the contribution of each courtyard morphological feature to thermal comfort, expressed as; Where: ∈ represents whether the -th morphological feature participates in the model prediction; represents the number of feature categories in the courtyard morphology; represents the SHAP value of the -th morphological feature.

[0029] In one embodiment, for S5 above, in S5, the non-linear influence mechanism of courtyard morphological features on winter and summer thermal comfort is revealed, including analyzing the contribution degree, influence mode of each feature on thermal comfort, and their interaction relationship. Based on the revealed influence mechanism, an optimization strategy for traditional courtyard morphology is proposed to achieve the improvement of thermal comfort. Specific examples of optimization strategies:

[0030] Adjust the courtyard morphological parameters: By reducing the sky view range (such as increasing the complexity of the courtyard boundary, setting up sunshade facilities, etc.), arranging in a north-south bias (to reduce east-west solar radiation), etc., adjust the courtyard morphological parameters to improve thermal comfort.

[0031] Example illustration: In traditional courtyards in Beijing, the SVF can be reduced by setting up obstacles such as high walls or hedges at the courtyard boundary, thereby reducing the impact of solar radiation on the courtyard temperature. At the same time, by arranging in a north-south bias, the solar radiation in winter can be fully utilized and the solar radiation in summer can be reduced.

[0032] Abundant vegetation: Plant an appropriate amount of vegetation such as trees and shrubs in the courtyard to provide shade and cooling effects. At the same time, the vegetation can also increase air humidity and reduce wind speed, thereby further improving thermal comfort.

[0033] Example illustration: In traditional courtyards in Beijing, some tree species with good shading effects (such as pagoda trees, ginkgo trees, etc.) can be selected for planting, and vegetation areas such as green belts or flower beds can be set up in the courtyard.

[0034] Optimize the building layout and structure: By optimizing the building layout and structure (such as increasing the building height, adjusting the building spacing, etc.), the ventilation and shading conditions in the courtyard can be improved.

[0035] Example illustration: In traditional courtyards in Beijing, the direct sunlight on low-rise buildings can be reduced by increasing the building height; at the same time, the ventilation conditions in the courtyard can be improved by adjusting the building spacing, thereby reducing the courtyard temperature.

[0036] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A data-driven evaluation method for the comfort impact of the morphological characteristics of traditional courtyard houses, characterized in that, The following steps are involved: S1: Data collection and organization: Collect morphological characteristics data of traditional courtyards and related thermal comfort data, integrate the built environment and natural environment, and construct a morphological index evaluation system; S2: Numerical simulation: Combine the data collected in S1 to build the ENVI-met model, simulate the microclimate of a typical day in winter and summer, obtain the UTCI thermal comfort index, and calculate the average value over the simulation domain and time to represent the thermal comfort level of each courtyard; S3: Machine learning model construction and analysis: Use the XGBoost model to conduct regression analysis on the relationship between courtyard morphological characteristics and thermal comfort index; S4: Interpretability Analysis: Using the SHAP interpretation method, the importance and local dependencies of morphological features are examined, and the nonlinear correlation patterns of individual variables are visualized. SHAP values are used to understand the direct impact of different morphological features on thermal comfort, as well as the synergistic or inhibitory effects between them. S5: Influence mechanism revealed and optimization strategy proposed: Comprehensively analyze the regression results of the machine learning model and the output results of the SHAP interpretation method to reveal the nonlinear influence mechanism of courtyard morphological characteristics on thermal comfort in winter and summer. Based on the revealed influence mechanism, an optimization strategy for the traditional courtyard morphology is proposed.

2. The comfort impact evaluation method for the morphological characteristics of traditional courtyards driven by data according to claim 1, wherein In S1, the morphological index evaluation system includes comprehensive indicators, built environment indicators and natural environment indicators. The comprehensive indicators include sky visibility angle, the built environment indicators include building coverage, shape index, aspect ratio, courtyard area, courtyard boundary ratio and street boundary ratio, and the natural environment indicators include vegetation coverage.

3. The comfort impact evaluation method for the morphological characteristics of traditional courtyards driven by data according to claim 1, characterized in that In S2, the microclimate of a typical winter and summer day is simulated to obtain the UTCI thermal comfort index, and the average value is calculated for the simulation domain and time. The specific steps are as follows: ENVI-met model construction: Based on the data collected in S1, a three-dimensional model of each traditional courtyard was constructed in ENVI-met software, and the model parameters were set; Microclimate simulation: Select the meteorological conditions of a typical winter or summer day as input and run the ENVI-met model to simulate the microclimate. During the simulation, the model calculates the microclimate parameters for each time step. Calculation of UTCI thermal comfort index: Calculate the UTCI thermal comfort index using the simulated microclimate parameters; Simulation domain and temporal average calculation: The UTCI values within the simulation domain were spatially averaged to represent the thermal comfort level of the entire courtyard, and the UTCI values within the simulation time period were temporally averaged to obtain the average thermal comfort level of the entire typical day.

4. A method for evaluating the comfort impact of traditional courtyard form characteristics driven by data according to claim 3, characterized in that, In the calculation of UTCI thermal comfort index, the formula for calculating UTCI thermal comfort index is expressed as ； ； Among them, represents the thermal comfort of the th courtyard in the th hour per unit time; represents the number of simulation units in the th courtyard; represents temperature; represents the temperature deviation caused by environmental or other factors; represents relative humidity; represents the temperature regulation time; represents wind speed; represents the th unit in this courtyard at the th hour, and the calculation formula of this value is expressed as; value, and the calculation formula of this value is expressed as; ； Among them, represents the thermal comfort index of the th courtyard; represents 24 hours of a typical day in winter or summer.

5. A comfort impact evaluation method for the morphological characteristics of traditional courtyards driven by data according to claim 1, characterized in that In S3, the XGBoost model is used to perform regression analysis on the relationship between courtyard morphological characteristics and thermal comfort index, including the following steps: Data preparation: The courtyard morphological characteristics and UTCI values were combined into a dataset, with the morphological characteristics as the independent variables and the UTCI values as the dependent variables; Data preprocessing: clean the data, remove missing values or outliers, and standardize or normalize the data; XGBoost model construction: setting model parameters; Model training: Divide the dataset into a training set and a test set. Use the training set data to train the XGBoost model, and optimize the objective function by continuously adding new decision trees. The formula of the objective function is , where is the loss function,[[]] is the actual value,[[]] is the predicted value,[[]] is the regularization term of the k-th decision tree. During the training process, monitor the changes in the loss function and evaluation metrics of the model; Model evaluation: Use the test set data to evaluate the performance of the model, and calculate the prediction accuracy and mean square error of the model; Regression analysis: Use the trained XGBoost model to perform regression analysis on the relationship between courtyard morphological characteristics and UTCI values, and analyze the feature importance and local dependence relationship output by the model.

6. The comfort impact evaluation method for the morphological characteristics of traditional courtyards driven by data according to claim 1, wherein In S4, understand the direct influence degree of different morphological characteristics on thermal comfort through SHAP values. The calculation method of SHAP values is expressed as: in, Indicates the SHAP value of each feature; Represents the feature set involved in prediction; Represents a collection of features; Indicates the number of features; Representing machine learning models for interpretation; is the model output value under the feature combination S; Representation subset Add features A collection of Representing a collection The number of elements in .

7. A method for evaluating the comfort impact of the morphological characteristics of traditional courtyards driven by data according to claim 6, characterized in that, The SHAP method calculates a baseline value based on the features explained by the model and the data distribution. It decomposes the explained variable of the model into SHAP values and a baseline value. The SHAP method assigns SHAP values to the indicators according to the contribution of each courtyard morphological feature to thermal comfort, expressed as; Wherein: ∈ indicates whether the th morphological feature participates in model prediction; represents the number of feature categories in the courtyard form; indicates the SHAP value of the th morphological feature.

8. A method for evaluating the comfort impact of the morphological characteristics of traditional courtyards driven by data according to claim 1, characterized in that, In S5, reveal the non-linear influence mechanism of courtyard morphological characteristics on winter and summer thermal comfort, including analyzing the contribution degree, influence method of each characteristic on thermal comfort, and the interaction relationship between them.

Images