Rural residence thermal environment optimization method based on microclimate simulation and resident preference

Abstract

The invention discloses a rural residence thermal environment optimization method based on microclimate simulation and resident preference, and the method mainly comprises the steps: building a three-dimensional microclimate model of a target site through thermal environment simulation software, setting different thermal environment design variables, and obtaining thermal environment performance indexes under a plurality of design scenes through simulation; determining key attributes and level values influencing the quality of the thermal environment reconstruction scheme, generating a selection set, and establishing a structured questionnaire; carrying out three-dimensional visualization and interactive design based on the obtained structured questionnaire, and obtaining interviewee questionnaire data; and based on the obtained interviewee questionnaire data, performing coupling analysis on resident preferences and thermal environment performance indexes obtained by microclimate simulation, and identifying an optimal design scene scheme considering the resident preferences and excellent thermal performance. According to the method, the limitation of a traditional method in the aspects of perception data expression, design parameter simulation, multi-target conflict coordination and the like is broken through, and a new path is provided for adaptive planning and design.

Description

Technical Field

[0001] This invention relates to the field of urban and rural human settlement environment design and evaluation technology, specifically to a method for optimizing the thermal environment of rural residences based on microclimate simulation and residents' preferences. Background Technology

[0002] In the field of urban and rural human settlement planning and architectural design, with the intensification of climate change and the frequent occurrence of high-temperature heat waves, traditional experience-based passive climate adaptation design methods face numerous challenges. Currently, improving the thermal comfort of the living environment has become a key issue that urgently needs to be addressed in the process of urban and rural renewal and rural revitalization. Especially in traditional settlements with regional cultural characteristics, how to regulate the thermal environment performance through natural solutions while maintaining the architectural style and residents' cultural identity has become a dual objective in design decisions.

[0003] In existing technologies, thermal environment simulation software (such as ENVI-met, CFD, etc.) is often used to evaluate the thermal performance of parameters such as spatial morphology, vegetation cover, and material properties, and the simulation results are used to guide design optimization. However, such methods often lack a systematic consideration of residents' preferences, cultural acceptance, and economic feasibility, resulting in a mismatch between "cold and heat adaptation" and "cultural acceptance" in the actual implementation of the design scheme, making it difficult to achieve the goal of truly people-centered environmental improvement.

[0004] On the other hand, while Discrete Choice Experiments (DCEs), as a quantitative method for preference identification, have been widely applied in fields such as transportation planning, health policy, and environmental economics, their application in urban and rural planning and building thermal environment design remains largely unexplored. Firstly, spatial environmental attributes are mostly physical morphological characteristics, making it difficult to construct experimental designs using standardized attribute-level approaches. Secondly, traditional DCE methods have not been integrated with microclimate simulation results, failing to construct an integrated assessment path encompassing "performance-preference-culture." Furthermore, existing research largely focuses on single-factor interventions, lacking multivariate collaborative optimization and system feedback analysis mechanisms, resulting in research findings that are difficult to provide actionable systemic recommendations for practical design.

[0005] Therefore, the existing technology has the following prominent problems: 1) The thermal environment design is out of sync with residents’ preferences: thermal simulation optimization does not fully consider residents’ acceptance in terms of culture, economy and maintenance, and it is difficult to balance “climate adaptability” and “cultural continuity”.

[0006] 2) Lack of integrated evaluation system: Existing research has failed to effectively integrate quantitative preference modeling (such as DCE) with microclimate simulation models, and lacks decision-making tools that bridge performance indicators with behavioral choices.

[0007] 3) Univariate intervention lacks a systematic feedback mechanism: Thermal environment design often focuses on adjusting a single factor, lacking multivariate coupling analysis and interaction effect identification, resulting in a one-sided optimization strategy.

[0008] In summary, existing urban and rural planning and architectural design methods still have significant shortcomings in terms of thermal adaptability, cultural acceptability, and systematic identification of residents' preferences. There is an urgent need for an integrated design optimization method that combines microclimate simulation, visual selection experiments, and statistical modeling to achieve a more empirically grounded and locally adapted path for environmental improvement. Summary of the Invention

[0009] This invention provides a method for optimizing the thermal environment of rural residences based on microclimate simulation and residents' preferences. It addresses the current lack of a systematic approach that integrates residents' preferences, cultural acceptance, and microclimate performance assessment in urban and rural planning and building thermal environment design. The invention proposes a thermal environment optimization decision-making method that integrates microclimate simulation, three-dimensional visualization, and discrete choice experiment (DCE) to achieve spatial design support and strategy recommendation under multi-objective guidance.

[0010] According to the first aspect, one embodiment provides a method for optimizing the thermal environment of rural residences based on microclimate simulation and residents' preferences, the method comprising: A three-dimensional microclimate model of the target site was constructed using thermal environment simulation software. Different thermal environment design variables were set, and thermal environment performance indicators under multiple design scenarios were obtained through simulation. Based on the discrete choice test method, the key attributes and level values ​​that affect the merits of thermal environment modification schemes are determined. Based on the determined key attributes and level values, a selection set is generated, and a structured questionnaire is established based on the obtained selection set. Based on the obtained structured questionnaire, 3D visualization and interactive design were performed, and questionnaire data from respondents were obtained; Based on the obtained questionnaire data from respondents, the intensity of residents' preferences for different design scenarios was quantified through preference modeling. By coupling the residents' preferences with the thermal environment performance indicators obtained from microclimate simulation, the optimal design scenario scheme that takes into account both residents' preferences and excellent thermal performance was identified.

[0011] Furthermore, a three-dimensional microclimate model of the target site was constructed using thermal environment simulation software. Different thermal environment design variables were set, and thermal environment performance indicators under multiple design scenarios were obtained through simulation, specifically including: The basic data for modeling the research plots are obtained through plot surveys, including typical vegetation parameters and building parameters; orthophotos of the research plots are obtained through drone aerial photography or information maps of the plots are obtained through surveying and mapping. Meteorological data, including air temperature, relative humidity, and wind speed, were obtained from the field measurements of the research plots. The meteorological data measured in the field were used to evaluate the accuracy of the model. A three-dimensional microclimate model of the target site was constructed using ENVI-met software and simulation was performed, including: site model construction, vegetation model construction, building material parameter setting, input of meteorological boundary conditions, and simulation operation; the model accuracy was evaluated based on the simulation results. The simulated output data includes air temperature, relative humidity, wind speed, and mean radiant temperature. These data are then imported into RayMan software for secondary calibration and calculation. In addition, thermal comfort indices, including physiological equivalent temperature and general thermal climate index, are calculated using standard human body parameters.

[0012] Furthermore, a three-dimensional microclimate model of the target site was constructed using thermal environment simulation software. Different thermal environment design variables were set, and thermal environment performance indicators under multiple design scenarios were obtained through simulation, specifically including: Based on the obtained thermal environment performance indicators, a comprehensive evaluation index, the composite outdoor thermal comfort index, was constructed using principal component analysis. Using the composite outdoor thermal comfort index as an input variable, the K-means clustering algorithm was used to divide all design scenarios. The clustering analysis classified the design scenarios into three categories, representing different thermal environment qualities: comfortable, transitional, and uncomfortable. The clustering results were verified by the silhouette coefficient and the Davidson-Bolding index.

[0013] Furthermore, the key attributes and level values ​​affecting the quality of thermal environment modification schemes were determined, specifically including: The key attribute identification process is divided into the following three stages: First, based on the research content, we reviewed the existing spatial preference research based on discrete choice experiments and found that: rural housing choice modeling needs to include both physical spatial attributes and perceived or ecological attributes. Secondly, by conducting in-depth interviews with residents, we can understand the rural housing environment factors that residents value highly, including those that directly affect comfort, maintenance costs, and aesthetic features. Finally, combining field research and policy literature analysis, the attribute set was further refined into relevant variables that fit current rural development, ecological concern and vernacular construction practices, and the key attributes were finally determined. Reasonable level values ​​were set for each attribute. These level values ​​were derived from field research, pre-survey, and literature review, taking into account both theoretical logic and practical situation. At the same time, the number of levels was controlled to match the age structure and cognitive level of rural residents, so as not to increase the complexity of the experimental design and the cognitive burden of the respondents.

[0014] Furthermore, the key attributes include: Building structure: including roof, facade, and floor materials; Spatial greening: including vegetation types, coverage, width / height, and configuration methods; Thermal performance: including the range of physiologically equivalent temperature or general thermal climate index values ​​derived from simulations; Usage and maintenance: including maintenance frequency and cost information; Perceive cultural characteristics: including the locality of materials and the consistency of traditional style.

[0015] Furthermore, a selection set is generated based on the identified key attributes and level values, and a structured questionnaire is built based on the obtained selection set, specifically including: Selection set generation: Using SPSS or Ngene software, based on defined attributes and corresponding level values, multiple statistically efficient and independent orthogonal selection sets were generated using orthogonal experimental design. Each selection set included three options: Option A, Option B, and "Do not select / Keep the status quo". Including the "Do not select / Keep the status quo" option effectively avoids preference overestimation bias caused by forced selection. A repeatability test was also included, where the first scenario in each selection set was presented again in random order at the end of the corresponding selection set. By comparing whether the two selections were consistent, invalid questionnaires with arbitrary or unstable answers were identified and eliminated, thereby further improving the quality of the data. Questionnaire Design: Based on the obtained selection set, a structured questionnaire was designed, consisting of the following three logically progressive parts: The first part is thermal comfort perception and living environment evaluation: Based on the simplified scale of ASHRAE thermal comfort standard, multiple sets of questions were used to understand the respondents’ thermal feelings about their current living environment, overall satisfaction, main environmental factors that caused dissatisfaction, and main cooling methods. The second part is the core component: using 3D scene diagrams to clearly explain the definitions and intuitive effects of various attributes and their different levels to respondents, in order to overcome the comprehension barriers of pure text descriptions and enhance the realism and immersion of the scenario; each respondent will complete 12 carefully designed sets of choices in turn; in each set of choices, respondents need to choose between two transformation plans with different attribute combinations and corresponding costs and an option of "not choosing / keeping the status quo", thereby revealing their implicit preference structure through multi-attribute trade-offs; The third part is the collection of socioeconomic characteristics information: After completing all the selection tasks, the questionnaire asks the respondents about their personal and family socioeconomic characteristics in a relatively natural way, including age, gender, education level, annual family income, length of residence, and current housing type. The corresponding variables will serve as key explanatory variables for subsequent econometric model analysis of preference heterogeneity, in order to gain a deeper understanding of the systematic differences in the renewal choices of groups with different economic capabilities, cultural backgrounds, and life experiences.

[0016] Furthermore, a selection set is generated based on the identified key attributes and level values, and a structured questionnaire is built based on the obtained selection set, specifically including: Questionnaire survey implementation design: The questionnaire survey dates should cover both weekdays and weekends to improve the coverage and representativeness of the sample; The daily surveys are scheduled to be conducted during multiple time periods when residents are more willing to be interviewed, reaching out to and covering family members with different daily routines, thus reducing selection bias caused by a single interview time. The survey employed one-on-one face-to-face interviews to ensure thorough communication and accurate understanding. Before the data collection work officially started, all investigators participating in the field survey received unified and standardized special training. All collected questionnaire data were thoroughly anonymized after entry and stored in encrypted dedicated storage devices, with access strictly limited to core research team members. Before each interview, the investigators will clearly and completely explain to potential participants the research objectives, main content, expected time, and their rights, and make it clear that participation is entirely voluntary and participants can withdraw at any time without reason. In the specific sampling and interview implementation stage, in each pre-selected settlement, a combination of systematic sampling or random sampling is used to determine the interviewees; based on the sampling results, 3 to 5 households are visited to reduce the clustering effect caused by excessive concentration of samples and improve the spatial distribution balance of the samples; finally, invalid questionnaires are removed as needed. Calculate the minimum sample size for discrete selection using Orme's "rule of thumb" to ensure that the sample size meets the basic requirements for the sample size of discrete selection experimental designs and to guarantee the statistical power of subsequent analyses.

[0017] Furthermore, based on the obtained structured questionnaire, 3D visualization and interactive design were performed, and respondent questionnaire data was obtained, specifically including: 3D visualization: Through sample plot survey and mapping, residential models are generated in SketchUp software and rendered using rendering software; each set of selected tasks is presented as a carefully rendered 3D scene map, which systematically changes key residential attributes and clearly marks the corresponding assumed costs. Interactive design: Import structured questionnaires into a web-based questionnaire system. Respondents can switch perspectives to view the solutions, rotate scene models, adjust thermal comfort level color gradation, perform interactive operations, and select preferences. Data feedback: Automatically stores respondents' selection behavior and their socioeconomic characteristics.

[0018] Furthermore, based on the obtained questionnaire data from respondents, preference modeling was used to quantify the intensity of residents' preferences for different design scenarios. By coupling residents' preferences with thermal environmental performance indicators obtained from microclimate simulation, the optimal design scenario scheme that balances residents' preferences with excellent thermal performance was identified, specifically including: The collected questionnaire data were preprocessed and coded. Each respondent's multiple selection data were matched with their unique socioeconomic characteristic data to construct a "long format" panel dataset for model estimation. Discrete choice model construction and estimation: Three types of discrete choice models were constructed using Stata software to form a gradient optimization model system to accurately capture residents' preferences: Model 1 is a conditional logit model without individual economic characteristic variables; Model 2 is a mixed logit model with individual economic characteristic variables; Model 3 is a panel mixed logit model that controls for repeated observation effects at the individual level. Goodness-of-fit test: A goodness-of-fit test is conducted on the above three types of discrete choice models, with the log-likelihood value as the core indicator. At the same time, the model is ensured to pass the chi-square test to verify the overall robustness and data fit of the model. Based on the optimal model that passes the test, namely the panel mixed logit model, resident preference analysis is carried out. Marginal willingness to pay analysis: To analyze the marginal value of different living environment attributes, a point estimate of the value of a unit change in a non-price choice attribute is calculated using a formula, namely the marginal willingness to pay. This measure is the ratio between the coefficient of a certain attribute in an individual's utility function and the coefficient of the price attribute.

[0019] in, Represented as attributes Marginal willingness to pay For attributes The corresponding utility coefficient, The utility coefficient of the price attribute; by ranking the marginal willingness to pay for different living environment attributes, the residents' preference for choosing living environment attributes is identified; Coupling Analysis and Results Output: Based on the estimation results of the panel mixed Logit model, the utility score for each design scenario is calculated. The utility score consists of the deterministic utility components in the model, and the specific calculation formula is as follows:

[0020] Among them, V njt x represents the deterministic utility that individual n obtains from option j in task t; k,jt β represents the specific level of the k-th attribute in scheme j; kThese are the fixed coefficients or average random coefficients of the k-th attribute estimated by the panel mixture Logit model; The utility scores of different design scenarios are compared and correlated with the composite outdoor thermal comfort index obtained through microclimate simulation and principal component analysis to identify the optimal design scenario that balances residents' preferences and excellent thermal performance.

[0021] According to the second aspect, one embodiment provides a rural residential thermal environment optimization system based on microclimate simulation and resident preferences, the system comprising: The microclimate simulation module is used to construct a three-dimensional microclimate model of the target site using thermal environment simulation software, set different thermal environment design variables, and obtain thermal environment performance indicators under multiple design scenarios through simulation. The questionnaire design module is used to determine the key attributes and level values ​​that affect the merits of thermal environment modification schemes based on the discrete choice experiment method, generate a selection set based on the determined key attributes and level values, and build a structured questionnaire based on the obtained selection set. The questionnaire data acquisition module is used to perform 3D visualization and interactive design based on the obtained structured questionnaires, and to acquire the questionnaire data of the respondents. The statistical analysis module is used to quantify the intensity of residents' preferences for different design scenarios based on the obtained questionnaire data, through preference modeling, and to identify the optimal design scenario scheme that takes into account both residents' preferences and excellent thermal performance by coupling the residents' preferences with the thermal environment performance indicators obtained from microclimate simulation.

[0022] This invention provides a method for optimizing the thermal environment of rural residences based on microclimate simulation and residents' preferences, which has the following beneficial effects: 1. Achieve deep coupling between thermal environment simulation and subjective preference assessment Existing thermal environment assessment technologies mostly rely on two-dimensional layers generated by simulation software such as ENVI-met, which are difficult to directly translate into cognitive input that residents can perceive, resulting in a gap between "simulation-design-use". This invention is the first to encode microclimate simulation results as structured factors (UTCI, PET level) and embed them into a 3D visual scene, linking them with a discrete choice experiment (DCE) system. This allows respondents to make choices based on scene understanding rather than technical charts, greatly improving the authenticity of preference assessment and scene perceptibility.

[0023] 2. Quantitative preference modeling based on experimental design outperforms subjective rating methods. Unlike existing extensive methods that rely on Likert scales or satisfaction surveys, this invention uses orthogonal experimental design + hybrid Logit model (Panel MXL) to statistically extract the marginal and interaction effects of various thermal environment attributes (such as green coverage, ground material, roof structure, etc.) on residents' choice intentions, thereby achieving quantitative modeling and personalized recommendations for urban and rural housing thermal environment attributes.

[0024] 3. The theoretical verification of the "minimum effective green volume" for thermal environment design is proposed. Based on simulation and DCE experimental results, this system identifies a nonlinear relationship between green coverage and thermal comfort improvement: thermal comfort improves significantly (PET decreases by 1.2°C) when the coverage increases from 60% to 70%, but the marginal benefit diminishes after exceeding 70%, suggesting that design should focus on the optimal green coverage range of "65–70%". These results provide a quantitative classification basis for urban and rural green space design.

[0025] 4. Develop a scalable and systematic urban and rural thermal adaptation design support tool. This invention provides a standardized process: [1] Thermal environment scenario construction → [2] Microclimate simulation → [3] Comfort level transcoding → [4] 3D scene generation → [5] Preference experiment design → [6] Preference estimation → [7] Personalized design optimization suggestions. It can be used not only for the transformation of traditional settlement thermal environment, but also widely extended to new rural construction, urban old community renovation, and optimization of living environment for heat-sensitive vulnerable groups. It has good versatility, scalability and module integration capabilities. Attached Figure Description

[0026] Figure 1 A flowchart illustrating a method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences, as provided in one embodiment of the present invention; Figure 2 Example diagrams of different design scenario schemes in a method for optimizing the thermal environment of rural housing based on microclimate simulation and residents' preferences, provided as an embodiment of the present invention; Figure 3 An example selection set diagram is provided for a method for optimizing the thermal environment of rural residences based on microclimate simulation and residents' preferences, as an embodiment of the present invention. Detailed Implementation

[0027] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of the invention. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to the present invention are not shown or described in the specification. This is to avoid obscuring the core parts of the invention with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0028] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.

[0029] The first embodiment of this invention provides a method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences. This method integrates microclimate simulation, three-dimensional visualization, and Discrete Choice Experiment (DCE), and is applicable to adaptive spatial transformation design of traditional settlements, urban and rural residences, and other open spaces under high-temperature conditions. The following is a combination of... Figure 1 Please provide a detailed explanation.

[0030] like Figure 1 As shown, in step S100, a three-dimensional microclimate model of the target site is constructed using thermal environment simulation software, different thermal environment design variables are set, and thermal environment performance indicators under multiple design scenarios are obtained through simulation.

[0031] Microclimate Simulation Module: This module uses ENVI-met or other CFD simulation software to construct a 3D microclimate model of the target site. It sets thermal environment design variables such as different building materials (e.g., roof, facade, ground), vegetation cover, and layout. Through simulation, it obtains key thermal comfort indices (e.g., physiological equivalent temperature PET, Universal Thermal Climate Index UTCI, air temperature, humidity, wind speed, etc.) under multiple scenario combinations. This module mainly consists of the following steps: 1.1 Sample plot selection and meteorological data acquisition 1.1.1 Sample plot survey Based on the research content, a detailed field survey was conducted for subsequent Envi-met model construction. This mainly included typical vegetation parameters (tree species, vegetation cover, tree height, crown width, leaf area index, etc.), building parameters (number of buildings, density, size, location, materials, height, etc.), and other spatial elements in the study plots. To facilitate Envi-met model construction, orthophotos of the study plots were obtained through drone aerial photography or information maps of the plots were acquired through surveying. This detailed visual and spatial data is crucial for a deeper understanding of the spatial structure of rural residential environments and for simulating microclimate effects.

[0032] 1.1.2 Meteorological Data Acquisition Field measurements are one of the main methods in microclimate research. Data from field measurements are primarily used to verify the accuracy of the Envi-met model. In related studies, simulated and measured values ​​are generally compared using microclimate factors such as air temperature, relative humidity, and wind speed. Therefore, this method mainly collects air temperature, relative humidity, and wind speed to verify the applicability and accuracy of the Envi-met simulation in the sample plot. According to the "Design Standard for Thermal Environment of Urban Residential Areas JGJ286-2013," a typical summer meteorological day refers to the day within the hottest month of a typical meteorological year where the daily average temperature, diurnal temperature range, humidity, and solar radiation are closest to the average values ​​of that month. Simulation analysis based on the meteorological parameters of this day and the proposal of optimization schemes are commonly used methods in current thermal comfort research. A typical summer meteorological day can be selected as the date for the Envi-met simulation by continuously observing the weather and referring to relevant meteorological station data.

[0033] Taking the Kestrel 5500 handheld weather station as an example, field measurements were conducted. The Kestrel 5500 handheld weather station is capable of measuring environmental parameters such as air temperature, relative humidity, and wind speed with high precision. Before starting the measurement work, the instrument needs to be calibrated to ensure accurate measurement results. Measurements were conducted under clear, windless weather conditions. To avoid human interference, all measuring instruments were supported by tripods, positioned 1.5m above the ground and 1m away from people. The sensors were also protected by protective covers to avoid the influence of solar radiation. When selecting the location, the accuracy of numerical simulations under different environments needed to be tested, so the measuring instruments needed to be placed in different spatial environments. The measuring instrument took measurements every hour, with each measurement lasting one minute. The average value within one minute was taken as the measured value to ensure the accuracy and stability of the actual measurement results.

[0034] 1.2 Model Construction and Simulation 1.2.1 Software Introduction ENVI-met is a computational fluid dynamics (CFD)-based microclimate numerical simulation software suitable for dynamic simulation and analysis of outdoor environments. Developed in 1998 by German scholars Daniela Bruse and Michal Bruse, its model integrates theories from multiple disciplines, including fluid mechanics, thermodynamics, soil science, and plant physiology, enabling a relatively realistic simulation of the interactions between the land surface, buildings, vegetation, and the atmosphere. ENVI-met achieves a refined evaluation of microclimate characteristics through coupled calculations of airflow and heat transfer within space, featuring relatively simple operation, high simulation accuracy, and strong reliability of results. Compared to other CFD-based simulation tools such as Phoenics and Fluent, ENVI-met focuses more on the study of local microclimates around cities and buildings, and is particularly suitable for simulating areas with high vegetation cover. The software includes a built-in vegetation module based on plant physiological mechanisms, accurately reflecting the role of plants in regulating heat and moisture in the microenvironment, giving it a significant advantage in analyzing rural residential environments with rich vegetation structures, such as the traditional Sichuan-style forest-enclosed settlements.

[0035] The study utilizes ENVI-met V 5.6.1 to simulate and analyze the outdoor microclimate of the woodland settlement. This software supports simulations of the impacts of different design elements (such as greening configuration, building form, and surface materials) on climatic factors like air temperature, relative humidity, and solar radiation at spatial scales of 0.5 meters and above and time scales of 1–5 seconds. It can also output climate data for any point within the area. Therefore, it can assist researchers and designers in the fields of architecture, planning, and landscape to assess the potential impacts of design schemes on human thermal comfort, pollutant dispersion, and other aspects, providing a scientific basis for sustainable urban design and environmental optimization.

[0036] 1.2.2 Model Parameter Settings The first step is site model construction. Using data obtained from previous surveys of the study plots via drones or satellite maps, a rapid model was constructed in Envi-met's Monde module to determine basic attributes such as latitude, longitude, and time zone. Subsequently, fine-tuning was performed in Envi-met's Space module. Considering the detail required for rural living environments, a 1-meter resolution spatial discretization model was used. The model's Z-axis height was set to twice the height of the tallest building on the site to ensure the integrity of the airflow simulation. A 10-layer nested mesh was used for the model boundary to reduce the impact of boundary effects and improve the accuracy of the simulation.

[0037] The second step involves constructing the vegetation model in the Alberto module and setting building material parameters in DB Manager. Rural living environments contain diverse plant configurations. However, some plant configurations, such as bamboo, have fragmented leaf shapes and tend to bend naturally after maturity, making them difficult to accurately represent in numerical simulations. Considering the complexity of morphological simulation, and to improve model controllability and simulation efficiency, trees with relatively regular shapes and easier parameterization can be used as substitutes to simplify the modeling process while preserving their ecological regulation functions. Based on data from previous sample plot surveys, parameters such as tree height, tree species, and crown width are set in detail to more clearly identify the contribution of plants to microclimate regulation. In the DB Manager module, building material parameters are set according to local conditions based on the living environment elements determined by discrete selection experiments.

[0038] The third step involves inputting meteorological boundary conditions into the `guide` module and running the simulation in the `Core` module. A typical summer day is selected as the simulation day, characterized by high temperature and high humidity. Hourly meteorological data, including air temperature, relative humidity, wind speed, and wind direction, is inserted into the `guide` simulation. This data is collected from relevant meteorological stations. For more detailed meteorological boundary conditions, a full simulation can be used to input them. Alternatively, a simple simulation can be used to simulate the daily climate process. Following relevant methods, a 5-7 hour warm-up period is set for each simulation to improve model stability and result reliability. Subsequently, the simulation is run in the `core` module using the meteorological boundary file generated by the `guide` module, and the corresponding results are output.

[0039] 1.2.3 Model Accuracy Verification Based on the corresponding results of the Core simulation, open the Leonardo visualization module. This module can be used to output smooth graphs of temperature, wind field, comfort, etc., and to extract the required data at any point and any time. Export the simulation data of the corresponding field measurement points for accuracy verification. In the accuracy verification of the Envi-met model, comparing the simulated and measured values ​​of temperature and relative humidity is a common method. These methods have been widely used, and their feasibility has been verified in various studies. Commonly used evaluation indicators for model accuracy include mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE). Both MAE and RMSE can reflect the average error of the model. RMSE is more commonly used, but it cannot reflect the relative magnitude of the average error. RMSE reflects the magnitude of the error between the simulated and actual values. The closer the value is to 0, the smaller the error between the simulated and measured values, and the more accurate the model. The expression is as follows (S i For measured values, M i (Simulated values)

[0040] To evaluate the closeness between simulated and measured data, this study, in addition to using RMSE as an evaluation index, also introduced the Willmott consistency index d, where the value of d ranges from 0 to 1. The closer the value of d is to 1, the better the fit between the measured and simulated data. Its expression is as follows (Si is the measured value, Mi is the simulated value):

[0041] Tsoka et al. mentioned that the acceptable RMSE for temperature should be less than 4.30℃, and the RMSE for relative humidity should be less than 10.2%. The smaller the RMSE value and the greater the d value (≥0.7), the stronger the correlation between the two sets of data. If the errors and consistency between the simulated and measured values ​​are within acceptable ranges, the simulation of microclimate and thermal comfort of the sample plot using Envi-met can be considered to have good accuracy. If they are not within acceptable ranges, further modifications to the model parameters are needed in section 1.2.2, Parameter Settings.

[0042] 1.2.4 Thermal comfort level export Using the BIO-met module built into Envi-met, thermal comfort indices such as physiological equivalent temperature and general thermal climate index are calculated based on discrete choice experimental questionnaires and standard human body parameters. These indices can be calculated numerically or by calling official algorithms (software implementation) based on human thermal balance / thermal regulation models. To ensure accuracy, potentially biased data on air temperature, relative humidity, wind speed, and mean radiant temperature directly output by ENVI-met are imported into RayMan software for secondary correction and calculation. The thermal comfort index results corrected by RayMan post-processing have reliable accuracy, providing a robust data foundation for subsequent analysis.

[0043] 1.3 Microclimate Indicator Output and Scene Generation To systematically quantify and compare different design scenarios (i.e., different combinations of design variables including architectural, vegetation, facade, and surface paving configuration parameters, such as...) Figure 2The overall thermal environment performance (as shown) was studied using standardized data extraction and aggregation methods. Multiple spatially representative sampling points were symmetrically and evenly distributed within the simulated area. These points covered typical microclimate spatial types, including building perimeters, courtyard centers, shady areas, and open zones, allowing for better capture of thermal environment differentiation within the residential environment caused by differences in underlying surface and spatial morphology. The arithmetic mean of the values ​​from all sampling points at each time point was calculated to determine the spatial average of each microclimate factor for each scenario over the entire time period. This effectively integrates the spatial heterogeneity within the forest-enclosed settlement, providing a unified and comparable scalar value to summarize the overall microclimate performance under this simulated scenario.

[0044] The spatial average effects of each microclimate factor across all scenarios over the entire time period are as follows: 1) Within a woodland enclosure, there are shaded (cool) and sun-exposed (hot) areas. Looking at only a single point will yield extremely limited results. The spatial average represents the average thermal state of the design across the entire physical space, answering the question, "As a whole, how much cooler is this design than that one?"

[0045] 2) Eliminating the interference of spatial heterogeneity: By averaging different locations such as "around the building, in the center of the courtyard, and under the shade of vegetation", you smooth out outliers. In this way, when comparing different schemes, the results will not be affected by the accidental shading of a specific location (such as a scheme just adding an extra tree at a certain point), but will reflect the influence of macro design strategies (such as overall layout and greening rate).

[0046] The simulation focuses on the objective physical impacts of changes in these physical living environment elements on the microclimate. Since the "willingness to pay" attribute represents residents' subjective value trade-offs and economic constraints, its level differences do not directly change physical parameters such as material albedo, heat capacity, or vegetation transpiration efficiency. Therefore, repetitive scenarios with identical material configurations but different levels of willingness to pay will not be included in the simulation. This ensures that the simulation work directly serves to reveal the independent and interactive mechanisms of spatial morphology, material properties, and vegetation cover on the thermal environment, providing a reliable objective environmental performance data foundation for subsequent coupled analysis with residents' subjective preferences and utility.

[0047] 1.4 Microclimate Index Analysis Methods 1.4.1 Multivariate ANOVA and Tukey HSD Post-hoc Test To systematically analyze the complex and multidimensional relationships between key attributes and thermal environment indicators such as air temperature, relative humidity, physiological equivalent temperature, and general thermal climate index, multivariate analysis of variance (MANOVA) was first employed as the primary analytical tool. Its core principle lies in the fact that when there is an intrinsic correlation between the dependent variables (here, four thermal environment indicators), MANOVA can test the overall joint effect of multiple independent variables (morphological attributes) on these related dependent variable sets. Compared to performing multiple univariate ANOVAs on each indicator individually, this method effectively controls the interference caused by the correlation between dependent variables and maintains the overall Type I error (i.e., the probability of incorrectly rejecting the null hypothesis) at the level of a single test, avoiding error inflation caused by multiple tests, thus ensuring the robustness of the overall conclusions. After confirming the existence of a significant joint effect, a further in-depth analysis was conducted using one-way ANOVA. This step aims to analyze the overall effect discovered by MANOVA, separating and quantifying the independent influence of each individual morphological attribute on each specific thermal environment indicator. Finally, for attributes that achieved significance in ANOVA, the study employed the Tukey HSD test for post-hoc multiple comparisons. The aim is to accurately identify the specific differences between different levels within the same attribute while controlling the overall error rate.

[0048] 1.4.2 Construction and Cluster Analysis of Composite Thermal Comfort Indicators To comprehensively evaluate the thermal environment performance under different design scenarios and overcome the complexity of comparing multiple indicators, principal component analysis (PCA) was used to construct a comprehensive evaluation index—the composite outdoor thermal comfort index. This index is constructed based on core microclimate indicators output from simulations, such as air temperature, relative humidity, physiological equivalent temperature, and general thermal climate indices. PCA analysis was performed in SPSS software to derive the first principal component formula. The weight distribution structure was then examined to see if it conformed to the expectations of classical thermal comfort theory, statistically verifying the theoretical rationality of the composite index. The specific calculation method is as follows: PCA analysis was used to extract the first principal component with an eigenvalue greater than 1 as the composite index. The mathematical expression of the composite outdoor thermal comfort index is a linear combination of various standardized microclimate indicators:

[0049] F1 represents the composite outdoor thermal comfort index, and Z... i The i-th thermal environment index (T) after standardization a ,RH, PET, UTCI), W i This is the component score coefficient output by SPSS. This index explains XX% (typically >80%) of the total variance of the original data and can represent the overall trend of thermal environment changes.

[0050] Subsequently, using this composite index value as an input variable, the K-means clustering algorithm was employed to classify all simulated scenarios. Cluster analysis clearly categorized the scenarios into three types, representing different thermal environment qualities: comfortable, transitional, and uncomfortable. To verify the quality and robustness of the clustering results, the silhouette coefficient and the Davidson-Bolding index were calculated. A silhouette coefficient closer to 1 indicates better clustering, and a Davidson-Bolding index closer to 0 indicates better clustering. These two indices jointly confirm whether the clustering analysis produced classification results with clear structure, tight intra-group distribution, and high inter-group separation, and whether it possesses high internal consistency and good interpretability, laying the foundation for further in-depth analysis of the characteristics of different scenario categories.

[0051] like Figure 1 As shown, in step S200, based on the discrete selection test method, the key attributes and level values ​​that affect the merits of the thermal environment modification scheme are determined, a selection set is generated based on the determined key attributes and level values, and a structured questionnaire is established based on the obtained selection set.

[0052] 2.1 Determining Attributes and Levels The key step in choosing the experimental method is to identify and evaluate the attributes and their level of importance of an item. When selecting relevant attributes, it is necessary to consider relevant policies and the degree of importance consumers place on those attributes. Therefore, the attribute identification process is divided into three stages. First, based on the research content, a review of existing spatial preference studies based on Discrete Choice Experiments (DCE) reveals that rural housing choice modeling typically includes both physical spatial attributes (such as materials and layout) and perceived or ecological attributes (such as greenery and comfort). Second, through in-depth interviews with residents, it is understood that residents highly value rural housing environmental factors, such as vegetation, roofing materials, building facades, and ground paving, which directly affect comfort, maintenance costs, and aesthetic features. Finally, combining field research and policy literature analysis, the attribute set is further refined into relevant variables that align with current rural development, ecological concerns, and vernacular construction practices, ultimately determining the key attributes. The attributes include, but are not limited to: building structure (roof, facade, ground material); green space (vegetation species, coverage, width / height, configuration); thermal performance (PET or UTCI value range derived from simulation); use and maintenance (maintenance frequency, cost indication); perceived cultural characteristics (material locality, consistency with traditional style). The level values ​​must be set with a reasonable range and gradient to comprehensively encompass the respondents' preference thresholds, ensuring that their choice behavior is fully expressed within the designed experimental framework. The levels should be derived from detailed field research, pre-surveys, and existing literature evidence. When assigning levels, it is essential to strictly adhere to the constraints of the theoretical model to ensure data logical consistency, while also closely aligning with current social realities and industry development trends to avoid cognitive dissonance among respondents due to unrealistic level settings, thus guaranteeing the reliability and forward-looking nature of the experimental results. Simultaneously, the age structure and cognitive level of rural residents need to be considered; the number of levels should not be excessive to avoid increasing the complexity of the experimental design and the cognitive burden on the respondents.

[0053] 2.2 Selection Set Design To ensure scientific rigor and efficient survey implementation, a standardized process for discrete choice experiments was adopted for choice set design. First, based on defined attributes and levels, theoretically, multiple possible combinations exist. Choice sets can be generated using SPSS or Ngene software. Taking SPSS as an example, its orthogonal experimental design function can generate multiple statistically efficient and independent orthogonal choice sets based on the attributes and number of levels. Each choice set contains three options: Option A, Option B, and a "Do not choose / Maintain the status quo" option. Including the "Do not choose" option not only more realistically simulates the possibility of "abandoning purchase" in market decision-making but also effectively avoids the bias of overestimating preferences that may result from forced selection. To test the seriousness and internal consistency of respondents' responses, this study included a replication test in the survey design. Specifically, the first scenario in each choice set was presented again at the end of the set in a random order. By comparing the consistency between the two selections, invalid questionnaires with arbitrary or unstable responses can be identified and eliminated, thereby further improving the quality of the data.

[0054] 2.3 Questionnaire Design for Discrete Choice Experiments To obtain high-quality data on residents' preferences, a structured questionnaire was designed. The questionnaire strictly adhered to the methodological requirements of discrete choice experiments and its design fully considered the cognitive habits of rural respondents, aiming to reduce response bias and ensure the reliability and validity of the data. The questionnaire typically consists of three logically progressive parts: (1) Thermal Comfort Perception and Living Environment Evaluation. This section aims to establish the contextual basis of the study and collect residents' subjective evaluations of the existing environment. Considering the special characteristics of rural residents and to reduce the burden of responses and understanding thresholds for respondents, a simplified scale based on the ASHRAE thermal comfort standard was used in the thermal comfort questionnaire. The questionnaire used multiple sets of questions to understand respondents' thermal perception of their current living environment, overall satisfaction, and main environmental factors causing dissatisfaction, as well as their main cooling methods. This evaluation provides data on residents' adaptive behaviors for the study.

[0055] (2) Core Module of Discrete Choice Experiment. This is the core component of the questionnaire, designed to construct a near-realistic "hypothetical market." 3D scene diagrams are used to clearly explain the definitions and intuitive effects of various attributes and their different levels to respondents, overcoming the comprehension barriers of pure text descriptions and enhancing the realism and immersion of the scenario. Each respondent will complete 12 carefully designed choice sets sequentially, such as... Figure 3 As shown in the figure. In each group, respondents had to choose between two renovation options with different combinations of attributes and corresponding costs and a "no choice" option, thereby revealing their implicit preference structure through multi-attribute trade-offs.

[0056] (3) Collection of socioeconomic characteristics. After completing all the selection tasks, the questionnaire asked respondents about their personal and family socioeconomic characteristics in a relatively natural way, including age, gender, education level, annual family income, length of residence, and current housing type. These variables will serve as key explanatory variables for subsequent econometric model analysis of preference heterogeneity, and will be used to gain a deeper understanding of the systematic differences in renewal choices among groups with different economic capabilities, cultural backgrounds, and life experiences.

[0057] 2.4 Questionnaire Design for Discrete Choice Experiments To obtain high-quality preference data, the survey dates should cover both weekdays and weekends to improve sample coverage and representativeness. Daily surveys should be conducted during multiple time slots—morning, noon, and evening—when residents are most receptive to interviews, reaching out to family members with different daily routines to further reduce selection bias that may be introduced by limited interview times. The survey will employ one-on-one face-to-face interviews to ensure thorough communication and accurate understanding. Before the formal commencement of data collection, all investigators participating in the field survey will receive standardized and specialized training. The training will cover questionnaire structure and core concept interpretation, interview communication skills, and ethical considerations to ensure a high degree of consistency in the understanding and communication of the questionnaire content among all investigators, minimizing information bias caused by differences in questioning methods or interpretations. This study will strictly adhere to academic ethical standards throughout. Before each interview, investigators will clearly and completely explain the research objectives, main content, expected timeframe, and the rights of potential participants, explicitly stating that participation is entirely voluntary and that participants can withdraw at any time without giving a reason. Informed consent was obtained in writing. For participants with reading difficulties or difficulty writing, verbal consent was provided and the entire process was recorded. The questionnaire design did not collect any personally identifiable information (such as name, ID number, detailed address, etc.). All collected questionnaire data was thoroughly anonymized after entry and stored on encrypted dedicated storage devices. Access was strictly limited to core members of the research team to ensure that the privacy of respondents and data security were fully protected.

[0058] In the specific sampling and interview implementation phase, a combination of systematic sampling and random sampling was used to determine the respondents in each pre-selected settlement. Based on the sampling results, 3 to 5 households were visited to reduce the clustering effect caused by excessive sample concentration and improve the spatial balance of the sample. Finally, invalid questionnaires were removed as needed.

[0059] This study uses Orme's "rule of thumb" to calculate the minimum sample size for discrete choice. The total number of choice sets is t, the number of choices in each set is a, and the maximum number of levels corresponding to each environmental factor is b, which is the minimum acceptable sample size (N) for this study. It is essential to ensure that the sample size meets the basic requirements for discrete choice experimental designs to guarantee the statistical power of subsequent analyses. The calculation is as follows:

[0060] like Figure 1 As shown, in step S300, three-dimensional visualization and interactive design are performed based on the obtained structured questionnaire, and the questionnaire data of the respondents are obtained.

[0061] 3D Visualization and Interactive Questionnaire Module: To overcome the limitations of traditional textual descriptions in expressing the attributes of rural living environments, this study innovatively adopted a choice set supported by visual images. Through plot surveys and mapping, residential models were generated in SketchUp software and rendered using rendering software such as Escape. Each choice task was presented as a meticulously rendered 3D scene diagram, systematically showing changes in key residential attributes such as roofs, facades, ground materials, and vegetation cover, with corresponding assumed costs clearly marked. This visual presentation greatly enhanced the realism and immersion of the scenario, helping respondents, especially rural residents and the elderly who are not familiar with architectural terminology, to more intuitively understand the differences between different options, thereby reducing their cognitive burden and enabling them to make choices closer to their true desires.

[0062] Interaction format: Import the discrete choice experiment questionnaire into a web-based questionnaire system, such as Wenjuanxing platform. Respondents can switch perspectives to view the solutions, rotate the scene model, view thermal comfort level color gradation coverage, perform interactive operations, and reselect preferences. Data feedback: Automatically stores respondents' selection behaviors and their background information (such as age, occupation, cultural identity, etc.).

[0063] like Figure 1 As shown, in step S400, based on the obtained questionnaire data of the respondents, the intensity of residents' preferences for different design scenarios is quantified by preference modeling, and the optimal design scenario scheme that takes into account both residents' preferences and excellent thermal performance is identified by coupling the residents' preferences with the thermal environment performance indicators obtained by microclimate simulation.

[0064] In this embodiment, the preference modeling and statistical analysis module contains the following details: 4.1 Questionnaire Data Coding To ensure the subsequent econometric model can correctly identify and process various types of information, systematic preprocessing and variable coding of the collected questionnaire data are necessary. The introduced explanatory variables are typically divided into three categories: first, scheme attribute variables describing the characteristics of each option, which differ between different options; second, individual characteristic variables characterizing information such as the respondents' socioeconomic background, which also differ among different respondents; and finally, dependent variable coding. If the different levels of an attribute cannot be presented in an equal interval relationship, this variable is treated as a categorical variable; otherwise, it is treated as a continuous variable. In this study, all individual characteristic variables except age (which is a continuous variable) are categorical variables.

[0065] Categorical variables in the residential environment attributes are virtually coded, typically taking values ​​of 0 or 1. A value of 1 indicates compliance with the specified level, while a value of 0 indicates non-compliance. The first level of each attribute is used as a reference group to reflect the average effect difference of other categories relative to this reference group. In the modeling process, residents' choices to maximize the utility of their residential environment are used as the dependent variable, assigned a value of 1 when a resident selects an option and 0 otherwise—a binary variable. Finally, the multiple sets of choices for each respondent are matched with their unique socioeconomic characteristic data to construct a "long-format" panel dataset for model estimation.

[0066] 4.2 Preference Modeling After obtaining the questionnaire data, we need to explain the underlying mechanisms of residents' choices, such as which living environment factors significantly influence these choices. Is this influence positive or negative, and to what extent? By utilizing the selection mechanisms discovered in the research, we can help predict residents' decisions in different environments, thus providing an objective and scientific basis for improving the living environment.

[0067] Due to the specific nature of the selected data, it is impossible to achieve the desired result using a general regression model. The logit model, the most basic form of discrete choice model, can explain individual choice behavior, calculate the probability of each option being selected, and predict choice behavior. Therefore, this study uses Stata software to construct a conditional logit model without individual economic characteristics (Model 1), a mixed logit model with individual economic characteristics (Model 2), and a panel mixed logit model that controls for repeated observations at the individual level (Model 3).

[0068] The conditional logit model, also known as the McFadden choice model, was first proposed by American economist Daniel McFadden in 1974. This model is based on several strict assumptions, primarily including the following three characteristics: ① It assumes that all respondents have consistent preference parameters, meaning only the overall average preference can be estimated; ② It assumes that unobserved preference components (i.e., the random error term ε) are independently and identically distributed, with the same utility and variance, and all covariances are zero. This means that if there is a correlation between unobserved factors, the model cannot handle such situations; ③ It assumes that the options satisfy the "independence of irrelevant options," meaning that the options are independent and uncorrelated, and the probability of selection does not change with the increase or decrease of the number of options.

[0069] When constructing the model, Model 1 is the basic model, which only considers the influence of attributes and their levels in the residential environment on choice behavior. Since it does not include individual economic characteristic variables, it cannot prove the existence of heterogeneous preferences among individuals. A conditional logit model is used in Stata software via the clogit command. In this study, if resident i chooses option j, then the utility Uij obtained by choosing option j can be expressed as: U ij =β1*X1 ij +β2*X2 ij +β3*X3 ij +...+β n *Xn ij +ε Where ε is a constant; β is the parameter estimate of each residential environment attribute variable (i.e., the key attributes determined in 2.1 above); and X1-Xn are the various attributes set.

[0070] The mixed logit model, in addition to including residential environment attributes, also comprehensively considers the socioeconomic status of the residents of the forest-enclosed settlement. The mixed logit model is used in Stata software via the `mixlogit` command. In this study, if resident i chooses option j, then the utility U obtained by choosing option j is... ij It can be represented as: U ij =β1*X1 ij +β2*X2 ij +β3*X3 ij +β4*a+β5*b+...+β n *Xn ij +ε Where ε is a constant; β is the parameter estimate of each residential environment attribute variable; X1-Xn are the various attributes set; and az represents the socioeconomic characteristics of the residents.

[0071] Meanwhile, in order to distinguish the utility differences between different options, the Alternative-Specific Constant (ASC) is introduced. ASC is a dummy variable, defined as 1 when the alternative is chosen, and 0 otherwise. When the coefficient of ASC is positive, it means that residents are more willing to choose among the given options.

[0072] To analyze the heterogeneity of willingness-to-pay preferences for alternative attributes, an individual variable Z with ASC interaction was used to improve the explanatory power of the model. In this case, the utility function is expressed as follows:

[0073] Among them, U ij: Utility value, i: represents the i-th respondent; j: represents the j-th option; X ij The specific values ​​of each attribute in scheme j (e.g., price, distance, environmental quality, etc.). This represents the degree of preference (weight) of respondents for each attribute. For example, the coefficient for price is usually negative, indicating that the higher the price, the lower the efficiency. The main effect of alternative constant (ASC) is the inherent average basic utility of alternative j without considering individual characteristics. The coefficient of the interaction term represents how an individual's socioeconomic attribute Z influences their preference for option j. This represents the personal attributes of the i-th respondent (such as age, income, education level, etc.).

[0074] During parameter estimation, due to the heterogeneity of residents, their preferences for alternative attributes may also be heterogeneous. When heterogeneity actually exists, imposing the assumption of homogeneity in preferences and responses will lead to biases in parameter and probability estimates. Furthermore, this study selected an alternative C option in the experimental set, which may cause the IIA (Independence from Irrelevant Alternatives) assumption to fail. A mixed logit model is used for parameter estimation. The probability of respondent i choosing alternative j can be expressed as:

[0075] in, It is the selection probability, representing the likelihood that the i-th respondent will choose option j (the probability value is between 0 and 1). β is the deterministic utility component of scheme j; β is the parameter vector; It is the probability density function of β. These are the parameters that describe the distribution.

[0076] Compared to the mixture logit, the panel mixture logit not only considers heterogeneity among individuals but also explicitly considers the correlation between multiple choices made by the same individual, meaning that the random parameters of the same individual remain consistent across different choice scenarios. The panel mixture logit model is used in Stata software via the `cmmixlogit` command. The probability of a respondent choosing a particular option estimated by the panel mixture logit model can be expressed as the joint probability of the same respondent i across all T choice tasks:

[0077] Where j(i,t) represents the scheme actually chosen by individual i in task t. The attribute vector of the selected scheme for individual i in the t-th task; Sum all J alternative solutions for the t-th task.

[0078] Unlike conditional logit models, when constructing mixed logit and panel mixed logit models, the coefficients of all variables are first set to random coefficients following a normal distribution, as condition 1. Based on this setting, Halton random sampling is used, with the number of samplings R=1000, to obtain the model estimation results under condition 1. Based on the model results from the previous step, the mean and standard deviation of the coefficients are corrected. The mean of the coefficients reflects the overall preference trend of respondents, while the standard deviation reflects the dispersion of individual preferences. To more accurately capture the differences in preferences among different individuals, a hypothesis test is performed on the estimated standard deviation. If the test result P<0.05, the null hypothesis is rejected, and the coefficient of that variable is retained as a random coefficient; the remaining variables that fail the test are set as fixed coefficients. The model is then re-estimated to improve the model's fit and estimation accuracy. The option category variable insurance is automatically set as a fixed variable, and the current option is set as the reference option. The regression command used is the cmmlogit command, and then individual characteristic variables such as gender, age, occupation, and income are added to the model and run using the casevar command.

[0079] 4.3 Statistical Analysis 4.3.1 Goodness-of-fit test and result analysis Log-likelihood is a fundamental indicator for evaluating the goodness of fit of discrete choice models. Essentially, it is the natural logarithm of the probability (i.e., the likelihood function) of observing actual sample data given model parameters. Its magnitude reflects the model's ability to interpret the data. A higher log-likelihood value (i.e., a smaller negative value) indicates that the probability distribution specified by the model is closer to the true distribution of the sample data, and the model fit is better. To ensure the discrete choice model passes the chi-square test (Prob>chi² = 0.0000), it is assumed that the model is robust overall. Subsequently, the discrete choice model parameters output by Stata software are used for resident preference analysis. The significance and sign of the coefficient β are used to assess residents' preferences for different living environment factors, serving as a basis for subsequent policy formulation.

[0080] By comparing the magnitude of the log-likelihood value and the AIC / BIC information criterion (the smaller the value, the better the model), researchers can identify the "optimal model" with the strongest explanatory power and the most robustness for sample data among models of different complexities, such as basic MNL, mixed Logit, and panel mixed Logit.

[0081] 4.3.2 Marginal Willingness to Pay Analysis To analyze the marginal value of different living environment attributes, a point estimate of the value of a unit change in a non-price choice attribute is calculated using a formula, namely the marginal willingness to pay (MWTP), which is usually used to measure the ratio between the coefficient of a certain attribute in an individual's utility function and the coefficient of the price attribute.

[0082]

[0083] in, Represented as attributes Marginal willingness to pay For attributes The corresponding utility coefficient, The utility coefficient of the price attribute was used to collect residents' choice preferences through a discrete choice model.

[0084] By ranking the marginal willingness to pay for different residential environment attributes, residents' preferences for their preferred residential environment attributes can be identified.

[0085] 4.3.3 Coupled Analysis of Residents' Preferences and Microclimate Simulation in Discrete Choice Experiments To systematically evaluate the matching relationship between residents' declarative preferences and the objective thermal environmental performance under different design scenarios, this study constructs a coupled analysis framework. The core of this framework lies in comparing preference data derived from behavioral choice models with performance data derived from physical simulations on the same dimension.

[0086] First, to quantify the strength of residents' systematic preferences for each scenario, this study calculated the "utility score" for each design scenario based on the estimation results of a panel-mixed Logit model. This score is composed of the deterministic utility component in the model, and the specific calculation formula is as follows:

[0087] Among them, V njt x represents the deterministic utility score that individual n obtains from option j in task t; k,jt The specific level of the k-th attribute in scheme j is obtained through a questionnaire from respondents; β kThe fixed or average random coefficients of this attribute are estimated by a panel mixture Logit model and then iteratively derived using the simulated maximum likelihood estimation method. This is typically done automatically using software such as Stata (command cmmixlogit) or Nlogit. During the calculation, the willingness-to-pay (WTP) attribute is excluded to ensure that the utility score purely reflects the residents' preference structure for the allocation of physical space, without being influenced by their direct willingness to pay. This method removes the random error term ε. njt This allows for the extraction of stable preference measures that represent the general tendencies of the group, thereby achieving comparability across all scenarios on a unified preference scale.

[0088] Subsequently, these scenario-level utility scores are compared and correlated with the composite outdoor thermal comfort index obtained through microclimate simulation and principal component analysis. This is typically achieved by constructing a two-dimensional physical-psychological coupled evaluation model: specifically, a scatter plot is drawn with the composite thermal comfort index obtained from microclimate simulation as the horizontal axis and the scenario utility scores calculated by the discrete choice model as the vertical axis, and the Pearson correlation coefficient is calculated to quantify the degree of synergy between the two. Then, a four-quadrant analysis method is used to partition the scatter plots. Points falling within the "high-temperature zone" represent the specific design scenarios simulated in ENVI-met. These scenarios are the final selected optimal design schemes that combine physical cooling benefits with residents' psychological preferences. This allows the identification of the optimal design scenario scheme (the best design scenario scheme) that balances residents' preferences and excellent thermal performance.

[0089] Strategy recommendation output module: Based on model results, this module constructs an adaptive thermal environment optimization design decision support system, applicable to urban and rural housing, settlement renovation, and planning evaluation scenarios. It supports the following functions: 1) Design attribute priority ranking map: Based on the statistical results of discrete choice experiments, the most preferred living environment factors and their corresponding willingness to pay were identified. Simultaneously, analysis of variance was used to identify the living environment factors with the greatest impact on microclimate. This provides a priority reference for design attributes in policy and program development.

[0090] 2) Identification of optimal solutions for different groups (such as the elderly, farmers, and young residents); In the discrete choice experiment module, by interactively analyzing individual economic characteristics with ASC terms, the heterogeneous preferences of different socioeconomic groups can be identified, which can provide differentiated reference suggestions for the formulation of policies and solutions.

[0091] 3) Temperature threshold – suggested mapping curve for greening configuration; microclimate simulation can analyze the impact of different levels of residential environmental factors, such as vegetation cover, on microclimate factors. Vegetation cover, as a key driver of the residential environment, is not necessarily better the higher it is. Through Tukey HSD post-hoc testing, the impact of different levels on microclimate can be compared. Combined with the marginal cost residents are willing to pay in discrete choice experiments, an optimal greening configuration acceptable to residents can be determined.

[0092] 4) Recommendation of a balance between thermal comfort and cultural continuity. By correlating the utility values ​​of discrete choice experiments with indicators such as thermal comfort, the most popular living environment options for residents can be determined. This derived "high-performance-high-preference" solution provides empirical evidence for scenarios such as urban and rural housing, settlement redevelopment, and planning evaluation.

[0093] The main innovations of the rural residential thermal environment optimization method based on microclimate simulation and residents' preferences proposed in this invention are as follows: 1. Semantic visual translation method and thermal comfort level coding mechanism of microclimate simulation results This invention proposes a semantic visual mapping method based on multi-source thermal environment simulation results, including: (1) constructing a unified thermal comfort index extraction mechanism compatible with multiple model outputs such as UTCI, PET, and WBGT; (2) introducing a hierarchical perception transcoding rule to translate complex thermal environment indices into multi-dimensional comfort level labels; and (3) mapping thermal sensitivity levels into multi-modal three-dimensional visual factors such as color gradation, texture, transparency, and dynamic particles, forming a thermal environment representation system that can be intuitively perceived in virtual simulation scenarios. This method can structure continuous numerical results into a psychologically discernible thermal perception spatial language and provide a unified standard input interface for 3D visual interaction experiments.

[0094] 2. Three-dimensional combined modeling and generation mechanism of microclimate elements and spatial configuration attributes A parameter-controllable spatial scene construction method was designed, which links the properties of components such as roofs, ground, and greenery with their thermal effects to generate a 3D scene, realizing a systematic and visual synthesis of building materials, vegetation structure, and thermal response. Technically, it overcomes the limitation of existing 3D modeling software in expressing environmental simulation attributes, and realizes the linked expression and parametric modeling of thermal environmental factors and spatial configuration.

[0095] 3. Deep integration technology of 3D visualization scene and selection experimental system A 3D display front-end for multi-attribute choice experiments (DCE) was developed and deployed synchronously with the preference experiment system, enabling respondents to make choices based on perceptible input in a real-world scenario. Technological breakthrough: This is the first time that experimental design methods have been deeply integrated with 3D thermal environment display, overcoming the shortcomings of traditional questionnaires in presenting thermal environment information and improving the authenticity and response rate of preference data.

[0096] 4. Interaction effect identification and personalized optimization algorithm module in multivariate scenarios By combining the DCE preference model with multi-attribute thermal environment simulation data, a reverse inference recommendation algorithm based on Logit model estimation results was constructed, which can output personalized thermal environment design suggestions (such as greening rate threshold, ground type optimization, etc.). Technically, it realizes a data reverse feedback closed loop from "resident behavior → preference weight → spatial design", which is an intelligent assisted design mechanism with design guidance capabilities.

[0097] 5. Scalable thermal environment decision support software platform architecture Innovation highlights: A prototype system for thermal environment preference assessment and feedback platform was built, which can be nested across multiple regions and connected to multiple climate models, supporting flexible deployment under different settlement types and climate conditions.

[0098] Technological Breakthrough: Establishing a seamless chain from "simulation → visualization → experimentation → optimization" to fill the gap in microclimate adaptive design systems based on subjective perception in urban and rural planning and architectural design.

[0099] Corresponding to the aforementioned method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences, this invention also discloses a system for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences, which specifically includes: The microclimate simulation module is used to construct a three-dimensional microclimate model of the target site using thermal environment simulation software, set different thermal environment design variables, and obtain thermal environment performance indicators under multiple design scenarios through simulation. The questionnaire design module is used to determine the key attributes and level values ​​that affect the merits of thermal environment modification schemes based on the discrete choice experiment method, generate a selection set based on the determined key attributes and level values, and build a structured questionnaire based on the obtained selection set. The questionnaire data acquisition module is used to perform 3D visualization and interactive design based on the obtained structured questionnaires, and to acquire the questionnaire data of the respondents. The statistical analysis module is used to quantify the intensity of residents' preferences for different design scenarios based on the obtained questionnaire data, through preference modeling, and to identify the optimal design scenario scheme that takes into account both residents' preferences and excellent thermal performance by coupling the residents' preferences with the thermal environment performance indicators obtained from microclimate simulation.

[0100] It should be noted that for a detailed description of the rural residential thermal environment optimization system based on microclimate simulation and residents' preferences provided in the embodiments of the present invention, please refer to the relevant description of the rural residential thermal environment optimization method based on microclimate simulation and residents' preferences provided in the embodiments of the present invention, which will not be repeated here.

[0101] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A method for optimizing the thermal environment of rural residences based on microclimate simulation and residents' preferences, characterized in that, The method includes: A three-dimensional microclimate model of the target site was constructed using thermal environment simulation software. Different thermal environment design variables were set, and thermal environment performance indicators under multiple design scenarios were obtained through simulation. Based on the discrete choice test method, the key attributes and level values ​​that affect the merits of thermal environment modification schemes are determined. Based on the determined key attributes and level values, a selection set is generated, and a structured questionnaire is established based on the obtained selection set. Based on the obtained structured questionnaire, 3D visualization and interactive design were performed, and questionnaire data from respondents were obtained; Based on the obtained questionnaire data from respondents, the intensity of residents' preferences for different design scenarios was quantified through preference modeling. By coupling the residents' preferences with the thermal environment performance indicators obtained from microclimate simulation, the optimal design scenario scheme that takes into account both residents' preferences and excellent thermal performance was identified.

2. The method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences as described in claim 1, characterized in that, A three-dimensional microclimate model of the target site was constructed using thermal environment simulation software. Different thermal environment design variables were set, and thermal environment performance indicators under multiple design scenarios were obtained through simulation. These included: The basic data for modeling the research plots are obtained through plot surveys, including typical vegetation parameters and building parameters; orthophotos of the research plots are obtained through drone aerial photography or information maps of the plots are obtained through surveying and mapping. Meteorological data, including air temperature, relative humidity, and wind speed, were obtained from the field measurements of the research plots. The meteorological data measured in the field were used to evaluate the accuracy of the model. A three-dimensional microclimate model of the target site was constructed using ENVI-met software and simulation was performed, including: site model construction, vegetation model construction, building material parameter setting, input of meteorological boundary conditions, and simulation operation; the model accuracy was evaluated based on the simulation results. The simulated output data includes air temperature, relative humidity, wind speed, and mean radiant temperature. These data are then imported into RayMan software for secondary calibration and calculation. In addition, thermal comfort indices, including physiological equivalent temperature and general thermal climate index, are calculated using standard human body parameters.

3. The method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences as described in claim 2, characterized in that, A three-dimensional microclimate model of the target site was constructed using thermal environment simulation software. Different thermal environment design variables were set, and thermal environment performance indicators under multiple design scenarios were obtained through simulation. These included: Based on the obtained thermal environment performance indicators, a comprehensive evaluation index, the composite outdoor thermal comfort index, was constructed using principal component analysis. Using the composite outdoor thermal comfort index as an input variable, the K-means clustering algorithm was used to divide all design scenarios. The clustering analysis classified the design scenarios into three categories, representing different thermal environment qualities: comfortable, transitional, and uncomfortable. The clustering results were verified by the silhouette coefficient and the Davidson-Bolding index.

4. The method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences as described in claim 1, characterized in that, The key attributes and level values ​​that influence the quality of thermal environment modification schemes should be determined, including: The key attribute identification process is divided into the following three stages: First, based on the research content, we reviewed the existing spatial preference research based on discrete choice experiments and found that: rural housing choice modeling needs to include both physical spatial attributes and perceived or ecological attributes. Secondly, by conducting in-depth interviews with residents, we can understand the rural housing environment factors that residents value highly, including those that directly affect comfort, maintenance costs, and aesthetic features. Finally, combining field research and policy literature analysis, the attribute set was further refined into relevant variables that fit current rural development, ecological concern and vernacular construction practices, and the key attributes were finally determined. Reasonable level values ​​were set for each attribute. These level values ​​were derived from field research, pre-survey, and literature review, taking into account both theoretical logic and practical situation. At the same time, the number of levels was controlled to match the age structure and cognitive level of rural residents, so as not to increase the complexity of the experimental design and the cognitive burden of the respondents.

5. The method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences as described in claim 4, characterized in that, The key attributes include: Building structure: including roof, facade, and floor materials; Spatial greening: including vegetation types, coverage, width / height, and configuration methods; Thermal performance: including the range of physiologically equivalent temperature or general thermal climate index values ​​derived from simulations; Usage and maintenance: including maintenance frequency and cost information; Perceive cultural characteristics: including the locality of materials and the consistency of traditional style.

6. The method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences as described in claim 1, characterized in that, A selection set is generated based on the identified key attributes and level values, and a structured questionnaire is built based on the obtained selection set, specifically including: Selection set generation: Using SPSS or Ngene software, based on defined attributes and corresponding level values, multiple statistically efficient and independent orthogonal selection sets were generated using orthogonal experimental design. Each selection set included three options: Option A, Option B, and "Do not select / Keep the status quo". Including the "Do not select / Keep the status quo" option effectively avoids preference overestimation bias caused by forced selection. A replication test was also set up, in which the first scenario in each selection set was presented again in random order at the end of the corresponding selection set. By comparing whether the two selections were consistent, invalid questionnaires with arbitrary or unstable answers were identified and eliminated, thereby further improving the quality of the data. Questionnaire Design: Based on the obtained selection set, a structured questionnaire was designed, consisting of the following three logically progressive parts: The first part is thermal comfort perception and living environment evaluation: Based on the simplified scale of ASHRAE thermal comfort standard, multiple sets of questions were used to understand the respondents’ thermal feelings about their current living environment, overall satisfaction, main environmental factors that caused dissatisfaction, and main cooling methods. The second part is the core component: using 3D scene diagrams to clearly explain the definitions and intuitive effects of various attributes and their different levels to respondents, in order to overcome the comprehension barriers of pure text descriptions and enhance the realism and immersion of the scenario; each respondent will complete 12 carefully designed sets of choices in turn; in each set of choices, respondents need to choose between two transformation plans with different attribute combinations and corresponding costs and an option of "not choosing / keeping the status quo", thereby revealing their implicit preference structure through multi-attribute trade-offs; The third part is the collection of socioeconomic characteristics information: After completing all the selection tasks, the questionnaire asks the respondents about their personal and family socioeconomic characteristics in a relatively natural way, including age, gender, education level, annual family income, length of residence, and current housing type. The corresponding variables will serve as key explanatory variables for subsequent econometric model analysis of preference heterogeneity, in order to gain a deeper understanding of the systematic differences in the renewal choices of groups with different economic capabilities, cultural backgrounds, and life experiences.

7. The method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences as described in claim 1, characterized in that, A selection set is generated based on the identified key attributes and level values, and a structured questionnaire is built based on the obtained selection set, specifically including: Questionnaire survey implementation design: The questionnaire survey dates should cover both weekdays and weekends to improve the coverage and representativeness of the sample; The daily surveys are scheduled to be conducted during multiple time periods when residents are more willing to be interviewed, reaching out to and covering family members with different daily routines, thus reducing selection bias caused by a single interview time. The survey employed one-on-one face-to-face interviews to ensure thorough communication and accurate understanding. Before the data collection work officially started, all investigators participating in the field survey received unified and standardized special training. All collected questionnaire data were thoroughly anonymized after entry and stored in encrypted dedicated storage devices, with access strictly limited to core research team members. Before each interview, the investigators will clearly and completely explain to potential participants the research objectives, main content, expected time, and their rights, and make it clear that participation is entirely voluntary and participants can withdraw at any time without reason. In the specific sampling and interview implementation stage, in each pre-selected settlement, a combination of systematic sampling or random sampling is used to determine the interviewees; based on the sampling results, 3 to 5 households are visited to reduce the clustering effect caused by excessive concentration of samples and improve the spatial distribution balance of the samples; finally, invalid questionnaires are removed as needed. Calculate the minimum sample size for discrete selection using Orme's "rule of thumb" to ensure that the sample size meets the basic requirements for the sample size of discrete selection experimental designs and to guarantee the statistical power of subsequent analyses.

8. The method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences as described in claim 1, characterized in that, Based on the obtained structured questionnaire, 3D visualization and interactive design were performed, and respondent questionnaire data was obtained, specifically including: 3D visualization: Through sample plot survey and mapping, residential models are generated in SketchUp software and rendered using rendering software; each set of selected tasks is presented as a carefully rendered 3D scene map, which systematically changes key residential attributes and clearly marks the corresponding assumed costs. Interactive design: Import structured questionnaires into a web-based questionnaire system. Respondents can switch perspectives to view the solutions, rotate scene models, adjust thermal comfort level color gradation, perform interactive operations, and select preferences. Data feedback: Automatically stores respondents' selection behavior and their socioeconomic characteristics.

9. The method for optimizing the thermal environment of rural residences based on microclimate simulation and resident preferences as described in claim 1, characterized in that, Based on the obtained questionnaire data from respondents, preference modeling was used to quantify the intensity of residents' preferences for different design scenarios. Furthermore, by coupling residents' preferences with thermal environment performance indicators obtained from microclimate simulation, the optimal design scenario scheme that balances residents' preferences with excellent thermal performance was identified, specifically including: The collected questionnaire data were preprocessed and coded, and each respondent's multiple selection data were matched with their unique socioeconomic characteristic data to construct a "long format" panel dataset for model estimation. Discrete choice model construction and estimation: Three types of discrete choice models were constructed using Stata software to form a gradient optimization model system to accurately capture residents' preferences: Model 1 is a conditional logit model without individual economic characteristic variables; Model 2 is a mixed logit model with individual economic characteristic variables; Model 3 is a panel mixed logit model that controls for repeated observation effects at the individual level. Goodness-of-fit test: A goodness-of-fit test is conducted on the above three types of discrete choice models, with the log-likelihood value as the core indicator. At the same time, the model is ensured to pass the chi-square test to verify the overall robustness and data fit of the model. Based on the optimal model that passes the test, namely the panel mixed logit model, resident preference analysis is carried out. Marginal willingness to pay analysis: To analyze the marginal value of different living environment attributes, a point estimate of the value of a unit change in a non-price choice attribute is calculated using a formula, namely the marginal willingness to pay. This measure is the ratio between the coefficient of a certain attribute in an individual's utility function and the coefficient of the price attribute. in, Represented as attributes Marginal willingness to pay For attributes The corresponding utility coefficient, The utility coefficient of the price attribute; by ranking the marginal willingness to pay for different living environment attributes, the residents' preference for choosing living environment attributes is identified; Coupling Analysis and Results Output: Based on the estimation results of the panel mixed Logit model, the utility score for each design scenario is calculated. The utility score consists of the deterministic utility components in the model, and the specific calculation formula is as follows: Among them, V njt x represents the deterministic utility that individual n obtains from option j in task t; k,jt β represents the specific level of the k-th attribute in scheme j; k These are the fixed coefficients or average random coefficients of the k-th attribute estimated by the panel mixture Logit model; The utility scores of different design scenarios are compared and correlated with the composite outdoor thermal comfort index obtained through microclimate simulation and principal component analysis to identify the optimal design scenario that balances residents' preferences and excellent thermal performance.

10. A rural residential thermal environment optimization system based on microclimate simulation and resident preferences, characterized in that, The system includes: The microclimate simulation module is used to construct a three-dimensional microclimate model of the target site using thermal environment simulation software, set different thermal environment design variables, and obtain thermal environment performance indicators under multiple design scenarios through simulation. The questionnaire design module is used to determine the key attributes and level values ​​that affect the merits of thermal environment modification schemes based on the discrete choice experiment method, generate a selection set based on the determined key attributes and level values, and build a structured questionnaire based on the obtained selection set. The questionnaire data acquisition module is used to perform 3D visualization and interactive design based on the obtained structured questionnaires, and to acquire the questionnaire data of the respondents. The statistical analysis module is used to quantify the intensity of residents' preferences for different design scenarios based on the obtained questionnaire data, through preference modeling, and to identify the optimal design scenario scheme that takes into account both residents' preferences and excellent thermal performance by coupling the residents' preferences with the thermal environment performance indicators obtained from microclimate simulation.

Images