Method for optimizing light-thermal balance of open-close roof sports building based on generative adversarial network and genetic algorithm

Abstract

The present application relates to the technical field of sports building light heat optimization, and more particularly to a method for balancing and optimizing the light heat environment of a retractable roof sports building based on a generative adversarial network and a genetic algorithm. The method comprises the following steps: obtaining retractable roof shape data; parameterizing the retractable roof shape data to generate a two-dimensional grayscale image of the retractable roof; performing fluid dynamics and light heat environment numerical simulation on the retractable roof shape data to obtain environment simulation data of the retractable roof; pairing the two-dimensional grayscale image of the retractable roof with the environment simulation data of the lower space of the retractable roof to obtain an input-output image pair; and using the generative adversarial network to predict the wind speed distribution of the input-output image pair to generate a wind speed distribution prediction value of the retractable roof. The present application realizes intelligent optimization of the light heat balance of a retractable roof sports building by combining a generative adversarial network and a genetic algorithm, and solves the deficiencies of traditional methods in flexibility, optimization accuracy and computational efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of light-heat optimization, and particularly relates to a light-heat balance optimization method for an openable roof sports building based on a generative adversarial network and a genetic algorithm. BACKGROUND

[0002] An openable roof, as a building structure with adaptive function, adjusts indoor temperature and air circulation by changing the opening degree of the roof, so as to achieve the light-heat balance of the building under different climate conditions. Traditional roof control strategies rely on fixed empirical rules and physical models, but with the diversification of building environment and climate conditions, these methods face problems such as insufficient flexibility and precision deviation. In order to improve the intelligence and adaptability of the control strategy, researchers have begun to explore advanced optimization methods based on generative adversarial networks (GAN) and genetic algorithms (GA) for intelligent control of openable roofs. The generative adversarial network can effectively simulate complex environmental changes through the training of the adversarial process of the generative model and the discriminative model. The genetic algorithm, as a simulation of natural selection optimization algorithm, can efficiently find the optimal solution in a large-scale search space. However, the existing optimization methods usually do not have a tool to intuitively judge the wind, light and thermal comfort environment, resulting in low complexity and precision of the light-heat balance optimization of the openable roof sports building. SUMMARY

[0003] Therefore, it is necessary to provide a light-heat balance optimization method for an openable roof sports building based on a generative adversarial network and a genetic algorithm, so as to solve at least one of the above technical problems.

[0004] To achieve the above-mentioned purpose, a light-heat balance optimization method for an openable roof sports building based on a generative adversarial network and a genetic algorithm is provided, and the method comprises the following steps:

[0005] Step S1: obtaining openable roof shape data; parameterizing modeling the openable roof shape data to generate a two-dimensional grayscale image of the openable roof; performing fluid dynamics and light-heat environment simulation on the sports building space covered by the openable roof to obtain wind-light-heat environment simulation data of the space covered by the openable roof;

[0006] Step S2: pairing the two-dimensional grayscale image of the openable roof and the wind-light-heat environment simulation data to obtain an input-output image pair; using a generative adversarial network to predict the wind speed distribution of the input-output image pair to generate a spatial wind speed distribution prediction value of the openable roof; and constructing a wind-light-heat environment gene library based on the spatial wind speed distribution prediction value and the light-heat environment simulation data;

[0007] Step S3: Obtain sports large space environment demand data; set wind, light and heat environment balance target according to the sports large space environment demand data, obtain wind, light and heat balance target parameters; based on the wind-light-heat environment gene library, the wind, light and heat balance target parameters are iteratively optimized by genetic algorithm to generate the Pareto optimal solution of the opening and closing roof shape parameters under the typical meteorological conditions; based on the roof shape parameter change opening ratio of the Pareto optimal solution, the opening ratio and the corresponding indoor environment are analyzed to generate the roof opening and closing ratio-ventilation thermal comfort light environment influence curve;

[0008] Step S4: Perform climate zone sensitivity analysis on the roof opening and closing ratio-ventilation thermal comfort light environment influence curve to generate a sensitivity analysis result; and dynamically adjust the opening ratio of the opening and closing roof according to the sensitivity analysis result to perform the opening and closing roof sports building wind, light and heat balance optimization and control operation.

[0009] The present application can precisely analyze the ventilation, illumination and thermal environment of the lower building space under the open-close roof by parameterizing modeling of the open-close roof shape data and fluid dynamics and light-thermal environment simulation of the lower space, and can optimize the wind-light-thermal balance, improve the comfort and energy efficiency of the sports space under the roof, and ensure the comfort of the building internal environment. The wind speed distribution prediction of the input-output image pair by the generative adversarial network can efficiently and accurately simulate the influence of the roof opening ratio change on the wind speed distribution, provide a scientific basis for the wind speed regulation of the roof, and further optimize the natural ventilation effect of the building. By constructing a wind-light-thermal environment gene library, the wind speed, illumination and thermal environment are optimized. The genetic algorithm iteratively optimizes the wind-light-thermal balance target parameters, so that the wind-light-thermal regulation of the open-close roof can be accurately adjusted according to different climate zones, building use requirements and environmental changes, ensuring the wind-light-thermal comfort and energy efficiency of the building. Based on the Pareto optimal solution, the roof opening ratio-ventilation thermal comfort light environment influence curve is generated, which provides detailed relationship between the roof opening ratio and ventilation, thermal comfort and light environment for designers, helps to make more scientific design decisions, and optimizes the ventilation, daylighting and thermal comfort of the open-close roof. Through climate zone sensitivity analysis, the morphological design of the open-close roof can be adjusted according to the characteristics of different climate zones, ensuring that the roof can be individually adjusted under various climate conditions to achieve the best wind-light-thermal balance effect. This targeted design not only improves the adaptability of the roof, but also provides protection for the energy saving effect of the building. According to the sensitivity analysis results, the opening ratio of the open-close roof is dynamically adjusted, so that the open-close roof can flexibly adjust the opening ratio under different meteorological conditions, optimize the balance of wind, light and thermal environment, and further improve the energy efficiency and comfort of the building. Therefore, the present application realizes intelligent optimization of the light-thermal balance of the open-close roof sports building by combining generative adversarial network and genetic algorithm, and solves the deficiencies of traditional methods in flexibility, optimization precision and calculation efficiency.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: obtaining open-close roof shape data;

[0012] Step S12: parameterizing modeling of the open-close roof shape data, defining the dynamic adjustment range of the roof opening and closing direction and opening ratio, and obtaining an open-close roof geometric model;

[0013] Step S13: converting the open-close roof geometric model into a grayscale image to generate an open-close roof two-dimensional grayscale image;

[0014] Step S14: Extract the roof shape of the two-dimensional gray image of the open-close roof, and use simulation software to perform CFD simulation and light-thermal environment simulation on the lower building space under the open-close roof, calculate the indoor and outdoor wind speed distribution and thermal comfort index, and obtain the indoor and outdoor wind speed distribution data and thermal comfort index data;

[0015] Step S15: Perform daylight analysis on the lower building space under the open-close roof to generate daylight factor distribution data; convert the indoor and outdoor wind speed distribution data, daylight factor distribution data and thermal comfort index data into charts respectively to generate wind speed distribution chart, daylight factor distribution chart and thermal comfort index thermal map;

[0016] Step S16: Integrate the wind speed distribution chart, daylight factor distribution chart and thermal comfort index thermal map into the wind-light-thermal environment simulation data of the space covered by the open-close roof.

[0017] The present application can clearly define the opening direction and opening proportion dynamic adjustment range of the roof by accurately obtaining and parameterizing the open-close roof shape data, ensure that the model has high flexibility and adjustability, and provide accurate basic data for subsequent simulation and optimization. Through the construction of the open-close roof geometric model, the opening proportion of the roof can be dynamically adjusted to meet the changes of different climates, environments and functional requirements, and ensure that the roof can optimize the ventilation and light-thermal environment under various conditions, further improve the comfort and energy utilization efficiency of the building. Through CFD simulation and light-thermal environment simulation, the indoor and outdoor wind speed distribution and thermal comfort index can be accurately calculated to provide detailed environmental data for building designers, help to adjust the open-close roof structure, optimize ventilation, lighting and thermal comfort, and improve the living and use experience of the building. Through daylight analysis to generate daylight factor, the daylighting performance of the open-close roof can be effectively evaluated to ensure that the building can make full use of natural light and reduce energy consumption of artificial lighting. At the same time, the analysis of daylight factor helps to more accurately optimize the opening proportion of the roof and improve the energy utilization efficiency of the building. The chart conversion step generates wind speed distribution chart and thermal comfort index thermal map, which helps designers to intuitively understand the ventilation effect and thermal comfort of the open-close roof under different conditions, and facilitates accurate design decisions. The thermal map and distribution chart can also provide scientific basis for the later maintenance and adjustment of the building.

[0018] Preferably, the daylight analysis on the lower building space under the open-close roof comprises:

[0019] The lower building space under the open-close roof is subjected to light extraction to obtain a light extraction area image of the lower space under the open-close roof; the sun elevation angle and azimuth angle of the light extraction area image of the lower space under the open-close roof are calculated to obtain the sun elevation angle and the sun azimuth angle;

[0020] The light propagation simulation of the daylighting area of the open-close roof is simulated based on the solar elevation angle and the solar azimuth angle, and simulation data of the solar light propagation path are generated; the irradiation intensity of each pixel point in the daylighting area image of the lower space of the open-close roof is calculated according to the simulation data of the solar light propagation path, and a local solar radiation intensity map is obtained;

[0021] The actual irradiation area ratio of the daylighting area image of the lower space of the open-close roof is calculated by using the local solar radiation intensity map, and a daylighting coefficient is obtained.

[0022] The present application can accurately identify which areas have good daylighting conditions under the sunlight by extracting the daylighting area image of the open-close roof, and further optimize the design of the roof opening, thereby improving the natural daylighting efficiency. By calculating the elevation angle and azimuth angle of the sun, the changes of the daylighting area can be dynamically simulated according to different seasons and times, ensuring that the daylighting analysis considers the changes of the solar angle, and generating more accurate solar radiation simulation data. Based on the calculation of the solar angle, the simulation of the light propagation path can comprehensively understand how the sunlight propagates in the roof opening area, helping designers better control the direction and intensity of daylighting. According to the simulation data of the solar light propagation path, the irradiation intensity of each pixel point is calculated, and a local solar radiation intensity map is obtained, which can comprehensively evaluate the daylighting conditions of different areas and provide accurate data support for space utilization and lighting design. The actual irradiation area ratio is calculated by using the local solar radiation intensity map, and the daylighting coefficient is generated, which can quantify the efficiency of daylighting, help designers adjust the design of the roof opening, maximize the natural daylighting effect, improve the indoor lighting quality, and reduce energy consumption.

[0023] Preferably, step S2 comprises the following steps:

[0024] Step S21: image pairing the two-dimensional gray image of the open-close roof and the wind speed distribution map of the lower space of the roof to obtain an input-output image pair; data set division is performed on the input-output image pair to generate a model training set, a model test set and a model validation set;

[0025] Step S22: constructing a Pix2Pix network framework; inputting the model training set, the model test set and the model validation set into the Pix2Pix network framework for training, testing and verification to obtain a Pix2Pix algorithm model; inputting a preset two-dimensional image of the open-close roof design into the Pix2Pix algorithm model for indoor wind speed prediction of the lower space of the open-close roof to generate an indoor predicted wind speed map of the lower space of the open-close roof;

[0026] Step S23: performing numerical matrix conversion on the indoor predicted wind speed map of the lower space of the open-close roof to obtain a spatial wind speed distribution prediction value of the open-close roof, and extracting the average wind speed value of the key area of the sports building from the spatial wind speed distribution prediction value of the open-close roof to obtain the average wind speed value of the key area.

[0027] Step S24: Parameter correlation of the opening ratio of the opening and closing roof according to the key area average wind speed value, and gene library construction using the correlated parameters and light and heat environment simulation data to obtain the wind-light-heat environment gene library.

[0028] The present application can ensure that the input and output data of wind speed prediction are matched by pairing the two-dimensional gray image of the opening and closing roof with the wind speed distribution map, and accurate training data is generated. Further, through data set division, the balance of training, testing and verification set is ensured, which helps to improve the reliability and accuracy of model training and provides strong data support for subsequent prediction. By constructing the Pix2Pix network framework, combining the image data of the opening and closing roof with the wind speed distribution map, accurate wind speed prediction can be realized. The Pix2Pix network uses the generative adversarial network (GAN) technology to learn the relationship between the input image and the target wind speed map, and generates high-quality wind speed prediction results to ensure the accuracy and usability of the prediction map. The prediction result is converted into a numerical matrix, and the average wind speed value of the key area is extracted. Through the numerical matrix conversion of the wind speed distribution map, the wind speed data of the key area in the sports building can be more accurately obtained, which helps to optimize the ventilation effect of the opening and closing roof, and provides data basis for subsequent wind-light-heat balance optimization. By parameter correlating the average wind speed value of the key area with the opening ratio, dynamic optimization and adjustment of the roof structure can be realized. This process uses the correlation of wind speed and light and heat environment data to establish a wind-light-heat environment gene library, providing an effective tool for architects to ensure that buildings can optimize ventilation and balance of light and heat environment under different climate conditions.

[0029] Preferably, step S22 comprises:

[0030] Constructing a Pix2Pix network framework, wherein the Pix2Pix network framework comprises a generator and a discriminator;

[0031] Setting training parameters for the Pix2Pix network framework to generate training parameter setting values, wherein the training parameter settings include learning rate setting, convolution layer number setting, epoch setting and optimizer setting;

[0032] Inputting the model training set into the Pix2Pix network framework to predict the wind speed through the generator in the Pix2Pix network framework, generating an initial opening and closing roof predicted wind speed map, and using the model test set to compare the initial opening and closing roof predicted wind speed map with the real result image through the discriminator in the Pix2Pix network framework;

[0033] Calculating the structural similarity and root mean square error of the initial opening and closing roof predicted wind speed map to obtain the accuracy of the wind speed prediction map;

[0034] The wind speed prediction graph accuracy is predicted through the model verification set, and a Pix2Pix algorithm model is obtained;

[0035] The preset open-close roof design two-dimensional image is input into the Pix2Pix algorithm model to predict the indoor wind speed of the lower space of the open-close roof, and an indoor predicted wind speed graph of the lower space of the open-close roof is generated.

[0036] The present application can generate accurate wind speed prediction graphs by using the generator in the Pix2Pix network framework for wind speed prediction and combining the discriminator for preliminary image comparison. This method not only can handle complex wind speed prediction tasks, but also can efficiently predict in multiple open-close roof states. By setting the training parameters (such as learning rate, convolution layer number, epoch and optimizer), the parameter optimization in the model training process is ensured, thereby improving the learning effect and accuracy of the network model, reducing the risk of overfitting, and enhancing the generalization ability of the model. By calculating the structural similarity (SSIM) and root mean square error (RMSE) of the initial prediction result graph, the difference between the predicted wind speed graph and the true result can be accurately evaluated, thereby providing a basis for further optimizing the model. This precision evaluation method ensures the high quality of the prediction results, further improving the reliability of the building wind environment analysis. By adjusting the training parameters to optimize the prediction accuracy and further optimizing the wind speed prediction graph through convergence check, it can ensure that the finally generated open-close roof wind speed prediction graph has high accuracy and stability, and can provide more accurate wind environment data for building design. The use of discriminators ensures that the difference between the generated wind speed prediction graph and the true wind speed graph is minimized, thereby improving the realism of the generated image, so that the model can better imitate the actual wind speed distribution graph, and is suitable for complex building wind environment analysis.

[0037] Preferably, step S3 comprises the following steps:

[0038] Step S31: acquiring sports large space environment demand data; setting wind-light-heat balance target parameters by setting wind-light-heat balance targets for the open-close roof space according to the sports large space environment demand data;

[0039] Step S32: performing diversity evaluation on the wind-light-heat balance target parameters through the NSGA-III algorithm, and dynamically adjusting the wind-light-heat balance target parameters based on the diversity evaluation results to generate wind-light-heat balance target adjustment parameters;

[0040] Step S33: iteratively optimizing the wind-light-heat balance target parameters based on the wind-light-heat environment gene library through the NSGA-III algorithm, and eliminating extreme solutions to generate a Pareto optimal solution;

[0041] Step S34: Based on the Pareto optimal solution, the opening ratio of the roof shape parameter change is set, a multiple linear regression equation is established for the different opening ratios of the retractable roof and the corresponding indoor environmental performance, and the retractable roof opening ratio-ventilation influence curve, the retractable roof opening ratio-natural lighting influence curve and the retractable roof opening ratio-thermal comfort influence curve are generated, and are integrated into the retractable roof opening ratio-ventilation thermal comfort light environment influence curve.

[0042] The present application can set the wind-light-heat balance target for the wind speed distribution prediction value of the lower space of the retractable roof by using the wind-light-heat environment gene library, and generate a set of wind-light-heat balance target parameters for the roof design. This step provides a scientific theoretical basis for subsequent optimization, ensuring that the design meets the comfort and energy saving needs under different environmental conditions. The wind-light-heat balance target parameters are evaluated for diversity by NSGA-III algorithm, and the wind-light-heat balance target parameters are dynamically adjusted according to the evaluation results. This method considers diversity and feasibility, and can flexibly adjust the design parameters according to the actual environmental requirements, ensuring that the design optimization has higher adaptability and stability. By iteratively optimizing the wind-light-heat balance target parameters and eliminating extreme solutions, the feasibility and stability of the optimization process are ensured. This process uses intelligent algorithms such as NSGA-III to not only optimize the wind-light-heat balance of the roof, but also eliminate unreasonable extreme solutions, thereby ensuring that the final design has optimal performance indicators. By establishing a multiple linear regression equation based on the Pareto optimal solution for the different opening ratios of the retractable roof and the corresponding indoor environmental performance, the influence curve between the retractable roof opening ratio and ventilation, natural lighting and thermal comfort is generated. This process can clearly show the relationship between design factors and actual effects, providing accurate theoretical basis for subsequent decision-making. By integrating these influence curves, the retractable roof opening ratio-ventilation thermal comfort light environment influence curve is generated, providing a comprehensive evaluation tool for actual building design.

[0043] Preferably, step S31 comprises:

[0044] Obtain sports large space environment demand data;

[0045] Define optimization targets according to the sports large space environment demand data, wherein the optimization targets include maximizing the daylighting coefficient, maximizing the natural ventilation speed, and minimizing the thermal comfort index;

[0046] Constrain the decision variables for maximizing the daylighting coefficient, maximizing the natural ventilation speed, and minimizing the thermal comfort index, and limit the spatial three-dimensional coefficient of the opening ratio.

[0047] Based on the optimization target and the decision variable, the wind speed distribution prediction value of the retractable roof is combined with the parameter to obtain the wind-light-heat balance target parameter.

[0048] The present application can ensure that the roof design takes into account indoor lighting, ventilation effect and comfort by maximizing the daylighting coefficient, maximizing the natural ventilation speed and minimizing the thermal comfort index optimization target setting. This comprehensive optimization makes the building achieve the best balance in energy consumption, residential comfort and environmental benefits, and is especially suitable for green buildings and energy-saving design. By constraining the decision variables (such as the spatial three-dimensional coefficient of the opening ratio), the maximum and minimum range of the roof opening can be controlled to ensure that the design does not deviate from the actual feasibility. For example, an excessively large opening ratio will increase the radiant heat and affect the thermal comfort, and an excessively small opening ratio cannot effectively ventilate or light, and the setting of the constraint condition helps to find the best solution within the feasibility range. Based on the combination of optimization targets and decision variables, the wind-light-heat balance target parameters can be accurately generated, which provides a theoretical basis for subsequent light-heat balance calculation, wind speed distribution prediction and lighting analysis, making the entire design process more detailed and scientific. By considering lighting, natural ventilation and thermal comfort at the same time, a scheme with multiple advantages can be developed at the initial stage of building design. This multi-objective optimization not only improves the functionality of building design, but also reduces energy consumption in future operation and reduces dependence on air conditioning and other facilities. This optimization method can be flexibly applied to different types of buildings, and the optimization targets can be adjusted according to specific needs. For example, for sports buildings, ventilation speed is more critical, while for residential buildings, lighting coefficient and thermal comfort are more important. The adaptability of this method can meet the needs of different projects.

[0049] Preferably, step S32 comprises the following steps:

[0050] Step S321: setting the population size, crossover rate and mutation rate of the NSGA-III algorithm, and performing non-dominated sorting on the wind-light-heat balance target parameters according to the population size, crossover rate and mutation rate to obtain a wind-light-heat balance sorting result;

[0051] Step S322: performing crowded distance calculation on the wind-light-heat balance sorting result to obtain wind-light-heat balance sample distance data;

[0052] Step S323: performing diversity evaluation on the wind-light-heat balance target parameters using the wind-light-heat balance sorting result and the wind-light-heat balance sample distance data, and performing dynamic parameter adjustment on the wind-light-heat balance target parameters based on the diversity evaluation result to generate wind-light-heat balance target adjustment parameters, wherein the formula of the diversity evaluation is as follows:

[0053]

[0054] In the formula, D totalFor the diversity evaluation result, α is the influence coefficient of the diversity metric of the target space adjustment on the total diversity evaluation, β is the influence coefficient of the physical space sample distance metric of the target space adjustment on the total diversity evaluation, D is the diversity metric of the population in the target space, s i For the first wind-solar-thermal balance sample, s j For the second wind-solar-thermal balance sample, dist(s i ,s j ) is the physical distance between s i and s j , N is the number of samples.

[0055] The present application can accurately measure the diversity of the wind-solar-thermal balance target parameters by calculating the crowding distance of the wind-solar-thermal balance ranking result and further evaluating based on the diversity evaluation formula. This accurate evaluation helps to ensure the balanced development of the design scheme in multiple target spaces, thereby avoiding local optimal solution and promoting global optimization. Through dynamic adjustment of the wind-solar-thermal balance target parameters, the design scheme can be updated and optimized in real time in the constantly changing design space. The diversity evaluation result as the basis for dynamic adjustment enables the wind-solar-thermal balance target parameters to timely adapt to new design requirements and environmental conditions, effectively avoiding falling into local optimum in the optimization process and ensuring the superiority of the final scheme. The use of the diversity evaluation and dynamic adjustment function of the NSGA-III algorithm enables the wind-solar-thermal balance target parameters to be flexible and adaptable in the face of different design requirements and environmental changes. This flexibility is particularly important for dealing with complex and uncertain building environments, ensuring the applicability and efficiency of the design scheme. In the diversity evaluation formula, the α and β coefficients are used to control the diversity metric of the target space and the sample distance metric of the physical space, respectively, which can reasonably balance the influence of different optimization objectives in the optimization process, thereby improving the comprehensive performance of the design scheme. In this way, the algorithm can effectively balance different design objectives (such as daylighting, ventilation, and thermal comfort) and avoid compromising other objectives due to the prominence of a certain objective. Through crowding distance calculation and diversity evaluation, suitable design schemes can be better selected, and the optimization process can be accelerated through dynamic adjustment, which not only improves the quality of the design scheme, but also reduces unnecessary design iterations, saves computing resources and time, and improves design efficiency.

[0056] Preferably, step S4 comprises the following steps:

[0057] Step S41: Perform climate zone sensitivity analysis on the roof opening ratio-ventilation thermal comfort light environment influence curve to generate sensitivity analysis results;

[0058] Step S42: Perform climate zone testing on the opening roof shape data based on the sensitivity analysis results to generate applicability range evaluation data;

[0059] Step S43: Perform dynamic control adjustment of the opening ratio by applying the range evaluation data to perform the opening-closing roof sports building view thermal balance optimization operation.

[0060] The present application can deeply understand the ventilation and light-thermal comfort performance of the roof under different climate conditions through climate zone sensitivity analysis of the roof-ventilation thermal comfort light environment influence curve, so that the design can adapt to the needs of different regions. According to the difference of climate zones, the design is adjusted to ensure that the building can achieve the best thermal comfort and ventilation effect under various climate conditions. Through the generation of application range evaluation data, the opening ratio of the opening-closing roof structure data can be accurately tested, so as to dynamically control the adjustment of the opening ratio. This adjustment ensures that the roof can achieve the best effect under different climate zones, especially in areas with variable climate, it can better control the temperature, humidity and air flow in the room, and improve the overall living experience. Dynamic control of the opening ratio adjustment helps to optimize the natural ventilation and lighting efficiency of the building, reduces the dependence on air conditioning and artificial lighting, and improves the energy efficiency of the building. This optimization scheme helps to reduce the energy consumption of the building, supports more environmentally friendly and low-carbon building design, and promotes sustainable development. Through sensitivity analysis and application range evaluation, the advantages and disadvantages of building design in different climate zones can be accurately identified. This analysis not only improves the accuracy of the design, but also reduces the uncertainty in the actual construction process, avoids excessive adjustment or errors, and ensures that the final design can achieve consistent thermal comfort performance and ventilation effect in various climate zones. Based on the results of sensitivity analysis, scientific decision support can be provided to the design team to help them more reasonably select the most suitable roof opening ratio and structure scheme for the target area. This decision basis helps to make the best design choice under complex and variable climate conditions to ensure the comfort and energy efficiency of the building.

[0061] Preferably, step S42 comprises the following steps:

[0062] Step S421: According to the sensitivity analysis results, the key parameters of the opening-closing roof shape data are identified to obtain sensitive parameter data; and dynamic response analysis is performed on the sensitive parameter data to obtain response feature data;

[0063] Step S422: Perform climate zoning mapping on the opening-closing roof shape data based on the response feature data to obtain zoning data; and perform climate element simulation on the opening-closing roof shape data based on the zoning data to obtain climate simulation data;

[0064] Step S423: Perform performance testing on the opening-closing roof shape data based on the climate simulation data to obtain testing data; and perform environmental adaptability range analysis on the opening-closing roof shape data using the testing data to generate application range evaluation data.

[0065] The present application can accurately capture sensitive parameters and morphological response characteristics in the design by identifying key parameters of the open-close roof and performing dynamic response analysis. These data provide a scientific basis for subsequent climate adaptability testing, ensuring that the roof design can be optimized and adjusted for different environmental conditions, thereby achieving optimal ventilation, lighting, and thermal comfort. Through climate zoning mapping and climate element simulation, step S42 realizes detailed climate adaptability testing of the open-close roof structure. These simulation data can provide quantitative performance predictions under climate conditions for the design team, helping to ensure that the design maintains stable performance in different climate zones, especially under extreme climate conditions. The multi-dimensional analysis methods such as response characteristic data analysis, climate zoning mapping, and climate element simulation help to deeply understand the multi-level performance of the open-close roof structure under different climates and environments, further optimize the design scheme, and ensure its consistent performance under different climate conditions. Based on the structural performance testing, step S42 can deeply analyze the environmental adaptability range of the open-close roof structure, ensuring that it maintains excellent performance in different climate zones. Through suitability evaluation, the design can optimize the opening ratio and structural form of the roof, so that it can achieve energy-saving and comfortable goals in different climate environments, thereby improving the sustainability of the building. Through sensitivity analysis and climate simulation, accurate design parameters and decision support are provided for designers, which help designers better understand the performance of the building under different climate conditions, thereby making more scientific and accurate design decisions and avoiding design errors. BRIEF DESCRIPTION OF DRAWINGS

[0066] Figure 1 Figure 1 is a schematic diagram of the step flow of the open-close roof sports building light-thermal balance optimization method based on the generative adversarial network and genetic algorithm;

[0067] Figure 2 Figure 2 is a schematic diagram of the detailed implementation steps of step S2 in the method; Figure 1

[0068] Figure 3 Figure 3 is a schematic diagram of the detailed implementation steps of step S3 in the method; Figure 1

[0069] The implementation of the present application, functional characteristics and advantages will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0070] The technical method of the present application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.​​

[0071] Furthermore, the accompanying drawings are included to provide a further understanding of the application, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the application and, together with the description, serve to explain the principles of the application. In the drawings:

[0072] It should be understood that, although terms such as "first," "second," and the like can be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element without departing from the scope of the example embodiments. The term "and / or" as used herein encompasses any and all combinations of one or more of the associated listed items.

[0073] To achieve the above object, there is provided Figures 1 to 3 A method for optimizing light-thermal balance of an openable roof of a sports building based on a generative adversarial network and a genetic algorithm, the method comprising the following steps:

[0074] Step S1: obtaining openable roof shape data; parameterizing modeling the openable roof shape data to generate a two-dimensional grayscale image of the openable roof; performing fluid dynamics and light-thermal environment simulation on a sports building space covered by the openable roof to obtain wind-light-thermal environment simulation data of the space covered by the openable roof;

[0075] Step S2: pairing the two-dimensional grayscale image of the openable roof and the wind-light-thermal environment simulation data to obtain an input-output image pair; using a generative adversarial network to predict the wind speed distribution of the input-output image pair to generate a spatial wind speed distribution prediction value of the openable roof; and constructing a wind-light-thermal environment gene library based on the spatial wind speed distribution prediction value and the light-thermal environment simulation data;

[0076] Step S3: obtaining sports large space environment demand data; setting wind-light-thermal environment balance targets for the openable roof space according to the sports large space environment demand data to obtain wind-light-thermal balance target parameters; iteratively optimizing the wind-light-thermal balance target parameters based on the wind-light-thermal environment gene library through a genetic algorithm to generate a Pareto optimal solution of the openable roof shape parameters under typical meteorological conditions; and performing regression analysis on the opening ratio and the corresponding indoor environment based on the change in the roof shape parameters of the Pareto optimal solution to generate a roof opening ratio-ventilation thermal comfort light environment influence curve;

[0077] Step S4: Climate zone sensitivity analysis is performed on the opening-closing roof ratio-ventilation thermal comfort light environment influence curve to generate a sensitivity analysis result; and the opening-closing roof is dynamically adjusted in terms of the opening ratio based on the sensitivity analysis result to perform the opening-closing roof sports building wind-light-thermal balance optimization operation.

[0078] The present application can accurately analyze the ventilation, illumination and thermal environment of the roof by parameterized modeling, fluid dynamics and light-thermal environment simulation of the opening-closing roof form data, optimize the wind-light-thermal balance, improve the comfort and energy efficiency of the space under the roof, and ensure the comfort of the building interior environment. The generated adversarial network can efficiently and accurately simulate the influence of the opening ratio change of the roof on the wind speed distribution, provide a scientific basis for the wind speed regulation of the roof, and further optimize the natural ventilation effect of the building. By constructing a wind-light-thermal environment gene library, the wind speed, illumination and thermal environment are optimized. The genetic algorithm iteratively optimizes the wind-light-thermal balance target parameters, so that the wind-light-thermal regulation of the opening-closing roof can be accurately adjusted according to different climate zones, building use requirements and environmental changes, ensuring the wind-light-thermal comfort and energy efficiency of the building. Based on the Pareto optimal solution, the roof opening-closing ratio-ventilation thermal comfort light environment influence curve is generated, which provides detailed relationship between the roof opening ratio and ventilation thermal comfort, light environment for designers, helping to make more scientific design decisions, and optimizing the ventilation and thermal comfort of the opening-closing roof. Through climate zone sensitivity analysis, the structure design of the opening-closing roof can be adjusted according to the characteristics of different climate zones to ensure that the roof can achieve the best wind-light-thermal balance effect under various climate conditions. This targeted design not only improves the adaptability of the roof, but also provides protection for the energy-saving effect of the building. According to the sensitivity analysis result, the opening ratio of the opening-closing roof structure is dynamically adjusted, so that the opening-closing roof can flexibly adjust the opening ratio under different climate conditions, optimize the balance of wind, light and heat environment, and further improve the energy efficiency and comfort of the building. Therefore, the present application realizes intelligent optimization of the light-thermal balance of the opening-closing roof sports building by combining the generated adversarial network and genetic algorithm, and solves the deficiencies of traditional methods in flexibility, optimization precision and calculation efficiency.

[0079] In the embodiment of the present application, as shown in the reference Figure 1 The opening-closing roof sports building light-thermal balance optimization method based on the generated adversarial network and genetic algorithm includes the following steps:

[0080] Step S1: Obtain the opening and closing roof shape data; parameterize the opening and closing roof shape data to generate a two-dimensional gray image of the opening and closing roof; perform fluid dynamics and light-thermal environment simulation on the sports building space covered by the opening and closing roof to obtain wind-light-thermal environment simulation data of the space covered by the opening and closing roof;

[0081] In the embodiments of the present application, by obtaining the design data of the opening and closing roof, including the geometric shape, size, material, opening ratio, opening and closing mode and other structural information of the roof, these data can be obtained through architectural design drawings, structural design files or 3D scanning data of existing buildings, and various physical parameters of the roof can be collected using laser scanning technology, optical imaging, measuring tools and other means. The obtained opening and closing roof shape data is integrated, and the integrity and accuracy of the data are ensured, and these data include the spatial coordinate system, geometric characteristics, opening position, size information and the like of the roof. Through CAD (Computer Aided Design) software or BIM (Building Information Modeling) platform, it is ensured that the data can be used as the basis for subsequent modeling. According to the opening and closing roof structure data, a two-dimensional or three-dimensional model of the opening and closing roof is generated by using parameterized modeling technology. This step can be performed by modeling software (such as AutoCAD, Rhino, SolidWorks, etc.). Parameterized design allows dynamic adjustment of the opening and closing ratio, size, angle and the like of the roof, thereby quickly generating different roof configurations. Each design parameter (such as opening ratio, roof inclination angle, number of layers, etc.) can be used as an input to the model. Through the modeled roof data, a two-dimensional grayscale image of the opening and closing roof is generated using rendering tools, and the image should show the visualization information of the geometric shape and opening position of the roof. In the image rendering process, ray tracing technology can be used to simulate the lighting and shadow of the roof, and a real grayscale image is generated, in which the grayscale value reflects the lighting intensity or opening and closing state of different areas. A three-dimensional geometric model of the opening and closing roof is constructed using CAD software (such as AutoCAD, SolidWorks) or BIM tools (such as Revit). The model should include the openable mechanism of the roof, the supporting structure and the influencing factors of the surrounding environment. For fluid dynamics simulation, the CFD software (such as ANSYS Fluent, OpenFOAM) is used to divide the grid of the space under the roof and its surrounding air domain. For light and heat environment simulation, the finite element method (such as ANSYS, COMSOL) is used for grid refinement to ensure accuracy. According to the actual situation, the fluid boundary conditions such as wind speed, air pressure and temperature are set. Different opening and closing angles of the opening and closing roof and different meteorological conditions (such as wind speed, air temperature, humidity, etc.) can be simulated. The CFD software is used to solve the air flow, wind speed distribution, wind pressure distribution, dynamic pressure and the like of the opening and closing roof under different opening and closing states. The stability, airflow penetration and possible airflow impact of the opening and closing roof are analyzed. Through fluid dynamics simulation, the wind speed distribution diagram, pressure distribution diagram and streamline diagram are obtained, and the ventilation performance, heat exchange effect and wind force influence on the roof structure of the opening and closing roof are analyzed. The light intensity, solar radiation, air temperature, humidity and other light and heat environment conditions are set. By simulating the light and heat flow distribution in different seasons and different time periods, more representative environmental data is obtained. Through finite element simulation, the heat transfer of the roof under different opening and closing angles is simulated.The temperature changes of the roof surface, structure and surrounding environment are analyzed by considering heat exchange modes such as radiation, convection and conduction. The roof surface temperature distribution, heat radiation distribution diagram and the influence of the opening and closing roof on indoor and outdoor temperatures are obtained. Further analysis can be made on the influence of the opening and closing roof on energy efficiency, indoor temperature control and comfort.

[0082] Step S2: image pairing is performed on the opening and closing roof two-dimensional gray-scale image and the wind-light-thermal environment simulation data to obtain an input-output image pair; a generative adversarial network is used to predict the wind speed distribution of the input-output image pair to generate a spatial wind speed distribution prediction value of the opening and closing roof; and a wind-light-thermal environment gene library is constructed based on the spatial wind speed distribution prediction value and the light-thermal environment simulation data;

[0083] In the embodiments of the present application, for each two-dimensional gray image, corresponding simulation data is sought, which is usually generated according to the geometric structure of the roof and environmental conditions. For example, the opening angle and the wind speed distribution corresponding to different positions of the roof need to be accurately matched. When matching, the size and resolution of the gray image should be consistent with the simulation data, and the matching relationship between them should be accurate. Each gray image should correspond to a set of wind speed distribution data to form an input-output image pair. Image matching techniques such as image registration can be used to ensure a one-to-one correspondence between the input image and the output simulation data. A generative adversarial network (GAN) model is used for wind speed distribution prediction. The GAN consists of a generator and a discriminator: the generator accepts a two-dimensional gray image of an open or closed roof as input and generates a wind speed distribution prediction image through a deep neural network. The discriminator judges the similarity between the generated wind speed distribution prediction image and the real wind speed distribution image, and provides feedback to help the generator gradually improve the quality of the generated image. A convolutional neural network (CNN) is used for feature extraction from the input two-dimensional gray image and for generating a predicted wind speed distribution image through deconvolution. A CNN architecture is also used to distinguish between generated wind speed distribution images and real images, and to continuously improve the discrimination ability by learning how to identify the features of real wind speed images. The performance of the generator is optimized by calculating the difference (such as structural similarity and root mean square error) between the generated wind speed image and the real image. The performance of the discriminator is optimized by calculating the true-false judgment error of the generated image and the real image. The input-output image pair (two-dimensional gray image of open or closed roof and simulated wind speed distribution data) is used to train the GAN model. During training, the generator continuously generates wind speed distribution prediction images, while the discriminator evaluates the authenticity of these images. By alternately optimizing the generator and the discriminator, the accuracy of wind speed distribution prediction is gradually improved. When the adversarial process of the generator and the discriminator reaches equilibrium, the model will be able to accurately generate wind speed distribution prediction images. The trained GAN model is used to input new two-dimensional gray images of open or closed roofs to generate corresponding wind speed distribution prediction images, which contain information such as wind speed distribution, airflow path, and wind speed intensity inside the roof. Wind speed distribution data under various open or closed roof structures are collected from the generated wind speed distribution prediction values, including wind speed distribution images under different opening ratios, angles, and wind speed conditions, as a basis for subsequent analysis. The wind speed distribution prediction image under each structure is associated with specific open or closed roof structure parameters such as opening ratio, inclination angle, and airflow path. The generated wind speed distribution prediction values are combined with the structure data of the open or closed roof, environmental conditions, and light-thermal environment simulation data to create a wind-light-thermal environment gene library containing various roof configurations and wind speed distribution characteristics and light-thermal characteristics.

[0084] Step S3: Obtain sports large space environment demand data; set wind, light and heat environment balance target according to the sports large space environment demand data, obtain wind, light and heat balance target parameters; based on the wind-light-heat environment gene library, the wind, light and heat balance target parameters are iteratively optimized by the genetic algorithm to generate the Pareto optimal solution of the opening and closing roof shape parameters under the typical meteorological conditions; based on the roof shape parameter change opening ratio of the Pareto optimal solution, the opening ratio and the corresponding indoor environment are analyzed to generate the roof opening and closing ratio-ventilation thermal comfort light environment influence curve;

[0085] In the embodiments of the present application, a plurality of wind-light-thermal balance objectives are set by combining the wind speed distribution prediction value in the wind-light-thermal environment gene library with the morphological characteristics of the roof and external environmental conditions. The main objectives include maximizing the daylighting coefficient, maximizing the natural ventilation speed, and minimizing the thermal comfort index. On the basis of the wind-light-thermal balance objectives, the structural parameters (such as the opening ratio, roof inclination angle, and wind speed distribution) of the opening and closing roof are constrained to ensure that these objectives are feasible within the design range. The spatial three-dimensional coefficient of the opening ratio is set as a constraint condition of the decision variable. For example, by optimizing the size and shape of the opening, the wind speed distribution, lighting conditions, and temperature control are ensured to achieve the best balance. Based on the above optimization objectives and decision variables, wind-light-thermal balance target parameters are generated through parameter combination analysis. These parameters can include the values of relevant indicators such as daylighting amount, ventilation speed, and thermal comfort index. A genetic algorithm (GA) is used to optimize the wind-light-thermal balance target parameters. The genetic algorithm searches for the best design parameters by simulating natural selection and genetic mechanisms. A plurality of solutions (i.e., different combinations of roof design parameters) are randomly generated as the initial individuals of the population. According to the different objectives set, a fitness function is designed to measure the quality of each solution. For example, the fitness function can score according to the degree of implementation of the wind-light-thermal balance objectives, and the optimal solution corresponds to the design of minimizing the thermal comfort index, maximizing the daylighting coefficient, and maximizing the natural ventilation speed. Through the selection mechanism, individuals with higher fitness are selected from the current population to generate new offspring through crossover (combining the characteristics of two individuals) and mutation (introducing small random changes). In each generation, individuals with lower fitness are eliminated, and individuals with higher fitness continue to reproduce. Through multiple generations of evolution, the best solution is gradually approached. Through multiple iterations of optimization, a set of optimal parameters is finally generated, which represent the specific scheme in the opening and closing roof design that can achieve the optimal wind-light-thermal balance. The output results of the genetic algorithm will generate a set of Pareto optimal solutions, which represent different design schemes that have achieved the best balance among multiple objectives. For example, one solution performs best in terms of lighting and ventilation, while another solution is more advantageous in terms of thermal comfort. The Pareto optimal solution is not a single optimal solution, but a set of solutions that cannot be further improved by any one objective without compromising the performance of other objectives. Based on the Pareto optimal solution, regression analysis is performed. Regression analysis models the relationship between the opening ratio in the morphological parameters of the roof and the wind speed distribution, daylighting amount, and thermal comfort index. Through the regression model, the influence of the opening ratio parameter on the wind speed distribution, daylighting, and thermal comfort can be quantified, thereby establishing an influence curve. According to the regression analysis results, a roof opening ratio-ventilation thermal comfort light environment influence curve is drawn.

[0086] Step S4: Perform climate zone sensitivity analysis on the roof opening ratio-ventilation thermal comfort light environment influence curve to generate sensitivity analysis results; and perform dynamic adjustment on the opening ratio of the opening and closing roof according to the sensitivity analysis results to perform the opening and closing roof sports building wind light and heat balance optimization and control operation. In the embodiment of the present application, the world is divided into multiple climate zones according to the climate conditions of different regions, such as tropical climate, temperate climate, and cold climate. The climate characteristics (such as temperature, humidity, precipitation, and wind speed) of each climate zone will have different effects on the design parameters of the opening and closing roof. On the basis of the roof opening ratio-ventilation thermal comfort light environment influence curve, the sensitivity of the roof form (such as opening ratio, ventilation opening distribution, roof angle, etc.) to the light and heat balance target (lighting, ventilation, thermal comfort) in different climate zones is analyzed. The parameters of each climate zone are simulated and adjusted to evaluate the influence of the opening and closing roof design on various performances (such as lighting, ventilation speed, and thermal comfort index) to generate sensitivity analysis results. For example, some climate zones rely more on natural ventilation, while others rely more on lighting. Therefore, the goal of the analysis is to determine which roof parameters have a greater impact on wind light and heat balance under different climate conditions. By changing each design parameter (such as opening ratio, ventilation opening position, etc.) one by one and observing its impact on wind light and heat balance targets, the sensitivity of the parameter is evaluated. Through multi-variable analysis by a system model, the optimal parameter combination of global performance is obtained. According to the climate zone sensitivity analysis results, the key parameters in the roof design are dynamically adjusted. In particular, the opening ratio, a key factor affecting ventilation and lighting, is optimized and adjusted. The goal of the adjustment is to ensure that the best wind light and heat balance is achieved under different climate conditions according to the needs of different climate zones, for example, in tropical climates, a larger opening ratio is needed to increase natural ventilation; while in cold climates, a smaller opening ratio is needed to keep the indoor warm. According to the real-time monitored environmental data (such as temperature, humidity, wind speed, etc.), the system automatically adjusts the opening ratio, and this adaptive adjustment can respond to changes in climate or changes in building internal demand. By introducing intelligent control algorithms (such as fuzzy control, neural network control, etc.), the opening ratio is precisely controlled according to environmental changes, optimizing the ventilation and lighting performance of the roof. During the dynamic adjustment of the opening ratio, comprehensive optimization is performed in combination with lighting, ventilation, and thermal comfort index to ensure that the wind light and heat balance target is achieved. By installing sensors and a data acquisition system, real-time monitoring of the lighting, ventilation, and thermal comfort of the roof is performed, and further adjustments are made according to the feedback data to ensure that the roof design always maintains the optimal light and heat balance state.

[0087] Preferably, step S1 comprises the following steps:

[0088] Step S11: Obtain opening and closing roof form data;

[0089] Step S12: Parametric modeling of the open-close roof form data, defining the dynamic adjustment range of the opening direction and opening ratio of the roof, obtaining the open-close roof geometric model;

[0090] Step S13: Gray image conversion of the open-close roof geometric model, generating the open-close roof two-dimensional gray image;

[0091] Step S14: Extracting the roof form of the open-close roof two-dimensional gray image, and using simulation software to perform CFD simulation and light-thermal environment simulation on the lower building space of the open-close roof, calculating the indoor and outdoor wind speed distribution and thermal comfort index, obtaining indoor and outdoor wind speed distribution data and thermal comfort index data;

[0092] Step S15: Daylight analysis of the lower building space of the open-close roof, generating daylight coefficient distribution data; converting the indoor and outdoor wind speed distribution data, daylight coefficient distribution data and thermal comfort index data into charts respectively, generating wind speed distribution chart, daylight coefficient distribution chart and thermal comfort index thermal map;

[0093] Step S16: Integrating the wind speed distribution chart, daylight coefficient distribution chart and thermal comfort index thermal map into the wind-light-thermal environment simulation data of the space covered by the open-close roof.

[0094] In the embodiments of the present application, data of the opening and closing roof shape is obtained from architectural design drawings, BIM models or other building information sources. This data usually includes the size, shape, support structure, material, opening and closing mechanism of the roof. If there is no ready-made data available, three-dimensional scanning or laser scanning technology can be used to obtain the geometric shape information of the roof. Geometric modeling of the roof is performed using modeling software such as AutoCAD, Rhino or Revit. The opening and closing direction of the roof (such as horizontal, vertical, etc.) and the opening ratio (such as opening percentage) can be dynamically adjusted according to actual needs. A dynamic model of the roof is established through a parametric design tool (such as the Grasshopper plug-in), ensuring that the opening and closing process of the roof can reflect the flexibility and adjustability of the opening and closing in reality. The model is converted into a software format suitable for simulation analysis (such as STL, OBJ format), providing geometric input for subsequent steps. The roof geometric model is converted into a two-dimensional image using Python software. The gray level of the image represents the morphological characteristics of the roof, with white representing the area without the roof (such as the opening), red representing the roof part, and gray and black representing the building entity part (such as the wall, the surrounding building). When generating the two-dimensional gray image, it is necessary to ensure that the resolution is high enough to preserve the details of the roof shape. The roof shape model is imported into CFD simulation software (such as ANSYS Fluent, OpenFOAM) for fluid dynamics simulation. During the simulation process, boundary conditions such as wind speed and air pressure are set to calculate the fluid dynamics characteristics such as wind speed distribution and wind pressure of the opening and closing roof under different opening and closing states. A thermal environment simulation software (such as COMSOL, EnergyPlus) is used to simulate the process of light, thermal radiation, convection and conduction. The influence of the roof shape on thermal comfort is calculated to obtain thermal comfort index data (such as temperature, humidity, thermal load, etc.). A light analysis tool (such as DIALux, Radiance) is used to analyze the daylighting of the space under the opening and closing roof. The influence of the opening ratio and opening angle of the roof on indoor daylighting is analyzed, and the daylight factor, i.e. the light intensity per unit area when the roof is open, is generated. The daylight factor is an important indicator of the uniformity of light distribution, which can reflect the optimization effect of the roof opening and closing state on indoor daylighting. Based on the CFD simulation results, wind speed data is converted into charts or heat maps to show the indoor and outdoor wind speed distribution under different opening and closing states of the roof. According to the light and thermal environment simulation results, a heat map of thermal comfort indicators is generated to visually display the influence of the opening and closing roof under different opening and closing states on indoor and outdoor thermal comfort. The results of daylight analysis are presented in the form of tables or charts to show the daylight factor under different opening angles and opening ratios. The wind speed distribution chart, thermal comfort indicator heat map and daylight factor are integrated to finally generate the CFD simulation data and light and thermal environment simulation data of the opening and closing roof.

[0095] Preferably, the daylight analysis of the space under the opening and closing roof comprises:

[0096] The light area of the lower part of the open-close roof is extracted to obtain a light area image of the lower part of the open-close roof; the height angle and the azimuth angle of the sun are calculated to obtain the height angle and the azimuth angle of the sun.

[0097] Based on the height angle and the azimuth angle of the sun, the light propagation simulation of the light area of the open-close roof is performed to generate the simulation data of the propagation path of the sunlight; the irradiation intensity of each pixel point in the light area image of the lower part of the open-close roof is calculated according to the simulation data of the propagation path of the sunlight to obtain a local solar radiation intensity map.

[0098] The actual irradiation area ratio of the light area image of the lower part of the open-close roof is calculated by using the local solar radiation intensity map to obtain a light coefficient.

[0099] In the embodiment of the application, the two-dimensional gray image of the open-close roof is processed by using image processing software or programming tools (such as MATLAB and Python OpenCV). The gray image is binarized, and the open area and the non-open area are distinguished according to the set color value (for example, the roof covering area is red, and the open area is white, the wall and the surrounding building are gray and black). A suitable threshold segmentation method (such as Otsu method) is used for automatic segmentation, or the threshold is manually adjusted to ensure accurate extraction of the light area. The solar height angle (α) represents the angle between the sunlight and the ground, and the calculation formula is: α = arcsin (sin (δ) · sin (φ) + cos (δ) · cos (φ) · cos (H)); wherein, δ is the solar declination angle, φ is the latitude of the observation point, and H is the hour angle (i.e. the time difference between the current time and the local noon). The solar azimuth angle (Az) represents the angle of the sun relative to the south direction, and the calculation formula is: where H is the hour angle, δ is the declination angle, and φ is the latitude of the observation point. The solar elevation and azimuth angles at different times and dates can be calculated using solar position calculation tools such as SolarCalc or custom programs. Based on the solar elevation and azimuth angles, perform a ray propagation simulation of the daylighting area. Use a ray tracing algorithm such as Monte Carlo Ray Tracing to simulate the propagation of light from the sun to the roof opening. Professional lighting simulation software such as Radiance, Daysim, and LDT tools can be used for ray propagation simulation, which can accurately simulate the propagation path of light in combination with the solar position and the geometry of the roof opening. Treat the sun as a point light source and set its elevation and azimuth angles. Calculate the propagation path of the light based on the geometry of the roof. During the ray propagation process, consider factors such as air scattering, reflection, and absorption to attenuate the light. According to the simulated solar light propagation path, calculate the irradiance intensity of each pixel in the image of the daylighting area of the retractable roof. For each pixel, calculate the number of solar rays passing through the pixel and the intensity of the light. Irradiance intensity is usually related to the incident angle of solar radiation, reflection, refraction, and attenuation effects. The irradiance intensity (I) can be calculated using the following formula: I = I0·cos(θ); where I0 is the solar intensity and θ is the angle between the light and the surface normal. Use the calculated irradiance intensity values to assign the corresponding light intensity values to each pixel and generate a local solar intensity map. By thresholding the local solar intensity map, filter out the effective irradiation area. The area with an irradiance intensity exceeding a certain threshold is the effective daylighting area. Calculate the area of the effective irradiation area in the local solar intensity map to obtain the actual irradiation area. The total daylighting area is the sum of the areas of all roof openings in the daylighting area image. The illumination factor (IF) is defined as the ratio of the actual irradiation area to the total daylighting area. The illumination factor generated as a lighting performance indicator for roof design can be further used for building design optimization.

[0100] As an example of the present application, reference is made to Figure 2 In this example, the step S2 includes:

[0101] Step S21: Image pairing of the retractable roof two-dimensional gray-scale image and the wind speed distribution map of the lower space of the roof is performed to obtain an input-output image pair. Data set division is performed on the input-output image pair to generate a model training set, a model test set, and a model validation set.

[0102] Step S22: constructing a Pix2Pix network framework; inputting a model training set, a model test set and a model validation set into the Pix2Pix network framework for training, testing and validation to obtain a Pix2Pix algorithm model; inputting a preset two-dimensional image of a design of an openable roof into the Pix2Pix algorithm model for indoor wind speed prediction of a lower space of the openable roof to generate an indoor predicted wind speed map of the lower space of the openable roof;

[0103] Step S23: performing numerical matrix conversion on the indoor predicted wind speed map of the lower space of the openable roof to obtain a predicted value of a spatial wind speed distribution of the openable roof, and extracting an average wind speed value of a key area of the sports building from the predicted value of the spatial wind speed distribution of the openable roof to obtain the average wind speed value of the key area;

[0104] Step S24: performing parameter correlation on an opening ratio of the openable roof according to the average wind speed value of the key area, and constructing a gene library by using the correlated parameters and light-thermal environment simulation data to obtain a wind-light-thermal environment gene library.

[0105] In the embodiments of the present application, the two-dimensional gray-scale image of the opening and closing roof and the corresponding wind speed distribution graph are paired one by one. Each two-dimensional gray-scale image corresponds to a wind speed distribution graph, wherein the two-dimensional gray-scale image is taken as input and the wind speed distribution graph is taken as output. It is necessary to ensure that the sizes of the images are consistent, and each pair of images represents the simulation results under the same opening and closing roof state. The image pairs are divided into data sets, usually 70% for the training set, 15% for the test set, and 15% for the verification set. The data set division can adopt a random division method to ensure the representativeness and diversity of the training set, test set and verification set. The training set is used for image pair data of model training, the test set is used for image pair data of model evaluation, and the verification set is used for image pair data of model verification and adjustment in the training process. Pix2Pix is an image-to-image conversion model suitable for image generation and wind speed prediction tasks. Its framework includes a generator and a discriminator: the generator accepts the two-dimensional gray-scale image of the opening and closing roof as input to generate the predicted wind speed distribution graph. The discriminator discriminates between the generated wind speed distribution graph and the real wind speed distribution graph, evaluates the authenticity of the generated image, and feeds back to the generator for optimization. Use a deep learning framework (such as TensorFlow, PyTorch) to build a Pix2Pix network. The network includes: an input layer: a two-dimensional gray-scale image of an opening and closing roof; a convolutional layer: used to extract feature information of the input image; a skip connection (Skip Connections): in order to preserve detailed information, the output of the shallow layer of the generator is spliced with the output of the deep layer; an output layer: generating a wind speed distribution graph. The divided training set data is input into the Pix2Pix model for training. By minimizing the loss function of the generator and the discriminator, the accuracy of the generated wind speed distribution graph is continuously optimized. The opening and closing roof predicted wind speed graph generated by the Pix2Pix network is converted into a numerical matrix. This can be done by mapping the image pixel value to the wind speed distribution to extract the wind speed value corresponding to each pixel point into a numerical matrix. The numerical matrix represents the wind speed distribution of the opening and closing roof under different opening and closing states, including the wind speed of different regions. Determine the key areas in the sports building (such as sports venues, spectator stands, entrances, etc.). Extract the wind speed matrix part of each key area and calculate the average wind speed value of the area. Through statistical methods (such as regression analysis) or machine learning methods (such as support vector machines, decision trees, etc.), the opening ratio and the average wind speed value of the key area are analyzed. This analysis can reveal the relationship between the opening ratio and the wind speed of the key area of the sports building. For example, analyze the wind speed trend under different opening ratios when the opening and closing roof opening ratio changes from 0% to 100%. Integrate the light and heat environment data related to the opening ratio (such as temperature, humidity, thermal comfort, etc.) into the parameter correlation model. Based on environmental factors such as wind speed, light, and thermal comfort, a multi-dimensional wind-light-heat environment gene library is established. Each data point in the gene library includes a mapping of an opening and closing roof state (opening ratio) and corresponding environmental characteristics.

[0106] Preferably, the step S22 comprises:

[0107] constructing a Pix2Pix network framework, wherein the Pix2Pix network framework comprises a generator and a discriminator;

[0108] training parameter setting of the Pix2Pix network framework, generating a training parameter setting value, wherein the training parameter setting comprises a learning rate setting, a convolutional layer number setting, an epoch setting and an optimizer setting;

[0109] inputting a model training set into the Pix2Pix network framework to perform wind speed prediction through the generator in the Pix2Pix network framework, generating an initial open-close roof predicted wind speed map, and using a model test set to compare the initial open-close roof predicted wind speed map with a real result image through the discriminator in the Pix2Pix network framework;

[0110] performing structure similarity and root mean square error calculation on the initial open-close roof predicted wind speed map to obtain wind speed prediction map accuracy;

[0111] performing prediction map convergence on the wind speed prediction map accuracy through a model validation set to obtain a Pix2Pix algorithm model;

[0112] inputting a preset open-close roof design two-dimensional image into the Pix2Pix algorithm model to perform indoor wind speed prediction of a lower space of a roof, and generating an open-close roof lower space indoor predicted wind speed map.

[0113] In the embodiment of the present application, the Pix2Pix network framework is based on the generator (Generator) and discriminator (Discriminator) in the generative adversarial network (GAN). The generator is responsible for generating the target image (i.e. the predicted wind speed map) from the input image, while the discriminator is responsible for evaluating the authenticity of the generated wind speed map, comparing the difference with the actual wind speed map and optimizing the generator. The generator usually adopts the U-Net architecture, which is a symmetrical encoder-decoder structure that can effectively process spatial information in the image and transfer detailed features at the jump connection to improve the accuracy of the prediction result. The input is a two-dimensional gray image of the open-close roof, wind speed distribution map, thermal comfort map and other information. All input images will be subjected to convolution operation for feature extraction, and the predicted wind speed distribution map will be generated. Selecting a suitable learning rate has an important influence on the convergence speed and performance of the network. Generally, setting a low learning rate (such as 0.0002) can prevent gradient explosion or disappearance, and the number of convolution layers of the generator and discriminator should be set according to the complexity of the network. Generally, the number of convolution layers of the generator can be from 4 to 8, and the number of convolution layers of the discriminator is similar. The more layers, the stronger the expression and processing capacity of the network. The number of iterations of training. It is generally recommended to set 20-50 epochs, which can be adjusted according to the training of the model. More epochs can give the network more opportunities for optimization. The Adam optimizer is used because it can effectively adjust the network parameters, especially when dealing with sparse gradients. The parameters are set as: beta_1=0.5, beta_2=0.999, and the learning rate is usually set to 0.0002. By inputting the image data in the training set into the generator in the Pix2Pix network framework. Each pair of input-output image pairs will be processed by the generator to generate an initial open-close roof predicted wind speed map from the geometric image of the open-close roof, wind speed distribution map and thermal comfort map and other information. The generator learns from the image pairs in the training set and generates the predicted wind speed map. The initial open-close roof predicted wind speed map deviates from the actual wind speed map. Using the data in the model test set, the discriminator compares the generated initial wind speed map with the real wind speed map. The discriminator evaluates the authenticity of the generated wind speed map and provides feedback information to the generator to promote the optimization of the generator. The following two common accuracy evaluation methods are used to evaluate the generated initial open-close roof predicted wind speed map: The structural similarity index (SSIM) is an index for measuring the similarity between two images, which focuses on the structural similarity and can effectively reflect the perceptual quality of the image. Using SSIM can compare the similarity between the generated wind speed map and the real wind speed map, and get a value in the range of [0, 1]. The closer the value is to 1, the more similar the two images are. The root mean square error is a commonly used method to calculate the difference between the predicted value and the true value, which can quantify the error degree of the generated image. The data in the model validation set is used to verify the accuracy of the generated wind speed map.By comparing the wind speed diagram in the verification set with the generated wind speed diagram, it is confirmed whether the model converges and whether the accuracy of the generated wind speed diagram meets the expectation. If the accuracy (such as SSIM and RMSE) of the model in the verification set meets the expectation and converges, it indicates that the wind speed prediction model is stable enough, and the final Pix2Pix algorithm model can be generated. The preset two-dimensional image of the openable roof design is input into the Pix2Pix algorithm model to predict the indoor wind speed of the lower space of the openable roof, and the predicted wind speed diagram of the lower space of the openable roof is generated.

[0114] As an example of the present application, reference is made to Figure 3 In this example, the step S3 includes:

[0115] Step S31: Obtain sports large space environment demand data; set the wind-light-heat balance target of the openable roof space according to the sports large space environment demand data to obtain the wind-light-heat balance target parameter;

[0116] Step S32: Perform diversity evaluation on the wind-light-heat balance target parameter through the NSGA-III algorithm, and perform dynamic parameter adjustment on the wind-light-heat balance target parameter based on the diversity evaluation result to generate the wind-light-heat balance target adjustment parameter;

[0117] Step S33: Perform iterative optimization on the wind-light-heat balance target parameter through the NSGA-III algorithm based on the wind-light-heat environment gene library, and eliminate extreme solutions to generate the Pareto optimal solution;

[0118] In the embodiments of the present application, by obtaining the sports large space environment demand data, the wind, light and thermal balance target is set based on the daylighting coefficient according to the wind speed, light intensity and thermal comfort data in the sports large space environment demand data under different opening and closing roof states. The setting of the wind, light and thermal balance target parameters includes: determining the ideal wind speed range to ensure the air circulation and comfort of the interior of the sports building. The appropriate daylighting coefficient is set to ensure that the natural lighting inside the building meets the design requirements. According to the thermal comfort standard (such as UTCI, PMV-PPD index), the appropriate thermal comfort target is set. The wind, light and thermal balance target parameters include the specific numerical range of the above targets, and these values are multivariate and interrelated. NSGA-III (Non-dominated Sorting Genetic Algorithm III) is a multi-objective optimization algorithm specially used for processing multi-objective problems, and is especially suitable for high-dimensional optimization tasks. The algorithm evaluates the quality of the solution through non-dominated sorting and crowding distance, and can effectively find the Pareto frontier solution. The wind, light and thermal balance target parameters are input, and the diversity of the target parameters is evaluated by using the NSGA-III algorithm. The results of the diversity evaluation will reveal the trade-off and distribution of wind speed, lighting and thermal comfort under different opening and closing roof design parameters. According to the diversity evaluation results, the wind, light and thermal balance target parameters are dynamically adjusted. For example, if the wind speed is too high and the lighting is insufficient, the algorithm will adjust the parameters to reduce the wind speed and increase the daylighting coefficient to achieve balance. The adjusted parameters are used as new wind, light and thermal balance target adjustment parameters, and these adjustments can be optimized through multiple iterations to approach the optimal target. Using the adjusted wind, light and thermal balance target parameters, multiple rounds of iterative optimization are performed by the NSGA-III algorithm to gradually adjust the parameter combination. Through each iteration, a new set of solutions is generated, and the quality of the solutions is evaluated (for example, through the distribution of the Pareto frontier). In each iteration, extreme solutions that do not meet the design target (for example, extremely low daylighting coefficient or extremely high wind speed) are removed to ensure that the optimization results are within an acceptable range. The removal of extreme solutions can be completed by setting a threshold or post-processing analysis. After multiple iterations, a set of Pareto optimal solutions is selected as the optimal design scheme. These solutions represent the best trade-off between wind speed, lighting and thermal comfort. Based on the Pareto optimal solution and the morphological data of the opening and closing roof, a regression model is used to establish the relationship between wind speed, lighting and thermal comfort and the roof opening and closing ratio parameters. The regression equation is: Y=β0+β1X1+β2X2+…+β n X n , where Y is the target variable (wind speed, lighting, thermal comfort), X1, X2, …, X n are design parameters (such as roof opening ratio), β0, β1, …, β nis the regression coefficient. According to the regression equation, the following three curves are generated: roof opening ratio-ventilation influence curve, which represents the influence of opening ratio on wind speed; roof opening ratio-natural lighting influence curve, which represents the influence of roof opening ratio parameter on indoor natural lighting; and roof opening ratio-thermal comfort influence curve, which represents the influence of roof opening ratio parameter on thermal comfort (e.g. temperature, humidity, UTCI value, etc.). The above three curves are integrated to generate a roof opening ratio-ventilation thermal comfort light environment influence curve, which comprehensively reflects the influence of wind speed, lighting and thermal comfort, and provides a comprehensive design reference. The three curves can be represented by a multi-dimensional function: E = f(v, l, t); where E is the comprehensive environmental influence, v is the wind speed, l is the lighting coefficient, and t is the thermal comfort.

[0119] Preferably, step S31 comprises:

[0120] Obtaining sports large space environment demand data;

[0121] Defining optimization targets according to the sports large space environment demand data, wherein the optimization targets include maximizing the lighting coefficient, maximizing the natural ventilation speed, and minimizing the thermal comfort index;

[0122] Conducting decision variable constraints on the maximizing the lighting coefficient, maximizing the natural ventilation speed, and minimizing the thermal comfort index, and limiting the spatial three-dimensional coefficient of the opening ratio;

[0123] Based on the optimization targets and the decision variables, parameter combination is performed on the wind speed distribution prediction value of the opening and closing roof to obtain wind-light-thermal balance target parameters.

[0124] In the embodiment of the present application, by acquiring the sports large space environment demand data, the wind-light-heat environment gene library and the daylighting coefficient can be combined to determine the optimization target, which are respectively maximizing the daylighting coefficient, maximizing the natural ventilation speed and minimizing the thermal comfort index, to ensure the balance between each target in the optimization process, and to make the roof structure meet the actual feasible design specification, and each target needs to be set with decision variable constraints: the decision variables include the opening ratio, opening direction, roof shape, roof inclination angle and other parameters that affect daylighting, ventilation and thermal comfort of the roof. The maximum and minimum values of the roof opening ratio are limited. According to the building structure and safety requirements, the opening ratio should not exceed a certain proportion, so as not to affect the stability or energy efficiency of the building. The spatial three-dimensional coefficient of the opening ratio is set to define the range of the roof opening ratio to control the design space in the optimization process. There is a certain trade-off between the ventilation rate and the daylighting coefficient. For example, too large opening will affect the wind speed or thermal comfort, and it is necessary to ensure the balance between ventilation and daylighting. The constraint of the maximum daylighting coefficient target is set: Cmax≤1, where Cmax represents the maximum daylighting coefficient. The constraint of the maximum ventilation speed is set: Vmax≥0, where Vmax is the maximum ventilation speed. The constraint of the minimum thermal comfort index is set: Hmin≤3, where Hmin is the minimum thermal comfort index (such as PMV value). Using the wind speed distribution prediction in the early stage (by CFD simulation and other methods), the wind speed distribution data of the roof under different opening ratios is obtained. The wind speed distribution data considers the influence of different opening shapes, roof angles and other factors on indoor and outdoor wind speed. The wind speed distribution prediction value is combined with the optimization target of daylighting coefficient and thermal comfort index to form the light-heat balance target parameter. The result of parameter combination considers the following aspects: the interaction between wind speed and temperature to ensure that the retractable roof structure design can maximize the ventilation speed and reduce heat accumulation. The mutual influence between daylighting coefficient and sunshine path to ensure that natural light is increased without affecting indoor heat radiation level. The correlation between thermal comfort index and roof opening ratio, shape to ensure that the final scheme meets the comfort requirement by calculating the thermal comfort value under each design scheme. In the optimization process, according to the combination of the objective function, the final wind-light-heat balance target parameter is obtained. The parameter set is used for the execution of subsequent optimization algorithms (such as NSGA-III algorithm) to ensure the balance between different optimization targets to obtain the optimal retractable roof design.

[0125] Preferably, step S32 comprises the following steps:

[0126] Step S321: setting the population size, crossover rate and mutation rate of the NSGA-III algorithm, and performing non-dominated sorting on the wind-light-heat balance target parameter according to the population size, crossover rate and mutation rate to obtain the wind-light-heat balance sorting result;

[0127] Step S322: Calculate the crowding distance of the wind-solar-thermal balance sorting result to obtain wind-solar-thermal balance sample distance data;

[0128] Step S323: Perform diversity evaluation on the wind-solar-thermal balance target parameter using the wind-solar-thermal balance sorting result and the wind-solar-thermal balance sample distance data, and perform dynamic parameter adjustment on the wind-solar-thermal balance target parameter based on the diversity evaluation result to generate wind-solar-thermal balance target adjustment parameter, wherein the formula of the diversity evaluation is as follows:

[0129]

[0130] In the formula, D total is the diversity evaluation result, a is the influence coefficient of the diversity measurement of the adjustment target space on the total diversity evaluation, β is the influence coefficient of the physical space sample distance measurement of the adjustment target space on the total diversity evaluation, D is the diversity measurement of the population in the target space, s i is the first wind-solar-thermal balance sample, s j is the second wind-solar-thermal balance sample, dist(s i ,s j ) is the physical distance between s i and s j , and N is the sample quantity.

[0131] In this embodiment of the invention, an appropriate population size N is set, which determines the number of individuals in each generation. The chosen population size should be able to explore the optimization space within a suitable time frame, while avoiding excessive computational complexity. The crossover rate Cr refers to the probability of performing a crossover operation in the population. An appropriate crossover rate can help the algorithm search a wider solution space. The mutation rate Mr refers to the probability of performing a mutation operation on an individual. Mutation operations help increase the diversity of the population and avoid the occurrence of local optima. According to the optimization objective of the wind-solar-thermal balance target parameters, the non-dominated sorting method is used to sort the population. Non-dominated sorting is a multi-objective optimization method that can distinguish the optimal solution (Pareto optimal solution) from other solutions through sorting. In this process, individuals in the population will be divided into different levels (layers). The first level is non-dominated solutions, the second level is solutions dominated by the first level, and so on. Crowding distance calculation is used to evaluate the distribution of solutions in the target space and avoid the population from clustering in a certain location. Solutions with larger crowding distances are generally considered to be more diverse and have a greater chance of survival in the next generation. For each non-dominated solution, the distance between it and its neighboring solutions is calculated to obtain the crowding distance Di. The calculation method is as follows: For each objective, individuals are sorted in ascending or descending order of objective value. For each individual, the distance between it and its left and right neighbors is calculated. Individuals with greater distances will have larger crowding distances. The crowding distance Di for each individual is obtained as a key factor in diversity assessment. Based on the wind-solar-heat balance ordination results and crowding distance data, the following formula is used for diversity assessment: In the formula, D total For the diversity assessment results, α is the influence coefficient of adjusting the diversity measure in the target space on the overall diversity assessment, β is the influence coefficient of adjusting the physical spatial sample distance measure in the target space on the overall diversity assessment, D is the diversity measure of the population in the target space, and dist(s) is the diversity measure of the population in the target space. i ,s j ) represents the wind-solar-thermal balance sample s i and s j The physical distance between them, s i For the first wind-solar-thermal balance sample, s j This is the second wind-solar-thermal balance sample. The physical distance between each sample and other samples is calculated, and crowding distance is used to enhance the evaluation effect. The diversity of the target space is weighted and summed with the distances in the physical space to obtain the diversity evaluation result of the solar-thermal balance target parameters. Based on the diversity evaluation result, the solar-thermal balance target parameters are dynamically adjusted. By adjusting the sample distances between the target space and the physical space, excessive concentration or duplication of solutions can be effectively avoided, maintaining the diversity of the solution space.

[0132] Preferably, step S4 comprises the following steps:

[0133] Step S41: Climate zone sensitivity analysis is performed on the roof opening ratio-ventilation thermal comfort light environment influence curve to generate sensitivity analysis results;

[0134] Step S42: Climate zone testing is performed on the opening roof shape data according to the sensitivity analysis results to generate applicability range evaluation data;

[0135] Step S43: Dynamic control adjustment is performed on the opening ratio through the applicability range evaluation data to perform the wind light thermal balance optimization operation of the opening roof sports building.

[0136] In the embodiments of the present application, different climate zones (such as tropical, temperate, and polar zones) are divided according to climatology data. The characteristics of each climate zone include temperature, humidity, wind speed, and sunlight, etc. Meteorological data of each climate zone is collected as the basic data source for analysis. The roof opening ratio-ventilation thermal comfort light environment influence curve shows the influence of roof morphology on indoor thermal comfort, especially for ventilation and lighting efficiency under different climate conditions. Sensitivity analysis methods (such as local sensitivity analysis or global sensitivity analysis) are used to evaluate the degree of influence of climate change on the thermal comfort light environment influence curve. The influence of opening ratio, roof material, and ventilation design on comfort is focused on, and the sensitivity analysis results are generated to provide the basis for optimizing the roof structure for each climate zone by quantifying the changes in thermal comfort in each climate zone. Repeat experiments are conducted for different climate zones to collect the changes in the roof-ventilation thermal comfort light environment influence curve under various climate conditions. Statistical methods are used to calculate the volatility of the curve to evaluate its stability and sensitivity. According to the sensitivity analysis results obtained in S41, a roof structure test scheme is developed for different climate zones, covering multiple parameters such as opening ratio. By constructing typical test cases for climate zones, the ventilation and lighting effects of the roof under different climate conditions are simulated. On the numerical simulation platform, a three-dimensional model of the opening and closing roof is established, and wind-light-thermal balance calculations are performed under different climate conditions to generate applicability data for the roof morphology. Through actual operation simulation, performance indicator data such as ventilation, lighting, and thermal comfort of the opening and closing roof in different climate zones are obtained. The main factors considered include thermal comfort index (such as PMV index), ventilation speed, and lighting efficiency. The test results of different climate zones are comprehensively analyzed to generate applicability evaluation data, which evaluates the applicability of the opening and closing roof structure in different climate zones. This includes the performance of thermal comfort, energy efficiency, and ventilation effect of the opening and closing roof under specific climate conditions. Based on the applicability evaluation data generated in S42, a dynamic control strategy is developed. For each climate zone and different thermal comfort requirements, the opening ratio of the opening and closing roof is adjusted to optimize ventilation, lighting, and thermal comfort. The adjustment of the opening ratio can be dynamically adjusted based on real-time climate data, such as increasing the opening ratio in hot weather to enhance ventilation, and reducing the opening ratio in cold weather to reduce heat loss. Using control theory and optimization algorithms (such as PID control, fuzzy control, etc.), the opening ratio of the roof is dynamically adjusted according to external environmental conditions (such as temperature, humidity, wind speed, etc.). Combined with the different needs of climate zones, a dynamic control scheme suitable for different seasons and weather conditions is developed to ensure the optimization of wind-light-thermal balance of the roof under various environmental conditions. The dynamic control system is implemented in actual buildings to collect real-time climate and indoor environment data, adjust the opening ratio of the opening and closing roof, and optimize the wind-light-thermal balance inside and outside the building.

[0137] Preferably, step S42 comprises the following steps:

[0138] Step S421: According to the sensitivity analysis result, the key parameter of the open-close roof shape data is identified, so as to obtain the sensitive parameter data; the dynamic response analysis is carried out on the sensitive parameter data, so as to obtain the response characteristic data;

[0139] Step S422: The open-close roof shape data is mapped by the response characteristic data, so as to obtain the partition data; the climate element simulation is carried out on the open-close roof shape data according to the partition data, so as to obtain the climate simulation data;

[0140] Step S423: The performance test is carried out on the open-close roof shape data based on the climate simulation data, so as to obtain the test data; the environmental adaptability range analysis is carried out on the open-close roof shape data by using the test data, so as to generate the applicable range evaluation data.

[0141] In the embodiments of the present application, the key parameters affecting the light-thermal balance of the openable roof in different climate zones are determined based on the sensitivity analysis results of step S41, including the opening ratio, roof material, wind speed, solar radiation intensity, indoor temperature, etc. The sensitivity analysis method (such as local sensitivity analysis, global sensitivity analysis) is used to evaluate the sensitivity of each parameter, and the key parameters that have the greatest impact on the roof performance are identified. Dynamic response analysis is performed on the sensitive parameter data to simulate the dynamic response of the openable roof under different climate conditions. The response characteristics of the openable roof are analyzed through dynamic simulation, such as the response of the roof ventilation volume when the opening ratio changes, the influence of the material thermal conductivity on the indoor temperature, etc. Response characteristic data is generated, such as the change trend of the roof ventilation efficiency, thermal comfort, etc. under different wind speeds and different opening ratios. According to the characteristics of different climate zones, the openable roof form data is mapped to the climate zones. The characteristics of each climate zone, such as temperature, humidity, precipitation, wind speed, etc. will have different effects on the ventilation effect and thermal comfort of the roof. The key parameters of the openable roof are combined with the characteristics of the climate zones to form climate zoning mapping data. The data of each zone should include the climate factors, wind speed, lighting demand, temperature and humidity conditions, etc. in the region. Based on the climate zoning mapping data, climate element simulation is performed, and the performance of the roof structure under different climate zones is simulated through climate simulation tools. The simulation content includes the influence of wind speed, solar radiation, air humidity, etc. under different climate conditions, and climate simulation data is generated, which reveals how the openable roof responds to different climate factors under different climate conditions, helping to better understand the adaptability of roof design to climate conditions. Climate simulation data is used to perform various performance tests on the openable roof, including thermal comfort, ventilation efficiency, lighting level, etc. The actual performance of the roof under different climate conditions is tested through numerical simulation, wind tunnel experiment, etc. For example, the relationship between the opening ratio and the ventilation efficiency of the roof is tested under simulated tropical climate; the influence of the thermal conductivity of the roof material on the thermal comfort is tested under cold climate, etc. According to the test results, environmental adaptability range analysis is performed to evaluate the adaptability of the openable roof under different climate conditions. For example, whether the ventilation and lighting efficiency of the openable roof in temperate climate is sufficient, whether the thermal comfort in cold climate meets the requirements, etc. Application range evaluation data is generated, which provides guidance for the design of the openable roof and determines its use range in different climate zones, helping to select the most suitable design scheme.

[0142] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, the scope of the present application being defined by the appended claims and not by the above description, therefore all variations falling within the meaning and scope of the equivalent elements of the application file are intended to be included within the present application.

[0143] The foregoing is considered as illustrative only of the principles of the application. Numerous modifications and changes will readily occur to those skilled in the art, and it is intended to embrace all such modifications and changes that fall within the scope of the application. Accordingly, the application is not to be restricted in scope to the specific embodiments disclosed herein but is to be accorded the full scope that the principles and novel features request appropriately granted.

Claims

1. A method for optimizing the light-thermal balance of an open-close roof sports building based on a generative adversarial network and a genetic algorithm, characterized in that, The method comprises the following steps: Step S1: acquiring the opening and closing roof shape data; Step S2: performing image pairing on the opening and closing roof two-dimensional gray image and the wind-light-thermal environment simulation data to obtain an input-output image pair; using a generative adversarial network to predict the wind speed distribution of the input-output image pair to generate a spatial wind speed distribution prediction value of the opening and closing roof; Step S3: acquiring the sports large space environment demand data; setting a wind-light-thermal balance target for the opening and closing roof space according to the sports large space environment demand data to obtain a wind-light-thermal balance target parameter; performing iterative optimization on the wind-light-thermal balance target parameter based on the wind-light-thermal environment gene library through a genetic algorithm to generate a Pareto optimal solution of the opening and closing roof shape parameter under typical meteorological conditions; performing regression analysis on the opening ratio and the corresponding indoor environment based on the change of the roof shape parameter of the Pareto optimal solution to generate a roof opening ratio-ventilation thermal comfort light environment influence curve; wherein, step S3 comprises the following steps: Step S31: acquiring the sports large space environment demand data; setting a wind-light-thermal balance target for the opening and closing roof space according to the sports large space environment demand data to obtain a wind-light-thermal balance target parameter; Step S32: performing diversity evaluation on the wind-light-thermal balance target parameter through an NSGA-III algorithm, and performing dynamic parameter adjustment on the wind-light-thermal balance target parameter based on the diversity evaluation result to generate a wind-light-thermal balance target adjustment parameter; wherein, step S32 comprises the following steps: Step S321: setting the population size, crossover rate and mutation rate of the NSGA-III algorithm, and performing non-dominated sorting on the wind-light-thermal balance target parameter according to the population size, crossover rate and mutation rate to obtain a wind-light-thermal balance sorting result; Step S322: performing crowded distance calculation on the wind-light-thermal balance sorting result to obtain wind-light-thermal balance sample distance data; Step S323: performing diversity evaluation on the wind-light-thermal balance target parameter using the wind-light-thermal balance sorting result and the wind-light-thermal balance sample distance data, and performing dynamic parameter adjustment on the wind-light-thermal balance target parameter based on the diversity evaluation result to generate a wind-light-thermal balance target adjustment parameter, wherein the formula of the diversity evaluation is as follows: Step S33: performing iterative optimization on the wind-light-thermal balance target adjustment parameter based on the wind-light-thermal environment gene library through the NSGA-III algorithm, and removing extreme solutions to generate a Pareto optimal solution; ​ ​ In the formula, D total is the diversity evaluation result, a is the influence coefficient of the diversity measurement of the target space on the total diversity evaluation, β is the influence coefficient of the physical space sample distance measurement of the target space on the total diversity evaluation, D is the diversity measurement of the population in the target space, s i , s j is the wind-solar-thermal balance sample, dist(s i , s j ) is the physical distance between s i and s j , and N is the sample quantity; ​ Step S34: Based on the Pareto optimal solution, the opening ratio of the roof shape parameter change is started, the multiple linear regression equation of the different opening ratio of the retractable roof and the corresponding indoor environmental performance is established, the retractable roof opening ratio-ventilation influence curve, the retractable roof opening ratio-natural lighting influence curve and the retractable roof opening ratio-thermal comfort influence curve are generated respectively, and are integrated into the retractable roof opening ratio-ventilation thermal comfort light environment influence curve; Step S4: Climate zone sensitivity analysis is performed on the retractable roof opening ratio-ventilation thermal comfort light environment influence curve, and a sensitivity analysis result is generated; according to the sensitivity analysis result, the opening ratio of the retractable roof is dynamically adjusted to perform retractable roof sports building wind light and heat balance optimization and control operation.

2. The method of claim 1, wherein the method is a method of optimizing the light-thermal balance of an open-close roof sports building based on a generative adversarial network and a genetic algorithm, characterized in that, Step S1 includes the following steps: Step S11: Obtain the retractable roof shape data; Step S12: Parameterize the retractable roof shape data, define the dynamic adjustment range of the roof opening direction and the opening ratio, and obtain the retractable roof geometric model; Step S13: Convert the retractable roof geometric model into a gray scale image to generate a two-dimensional gray scale image of the retractable roof; Step S14: Extract the roof shape of the two-dimensional gray scale image of the retractable roof, and use simulation software to perform CFD simulation and light and heat environment simulation on the lower building space under the retractable roof, calculate the indoor and outdoor wind speed distribution and thermal comfort index, and obtain indoor and outdoor wind speed distribution data and thermal comfort index data; Step S15: Perform daylight analysis on the lower building space under the retractable roof to generate daylight coefficient distribution data; convert the indoor and outdoor wind speed distribution data, daylight coefficient distribution data and thermal comfort index data into charts respectively to generate wind speed distribution chart, daylight coefficient distribution chart and thermal comfort index heat map; Step S16: Integrate the wind speed distribution chart, daylight coefficient distribution chart and thermal comfort index heat map into wind light and heat environment simulation data of the space covered by the retractable roof.

3. The method of claim 2, wherein the method is characterized by, The daylight analysis on the lower building space under the retractable roof includes: Performing light extraction on the lower building space under the retractable roof to obtain a retractable roof lower space light extraction image; calculating the sun elevation angle and azimuth angle of the retractable roof lower space light extraction image to obtain the sun elevation angle and the sun azimuth angle; Based on the sun elevation angle and the sun azimuth angle, the light propagation simulation of the retractable roof light extraction area is performed to generate sun light propagation path simulation data; the irradiation intensity of each pixel point in the retractable roof lower space light extraction image is calculated according to the sun light propagation path simulation data to obtain a local solar radiation intensity map; Using the local solar radiation intensity map, the actual irradiation area ratio of the retractable roof lower space light extraction image is calculated to obtain the daylight coefficient.

4. The method of claim 1, wherein the method is a method of optimizing the light-thermal balance of an open-close roof sports building based on a generative adversarial network and a genetic algorithm, characterized in that, Step S2 includes the following steps: Step S21: Image pairing is performed on the retractable roof two-dimensional gray scale image and the wind speed distribution chart of the lower space under the retractable roof to obtain an input-output image pair; data set division is performed on the input-output image pair to generate a model training set, a model test set and a model validation set; Step S22: constructing a Pix2Pix network framework; inputting a model training set, a model test set, and a model validation set into the Pix2Pix network framework for training, testing, and validation to obtain a Pix2Pix algorithm model; inputting a preset two-dimensional image of a design of the openable roof into the Pix2Pix algorithm model for indoor wind speed prediction of a lower space of the openable roof to generate an indoor predicted wind speed map of the lower space of the openable roof; Step S23: performing numerical matrix conversion on the indoor predicted wind speed map of the lower space of the openable roof to obtain a predicted value of a spatial wind speed distribution of the openable roof, and extracting an average wind speed value of a key area of the sports building through the predicted value of the spatial wind speed distribution of the openable roof to obtain the average wind speed value of the key area; Step S24: performing parameter correlation on an opening ratio of the openable roof according to the average wind speed value of the key area, and constructing a gene library by using the correlated parameters and light-thermal environment simulation data to obtain a wind-light-thermal environment gene library.

5. The method of claim 4, wherein the method is characterized by, Step S22 includes: constructing a Pix2Pix network framework, wherein the Pix2Pix network framework includes a generator and a discriminator; performing training parameter setting on the Pix2Pix network framework to generate a training parameter setting value, wherein the training parameter setting includes learning rate setting, convolution layer number setting, epoch setting, and optimizer setting; inputting the model training set into the Pix2Pix network framework, performing wind speed prediction by the generator in the Pix2Pix network framework to generate an initial openable roof predicted wind speed map, and performing real result image comparison on the initial openable roof predicted wind speed map by the discriminator in the Pix2Pix network framework through the model test set; performing structure similarity and root mean square error calculation on the initial openable roof predicted wind speed map to obtain wind speed prediction map accuracy; performing prediction map convergence on the wind speed prediction map accuracy by the model validation set to obtain the Pix2Pix algorithm model; inputting the preset two-dimensional image of the design of the openable roof into the Pix2Pix algorithm model for indoor wind speed prediction of a lower space of the openable roof to generate an indoor predicted wind speed map of the lower space of the openable roof.

6. The method of claim 1, wherein the method is a method of optimization of the light-thermal balance of an openable roof sports building based on a generative adversarial network and a genetic algorithm, characterized in that, Step S31 includes: obtaining sports large space environment demand data; defining optimization targets according to the sports large space environment demand data, wherein the optimization targets include maximizing a daylight coefficient, maximizing a natural ventilation speed, and minimizing a thermal comfort index; performing decision variable constraint on the maximizing daylight coefficient, the maximizing natural ventilation speed, and the minimizing thermal comfort index to limit a spatial three-dimensional coefficient of the opening ratio; performing parameter combination on the openable roof wind speed distribution predicted value based on the optimization targets and the decision variables to obtain wind-light-thermal balance target parameters.

7. The method of claim 1, wherein the method is a method of optimization of the light-thermal balance of an openable roof sports building based on a generative adversarial network and a genetic algorithm, characterized in that, Step S4 includes the following steps: Step S41: performing climate zone sensitivity analysis on the roof opening ratio-ventilation thermal comfort light environment influence curve to generate a sensitivity analysis result; Step S42: performing climate zone testing on the openable roof shape data according to the sensitivity analysis result to generate applicable range evaluation data; Step S43: performing dynamic control adjustment on the opening ratio through the applicable range evaluation data to perform the wind-light-thermal balance optimization operation of the openable roof sports building.

8. The method of claim 7, wherein the method is characterized by, Step S42 comprises the following steps: Step S421: According to the sensitivity analysis result, the key parameters of the open-close roof shape data are identified, so as to obtain the sensitive parameter data; the dynamic response analysis is carried out on the sensitive parameter data, so as to obtain the response characteristic data; Step S422: The response characteristic data is used for climate zoning mapping of the open-close roof shape data, so as to obtain the zoning data; and the zoning data is used for climate element simulation of the open-close roof shape data, so as to obtain the climate simulation data; Step S423: The performance test is carried out on the open-close roof shape data based on the climate simulation data, so as to obtain the test data; and the environmental adaptability range analysis is carried out on the open-close roof shape data by using the test data, so as to generate the applicable range evaluation data.

Images