Building low-carbon optimization design method based on energy consumption-thermal comfort correlation model

Abstract

According to the method, a set of parameterized simulation optimization design method integrating passive low carbon emission and high comfort is constructed for the design stage of a building scheme, and the method is a building low-carbon optimization design method based on an energy consumption-thermal comfort correlation model and comprises the steps that based on a parameterized building performance optimization model, an energy consumption-thermal comfort correlation model is established; establishing a parameterized model, performing microenvironment simulation by using a plug-in, and collecting data; sensitivity analysis of the parameterized model: carrying out sensitivity analysis by adopting random forest regression, gradient enhanced regression and decision tree regression, and discussing a relationship between a design variable and an environment target; and performing multi-target optimization based on performance simulation to obtain a compromise optimal result under multiple targets. According to the method, corresponding geometric and enclosure structure design parameter strategies can be provided from the aspects of building greenhouse gas emission, indoor and outdoor thermal environment comfort and the like, efficient design iteration and decision support is provided for designers, and sustainable development of green communities is promoted.

Description

Technical Field

[0001] The present invention belongs to the technical field of building performance optimization, and in particular relates to a building low-carbon optimization design method based on an energy consumption-thermal comfort correlation model. Background Art

[0002] As the basic unit of an ecological city, the ecological residential community (ecological residential community / ecological community) is of great significance in improving the living environment, reducing building energy consumption, and promoting the development of an ecological city. It is a sustainable human settlement model. The construction of an ecological residential community is multifaceted and cannot be separated from the creation of a comfortable community microclimate. Microclimate is a basic component of the human living environment. It directly affects people's living environment and can be adjusted through technical means to adapt to current climatic conditions, create a comfortable living environment, and reduce building energy consumption. With the continuous advancement of architectural parametric technology, computer simulation can analyze the impact of climate on cities, buildings, and humans, and is one of the important methods in urban microclimate research. Technical intervention can promote adaptation to current climatic conditions, ensure comfort, and reduce building energy demand through simulation and optimization. With the advancement of architectural parametric technology, computer simulation has become a tool for analyzing the impact of climate on cities, buildings, and residents. This technology is crucial in urban microclimate research. In order to effectively reduce greenhouse gas emissions and create a sustainable building environment, architects need to better understand how microclimate affects building energy consumption and conduct accurate microclimate simulations for sustainable urban planning and development.

[0003] For the objective function of building greenhouse gas emissions, the key is to understand how the microclimate affects building energy consumption. For the objective function of human comfort, the key is how the building performance can best adapt to the microclimate. Therefore, Taking human comfort and building greenhouse gas emissions as the objective function, its core lies in the in-depth analysis of the relationship between the climate characteristics of the building and the architectural form, as well as the details of the building parameters and design, such as orientation, layout and structural form, and exploring its effect on carbon emissions and human comfort by adjusting factors such as the building form.

[0004] The bottlenecks facing community construction at present are: 1) Lack of quantitative definition of community microclimate indicators. Understanding the deep mechanism and process mechanism of the microclimate cycle process can effectively improve the outdoor comfort of the community and reduce the carbon emission intensity of community buildings. 2) Lack of quantitative inspection and analysis methods for microclimate that coordinate the external space of the community and the internal space of the building. As a result, it is difficult to directly transform relevant research results into operational design actions. 3) Lack of precise regulation of the community microclimate design process. The design of community external space under the constraints of microclimate comfort needs to be further refined on the basis of the control detailed regulations. 4) Existing optimization methods usually only provide global design guidance, lack of adaptive adjustments for specific design conditions and environmental requirements, and vague decision support directions, so that the design scheme often cannot achieve the best adaptive ventilation effect, resulting in weak design carrying capacity and difficulty in combining well with actual design needs. Summary of the invention

[0005] This paper takes the relevant specifications of the community's indoor and outdoor physical environment as constraints, human comfort and building carbon emissions as objective functions, studies the outdoor microenvironment of the ecological community, proposes a multi-objective optimization plan, and provides corresponding basis and measures for the subsequent construction of the ecological community.

[0006] The specific technical solutions are as follows: A low-carbon building optimization design method based on an energy consumption-thermal comfort correlation model, including parameterized building performance optimization, building microenvironment simulation, sensitivity analysis of parameterized models, and multi-objective optimization based on performance simulation.

[0007] A community multi-objective optimization system and method based on microenvironment simulation includes the following steps: Step 1: Establish a community building benchmark model and collect modeling parameters based on real buildings. The modeling parameters include building type, building geometry parameters, building maintenance structure design parameters, and community morphology parameters.

[0008] Step 2: Conduct parametric building environment and energy consumption modeling simulation and optimization, including establishing an energy consumption-thermal comfort correlation model, calculating the building's annual carbon emissions, simulating the building's microenvironment and evaluating indoor and outdoor thermal comfort, and iterative simulation and optimization of the model.

[0009] Step 3: Calibrate the simulation results according to the measured data in the community, perform parameterized simulation and analysis based on the calibration model, and continuously adjust the input parameters of uncertainty through multiple simulations to reduce cognitive uncertainty.

[0010] Step 4: Correlation analysis between building form parameters and performance indicators, local sensitivity analysis of different design parameters of building form, and analysis of the relationship and influence intensity between design parameters and building energy demand.

[0011] Step 5: Establish a prediction model based on artificial neural network, obtain a small number of statistically representative samples for training through sampling method to minimize the number of building performance simulations in the direct optimization process, establish ANN meta-model, and dynamically couple it with the optimization algorithm to find the Pareto optimal solution for the optimization goal.

[0012] Step 6: Construction of a multi-objective optimization platform: Perform particle swarm single-objective optimization according to the design variables and objective functions; call the design parameters of the building physics model and the objective function of the natural energy ventilation model through the potential optimization model, and use the particle swarm optimization algorithm PSO to iteratively search for the optimal design solution in the multi-dimensional design space, and obtain the design parameter values ​​under the conditions of optimal community carbon emissions, optimal percentage of uncomfortable indoor thermal environment time, and optimal percentage of comfortable outdoor thermal environment time.

[0013] Step 7: Design decision support, conduct architectural form design optimization comparison, perform thermal environment simulation on the baseline model and the optimized model, and analyze the visual comparison results to find that the human thermal comfort data presented by the optimized model environment is better than that of the baseline model.

[0014] Furthermore, the building type parameters described in step 1 include the number of building floors, the length of the front facade (S / N direction), the total building width, the standard floor height, the total building height, the building width-to-length ratio, the window-to-wall ratio, the orientation, and the annual meteorological data set of the area; the building geometry parameters include the volume, the total surface area, the total floor area, and the body coefficient; the building envelope design parameters include the exterior wall heat transfer coefficient (average value), the ground heat transfer coefficient (average value), the roof heat transfer coefficient (average value), the window heat transfer coefficient (average value), and the solar heat gain coefficient (shading coefficient); the community morphology parameters include the community plot ratio, the community greening rate, the building street height-to-width ratio, the tree spacing, the tree crown size, and the street material.

[0015] Furthermore, in step 2, the Rhino / Grasshopper parametric platform is used in the process of establishing the energy consumption-thermal comfort correlation model, and the Ladybug and honeybee environmental analysis plug-ins are used. In the Grasshopper program, it is only necessary to drag the parameter command component onto the canvas and connect the input and output of the components in different logical sequences to create the program. Complex geometry can be generated through mathematical functions. Under the predefined modeling logic, the complex model can be driven and quickly changed according to the environmental performance algorithm.

[0016] Performance analysis: Grasshopper plug-ins Ladybug and Honeybee connect 3D computer-aided design (CAD) interfaces to light environment analysis software Daysim and Radiance, as well as the proven simulation engine EnergyPlus. EnergyPlus is a dynamic building energy consumption simulation software that provides integrated (load and system) simulation to achieve accurate energy, temperature and comfort predictions. On the basis of modeling and performance analysis, Octopus, as a plug-in for Grasshopper graphical parametric modeling environment, can relatively easily achieve the optimization search of building environmental parameters. The constructed energy consumption-thermal comfort correlation model can realize the interactive operation and optimization integration of building model and environmental analysis. The changes in geometric model parameters in Grasshopper will be updated in real time to the environmental analysis software process. The optimization engine drives the iterative simulation of the model, and records different geometric and environmental input parameters and the corresponding output result parameters of the analysis target to generate an "input-output" parameter table.

[0017] Furthermore, the indoor and outdoor thermal comfort assessment in step 2 uses the PMV model to calculate the annual discomfort hours percentage (DH) as an indicator. The data acquisition process is as follows: the built parameterized model is input into the OpenStudio component, and EnergyPlus can be linked to perform energy consumption simulation calculations and read the simulation results, including heating and cooling energy requirements, lighting and equipment power consumption, etc., as well as indoor air dry bulb temperature, average radiation temperature and relative humidity parameters required for PMV calculation. The indoor air dry bulb temperature, average radiation temperature, wind speed and relative humidity are calculated through OpenStudio and input into the PMV calculation module. At the same time, different human metabolic rates and clothing thermal resistance values ​​are used to obtain the PMV value of the indoor thermal environment throughout the year under different design scenarios.

[0018] Furthermore, the iterative simulation in step 2 is driven by parametric modeling and single simulation calculation. Colibri, as the engine of iterative calculation, can drive the change of decision variables, record and collect the calculation results of dependent variables of different parameter variables, and generate an "input-output" table. The input end of the Octopus optimization plug-in is the decision variable, and the calculation results of the objective function are recorded at the same time, forming a closed-loop logic to search for the Pareto optimal solution set.

[0019] Furthermore, in step 2, the energy consumption of the building is generated by heat conduction, heat convection and heat radiation between the building and the environment. The specific process of building energy consumption calculation is as follows: the first step is to calculate the net energy demand of the building, that is, the energy required to meet the building specifications and provide indoor thermal environment control. The second step is to determine the transfer energy of the building, that is, the energy of the building in actual use, including heating, cooling, hot water, lighting and ventilation, as well as auxiliary energy required for fans, pumps, etc. The third step is to obtain the overall energy usage and related performance indicators based on the results of step 2. The annual energy demand of the building is defined as the sum of the cooling and heating loads of all apartments. The energy demand of other household hot water, electrical equipment, etc. is not included. The summer cooling time and winter heating time are set according to the requirements of different climate zones.

[0020] In this paper, the HVAC system performance coefficient is assumed to be 1, so the energy demand can be directly extracted from the EnergyPlus results of the simulation. The parameters for the energy demand calculation are determined based on the heating and cooling set point temperatures, assuming that no heat recovery device is implemented in the HVAC system. Therefore, the objective function of the annual building energy demand can be calculated as formula (1): (1) Where BED represents the annual building energy demand per unit building area [kWh / m 2 ], the calculation of building energy demand only considers heating and cooling demand, and does not consider other aspects such as lighting, is the cooling demand of the i-th floor of the building, It is the building Heating demand of the floor, is the total number of floors of the building, and A is the total area of ​​each floor of the air-conditioned zone of the building.

[0021] Furthermore, the whole process of model verification calibration and uncertainty analysis in step 3 is as follows: 1) Basic model validation, following the data and assumptions; 2) Classification of sources of uncertainty (modeling or experiment) and definition of their mathematical structure (i.e., probability distribution function or data interval); 3) Modeling using experimental input data and performing uncertainty propagation of simulations; 4) Comparison of simulation results with measured data; 5) Residual analysis; 6) Model calibration. When comparing the simulation results with the experimental measurements thereafter, both the measurement uncertainty of the experimental data (random) and the uncertainty propagation of the model in the model data (combined accidental and cognitive) are taken into account. On this basis, residual analysis is performed, i.e., the model calibration is performed using the experimental error estimate obtained from the difference between the simulated and measured model outputs. The goodness of fit of the regression model can be expressed by the coefficient of determination R, which can take values ​​ranging from 0 to 1 (or 0 to 100% if expressed as a percentage), where R=1 indicates that the data fit is perfect. R² is calculated as in formula (2); (2) in It is monitoring data. is the simulated data and n is the number of monitored data points. R² is an important indicator of goodness of fit, but it is not the only one to consider. You can also consider the mean absolute percentage error (MAPE) to calculate the average absolute value of the difference between the simulated data and the predicted data, normalized relative to the measured data itself. MAPE is calculated as formula (3): (3) In addition, in the prior art of model calibration procedures, two other metrics are used, namely NMBE and Cv (RMSE). NMBE (normalized mean error) is the sum of the differences between the predicted values ​​and the simulated energy consumption at time intervals (e.g., monthly, hourly) within the calculation period, and then the difference is divided by the sum of the predicted energy consumption. As shown in formula (4): (4) A positive value of NMBE means that the model estimates energy consumption too much, whereas a negative value means an underestimation. The root mean square error (RMSE) is a measure of the sample deviation of the difference between the simulated value and the model predicted value. Cv(RMSE) is the coefficient of variation of the RMSE and is calculated as the mean of the RMSE normalized to the predicted value. Cv(RMSE) represents a standardized measure of the variability between the predicted and simulated data. It specifies the overall uncertainty of the building energy consumption forecast and reflects the error size and dispersion, as shown in formula (5). A lower Cv(RMSE) value indicates a better calibrated model. (5).

[0022] Furthermore, in step 4, the specific operation process of local sensitivity analysis is as follows: Perform small-scale perturbations on each design variable in the optimal design parameter combination, observe the changes in the objective function value, and calculate the new design parameter combination after the perturbation. The corresponding objective function value , the change of the objective function is , for each design variable, calculate the local sensitivity coefficient , (6) If the variable units are different or need to be normalized, the normalized sensitivity coefficient can be calculated: (7) According to the size of the local sensitivity coefficient, the relative importance of each design variable to the objective function is judged. Variables with larger coefficients have a greater impact on the objective function and need to be given more attention in the design.

[0023] The specific operation process of global sensitivity analysis is as follows: Generate a sample set based on all design samples generated in the PSO process, or use Latin hypercube sampling LHS or Sobol sequence to generate a sample set of design variables, covering the entire design space. Assume that the design variables are , the sample set is , where m is the number of samples, calculate the global sensitivity index of each design variable, and for each sample , calculate its objective function value , record the objective function values ​​of all samples: (8) Sobol sensitivity index calculation, first-order sensitivity index Reflects a single design variable The independent effect of is calculated as: (9) in, In the design variable The expectation of the objective function value under certain conditions; is the total variance of the objective function; the total sensitivity index Reflecting design variables The overall effect of , including its interaction effects with other variables, is calculated as: (10) in, In addition to All design variables except .

[0024] Furthermore, in step 5, the process of establishing the meta-model of the artificial neural network is as follows: 1) Establish representative input samples to train and validate the meta-model; 2) Simulate the selected sample input data to obtain the corresponding output data; 3) Use input and output sample data to train the meta-model.

[0025] The neural network model established by the present invention takes into account the architectural form design factors. The input neurons include design parameters such as the building shape coefficient, the spacing between different buildings, the building aspect ratio and the building orientation. Based on the simulation results of the established typical model as the database basis, 2000 groups of data are selected for each city as the sample set to train the prediction model. The output neuron parameters of the model are 2, namely the annual energy demand per unit area of ​​the building and the annual percentage of the indoor thermal environment comfort. As an information processing system, the artificial neural network (ANN) is mainly derived from the imitation of the structure and function of the human brain. It is established based on the collection of connection units or nodes of artificial neurons. Neurons have a certain threshold and are usually adjusted and weighted as learning. The increase or decrease of the weight affects the strength of the signal at the connection. In the actual modeling process, the topological structure of the neural network is limited, and the known input and output data are used as learning and test samples to train and test the network model, so that an accurate network model can be obtained.

[0026] Furthermore, in step 5, Pareto is a classic model for multi-objective optimization. Its core idea is to find the optimal solution under the premise of minimum conflict of objectives. The Pareto optimal solution is a set that contains all solutions that are not better than any other solution. If the minimum value of the objective is required, then two feasible solutions , when formula (11) holds, It is called Pareto optimality : (11) Formula 3-5 shows that all corresponding The objective function is no greater than The objective function value of f( ) has a value that is absolutely smaller than f( ). When the objective function needs to find the maximum solution, the expression is changed to formula (12): (12) Furthermore, in step 6, the particle swarm single-objective optimization process is as follows: In the implementation process of the particle swarm optimization algorithm, the particle swarm is first initialized, and the initial particle group is randomly generated. Each particle represents a combination of design variables. Then the objective function is calculated, and the objective function value of each particle is calculated and used as the initial fitness value. The speed and position of each particle are updated by the following formula, Speed ​​update formula: (13) Position update formula: (14) Where: Indicates Generation of particles The speed determines the moving direction and step size of the particle in the search space; Represents particles In the The design variable combination of the generation, that is, the current position of the particle; Represents the inertia weight, which controls the influence of the particle's current velocity on its next step velocity; Represents the cognitive coefficient, which indicates the particle's movement to its best historical position The acceleration factor of approach; Indicated in The random numbers within the range introduce randomness into the process of particles moving to their optimal positions, increasing the randomness and diversity of the search process; Represents particles The historical optimal position of is the optimal combination of design variables found by the particle in all past iterations, which is the best solution retained by the particle based on its own historical experience; Represents the social coefficient, which indicates the particle's global optimal position in the group The acceleration factor of approach, this parameter determines the degree of dependence of the particle on the group experience; Indicated in A random number within a certain range introduces randomness into the process of particles moving to the global optimal position, helping to prevent the group from falling into the local optimal position. Represents the global optimal position, that is, the optimal solution among the historical optimal positions of all particles in the current generation. This position is the best solution found among all particles and is the goal that all particles tend to; Indicates the number of generations of the current iteration, starting from 0 and increasing gradually; Represents particles In the The speed of the generation is updated by combining inertia, individual experience and group experience; Represents particles In the The position of the next generation is obtained by adding the position of the previous generation and the updated speed. Convergence judgment is performed through iterative control. When the preset maximum number of iterations is reached or the change in the objective function value is less than the set threshold, the algorithm outputs the optimal design variable combination when the convergence condition is met.

[0027] The beneficial effects of the present invention are: The present invention integrates multiple evaluation indicators, parametric simulation, human comfort simulation, carbon emission simulation calculation, optimization algorithm, and sensitivity analysis, summarizes the low-carbon design parameters and rules of typical urban community buildings, realizes efficient design iteration and decision support, and promotes the sustainable development of ecological residential communities. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A flow chart of a building low-carbon optimization method based on an energy consumption-thermal comfort correlation model of the present invention; Figure 2 A schematic diagram of a building low-carbon optimization system based on an energy consumption-thermal comfort correlation model of the present invention; Figure 3 This is a schematic diagram of the modeling process from a surface to a geometric body in the present invention; Figure 4 This is a schematic diagram of local generation of residential space form based on Grasshopper in the present invention; Figure 5 This is a schematic diagram of the Pearson correlation matrix of city A in the present invention; Figure 6 This is a schematic diagram of the Pearson correlation matrix of city B in the present invention; Figure 7 For the present invention, the Pearson correlation matrix of C city is intended; Figure 8 This is a schematic diagram of the Pearson correlation matrix of city D in the present invention; Fig. 9 This is a schematic diagram of the Pearson correlation matrix of city E in the present invention; Fig.10 This is a schematic diagram of the best results of five cities in the present invention; Fig.11 This is a schematic diagram of wind speed simulation analysis in City A in the present invention; the left side is the baseline model, and the right side is the optimized model; Fig.12 This is a schematic diagram of the UTCI simulation analysis of City A in the present invention; the left side is the baseline model, and the right side is the optimized model. DETAILED DESCRIPTION

[0029] The present invention is further described below in conjunction with the accompanying drawings, but the protection scope of the present invention is not limited thereto. All other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of the present invention.

[0030] Example: like Figure 2As shown in the figure, a low-carbon building optimization design system based on the energy consumption-thermal comfort correlation model includes four aspects: data collection input, microenvironment simulation, data analysis, and result output. Among them, the data input needs to obtain the building morphology parameters, geometric parameters, maintenance structure design parameters, community morphology parameters, and climate data of the building. In the microenvironment simulation stage, it is necessary to establish a parametric building physical model component of the environmental control parameters, perform wind-heat coupling simulation, and building performance simulation. In the analysis stage, it is necessary to establish a prediction model and apply the particle swarm optimization algorithm to conduct sensitivity analysis on various design factors of the community, building performance, and indoor and outdoor comfort. In the result, the optimal solution under the multi-objective optimization conditions is obtained, and the wind-heat environment before and after the optimization is simulated to verify its effectiveness.

[0031] The steps for data collection are as follows: Obtain community building morphological parameters, building envelope structures and community morphological parameters, wherein the building morphological parameters include building type parameters and geometric parameters, wherein the building type parameters include the number of building floors, the length of the front facade (S / N direction), the total width of the building, the standard floor height, the total building height, the building width-to-length ratio, the window-to-wall ratio, the orientation, and the annual meteorological data set of the area; the building geometric parameters include the volume, the total surface area, the total floor area, and the body coefficient; the building envelope structure design parameters include the exterior wall heat transfer coefficient (average value), the ground heat transfer coefficient (average value), the roof heat transfer coefficient (average value), the window heat transfer coefficient (average value), and the solar heat gain coefficient (shading coefficient); the community morphological parameters include the community volume ratio, the community greening rate, the building street height-to-width ratio, the tree spacing, the tree crown size, and the street material.

[0032] The whole system is divided into several different subsystems, including geometry generation, system control part, simulation part and dynamic driving engine. These defined properties are created through the interaction between scripts. The subsystems are related to each other through the interdependency of scripts, and the parameters of each subsystem can be related to the parameters in other subsystems. Finally, a parametric simulation procedure of energy consumption-thermal comfort association is established using numerical simulation and dynamic search tools.

[0033] Reference Figure 1 ,The operation steps of establishing the energy consumption-thermal comfort correlation parameterized model are detailed as follows: S1. Climate data acquisition: Using the ladybug / honeybee component, the weather file on the computer can be linked to ladybug / honeybee through Boolean switching control. Setting Boolean to True can browse the weather file on the system.

[0034] S2. Architectural form parameters: Architectural form parameters are used to define the associative parametric model and provide all shape-driven parameters and their ranges.

[0035] The logic of geometric modeling is as follows: First, a face is constructed in Grasshopper, and then the face is extruded into a block. Next, it is copied and moved from one block to several blocks, depending on the number controlled by the floor parameter. Finally, the building geometry can be rotated according to the azimuth parameter, such as Figure 3 shown.

[0036] S3. Enclosure structure design parameter setting: Building component structures include roof, exterior wall, ground floor, ceiling, interior floor and window properties, which are assigned different material parameters.

[0037] The heat transfer coefficient of the material is calculated as shown in formula (15): (15) Where Rse is the fixed value of the external thermal resistance, Rsi is the fixed value of the internal thermal resistance, and R1+R2...+Rn is the sum of all thermal resistance values ​​of building materials in the building component. The calculation formula of R1, R2,..., Rn is formula (16): (16) Where k is the thermal conductivity of the building material and d is the material thickness. By inverse calculation, the change of heat transfer coefficient can be controlled by adjusting the thickness of the insulation layer in the GH platform.

[0038] S4. Open Studio Simulation: Once set up, the OpenStudio button in honeybee will export the HBZones into an OpenStudio file, which can then be run through EnergyPlus, which outputs a simulation report including the file path to the IDF file and the CSV file from the EnergyPlus run.

[0039] S5. Reading of energy demand calculation results: Reading EnergyPlus simulation results through the generated csv. file.

[0040] S6. Iterative simulation: The plugin Colibri can help convert definitions in Grasshopper to Excel The goal of this plugin is to easily generate Excel data tables in Grasshopper. The Iterator component loops and connects the sliders and drives the input values ​​of Grasshopper. Colibri's Iterator component allows users to specify the number of steps taken by each slider. It allows users to control the size of their design space and optionally specify granularity on each input vector within the design space, all without editing the actual slider in Grasshopper. As the Iterator iterates upstream, the Aggregator component collects all the data needed by DesignExplorer from the Grasshopper definition. It collects inputs from the iterator and outputs (performance indicators) from the Grasshopper definition and is responsible for generating images, naming images, and writing all data to the data.csv file.

[0041] The operational steps of data analysis are detailed as follows: 1) Sensitivity analysis of design factors and objective functions, using the spearman coefficient to analyze the influence relationship between two factors; 2) Establish a predictive model and use machine learning to simplify the optimization process.

[0042] Based on the above energy consumption-thermal comfort correlation parameterized model, we now take the XX community prototype (Table 1) as an example to establish a benchmark model for microenvironment parameterized simulation. According to the parameters of the XX community, we establish specific residential microenvironment simulation model design parameters and definitions and ranges, as shown in Table 2. The residential space form generated according to this parameter is as follows: Figure 4 shown.

[0043] Table 1. XX community model parameters

[0044] Table 2. Types and definitions of design parameters

[0045] During the random simulation, a data set was obtained for each city. Based on the research and analysis of the thermal environment of low-rise old communities in five typical cities, A, B, C, D, and E, a total of 3,659 data sets were obtained. The Pearson correlation matrix is ​​provided in Figure 5-9. The above sensitivity analysis can be used as a reference for community climate adaptability construction under the climate conditions of each city.

[0046] The combined set of sampled building design parameters and building performance data is divided into two different categories: training dataset and validation dataset. The former is the basis for training the machine learning model, while the latter is used to determine the accuracy of the prediction model. The machine learning model can achieve typical prediction accuracy by minimizing the mean square error between the simulation results and the prediction results of the training dataset, while maximizing the correlation coefficient between the simulation results and the prediction results of the validation dataset. Table 3-5 shows the performance of the developed models in predicting GHE, UTCI, and DH. The model performance of the training and testing datasets is presented here. Three performance criteria, namely, mean absolute error (MAE), mean square error (MSE), and coefficient of determination (R2), are determined and evaluated in this study. The experimental results show that all the developed models can estimate the desired output with high prediction accuracy. During the training phase, the R² of the RFR, GBR, and DTR models ranged between 99% and 100%, respectively, which indicates that the prediction effect of the neural network is good.

[0047] Table 3. Fitting index of forecast model (RFR) for each city

[0048] Table 4. Fitting index of prediction model (GBR) for each city

[0049] Table 5. Fitting index of prediction model (DTR) for each city

[0050] The optimal solution is obtained through parametric performance simulation and optimization of all design parameters based on NSGA-2 algorithm, e.g. Fig.10 Table 6 lists the optimal values ​​of design parameters for each city based on climate conditions, taking into account factors such as community greenhouse gas emissions, percentage of indoor discomfort time, and percentage of outdoor comfort time. Table 7 lists the results of the corresponding optimization objectives.

[0051] Table 6. Optimal design parameters for each typical city

[0052] UTCI(*): optimal percentage of comfortable time for outdoor thermal environment; DH(*): optimal percentage of uncomfortable time for indoor thermal environment; GHE(*): optimal annual carbon emissions of the building.

[0053] Table 7. Performance indicators of the optimal solution

[0054] UTCI(*): optimal percentage of comfortable time for outdoor thermal environment; DH(*): optimal percentage of uncomfortable time for indoor thermal environment; GHE(*): optimal annual carbon emissions of the building.

[0055] This example evaluates the impact of optimized solutions on outdoor thermal comfort, indoor thermal discomfort, and greenhouse gas emissions for multiple cities compared to a baseline model. In City A, significant greenhouse gas reductions were achieved with minimal compromise in indoor discomfort. Optimization in City B improved environmental performance in all aspects, while the hot climate in City C limited outdoor comfort. City D significantly reduced greenhouse gas emissions and improved indoor comfort, while City E balanced emissions and comfort. Overall, the best solution in each city significantly improved environmental performance, demonstrating the potential for community layout and design transformation. The study's annual indicators provide decision support for the planning and design of residential areas in similar climate regions, highlighting the importance of design-stage optimization for thermal comfort. Thermal environment simulation using CFD highlights the advantages of early wind environment assessment and natural ventilation design.

[0056] Take city A as an example. Fig.11 A comparison between wind speed simulations of the baseline model and the optimized model for City A is shown. Fig.12 The comparison and analysis of UTCI of the baseline model and the optimized model in City A are shown. The wind environment and UTCI distribution of the hottest week are calculated using the Ladybug plugin. The visualization analysis shows that the human thermal comfort of the optimized model exceeds that of the baseline model in City A.

Claims

1. A low-carbon optimization design method for buildings based on an energy consumption-thermal comfort correlation model, characterized in that: Based on the optimization system, the optimization system includes building physical model components, microenvironment simulation components, sensitivity analysis components and multi-objective optimization components; The building physical model component is used to obtain the building's geometric model, attribute parameters, construction parameters, material parameters, load parameters and environmental control parameters; The microenvironment simulation model component is used to perform building performance simulation and wind-heat comfort simulation to evaluate building performance and climate comfort; The sensitivity analysis component is used to perform sensitivity analysis on various design factors and building performance and indoor and outdoor comfort; The multi-objective optimization model component uses a particle swarm optimization algorithm to optimize design parameters according to the simulation results of the building performance model to achieve the best building performance and comfort.

2. A low-carbon optimization design method for buildings based on an energy consumption-thermal comfort correlation model as claimed in claim 1, comprising the following steps: Step 1: Establish a community building benchmark model and collect modeling parameters based on actual measured communities. The modeling parameters include building type parameters, building geometry parameters, building maintenance structure design parameters, and community morphology parameters. Step 2: Conduct parameterized building environment and energy consumption modeling simulation and optimization, including establishing an energy consumption-thermal comfort correlation model, calculating the building's annual carbon emissions, simulating the building's microenvironment, evaluating indoor and outdoor thermal comfort, and iterative simulation and optimization of the model; Step 3: Calibrate the simulation results based on measured community data, perform parameterized simulation and analysis based on the calibrated model, and adjust the input parameters of uncertainty to reduce epistemic uncertainty; Step 4: Correlation analysis between building form parameters and performance indicators: Conduct local sensitivity analysis on different design parameters of building form, and analyze the relationship and impact intensity between design parameters and building energy demand; Step 5: Establish a prediction model based on artificial neural network, obtain samples for training through sampling method, establish ANN meta-model, and dynamically couple it with the optimization algorithm to find the Pareto optimal solution for the optimization target; Step 6: Construction of a multi-objective optimization platform based on building environment simulation: Perform particle swarm single-objective optimization according to design variables and objective functions; call the design parameters of the building physics model and the objective function of the natural energy ventilation model through the potential optimization model, and use the particle swarm optimization algorithm PSO to iteratively search for the optimal design solution in the multi-dimensional design space, and obtain the design parameter values ​​under the conditions of optimal community carbon emissions, optimal percentage of uncomfortable indoor thermal environment time, and optimal percentage of comfortable outdoor thermal environment time; Step 7: Design decision support: Perform architectural form design optimization comparison, perform thermal environment simulation on the baseline model and the optimized model, and compare the results through visualization.

3. A low-carbon building optimization design method based on an energy consumption-thermal comfort correlation model as claimed in claim 2, characterized in that: In step 1, the building type parameters include the number of building floors, the length of the front facade, the total building width, the standard floor height, the total building height, the building width-to-length ratio, the window-to-wall ratio, the orientation, and the annual meteorological data set of the region; the building geometry parameters include the volume, the total surface area, the total floor area, and the body coefficient; The design parameters of the building envelope include the heat transfer coefficient of the exterior wall, the heat transfer coefficient of the ground, the heat transfer coefficient of the roof, the heat transfer coefficient of the window, and the solar heat gain coefficient; Community morphological parameters include community plot ratio, community greening ratio, building-street height-to-width ratio, tree spacing, tree crown size, and street materials.

4. A low-carbon building optimization design method based on an energy consumption-thermal comfort correlation model as claimed in claim 2, characterized in that: In step 2, the energy consumption-thermal comfort correlation model is constructed based on the Rhino / Grasshopper parametric platform and is linked to the open source building energy consumption simulation software platform OpenStudio through Ladybug / Honeybee. It can realize the interactive operation and optimization integration of building models and environmental analysis. The changes in geometric model parameters in Grasshopper will be updated in real time to the environmental analysis software process. The iterative simulation of the model is driven by the optimization engine, and different geometric and environmental input parameters and the output result parameters of the corresponding analysis targets are recorded to generate an input-output parameter table.

5. The low-carbon optimization design method for buildings based on the energy consumption-thermal comfort correlation model according to claim 2, characterized in that: In step 2, the indoor and outdoor thermal comfort evaluation uses the predicted mean value PMV model to calculate the annual uncomfortable time percentage DH as an indicator; the data acquisition process is as follows: The constructed community building benchmark model is input into the OpenStudio component, linked to EnergyPlus for energy consumption simulation calculation, and the simulation results are read. The indoor air dry-bulb temperature, mean radiant temperature, wind speed and relative humidity are calculated through the OpenStudio component and input into the calculation module of the PMV model. At the same time, different human metabolic rates and clothing thermal resistance values ​​are used to obtain the PMV value of the indoor thermal environment throughout the year under different design scenarios.

6. A low-carbon building optimization design method based on an energy consumption-thermal comfort correlation model as claimed in claim 2, characterized in that: In step 2, the iterative simulation is driven based on parametric modeling and single simulation calculation. The iterative calculation uses Colibri as the engine to drive the change of decision variables, and records and collects the calculation results of dependent variables of different parameter variables to generate an input-output table. The input end of the Octopus optimization plug-in is the decision variable, and the calculation results of the objective function are recorded at the same time to form a closed-loop logic and search for the Pareto optimal solution set.

7. A low-carbon building optimization design method based on an energy consumption-thermal comfort correlation model as claimed in claim 2, characterized in that: In the energy consumption-thermal comfort correlation model established in step 2, the building energy consumption process is as follows: 1) Calculate the net energy demand of the building; 2) Determine the building's transferred energy; 3) Based on the delivered energy determined in step 2), obtain the overall energy usage and related performance indicators.

8. The low-carbon optimization design method for buildings based on the energy consumption-thermal comfort correlation model according to claim 2, characterized in that: In step 3, the process of model verification calibration and uncertainty analysis is as follows: 1) Validate the basic model based on data and assumptions; 2) Classification of uncertainty sources and definition of their mathematical structure; 3) Use experimental input data to model and propagate uncertainty in simulation; 4) Compare the simulation results with the measured data; 5) Residual analysis; 6) Model calibration: When comparing the simulation results with the experimental measurements, the measurement uncertainty of the experimental data and the propagation of the model uncertainty in the model data are considered, on this basis, a residual analysis is performed.

9. A low-carbon building optimization design method based on an energy consumption-thermal comfort correlation model as claimed in claim 2, characterized in that: In step 4, The specific operation process of local sensitivity analysis is as follows: 1) Perturb each design variable in the optimal design parameter combination, observe the change in the value of the objective function, and calculate the new design parameter combination after the disturbance The corresponding objective function value , the change of the objective function is ;in, , ,..., Indicates the change in input quantity to; : represents the set of design variables in the parameter combination; :Represents the design variables the amount of disturbance or change applied; 2) For each design variable, calculate the local sensitivity coefficient : ； If the variable units are different or need to be normalized, calculate the normalized sensitivity coefficient: ； in, represents the current value of the design variable, Represents the design variable The change or disturbance value; 3) According to the size of the local sensitivity coefficient, determine the relative importance of each design variable to the objective function; The specific operation process of global sensitivity analysis is as follows: 1) Generate a sample set based on all design samples generated in the PSO process, and use Latin hypercube sampling LHS or Sobol sequence to generate a sample set of design variables, covering the entire design space. Assume that the design variables are , the sample set is , where m is the number of samples, calculate the global sensitivity index of each design variable, and for each sample , calculate its objective function value , record the objective function values ​​of all samples: ； 2) Sobol sensitivity index calculation, where the first-order sensitivity index Reflects a single design variable The independent effect of is calculated as: ； in, In the design variable The expectation of the objective function value under certain conditions; is the total variance of the objective function; 3) Overall sensitivity index Reflecting design variables The overall effect of , including its interaction effects with other variables, is calculated as: ； in, In addition to All design variables except .

10. A low-carbon building optimization design method based on an energy consumption-thermal comfort correlation model as claimed in claim 7, characterized in that: In step 6, the multi-objective optimization process based on building environment simulation is as follows: Based on the building simulation tool, a Monte Carlo simulation framework that uses random sampling to solve mathematical problems is established to perform uncertainty analysis and search for input parameters. Automated means are used to solve the problem of difficult to determine input parameters, and a reverse search process is used for modeling search. The calculation formula is: ； where y is a different decision variable The function can be divided into two groups: design parameter variables and scenario variables. Parametric variables representing architectural design; scene variables Include boundary conditions related to building operation and climate parameters; f corresponds to an energy simulation tool that calculates the value of a given decision variable x based on a physical function; Using functions f , forward modeling can find y for a given x, and reverse modeling can find multiple x for a given y, expressed as follows: Design parameters :performance; y: Performance Design parameters.

Images