Green building multi-objective optimization design method and device, electronic equipment and readable storage medium

A multi-objective optimization model for green buildings was constructed using the BO-CatBoost-NSGA-Ⅲ algorithm. This model solved the problem of optimizing the nonlinear relationship between energy consumption, carbon emissions, and thermal comfort, achieving high-precision prediction and optimization, reducing energy consumption and carbon emissions, and improving building comfort.

CN118916965AInactive Publication Date: 2024-11-08WUHAN INSTITUTE OF MARINE ELECTRIC PROPULSION (THE 712TH RESEARCH INSTITUTE OF CHINA STATE SHIPBUILDING CORP LTD)

Patent Information

Application Number
CN202411056165.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-08
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing technologies struggle to effectively optimize the complex nonlinear relationship between energy consumption, carbon emissions, and thermal comfort in green buildings, resulting in a lack of precision and reliability in building design.

Method used

The BO-CatBoost-NSGA-Ⅲ algorithm is adopted to construct a multi-objective optimization model. Building data is simulated using BIM-DB software. By combining Bayesian optimization and CatBoost prediction models, a high-precision prediction model for building energy consumption, carbon emissions, and thermal comfort is established. The importance of input variables is scored using the random forest algorithm, and multi-objective genetic optimization is performed using the NSGA-Ⅲ algorithm. The optimal solution is determined by the ideal point method.

Benefits of technology

It achieves high-precision prediction and optimization of building energy consumption, carbon emissions and thermal comfort, ensuring the accuracy of prediction results and the reliability of optimization process, reducing building energy consumption and carbon emissions, while improving thermal comfort.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916965A_ABST
    Figure CN118916965A_ABST
Patent Text Reader

Abstract

The invention discloses a multi-objective optimization design method for a green building, and the method comprises the steps: carrying out the simulation of the green building through BIM-DB software, obtaining a building data sample set through an orthogonal test, carrying out the optimization of a hyper-parameter of a gradient lifting algorithm through Bayesian optimization, building a BO-CatBoost prediction model, and predicting the energy consumption, carbon emission and thermal comfort of the building. Building a BO-CatBoost-NSGA-III multi-objective optimization model, taking the built BO-CatBoost prediction model function as a non-dominated sorting genetic algorithm fitness function with an elitist strategy, optimizing the energy consumption, carbon emission and thermal comfort of the building, and determining the optimal solutions of the energy consumption, carbon emission and thermal comfort of the building through an ideal point method according to the obtained Pareto optimal solution set; the invention further discloses a device, electronic equipment and a readable storage medium. According to the method, reference can be provided for parameter setting of the green building, and a thought is provided for multi-objective optimization of the green building.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of building construction, and specifically relates to a green building multi-objective optimization method, device, electronic device and readable storage medium based on BO-CatBoost-NSGA-III. Background Art

[0003] The construction industry should develop in the direction of energy conservation and sustainability, which has been recognized and supported by various countries and researchers. The theory of green building has become an important direction for the development of the construction industry and has attracted great attention. Green building energy conservation research can achieve energy conservation and water conservation, low carbon emissions, cost reduction and comfort improvement.

[0004] Therefore, a BO-CatBoost-NSGA-Ⅲ multi-objective optimization method and equipment for green buildings were developed to provide a reference for the parameter setting of green buildings. Summary of the invention

[0005] In view of the above defects or improvement needs of the prior art, one of the objectives of the present invention is to provide a multi-objective optimization design method for green buildings.

[0006] The technical solution adopted by the present invention to solve the technical problem is: a multi-objective optimization design method for green buildings, comprising the following steps:

[0007] Step 1: Use BIM-DB software to simulate green buildings, select 10 parameters including window-to-wall ratio, exterior wall heat transfer coefficient, roof heat transfer coefficient, exterior window heat transfer coefficient, exterior wall solar radiation absorption coefficient, roof solar radiation absorption coefficient, exterior window solar heat gain coefficient, air tightness, air conditioning heating setting temperature and air conditioning cooling setting temperature as input variables, and take 3 parameters including building energy consumption, carbon emissions and thermal comfort as output variables, and use orthogonal experiments to obtain a sample set of building data;

[0008] Step 2: Use Bayesian optimization to optimize the hyperparameters of the gradient boosting algorithm, build a BO-CatBoost prediction model, divide the data sample set into a training sample set and a test sample set, and use the training sample set and the test sample set to optimize the parameters of the BO-CatBoost prediction model and analyze the prediction accuracy. Use the mean square error and determination coefficient to evaluate the accuracy of the BO-CatBoost prediction model, predict building energy consumption, carbon emissions, and thermal comfort, and score the importance of input variables based on the random forest algorithm.

[0009] Step 3: Construct a BO-CatBoost-NSGA-Ⅲ multi-objective optimization model, and use the established BO-CatBoost prediction model function as the fitness function of the non-dominated sorting genetic algorithm with an elite strategy to optimize building energy consumption, carbon emissions, and thermal comfort.

[0010] Step 4: Based on the obtained Pareto optimal solution set, the optimal solution of building energy consumption, carbon emissions and thermal comfort is determined by the ideal point method.

[0011] Furthermore, in step 2, the BO-CatBoost prediction model is constructed by the following steps:

[0012] (2.1) Build a regression model based on CatBoost. First, count the categorical features, calculate the frequency of a category feature, consider the hyperparameters to generate new numerical features, and train the regression model with hyperparameters learning_rate, colsample_bytree, minibatch_frac and n_estimators in turn.

[0013] (2.2) The Bayesian algorithm is used to optimize the learning_rate, minibatch_frac, colsample_bytree, and n_estimators of the CatBoost regression model. The learning_rate optimization range is set to 0.05-0.2, and the minibatch_frac optimization range is set to (0,1]. The model accuracy is verified by combining the 5-fold cross-validation method to determine the parameter combination with the highest accuracy as the optimal parameter of the CatBoost regression model.

[0014] (2.3) Establishment and evaluation of BO-CatBoost prediction model: Based on the hyperparameter optimization results, the training set was used for learning simulation, and CatBoost building energy consumption, carbon emission and thermal comfort prediction models were established respectively. The prediction models of the training sample set were then tested using the test sample set. The vertical displacement and horizontal displacement prediction models of bridge piers based on the random forest algorithm were established respectively to obtain the prediction results.

[0015] Furthermore, in step 2, by introducing the mean square error MSE and the absolute coefficient R 2 There are two commonly used indicators to evaluate the accuracy of BO-CatBoost prediction model, MSE and R 2 It is calculated by the following formula:

[0016]

[0017]

[0018] Where n is the number of samples, is the predicted value, y i is the actual observed value, is the average of the actual observed values.

[0019] Furthermore, the step three uses the established BO-CatBoost prediction model function as the NSGA-III fitness function to optimize building energy consumption, carbon emissions and thermal comfort, specifically including the following steps:

[0020] (3.1) Determine the objective function: Introduce the relationship between the input variables of BO-CatBoost regression prediction fitting and building energy consumption, carbon emissions and thermal comfort as the fitness function in the multi-objective genetic algorithm, which can clearly reflect the complex nonlinear relationship between building design characteristic parameters and building energy consumption, carbon emissions and thermal comfort; the objective functions f1, f2 and f3 of building energy consumption, carbon emissions and thermal comfort are determined by the CatBoost prediction regression equation:

[0021]

[0022] In the formula, x1, x2, x3, x4, x5, x6, x7, x8, x9 and x 10 They are window-to-wall ratio, exterior wall heat transfer coefficient, roof heat transfer coefficient, exterior window heat transfer coefficient, exterior wall solar radiation absorption coefficient, roof solar radiation absorption coefficient, exterior window solar heat gain coefficient, air tightness, air conditioning heating setting temperature, and air conditioning cooling setting temperature;

[0023] (3.2) Determine the constraint range: In order to make the optimization design of construction parameters have practical significance and ensure that the obtained construction parameter setting scheme is more reasonable and feasible without safety hazards, it is necessary to set constraints on the values ​​of decision parameters based on engineering requirements and existing project data. The constraints are generally expressed as:

[0024]

[0025] In the formula and Respectively represent the lower and upper limits of the operating parameter values;

[0026] (3.3) Multi-objective optimization based on NSGA-III:

[0027] (3.4) Ideal point method to obtain the optimal solution: The ideal point method is used to select the optimal solution from the Pareto optimal solution set, so that each target value is as close to the ideal value as possible; the point composed of the target values ​​corresponding to each single target when it is optimal That is the ideal point; after determining the coordinates of the ideal point E, the distances from each point in the Pareto optimal solution set to the ideal point are calculated as follows:

[0028]

[0029] The point (x pareto ,y pareto ,z pareto ) are the coordinates of each point in the Pareto optimal solution set, and the point with the shortest distance is the optimal solution selected from the Pareto optimal solution set.

[0030] Furthermore, the step (3.3) specifically includes the following steps:

[0031] 1) Set the initial state of the decision variables, randomly generate decision variables according to the given variable boundary restrictions, and thus generate an initialized population with N individuals;

[0032] 2) Use the three mechanisms of selection, crossover, and mutation to operate on a population of size N to obtain the first generation of offspring populations of the same size as the parent population;

[0033] 3) Combine the offspring population with its parent population to form a mixed population, perform fast non-dominated sorting on the individuals in the mixed population, and select individuals from different dominance levels to join the next generation;

[0034] 4) Generate reference points according to subjective preferences or unified settings, connect the origin and the reference points to form a reference line, and then perform normalization transformation on the objective function and the algorithm population; calculate the vertical distance between each individual and the reference line, and associate the individual with the smallest vertical distance with the reference point;

[0035] 5) Find the reference point with the least number of associated solutions, and select the solution according to the number of associated solutions of the reference point; if the number of solutions is not 0, randomly select an individual; if the number of solutions is 0, select the individual with the smallest vertical distance to the reference point; repeat the solution selection process until the population size is N, and output the Pareto optimal solution set.

[0036] The second object of the present invention is to provide a green building multi-objective optimization design device based on BO-CatBoost-NSGA-Ⅲ, including: a first main module, used to collect data to establish an original sample set, using building design feature parameters as input variables, and building energy consumption, carbon emissions and thermal comfort as output variables; a second main module, used to build and train a BO-CatBoost prediction model, using the BO-CatBoost prediction model to predict building energy consumption, carbon emissions and thermal comfort, and scoring the importance of input variables based on the RF algorithm; a third main module, used to build a BO-CatBoost-NSGA-Ⅲ multi-objective optimization model, using the established BO-CatBoost prediction model function as the NSGA-Ⅲ fitness function, and optimizing building energy consumption, carbon emissions and thermal comfort; according to the obtained Pareto optimal solution set, the optimal solution of building energy consumption, carbon emissions and thermal comfort is determined by the ideal point method.

[0037] The third object of the present invention is to provide an electronic device, comprising: at least one processor, at least one memory and a communication interface communicatively connected to the processor; wherein the processor, the memory and the communication interface communicate with each other; the memory stores program instructions executable by the processor, and the processor calls the program instructions to execute a multi-objective optimization design method for a green building based on BO-CatBoost-NSGA-Ⅲ.

[0038] The fourth object of the present invention is to provide a non-transitory computer-readable storage medium, on which computer instructions are stored, and the computer instructions enable the computer to execute the green building multi-objective optimization method based on BO-CatBoost-NSGA-Ⅲ provided by any one of the above-mentioned various implementation methods.

[0039] In general, the above technical solution conceived by the present invention has the following technical advantages compared with the prior art:

[0040] 1. The present invention uses the BO-CatBoost algorithm to establish a high-precision prediction model for building energy consumption, carbon emissions and thermal comfort, and adopts the RF algorithm to score the importance of input variables to ensure that the prediction results of building energy consumption, carbon emissions and thermal comfort are more accurate and reliable.

[0041] 2. The present invention replaces the traditional mathematical function with the BO-CatBoost prediction model function as the genetic algorithm fitness function for multi-objective optimization, accurately mapping the complex nonlinear relationship between building energy consumption, carbon emissions and thermal comfort and building design characteristic parameters, and achieving more precise optimization.

[0042] 3. The present invention uses NSGA-Ⅲ to establish a multi-objective optimization model. Compared with the traditional genetic algorithm, the NSGA-Ⅲ algorithm has an elite strategy, which ensures the superiority of individuals and the diversity of the population, with high optimization accuracy and fast convergence speed.

[0043] 4. The present invention determines the coefficient R 2 The performance analysis of the prediction model using two commonly used indicators, mean square error (MSE) and mean square error (MSE), verified the effectiveness and correctness of the BO-CatBoost model in predicting building energy consumption, carbon emissions and thermal comfort.

[0044] 5. The present invention adopts the ideal point method to select the optimal decision-making plan, and determines a set of optimal solutions that make the multi-objective function reach the optimal state from the Pareto frontier solution set obtained by the NSGA-Ⅲ algorithm. It comprehensively considers the states of the three objective functions, has a simple acquisition method, and the obtained results are more objective. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A flow chart of a method and device for optimizing the displacement of bridge piers during shield tunneling construction based on BO-RF-NSGA-III provided in an embodiment of the present invention;

[0046] Figure 2 A schematic diagram of the structure of a green building multi-objective optimization method and device based on BO-CatBoost-NSGA-III provided in an embodiment of the present invention;

[0047] Figure 3 A schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention;

[0048] Figure 4 A diagram of energy consumption simulation results provided by an embodiment of the present invention;

[0049] Figure 5 A search result diagram of building energy consumption hyperparameters provided by an embodiment of the present invention;

[0050] Figure 6 A graph showing the search results for carbon emission hyperparameters provided in an embodiment of the present invention;

[0051] Figure 7 A thermal comfort hyperparameter search result diagram provided by an embodiment of the present invention;

[0052] Figure 8 A schematic diagram of prediction results of a building energy consumption training sample set provided by an embodiment of the present invention;

[0053] Fig. 9 A schematic diagram of prediction results of a building energy consumption test sample set provided by an embodiment of the present invention;

[0054] Fig.10A schematic diagram of prediction results of a carbon emission training sample set provided by an embodiment of the present invention;

[0055] Fig.11 A schematic diagram of prediction results of a carbon emission test sample set provided by an embodiment of the present invention;

[0056] Fig.12 A schematic diagram of prediction results of a thermal comfort training sample set provided in an embodiment of the present invention;

[0057] Fig.13 A schematic diagram of prediction results of a thermal comfort test sample set provided by an embodiment of the present invention;

[0058] Fig.14 A ranking diagram of the importance of influencing factors provided by an embodiment of the present invention;

[0059] Fig.15 A schematic diagram of the Pareto frontier ideal point effect provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0061] like Figure 1 As shown, an embodiment of the present invention provides a green building multi-objective optimization method based on BO-CatBoost-NSGA-Ⅲ, which mainly includes the following steps.

[0062] Step 1: Data acquisition and preprocessing.

[0063] The green building is simulated using BIM-DB software and the building data sample set is obtained using orthogonal experiments.

[0064] A total of 10 parameters, including window-to-wall ratio, exterior wall heat transfer coefficient, roof heat transfer coefficient, exterior window heat transfer coefficient, exterior wall solar radiation absorption coefficient, roof solar radiation absorption coefficient, exterior window solar heat gain coefficient, air tightness, air-conditioning heating setting temperature and air-conditioning cooling setting temperature, were taken as input variables, and three parameters, including building energy consumption, carbon emissions and thermal comfort, were taken as output variables. An orthogonal experiment was used to obtain a sample set of building data.

[0065] As a preferred embodiment, 70 sets of model simulation experiments are designed in this embodiment, and these 70 sets of simulation results are used as the data sample set of the BO-CatBoost prediction model. The specific level division of the input variables is shown in the following table, and the energy consumption simulation results are shown in Figure 4.

[0066] The data samples are normalized, and 80% of the samples are randomly selected to form a training sample set for training the model. In order to test the generalization performance of the model, the remaining 20% ​​of the samples are used as a test sample set to verify the model effect.

[0067] The specific level division of input variables is shown in the following table.

[0068]

[0069] Data preprocessing: Data normalization can scale the input data within a certain range, thereby eliminating the impact of the dimensions of different indicators on prediction efficiency and accuracy; considering that normalizing the data to the [-1, 1] interval is better than the [0, 1] interval to avoid drowning the characteristics of the input vector, this patent application chooses to scale the data samples for different targets to the [-1, 1] interval.

[0070] Step 2: Determine the hyperparameters of the CatBoost prediction model.

[0071] Bayesian optimization (BO) was used to optimize the hyperparameters of the gradient boosting algorithm (Categorical Boosting, CatBoost), and a BO-CatBoost prediction model was constructed. The data sample set was randomly divided into a training sample set and a test sample set, and the parameter optimization results were input into the model. A training model was established based on Python, and the training sample set and the test sample set were used to optimize the parameters and analyze the prediction accuracy of the BO-CatBoost prediction model. The mean square error and determination coefficient were used to evaluate the accuracy of the BO-CatBoost prediction model, and the building energy consumption, carbon emissions and thermal comfort were predicted. The importance of the input variables was scored based on the Random Forest (RF) algorithm. The specific steps are as follows.

[0072] (2.1) Build a regression model based on CatBoost. CatBoost consists of Categorical and Boosting, which can process categorical features efficiently and reasonably.

[0073] First, we count the categorical features and calculate the frequency of a category feature. We then consider the hyperparameters to generate new numerical features. Hyperparameters will affect the robustness of the regression prediction results and need to be optimized. We train the regression model with the hyperparameters learning_rate, colsample_bytree, minibatch_frac, and n_estimators in turn.

[0074] (2.2) In order to make the CatBoost regression model obtain better prediction results, this paper uses the Bayesian algorithm to optimize the four hyperparameters of the CatBoost regression model: learning_rate, minibatch_frac, colsample_bytree, and n_estimators.

[0075] learning_rate, called learning rate, is a continuous parameter used to reduce the gradient step size. This parameter will affect the total training time. The smaller its value, the more iterations required for training. The optimization range of this parameter is set between 0.05 and 0.2.

[0076] colsample_bytree, that is, the subsample ratio of the column when constructing each tree, its main function is to control the complexity of the tree model.

[0077] minibatch_frac is a parameter used to control the proportion of training samples used in each iteration. Its value range is (0,1], which indicates the proportion of training samples used in each iteration and can be used to control the speed and memory usage of the training process.

[0078] n_estimators is the number of tree models, which is used to limit the growth of the tree algorithm. The setting range of this parameter will affect the establishment of the tree model and its regression fitting performance.

[0079] The 5-fold cross-validation method is used to verify the model accuracy, so as to determine the parameter combination with the highest accuracy as the optimal parameter of the CatBoost prediction model.

[0080] (2.3) Establishment and evaluation of BO-CatBoost prediction model: Based on the hyperparameter optimization results, the training set was used for learning simulation, and CatBoost building energy consumption, carbon emission and thermal comfort prediction models were established respectively. The prediction models of the training sample set were then tested using the test sample set. The vertical displacement and horizontal displacement prediction models of bridge piers based on the random forest algorithm were established respectively to obtain the prediction results.

[0081] In this embodiment, the mean square error MSE and the absolute coefficient R are introduced. 2 There are two commonly used indicators to evaluate the accuracy of the BO-CatBoost prediction model; MSE can better reflect the deviation between the predicted value and the observed value, and R 2 It is used to measure the degree of fit between the predicted value and the observed value. The smaller the MSE value, the better the model performance. 2 The closer it is to 1, the higher the model prediction accuracy:

[0082]

[0083]

[0084] Where n is the number of samples, is the predicted value, y i is the actual observed value, is the average of the actual observed values.

[0085] Based on the random forest algorithm, the importance of 10 building design feature parameters was scored with building energy consumption, carbon emissions and thermal comfort as prediction indicators. After determining the important factors for building energy consumption, carbon emissions and thermal comfort, these factors should be fully considered in the design, more attention should be paid to them, and appropriate building materials should be selected to obtain ideal results.

[0086] In order to make the CatBoost prediction model obtain better prediction results, this patent application uses Bayesian optimization BO to optimize the three hyperparameters of the model, namely, tree depth (depth), L2 regularization parameter (12_leaf_reg) and learning rate (learning_rate), and combines the 5-fold cross-validation method to verify the model accuracy, thereby determining the hyperparameter combination with the highest accuracy as the optimal hyperparameter of the CatBoost prediction model.

[0087] Before optimizing the hyperparameters, we first give the initial ranges of the hyperparameters. The initial ranges of the three parameters depth, l2_leaf_reg and learning_rate are [4, 10], [1, 15] and [0.03, 0.05] respectively. The hyperparameter search results for building energy consumption, carbon emissions and thermal comfort are as follows: Figure 5 , Figure 6 , Figure 7 shown.

[0088] from Figure 5 It can be seen that when depth = 8, l2_leaf_reg = 1, and learning_rate = 0.03, the CatBoost prediction model of building energy consumption has the best performance.

[0089] from Figure 6It can be seen that when depth = 4, l2_leaf_reg = 1, and learning_rate = 0.05, the CatBoost prediction model of carbon emissions has the best performance.

[0090] from Figure 7 It can be seen that when depth = 4, l2_leaf_reg = 3, and learning_rate = 0.05, the CatBoost prediction model of thermal comfort has the best performance.

[0091] Prediction result analysis: Based on the parameter optimization results, the training sample set was used for learning simulation and the test sample set was used for verification. The vertical displacement and horizontal displacement prediction models of the RF bridge pier were established respectively. The prediction results were as follows: Figure 8 , Fig. 9 , Fig.10 , Fig.11 , Fig.12 , Fig.13 shown.

[0092] Depend on Figure 8 and Fig. 9 It can be seen that the CatBoost model can well predict the changes in building energy consumption. The mean square error MSE of the training sample set for predicting building energy consumption based on the CatBoost model is 0.00380, and the determination coefficient R 2 is 0.982, the mean square error MSE of the test sample set is 0.0228, and the determination coefficient R 2 It is 0.875, which means that the model has fully learned the rules between the input variables and output variables of the sample set, and the fitting effect is good.

[0093] from Fig.10 and Fig.11 It can be found that the CatBoost model can make a good prediction of the changes in carbon emissions. The mean square error MSE of the training sample set for carbon emission prediction based on the CatBoost model is 0.00972, and the determination coefficient R 2 is 0.957, the mean square error MSE of the test sample set is 0.00792, and the determination coefficient R 2 It is 0.957, which means that the model has fully learned the rules between the input variables and output variables of the sample set and can obtain prediction results with high accuracy.

[0094] from Fig.12 and Fig.13 It can be found that the CatBoost model can make a good prediction of the change of thermal comfort. The mean square error MSE of the training sample set for thermal comfort prediction based on the CatBoost model is 0.0609, and the determination coefficient R 2is 0.866, the mean square error MSE of the test sample set is 0.0534, and the determination coefficient R 2 It is 0.853, which means that the model has fully learned the rules between the input variables and output variables of the sample set and can obtain prediction results with high accuracy.

[0095] Importance analysis of architectural design characteristic parameters based on RF: In order to measure the importance of input variables, that is, the impact of input variable changes on the Gini coefficient, noise interference is randomly added to each input factor, and then the change in model accuracy is observed, and the magnitude of the decrease in accuracy is used to measure the importance of the characteristic variable. This patent application uses random forest to evaluate the importance based on OBB error, scores the importance of each input indicator, and uses the RF package to build a model to achieve the importance scoring process of different influencing factors. The importance ranking results are as follows: Fig.14 .

[0096] According to the importance ranking diagram of input variables, considering the building structure design factors at the same time, airtightness, solar heat gain coefficient of external windows, and heat transfer coefficient of external walls are several factors with high importance in affecting building performance; window-to-wall ratio, heat transfer coefficient of external windows, solar radiation absorption coefficient of external walls, air conditioning heating setting temperature, and air conditioning cooling setting temperature are factors with low impact on the prediction results. The reason is that in areas with hot summers and cold winters, the temperature difference between indoor and outdoor is large in winter, and high airtightness prevents heat loss and reduces heat loss; the temperature difference between indoor and outdoor is small in summer, and improved airtightness has little effect on energy consumption; high airtightness in the transition season is not conducive to ventilation and indoor air quality, so airtightness is the key factor affecting thermal comfort and energy consumption, ranking first and second respectively.

[0097] In areas with hot summers and cold winters, the problem of summer cooling needs to be considered. The solar radiation absorption coefficient of the external windows affects the ventilation and air conditioning operation load of the building, causing a large air conditioning energy loss, and thus becomes a key factor affecting building energy consumption. In the process of orthogonal experiments, the relevant specification documents were referred to for the air conditioning setting temperature, so that the air conditioning range is within a smaller reasonable range. This setting reduces the importance of air conditioning to building energy consumption and carbon emissions. After determining the important factors for building energy consumption, carbon emissions and thermal comfort, these factors should be fully considered in the design, given more attention, and appropriate building materials should be selected to achieve ideal results.

[0098] Step three, construct a BO-CatBoost-NSGA-Ⅲ multi-objective optimization model, and use the established BO-CatBoost prediction model function as the fitness function of the non-dominated sorting genetic algorithm-Ⅲ (NSGA-Ⅲ) with an elite strategy to optimize building energy consumption, carbon emissions and thermal comfort.

[0099] (3.1) Determine the objective function and variable constraint range.

[0100] Determine the objective function: Introduce the relationship between the input variables of BO-CatBoost regression prediction fitting and building energy consumption, carbon emissions and thermal comfort as the fitness function in the multi-objective genetic algorithm, which can clearly reflect the complex nonlinear relationship between the building design characteristic parameters (input variables) and building energy consumption, carbon emissions and thermal comfort.

[0101] Based on the CatBoost prediction model, the NSGA-III algorithm is used to perform multi-objective optimization on building energy consumption, carbon emissions and thermal comfort. The core problem to be solved by multi-objective optimization is to balance the contradictions between these objectives. Under the condition of optimal thermal comfort, building energy consumption and carbon emissions should be minimized. The corresponding objective function of the optimization model, that is, the NSGA-III fitness function, is obtained by the CatBoost prediction model functions of building energy consumption, carbon emissions and thermal comfort. f1, f2 and f3 are:

[0102]

[0103] Where x1, x2, x3, x4, x5, x6, x7, x8, x9 and x 10 They are window-to-wall ratio, exterior wall heat transfer coefficient, roof heat transfer coefficient, exterior window heat transfer coefficient, exterior wall solar radiation absorption coefficient, roof solar radiation absorption coefficient, exterior window solar heat gain coefficient, air tightness, air conditioning heating setting temperature and air conditioning cooling setting temperature.

[0104] (3.2) Set constraints on the values ​​of decision parameters.

[0105] Determine the constraint range: In order to make the optimization design of construction parameters have practical significance and ensure that the obtained construction parameter setting scheme is more reasonable and feasible without safety hazards, it is necessary to set constraints on the values ​​of decision parameters according to engineering requirements and existing project data. The constraints are generally expressed as In the formula and They represent the lower and upper limits of the operating parameter values ​​respectively.

[0106] When performing parameter optimization calculations, initial data, i.e., population, is first required. The decision range of the population needs to be set to ensure that the initial population has practical significance. In order to avoid the large difference between the optimized shield machine parameter results and the actual engineering parameters of the shield machine, resulting in a large parameter adjustment span and possible safety hazards, the range of the measured data is used as the main reference to set the range of the initial parameters. The variable constraints set are shown in the following table.

[0107]

[0108] (3.3) Multi-objective optimization based on NSGA-III.

[0109] 1) Set the initial state of the decision variables and randomly generate decision variables according to the given variable boundary restrictions, thereby generating an initialized population with N individuals.

[0110] 2) Use the three mechanisms of selection, crossover, and mutation to operate on a population of size N to obtain the first generation of offspring populations of the same size as the parent population.

[0111] 3) Combine the offspring population with its parent population to form a mixed population, perform fast non-dominated sorting on the individuals in the mixed population, and select individuals from different dominance levels to join the next generation.

[0112] 4) Generate reference points based on subjective preferences or unified settings, connect the origin and the reference points to form a reference line, and then normalize the objective function and the algorithm population. Calculate the vertical distance between each individual and the reference line, and associate the individual with the smallest vertical distance with the reference point.

[0113] 5) Find the reference point with the least number of associated solutions, and select the solution according to the number of associated solutions of the reference point. If the number of solutions is not 0, select an individual at random; if the number of solutions is 0, select the individual with the smallest vertical distance to the reference point. Repeat the solution selection process until the population size is N, and output the Pareto optimal solution set.

[0114] (3.4) The ideal point method is used to obtain the optimal solution.

[0115] The ideal point method is used to select the optimal solution from the Pareto optimal solution set, so that its target values ​​are as close to the ideal value as possible; the n groups of solutions in the Pareto optimal solution set are normalized so that the optimal value of the objective function is evenly distributed between its related minimum and maximum values, eliminating the influence of different objective function dimensions; the point composed of the target values ​​corresponding to each single objective when it is optimal That is the ideal point; after determining the coordinates of the ideal point E, the distances from each point in the Pareto optimal solution set to the ideal point are calculated as follows:

[0116]

[0117] The point (x pareto ,y pareto ,z pareto ) are the coordinates of each point in the Pareto optimal solution set, and the point with the shortest distance is the optimal solution selected from the Pareto optimal solution set.

[0118] Taking building energy consumption, carbon emissions and thermal comfort as the three optimization objectives, the population type of the CatBoost-NSGA-Ⅲ algorithm is set to double vector, the number of targets of the genetic algorithm is set to 3, the population size is set to 100, and the maximum evolutionary generation is set to 6. After determining the above algorithm parameter settings, multi-objective optimization is performed according to the steps of the NSGA-Ⅲ algorithm to obtain the optimal Pareto frontier, which contains 40 pairs of optimal solutions, such as Fig.15 shown.

[0119] When the architectural design characteristic parameters are basically consistent with those of the optimal solution, the building energy consumption, carbon emissions and thermal comfort are also basically consistent, which verifies the referenceability of the multi-objective optimization results. Therefore, when thermal comfort is as good as possible, building energy consumption and carbon emissions are as low as possible, and the architectural design characteristic parameters of the optimal solution can be used as reference values, providing a reference for the parameter setting of green buildings.

[0120] Step 4: Based on the obtained Pareto optimal solution set, use the ideal point method to determine the building design characteristic parameters that meet the optimal building energy consumption, carbon emissions and thermal comfort.

[0121] From the optimal solution set results, we know that the coordinates of the ideal point are E (125727.00, 74337.20, 0.01). By using the three-target distance calculation formula and the ideal point method, we know that the 26th group of Pareto optimal solution points P (126470.00, 74716.60, 0.18) has the shortest distance from the ideal point E, which is 0.1868441. That is, the 26th group of Pareto optimal solutions is the only optimal solution for this optimization.

[0122] The design solution corresponding to point P is compared with the original solution, and the optimization of the design solution is shown in the following table.

[0123]

[0124] The comparison of optimization target values ​​is shown in the following table.

[0125]

[0126] From the above table, we can see that the window-to-wall ratio, exterior wall heat transfer coefficient, roof heat transfer coefficient, exterior window heat transfer coefficient, roof solar radiation absorption coefficient, exterior window solar heat gain coefficient, air tightness and air conditioning heating setting temperature of the optimized building are all smaller than those of the initial building, while the exterior wall solar radiation absorption coefficient and air conditioning cooling setting temperature are larger than those of the initial building. In terms of natural dimension, building energy consumption decreased by 7.64%, about 10467.29MWh; in terms of environmental dimension, carbon emissions decreased by 5591.31t, about 6.96%; in terms of social dimension, thermal comfort dropped to 0.18, a decrease of 63.95%, closer to thermal neutrality. After multi-objective optimization, building energy consumption and carbon emissions decreased, the annual average indoor thermal comfort was significantly improved, and the overall building was optimized.

[0127] The present invention uses BIM-DB software to simulate green buildings and uses orthogonal experiments to obtain a sample set of building data; uses BO to optimize the hyperparameters of CatBoost, constructs a BO-CatBoost prediction model, predicts building energy consumption, carbon emissions and thermal comfort, and scores the importance of input variables based on the RF algorithm; constructs a BO-CatBoost-NSGA-Ⅲ multi-objective optimization model, uses the established BO-CatBoost prediction model function as the NSGA-Ⅲ fitness function, and optimizes building energy consumption, carbon emissions and thermal comfort. According to the obtained Pareto optimal solution set, the optimal solution of building energy consumption, carbon emissions and thermal comfort is determined by the ideal point method. The present invention can achieve energy saving, low carbon emissions, cost reduction and improved comfort of buildings without relying on manual control.

[0128] Reference Figure 2 , Figure 3 As shown, a green building multi-objective optimization design device disclosed in the present invention includes: a first main module: selecting building design feature parameters as input variables, selecting building energy consumption, carbon emissions and thermal comfort as output variables, and collecting data to establish an original sample set; a second main module, using the BO-CatBoost prediction model to predict building energy consumption, carbon emissions and thermal comfort, and scoring the importance of input variables based on the RF algorithm; a third main module, using the NSGA-Ⅲ algorithm to perform multi-objective optimization decisions on building energy consumption, carbon emissions and thermal comfort.

[0129] An electronic device disclosed in the present invention comprises: at least one processor, at least one memory and a communication interface communicatively connected to the processor; wherein the processor, the memory and the communication interface communicate with each other; the memory stores program instructions executable by the processor, and the processor calls the program instructions to execute a multi-objective optimization design method for green buildings based on BO-CatBoost-NSGA-Ⅲ.

[0130] The present invention discloses a non-transitory computer-readable storage medium, on which computer instructions are stored, and the computer instructions enable the computer to execute a green building multi-objective optimization method based on BO-CatBoost-NSGA-Ⅲ provided by any one of the above-mentioned various implementations.

[0131] It is easy for those skilled in the art to understand that the above description is only a preferred use case of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-objective optimization design method for green buildings, characterized by: The following steps are included Step 1: Use BIM-DB software to simulate green buildings, select window-to-wall ratio, exterior wall heat transfer coefficient, roof heat transfer coefficient, exterior window heat transfer coefficient, exterior wall solar radiation absorption coefficient, roof solar radiation absorption coefficient, exterior window solar heat gain coefficient, air tightness, air conditioning heating setting temperature and air conditioning cooling setting temperature as input variables, and use building energy consumption, carbon emissions and thermal comfort as output variables, and use orthogonal experiments to obtain a sample set of building data; Step 2: Use the Bayesian optimization algorithm to optimize the hyperparameters of the gradient boosting algorithm, build a BO-CatBoost prediction model, divide the data sample set into a training sample set and a test sample set, optimize the parameters of the BO-CatBoost prediction model and analyze the prediction accuracy. Use the mean square error and determination coefficient to evaluate the accuracy of the BO-CatBoost prediction model, and predict building energy consumption, carbon emissions, and thermal comfort. Step 3: Construct a BO-CatBoost-NSGA-Ⅲ multi-objective optimization model, and use the established BO-CatBoost prediction model function as the fitness function of the non-dominated sorting genetic algorithm with an elite strategy to optimize building energy consumption, carbon emissions, and thermal comfort. Step 4: Based on the obtained Pareto optimal solution set, the optimal solution of building energy consumption, carbon emissions and thermal comfort is determined by the ideal point method.

2. A multi-objective optimization design method for green buildings according to claim 1, characterized in that: In step 2, the BO-CatBoost prediction model is constructed through the following steps: (2.1) Build a regression model based on CatBoost. First, count the categorical features, calculate the frequency of a category feature, consider the hyperparameters to generate new numerical features, and train the regression model with hyperparameters learning_rate, colsample_bytree, minibatch_frac and n_estimators in sequence. (2.2) The Bayesian algorithm is used to optimize the learning_rate, minibatch_frac, colsample_bytree, and n_estimators of the CatBoost regression model. The learning_rate optimization range is set to 0.05-0.2, and the minibatch_frac optimization range is set to (0,1]. The model accuracy is verified by combining the 5-fold cross-validation method, and the parameter combination with the highest accuracy is determined as the optimal parameter of the regression model. (2.3) Based on the hyperparameter optimization results, the training set is used for learning simulation, and CatBoost building energy consumption, carbon emission and thermal comfort prediction models are established respectively. The prediction models of the training sample set are then tested using the test sample set. The vertical displacement and horizontal displacement prediction models of bridge piers based on the random forest algorithm are established respectively to obtain the prediction results.

3. A multi-objective optimization design method for green buildings according to claim 2, characterized in that: In the step 2, the mean square error MSE and the absolute coefficient R 2 Evaluate the accuracy of the BO-CatBoost prediction model, including MSE and R 2 It is calculated by the following formula: Where n is the number of samples, is the predicted value, y i is the actual observed value, is the average of the actual observed values.

4. A multi-objective optimization design method for green buildings according to claim 1, 2 or 3, characterized in that: The step three specifically includes the following steps: (3.1) The relationship between the input variables of BO-CatBoost regression prediction fitting and building energy consumption, carbon emissions and thermal comfort is introduced as the fitness function in the multi-objective genetic algorithm. The NSGA-III fitness function of the CatBoost building energy consumption, carbon emissions and thermal comfort prediction model is determined by the CatBoost prediction regression equation. The objective functions f1, f2 and f3 are: Where x1, x2, x3, x4, x5, x6, x7, x8, x9 and x 10 They are window-to-wall ratio, exterior wall heat transfer coefficient, roof heat transfer coefficient, exterior window heat transfer coefficient, exterior wall solar radiation absorption coefficient, roof solar radiation absorption coefficient, exterior window solar heat gain coefficient, air tightness, air conditioning heating setting temperature, and air conditioning cooling setting temperature; (3.2) Setting constraints on the values ​​of decision parameters In the formula and Respectively represent the lower and upper limits of the operating parameter values; (3.3) Multi-objective optimization based on NSGA-III; (3.4) The ideal point method is used to select the optimal solution from the Pareto optimal solution set, so that each objective value is as close to the ideal value as possible; the point composed of the objective values ​​corresponding to each single objective when it is optimal is the ideal point method formula; after determining the coordinates of the ideal point E, the distances from each point in the Pareto optimal solution set to the ideal point are calculated as follows: The point (x pareto ,y pareto ,z pareto ) are the coordinates of each point in the Pareto optimal solution set, and the point with the shortest distance is the optimal solution selected from the Pareto optimal solution set.

5. A multi-objective optimization design method for green buildings according to claim 4, characterized in that: In the step (3.3), the NSGA-III algorithm is used to realize the intelligent optimization problem of shield tunnel optimization design, and the specific implementation steps are as follows: 1) Set the initial state of the decision variables, randomly generate decision variables according to the given variable boundary restrictions, and thus generate an initialized population with N individuals; 2) Use the three mechanisms of selection, crossover, and mutation to operate on a population of size N to obtain the first generation of offspring populations of the same size as the parent population; 3) Combine the offspring population with its parent population to form a mixed population, perform fast non-dominated sorting on the individuals in the mixed population, and select individuals from different dominance levels to join the next generation; 4) Generate reference points based on subjective preferences or unified settings, connect the origin and the reference points to form a reference line, and then normalize the objective function and the algorithm population. Calculate the vertical distance between each individual and the reference line, and associate the individual with the smallest vertical distance with the reference point; 5) Find the reference point with the least number of associated solutions, and select the solution according to the number of associated solutions of the reference point. If the number of solutions is not 0, select an individual at random; if the number of solutions is 0, select the individual with the smallest vertical distance to the reference point. Repeat the solution selection process until the population size is N, and output the Pareto optimal solution set.

6. A green building multi-objective optimization design device, characterized in that: include: The first main module collects data to establish the original sample set, using architectural design characteristic parameters as input variables and building energy consumption, carbon emissions and thermal comfort as output variables; The BO-CatBoost prediction model is used to predict building energy consumption, carbon emissions and thermal comfort, and the second main module scores the importance of input variables based on the RF algorithm; the third main module uses the NSGA-Ⅲ algorithm to make multi-objective optimization decisions on building energy consumption, carbon emissions and thermal comfort.

7. An electronic device, characterized in that: include: At least one processor, at least one memory and a communication interface communicatively connected to the processor; wherein the processor, the memory and the communication interface communicate with each other; The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the method according to any one of claims 1 to 5.

8. A readable storage medium, characterized in that: The readable storage medium stores computer instructions, and the computer instructions enable a computer to execute the method according to any one of claims 1 to 5.

Citation Information

Patent Citations

  • Shield under-crossing tunnel optimization design method and device based on CatBoost-NSGA-III

    CN114969953A

  • Shield under-crossing construction pier displacement optimization method and equipment based on BO-RF-NSGA-III

    CN116776410A

  • Zero-carbon building multi-objective optimization method based on machine learning hybrid modeling

    CN117574785A

  • Tunneling optimization control method and device of shield tunneling machine

    CN117759269A

  • System anomaly detection method and device and storage medium

    CN117786576A

Cited By

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

    CN119940159A

  • Online proxy model assisted multi-objective optimization method for green hydrogen coupling hydrogenation process

    CN120199351A

  • GAN and RF-NSGA integration algorithm-based gymnasium wind-solar-energy collaborative optimization design method

    CN121389288A

  • Coordinated optimization design method of gymnasium wind environment based on gan and rf-nsga integrated algorithm

    CN121389288B

  • BIM building facade dynamic photo-thermal balance design method and system

    CN121615514A