A method for evaluating the energy-saving effect of green buildings based on BP neural networks
By constructing a green building energy-saving effect evaluation method based on BP neural network, the shortcomings of existing evaluation systems and the problem of subjectivity are solved, and multi-dimensional quantification and high-precision evaluation are achieved, which can adapt to technological iteration and multi-scenario applications.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUANGGANG NORMAL UNIV
- Filing Date
- 2026-01-28
- Publication Date
- 2026-06-02
AI Technical Summary
The existing green building evaluation system has shortcomings in its indicator system, strong subjectivity in its evaluation methods, and lack of unified standards, making it difficult to achieve objective and accurate assessment and comparability.
A green building energy-saving effect evaluation method based on BP neural network is constructed. Fourteen secondary evaluation indicators are selected by Delphi method, weights are determined by AHP method, and BP neural network is used for self-optimization to establish evaluation model.
It has implemented a multi-dimensional quantitative indicator system, which improves the objectivity and accuracy of the evaluation, with an error of no more than 5%. It is adaptable and scalable, and supports technology iteration.
Smart Images
Figure CN122134169A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method for evaluating the energy-saving effect of green buildings based on BP neural networks, which belongs to the field of green building technology evaluation. Background Technology
[0002] With the continuous expansion of my country's construction industry, the annual increase in building area is enormous. However, a large number of residential and public buildings face significant resource depletion and environmental pollution problems throughout their entire life cycle of planning, construction, and operation. As a pillar industry of the national economy, the construction industry's dual characteristics of "high pollution and high energy consumption" have put it in a dilemma between development needs and environmental policy constraints, making the transformation to green building an inevitable path for the industry's sustainable development.
[0003] Green building, with its core focus on the environmental friendliness of the entire building material, construction technology, and operation model, has been proven in practice to be an effective path that balances industry development and ecological protection. By the end of 2020, more than 4,500 green building projects had been launched nationwide, with a completed green building area exceeding 50,000 square meters, fully demonstrating the practical value and industry recognition of green building. However, in actual implementation, the development of green building still faces many technical bottlenecks, among which the imperfection of the evaluation system is one of the key constraints.
[0004] Currently, my country's green building evaluation system faces three core problems: First, the indicator system has shortcomings. Energy consumption evaluation only focuses on building energy consumption and utilization efficiency, lacking quantitative indicators for key dimensions such as renewable energy utilization rate and building-environment interaction. Second, the evaluation methods mainly combine qualitative and quantitative approaches, resulting in strong subjectivity and insufficient reliability of data sources, making it difficult to achieve objective and accurate assessment. Third, there are significant differences in energy-saving standards for green buildings between domestic and international markets, and the lack of a unified evaluation standard and certification system leads to a lack of comparability in energy-saving evaluation results for different regions and projects, severely hindering the overall green transformation process of the industry. Summary of the Invention
[0005] To address the problems existing in the prior art, this invention provides a green building energy-saving effect evaluation method based on BP neural network. This invention can accurately determine the index weights using a multi-dimensional quantitative index system and the AHP method, and can ensure high accuracy of the evaluation results through the strong data processing and self-optimization capabilities of BP neural network.
[0006] To achieve the above objectives, the technical solution provided by this invention is: a method for evaluating the energy-saving effect of green buildings based on BP neural networks, comprising the following steps:
[0007] S1. Using the Delphi method and comparative analysis, 14 secondary evaluation indicators were selected from five dimensions: energy-saving design, environmental energy consumption, building lighting, building ventilation, and building sunshine. The quantitative standards for each secondary evaluation indicator were clarified, and the energy-saving evaluation indicators were quantified. A green building energy-saving effect evaluation indicator system was constructed. Among them, energy-saving design, environmental energy consumption, building lighting, building sunshine, and building ventilation are primary evaluation indicators.
[0008] S2. The weights of the secondary evaluation indicators for energy-saving effects of green buildings are determined by the AHP method to eliminate subjective bias in the traditional weight allocation.
[0009] S3. Set standards for evaluating the energy efficiency of green buildings, establish an evaluation model for the energy efficiency of green buildings based on a BP neural network, and then determine the energy efficiency level of green buildings through the evaluation model.
[0010] A further improvement to the above scheme is as follows:
[0011] The 14 secondary evaluation indicators in step S1 are as follows: design quality, building envelope energy-saving quality and energy-saving material utilization in the energy-saving design dimension; lighting power density, comprehensive energy utilization and renewable energy utilization in the environmental energy consumption dimension; indoor lighting, external window anti-glare and underground lighting in the building lighting dimension; wind environment, indoor natural ventilation and indoor air quality in the building ventilation dimension; and building sunlight conditions and external shading effect in the building sunlight dimension.
[0012] Step S2 specifically includes:
[0013] (1) Constructing a judgment matrix using the 1-9 scale: Using numbers 1-9 to evaluate the relative importance of two secondary evaluation indicators, where 1 indicates that the two secondary evaluation indicators are of equal importance, and 9 indicates that the importance of one secondary evaluation indicator is much greater than that of the other; among which, the secondary evaluation indicators and The judgment scale is Secondary evaluation indicators and The judgment scale is ;
[0014] (2) Establish AHP quantification standard: The formula for calculating the weight vector is: In the formula, Represents the total number of secondary evaluation indicators. Represents secondary evaluation indicators;
[0015] The above formula Normalization, i.e., finding the second half ; by eigenvectors The weights of the indicators are calculated using the following formula:
[0016] ;
[0017] In the formula, This represents the largest eigenvalue of the judgment matrix. It is its corresponding feature vector;
[0018] (3) Consistency check of the judgment matrix: When constructing the judgment matrix, a consistency index is calculated. and consistency ratio The consistency of the judgment matrix is evaluated and verified.
[0019] ; The average random consistency index is determined by referring to the standard table.
[0020] The formula for the consistency ratio is:
[0021] ;
[0022] Let the threshold be 0.1, that is, when If the condition is met, the judgment matrix is considered consistent; otherwise, the judgment matrix is reconstructed until the consistency requirement is met. .
[0023] Step S3 specifically includes:
[0024] (1) Determine the parameters of the BP neural network model, treat each secondary evaluation index as a neuron, and adjust the weights of each neuron. And thresholds, allocate and adjust the weights of each secondary evaluation indicator;
[0025] (2) Using the quantitative data of 14 secondary evaluation indicators of green building energy-saving effect as input data, the output value of the BP neural network model is between 0 and 1. Based on the output value of the BP neural network model, four different evaluation levels are divided. The final output value is 0.8-1, which is an excellent evaluation level; 0.7-0.8, which is a good evaluation level; 0.6-0.7, which is a medium evaluation level; and 0-0.6, which is a poor evaluation level.
[0026] (3) Using the quantitative data of 14 secondary evaluation indicators of green building projects as samples, the training set and the test set were divided, and the weights of the neurons were adjusted. The BP neural network model is optimized with thresholds to ensure that the relative error between the test value and the target value is no more than 5% and that the evaluation level is consistent. Then, the energy-saving level of green buildings is determined using the optimized BP neural network model.
[0027] As can be seen from the above technical solution, the present invention provides a green building energy-saving effect evaluation method based on BP neural network. Through the Delphi method and comparative analysis, 14 evaluation indicators are selected from five dimensions—energy-saving design, environmental energy consumption, building lighting, building ventilation, and building sunshine—to construct a quantitative indicator system. Then, the AHP method is used to determine the indicator weights and a consistency test is performed to eliminate subjective bias. Finally, a BP neural network evaluation model is established, using the quantitative data of the 14 indicators as input. Through training and optimization, the model is used to determine the green building energy-saving level. The present invention has the following advantages over existing technologies:
[0028] (1) The technical solution adopted in this invention constructs a quantitative indicator system covering multiple dimensions such as energy-saving design and environmental energy consumption, which supplements the shortcomings of traditional evaluation indicators and clarifies the quantitative standards of each indicator, thereby improving the comprehensiveness and operability of the evaluation.
[0029] (2) The technical solution adopted in this invention eliminates the subjective bias of weight allocation through the AHP method, and combines the self-optimization and strong data processing capabilities of the BP neural network to make the relative error of the evaluation result no more than 5%, which significantly improves the objectivity and accuracy of the evaluation.
[0030] (3) The technical solution adopted in this invention reserves an interface for evaluating new energy technologies, supports incremental sample training, and can be flexibly expanded with the iteration of green building technologies. It solves the problem that traditional evaluation methods become ineffective as technology is updated, and has good adaptability and scalability. Attached Figure Description
[0031] Figure 1 Flowchart of Green Building Comprehensive Benefit Evaluation Method;
[0032] Figure 2 Neural network training results;
[0033] Figure 3 Line chart comparing test values and target values for projects 8-10. Detailed Implementation
[0034] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments, but the scope of protection of the present invention is not limited to the following embodiments.
[0035] The present invention provides a method for evaluating the energy-saving effect of green buildings based on a backpropagation neural network, the flowchart of which is shown below. Figure 1 As shown, it includes the following steps:
[0036] S1. Using the Delphi method and comparative analysis, 14 secondary evaluation indicators were selected from five dimensions: energy-saving design, environmental energy consumption, building lighting, building ventilation, and building sunshine. The quantitative standards for each secondary evaluation indicator were clarified, and the energy-saving evaluation indicators were quantified. A green building energy-saving effect evaluation indicator system was constructed. Among them, energy-saving design, environmental energy consumption, building lighting, building sunshine, and building ventilation are primary evaluation indicators.
[0037] The 14 secondary evaluation indicators are as follows: design quality, building envelope energy efficiency, and energy-saving material utilization in the energy-saving design dimension; lighting power density, comprehensive energy utilization, and renewable energy utilization in the environmental energy consumption dimension; indoor lighting, glare prevention of exterior windows, and underground lighting in the building lighting dimension; wind environment, indoor natural ventilation, and indoor air quality in the building ventilation dimension; and building sunlight conditions and external shading effect in the building sunlight dimension.
[0038] (1) Key influencing factors and sources of each primary evaluation indicator: Key influencing factors and sources of energy-saving design are shown in Table 1; Key influencing factors and sources of environmental energy consumption are shown in Table 2; Key influencing factors and sources of building daylighting are shown in Table 3; Key influencing factors and sources of building sunshine are shown in Table 4; Key influencing factors and sources of building ventilation are shown in Table 5;
[0039] Table 1 shows the key influencing factors and their sources in energy-saving design.
[0040]
[0041] Table 2 Key influencing factors and their sources of environmental energy consumption
[0042]
[0043] Table 3 Key influencing factors and their sources of building daylighting
[0044]
[0045] Table 4 Key influencing factors and their sources of building daylighting
[0046]
[0047] Table 5 Key influencing factors and their sources in building ventilation
[0048]
[0049] (2) Collect energy-saving evaluation index parameters and quantify parameters: Energy-saving design parameters are quantified in Table 6; environmental energy consumption parameters are quantified in Table 7; building lighting parameters are quantified in Table 8; building sunshine parameters are quantified in Table 9; building ventilation parameters are quantified in Table 10.
[0050] Table 6 Quantification of Energy-Saving Design Parameters
[0051]
[0052] In this embodiment, the design quality can be categorized as follows: Excellent: Reaching the international leading level, innovative and representative, and able to lead the industry development; Good: In a leading position domestically, with high innovation capability and representativeness, and has made positive contributions to the industry development; Medium: At an average level among similar projects, with a certain degree of innovation and replicability, and has made positive contributions to regional construction; Poor: Basically meets the requirements of building design codes and belongs to ordinary design; Poor: The design has certain problems and is difficult to meet the qualified standard.
[0053] Building envelope energy-saving technologies can be divided into five levels: Level 5: Utilizing the most advanced energy-saving technologies, achieving extremely high energy utilization efficiency and significant energy-saving effects; Level 4: Based on existing energy-saving technologies, employing relatively mature new technologies and materials to achieve excellent energy-saving standards; Level 3: Using conventional methods for renovation and maintenance to achieve appropriate energy-saving effects without affecting the structural safety of the building during use; Level 2: Achieving lower energy-saving targets, only reducing energy consumption and waste generation through some simple and easy-to-implement measures; Level 1: No energy-saving renovation or maintenance has been carried out.
[0054] Energy-saving material utilization levels are classified as follows: Level 1: High-efficiency energy-saving materials and technologies are used, with a utilization rate of over 90%; Level 2: Newer and more mature energy-saving materials and technologies are used, with a utilization rate between 80-90%; Level 3: Medium-quality energy-saving materials and technologies are used, with a utilization rate between 70-80%; Level 4: Only a small amount of basic energy-saving materials and technologies are used, with a utilization rate between 60-70%; Level 5: The utilization rate of energy-saving materials is below 60%.
[0055] Table 7 Quantification of Environmental Energy Consumption Parameters
[0056]
[0057] Table 8 Quantification of Building Daylighting Parameters
[0058]
[0059] Table 9 Quantification of Building Sunlight Parameters
[0060]
[0061] Table 10 Quantification of Building Ventilation Parameters
[0062]
[0063] (3) Construct a green building energy conservation evaluation index system: Based on the above compliance parameters and the five evaluation indicators and 14 secondary evaluation indicators related to green building energy conservation evaluation, a green building energy conservation level evaluation index system is established. The summary of the indicators is shown in Table 11.
[0064] Table 11 Summary of Indicators
[0065]
[0066] S2. The weights for evaluating the energy-saving effects of each green building are determined using the AHP method, eliminating subjective biases in traditional weight allocation.
[0067] (1) Constructing a judgment matrix using the 1-9 scale: Using numbers 1-9 to evaluate the relative importance of two secondary evaluation indicators, where 1 indicates that the two secondary evaluation indicators are of equal importance, and 9 indicates that the importance of one secondary evaluation indicator is much greater than that of the other; among which, the secondary evaluation indicators and The judgment scale is Secondary evaluation indicators and The judgment scale is The scaling of the judgment matrix and its definition are shown in Table 12.
[0068] Table 12 Matrix Scale and Definition Table
[0069]
[0070] (2) Establish AHP quantification standard: The formula for calculating the weight vector is: In the formula, Represents the total number of secondary evaluation indicators. Represents secondary evaluation indicators;
[0071] The above formula Normalization, i.e., finding the second half ; by eigenvectors The weights of the indicators are calculated using the following formula:
[0072] ;
[0073] In the formula, This represents the largest eigenvalue of the judgment matrix. It is its corresponding feature vector;
[0074] (3) Consistency check of the judgment matrix: When constructing the judgment matrix, a consistency index is calculated. and consistency ratio The consistency of the judgment matrix is evaluated and verified.
[0075] ; The average random consistency index is determined by the standard reference table; see Table 13 for the standard reference table.
[0076] Table 13. Values of the Average Random Consistency Index (RI)
[0077]
[0078] The formula for the consistency ratio is:
[0079] ;
[0080] Let the threshold be 0.1, that is, when If the condition is met, the judgment matrix is considered consistent; otherwise, the judgment matrix is reconstructed until the consistency requirement is met. .
[0081] S3. Set standards for evaluating the energy efficiency of green buildings, establish an evaluation model for the energy efficiency of green buildings based on a BP neural network, and then determine the energy efficiency level of green buildings through the evaluation model.
[0082] (1) Determine the parameters of the BP neural network model, treat each secondary evaluation index as a neuron, and adjust the weights of each neuron. And thresholds, allocate and adjust the weights of each secondary evaluation indicator;
[0083] (2) Using the quantitative data of 14 secondary evaluation indicators of green building energy-saving effect as input data, the output value of the BP neural network model is between 0 and 1. Based on the output value of the BP neural network model, four different evaluation levels are divided. The final output value is 0.8-1, which is an excellent evaluation level; 0.7-0.8, which is a good evaluation level; 0.6-0.7, which is a medium evaluation level; and 0-0.6, which is a poor evaluation level.
[0084] (3) Using the quantitative data of 14 secondary evaluation indicators of green building projects as samples, the training set and the test set were divided, and the weights of the neurons were adjusted. The BP neural network model is optimized with thresholds to ensure that the relative error between the test value and the target value is no more than 5% and that the evaluation level is consistent. Then, the energy-saving level of green buildings is determined using the optimized BP neural network model.
[0085] In this embodiment, examples of green building projects in a certain region were collected and summarized to understand the construction status of these green buildings. A total of 10 representative projects were collected in this study. Following the steps described above, the corresponding indicators were quantified to obtain scores for each secondary evaluation indicator of the project. The research results were then incorporated into a neural network evaluation system to obtain the final evaluation result.
[0086] By analyzing the relevant data of one of the projects, the 14 indicators of the established evaluation index system were parameterized and each indicator was evaluated, and the following results were obtained as shown in Table 14.
[0087] Table 14 Building Technical Indicators
[0088]
[0089] Based on the actual situation of the project cases, each indicator was quantified according to the indicator scoring table, and the scores of each indicator of the project were obtained. Then, the basic information of the 14 indicators of the 10 projects collected was summarized in the same way, as shown in Tables 15 and 16.
[0090] Table 15 Summary of Scores for Items 1-5
[0091]
[0092] Table 16 Summary Table of Scores for Items 5-10
[0093]
[0094] The quantitative scores of the 14 indicators for the 10 projects are summarized in Table 17.
[0095] Table 17 Summary of Quantitative Results for Items 1-10
[0096]
[0097] Based on the summary table of the quantitative results of the above indicators, a comprehensive quantitative score table was obtained, and the results are shown in Table 18.
[0098] Table 18 Summary of Comprehensive Energy Saving Evaluation Results for Projects 1-10
[0099]
[0100] Based on the comprehensive energy-saving evaluation results, the energy-saving evaluation model based on the BP neural network was run and calculated. According to the number of secondary indicators, the input layer of the BP neural network is 14; the number of hidden layer nodes is determined by the formula... Calculation, where This is a constant, set to 1 in this embodiment, meaning the number of hidden layer nodes is 4. Then, the data from the first 7 items is used as the training set for the neural network model, and the data from items 8-10 is used as the test set. The training results are shown below. Figure 2 .
[0101] Save the trained neural network model and input the data from Project 8-10 for computation. Export the results; these exported results are the test values. The values calculated in the above steps are used as the target values. Compare the test values with the target values. Figure 3 Results are analyzed in Tables 19 and 20.
[0102] Table 19 Comparison of test values and target values for items 8-10
[0103]
[0104] Table 20 Test Results and Evaluation Levels
[0105]
[0106] The data in Table 20 shows the difference between the test value and the target value. Although there is a slight error in the calculation results, the final evaluation results are consistent. Based on this, it can be concluded that the neural network evaluation method is applicable to the energy-saving evaluation of green buildings.
[0107] In summary, this invention provides a green building energy-saving effect evaluation method based on BP neural network, which not only breaks through the limitations of traditional evaluation methods in terms of coverage dimensions and subjective interference, but also has the flexibility of technological iteration and multi-scenario application. It realizes the upgrade of green building energy-saving effect evaluation from "experience-driven" to "data-driven", and can provide quantitative standards for the implementation of energy-saving standards in the green building industry. It has significant scientific research innovation value and engineering promotion prospects.
Claims
1. A method for evaluating the energy-saving effect of green buildings based on BP neural networks, characterized in that, Includes the following steps: S1. Using the Delphi method and comparative analysis, 14 secondary evaluation indicators were selected from five dimensions: energy-saving design, environmental energy consumption, building lighting, building ventilation, and building sunshine. The quantitative standards for each secondary evaluation indicator were clarified, and the energy-saving evaluation indicators were quantified. A green building energy-saving effect evaluation indicator system was constructed. Among them, energy-saving design, environmental energy consumption, building lighting, building sunshine, and building ventilation are primary evaluation indicators. S2. The weights of the secondary evaluation indicators for energy-saving effects of green buildings are determined by the AHP method to eliminate subjective bias in the traditional weight allocation. S3. Set standards for evaluating the energy efficiency of green buildings, establish an evaluation model for the energy efficiency of green buildings based on a BP neural network, and then determine the energy efficiency level of green buildings through the evaluation model.
2. The method for evaluating the energy-saving effect of green buildings based on BP neural networks according to claim 1, characterized in that, The 14 secondary evaluation indicators in step S1 are as follows: design quality, building envelope energy-saving quality and energy-saving material utilization in the energy-saving design dimension; lighting power density, comprehensive energy utilization and renewable energy utilization in the environmental energy consumption dimension; indoor lighting, external window anti-glare and underground lighting in the building lighting dimension; wind environment, indoor natural ventilation and indoor air quality in the building ventilation dimension; and building sunlight conditions and external shading effect in the building sunlight dimension.
3. The method for evaluating the energy-saving effect of green buildings based on BP neural networks according to claim 1, characterized in that, Step S2 specifically includes: (1) Constructing a judgment matrix using the 1-9 scale: Using numbers 1-9 to evaluate the relative importance of two secondary evaluation indicators, where 1 indicates that the two secondary evaluation indicators are of equal importance, and 9 indicates that the importance of one secondary evaluation indicator is much greater than that of the other; among which, the secondary evaluation indicators and The judgment scale is Secondary evaluation indicators and The judgment scale is ; (2) Establish AHP quantification standard: The formula for calculating the weight vector is: In the formula, Represents the total number of secondary evaluation indicators. Represents secondary evaluation indicators; The above formula Normalization, i.e., finding the second half ; by eigenvectors The weights of the indicators are calculated using the following formula: ； In the formula, This represents the largest eigenvalue of the judgment matrix. It is its corresponding feature vector; (3) Consistency check of the judgment matrix: When constructing the judgment matrix, a consistency index is calculated. and consistency ratio The consistency of the judgment matrix is evaluated and verified. ; The average random consistency index is determined by referring to the standard table. The formula for the consistency ratio is: ； Let the threshold be 0.1, that is, when If the condition is met, the judgment matrix is considered consistent; otherwise, the judgment matrix is reconstructed until the consistency requirement is met. .
4. The method for evaluating the energy-saving effect of green buildings based on BP neural networks according to claim 1, characterized in that, Step S3 specifically includes: (1) Determine the parameters of the BP neural network model, treat each secondary evaluation index as a neuron, and adjust the weights of each neuron. And thresholds, allocate and adjust the weights of each secondary evaluation indicator; (2) Using the quantitative data of 14 secondary evaluation indicators of green building energy-saving effect as input data, the output value of the BP neural network model is between 0 and 1. Based on the output value of the BP neural network model, four different evaluation levels are divided. The final output value is 0.8-1, which is an excellent evaluation level; 0.7-0.8, which is a good evaluation level; 0.6-0.7, which is a medium evaluation level; and 0-0.6, which is a poor evaluation level. (3) Using the quantitative data of 14 secondary evaluation indicators of green building projects as samples, the training set and the test set were divided, and the weights of the neurons were adjusted. The BP neural network model is optimized with thresholds to ensure that the relative error between the test value and the target value is no more than 5% and that the evaluation level is consistent. Then, the energy-saving level of green buildings is determined using the optimized BP neural network model.