Large-scale public building energy consumption prediction method based on XGBoost model

Abstract

The invention relates to a large-scale public building energy consumption prediction method based on an XGBoost model, and the method comprises the steps: classifying public buildings according to climate regions and purposes, collecting the energy consumption data of all types of public buildings, and constructing a building energy consumption sample data set; analyzing influence factors of the energy consumption of the public building; analyzing the significance of each influence factor on the building energy consumption by adopting a stepwise regression method, and screening the influence factors; respectively constructing energy consumption calculation regression models of various public buildings; establishing an energy consumption prediction correction model; and the energy consumption calculation regression model and the energy consumption prediction correction model are used to calculate an energy consumption prediction value of the public building. According to the method, prediction errors can be effectively reduced by utilizing a dual prediction and correction mechanism, the reliability of a prediction result is improved, the energy consumption value can be calculated for large public buildings with different purposes in different climate regions, the dependence on building sample data and expert evaluation conclusions is reduced, and the method can be suitable for intelligent building scenes.

Description

Technical Field

[0001] The present invention belongs to the field of building energy conservation, and specifically relates to a method for predicting energy consumption of large public buildings based on an XGBoost model. Background Art

[0002] During a building's operation, changes occur in the indoor environment, usage patterns, management methods, and construction materials, all of which affect the building's energy consumption. Conducting energy efficiency evaluation and diagnosis of existing public buildings is a practical way to understand their actual operation and performance, and serves as the front-end foundation for implementing energy-saving retrofits. Currently, public building energy efficiency evaluation methods are primarily divided into two categories based on the subject of analysis and evaluation: actual energy consumption evaluation methods that do not require building design parameters and are based on actual building energy consumption data; and pre-performance simulation evaluation methods that utilize building design and simulation parameter data.

[0003] 1) Actual energy consumption evaluation method: The actual energy consumption evaluation method includes energy consumption value evaluation based on direct measurement data, regression method and benchmark method. The outstanding feature of this type of method is that it is based on the actual energy consumption index or benchmark value of the building in the base year or base period, and compares it with the energy consumption index of the base period to obtain a relative value, which can reflect the actual energy consumption level of the building and has outstanding benchmarking.

[0004] 2) Ex-ante simulation evaluation method: Ex-ante performance simulation evaluation methods mainly include dynamic simulation method, on-site testing method and expert evaluation method. The main idea of this type of method is to simulate the performance of the building through ex-ante building performance simulation software, and determine the performance of the building by comparing the simulation results with the target value.

[0005] The above two methods represent two development stages of building operation performance evaluation. Both methods have played a certain role in the energy-saving evaluation and renovation of public buildings in my country and have made relevant progress.

[0006] Evaluation models using a benchmark approach incorporate "benchmark buildings" into the evaluation, which can avoid comparability between individual buildings. However, the limitations of "benchmark buildings" lead to certain errors in the evaluation results. Furthermore, as indoor environments, usage patterns, management methods, and building materials change, this evaluation method is overly general and lacks specificity. Field testing or dynamic simulation methods can provide precise evaluations of building energy performance, but they require dedicated data collection and monitoring systems and high professional expertise, resulting in high costs and time-consuming data collection. Evaluation models based on expert evaluation are highly dependent on experts and are subject to significant subjectivity. Actual energy consumption evaluation models rely on actual operating data, but these can be inaccurate due to external factors such as irregular operations and the effectiveness of building parameters. Ex ante performance simulation methods have particular practical significance in cold regions where monitoring systems are lacking. However, due to the stringent requirements for the accuracy of simulated data, the evaluation results may differ significantly from actual results.

[0007] The existing building energy consumption prediction methods have the following shortcomings: 1. The evaluation model using the benchmark method has certain errors in the evaluation results due to the limitations of the "benchmark building". Moreover, as the indoor environment, usage, management methods and building materials of the building change, this evaluation method is too general and not targeted.

[0008] 2. Using on-site testing or dynamic simulation methods requires accurate evaluation of the building's energy performance. However, due to the need to establish a dedicated data collection and monitoring system and the high professional requirements of personnel, the cost is high and data collection takes a long time.

[0009] 3. The evaluation model based on expert evaluation is highly dependent on experts and is highly subjective.

[0010] 4. The actual energy consumption evaluation model is based on actual operation data, but the evaluation is not accurate due to external factors such as irregular operation and the effectiveness of the building's own parameters.

[0011] 5. The use of ex ante performance simulation has great practical significance in cold regions where there is a lack of detection systems. However, due to the strict requirements on the accuracy of the simulation data, the evaluation results may differ significantly from the actual results. Summary of the Invention

[0012] The purpose of the present invention is to address the above-mentioned problems and provide a large-scale public building energy consumption prediction method based on the XGBoost model. The method classifies public buildings according to different climate zones and different uses, analyzes the significant influencing factors of energy consumption of various types of public buildings, and constructs energy consumption calculation regression models for various types of public buildings to calculate the regression prediction value of building energy consumption. The AI model is used to further correct the regression prediction value of building energy consumption to obtain accurate building energy consumption evaluation results.

[0013] In order to achieve the above object, the technical solution provided by the present invention is: The energy consumption prediction method for large public buildings based on the XGBoost model includes the following steps: Step 1: Classify public buildings by climate zone and purpose, collect energy consumption data of various types of public buildings, and construct a building energy consumption sample dataset; Step 2: Analyze the factors affecting energy consumption of public buildings; Step 3: Use the stepwise regression method to analyze the significance of each influencing factor on building energy consumption, and screen the influencing factors of various types of public buildings; Step 4: Based on the results of step 3, construct energy consumption calculation regression models for various types of public buildings; Step 5: Based on the energy consumption calculation regression model, an energy consumption prediction correction model is established; and the energy consumption prediction value of the public building is calculated using the energy consumption calculation regression model and the energy consumption prediction correction model.

[0014] Preferably, public buildings are classified according to climate zones and uses. The specific classifications include: commercial office buildings in cold climate zones, commercial office buildings in hot summer and cold winter climate zones, administrative unit office buildings in cold climate zones, administrative unit office buildings in hot summer and cold winter climate zones, hotel buildings in cold climate zones, hotel buildings in hot summer and cold winter climate zones, hospital buildings in cold climate zones, hospital buildings in hot summer and cold winter climate zones, shopping center-type buildings in cold climate zones, shopping center-type buildings in hot summer and cold winter climate zones, cultural buildings in cold climate zones, and cultural buildings in hot summer and cold winter climate zones.

[0015] Furthermore, the stepwise regression method was used to analyze the significance of the impact of various factors on building energy consumption, including: 1) Use various energy consumption influencing factors to conduct linear regression statistics on energy consumption indicators and perform F value test; 2) Based on the effect of each influencing factor on the energy consumption index, introduce them into the regression equation in descending order; the effect of each influencing factor on the energy consumption index is determined by the size of the partial correlation coefficient between each factor and the energy consumption index, or by the F test value of the regression between each factor and the energy consumption index; 3) When the introduced factors become insignificant due to the subsequent introduction of other factors, they are eliminated; 4) Repeat steps 2) and 3) until no insignificant factors are eliminated from the regression equation and no significant factors can be introduced into the regression equation.

[0016] Furthermore, in step 4, the energy consumption calculation regression model for commercial office buildings in cold climate zones is: y = 2.08530 + 0.07332 x 1+ 0.45082 x 2+2.31729 x 4+1.67997 x 5+1.4367 x 6-2.07138 x 7; Where y represents the building energy consumption per unit area; x 1 is the population density; x 2 is the daily operation time of the building; x 4 is the area ratio of the dining area; x 5 is the proportion of commercial area; x 6 is the area ratio of the data center; x 7 is the percentage of garage area.

[0017] The regression model for calculating energy consumption of commercial office buildings in hot summer and cold winter climate zones is: y =1.54 +0.00026226 x 0+1.8367 x 1+0.02375 x 2+1.755 x 6+100.75 x 12 ; Where y represents the building energy consumption per unit area; x 0 is the building area; x 1 is the population density; x 2 represents the daily operation time of the building; x 6 represents the area ratio of the data center; x 12 It is the product of the air-conditioning area ratio and the natural logarithm of the number of days the air-conditioning is in operation.

[0018] Furthermore, the regression model for calculating energy consumption of administrative office buildings in cold climate zones is: y =2.0853 +0.07332 x 1+0.45082 x 2+2.31729 x 4+1.43670 x 6-2.07138 x 7; Where y represents the building energy consumption per unit area; x 1 is the population density; x 2 is the daily operation time of the building; x 4 is the area ratio of the dining area; x 6 is the area ratio of the data center; x 7 is the proportion of underground garage area; The regression model for calculating energy consumption of administrative office buildings in hot summer and cold winter climate zones is: ; In the formula is the building area, x 1 is the population density; is the number of operating hours per day, It is the product of the air-conditioned area ratio and the natural logarithm of the air-conditioned degree days.

[0019] Furthermore, the regression model for calculating energy consumption of hotel buildings in cold climate zones is: ; Where y represents the building energy consumption per unit area; x 5 is the commercial area, i.e. the proportion of shop area; The number of cold storages per 100 square meters; It represents the product of the number of guest rooms per 100 square meters and the occupancy rate; The regression model for calculating energy consumption of hotel buildings in hot summer and cold winter climate zones is: ; Where y represents the building energy consumption per unit area; x 1 is the population density; Indicates the annual heating days of the building; Indicates the annual cooling days of the building; It is the product of the number of rooms per 100 square meters of building area and the annual occupancy rate; Indicates whether the hotel offers on-site laundry services.

[0020] Furthermore, the energy consumption calculation regression model of hospital buildings in cold climate zones is: ; Where y represents the building energy consumption per unit area; is the building area, The number of beds provided is the number of inpatient beds; The number of surgeries per year; is the number of hospital staff; The regression model for calculating energy consumption of hospital buildings in hot summer and cold winter climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the number of beds provided / actual number of open beds; represents the number of outpatient visits per year; It is the heating degree days HDD; Indicates the number of air-conditioning degree days CDD.

[0021] Furthermore, the energy consumption calculation regression model for shopping mall buildings in cold climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the daily running time; Indicates the proportion of air-conditioned area, that is, the cooling area per 10,000 square meters; The area of the dining area; It is the ratio of garage area, i.e. parking area per thousand square meters; Indicates warehouse area per 10,000 square meters; The energy consumption calculation regression model for shopping mall buildings in hot summer and cold winter climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the number of employees per thousand square meters / personnel density; Indicates the daily running time; It represents the sum of the number of freezers and refrigerators per thousand square meters; It is the product of the number of days the air conditioner is in operation and the air conditioning area ratio.

[0022] Furthermore, the energy consumption calculation regression model for cultural buildings in cold climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the density of people; It is the product of the number of heating days and the proportion of heating area; It represents the product of the number of days the air conditioner is in operation and the proportion of the cooling area; Indicates the proportion of constant temperature and humidity area; The energy consumption calculation regression model for cultural buildings in hot summer and cold winter climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the daily opening hours.

[0023] Preferably, in step 5, an energy consumption prediction correction model is established using the XGBoost model, wherein the input data items for establishing the energy consumption prediction correction model include the energy consumption regression calculation value y, the climate type C t 、Use Classificationc , building area X0, population density X1, daily operation time of the building X2, air-conditioning area ratio X3, catering area ratio X4, commercial area ratio X5, data room area ratio X6, garage area ratio X7, heating days per year X8, air-conditioning operation days per year X9, number of beds provided X 10 .

[0024] Compared with the prior art, the present invention has the following beneficial effects: 1) This invention constructs an automatic calculation and evaluation model for the energy consumption of large public buildings. It can calculate the energy consumption values of large public buildings with different uses in different climate zones. It does not rely on the actual operating data of the buildings. It can serve as a reference for the energy consumption design of public buildings and can also be used for the actual energy consumption evaluation of public buildings.

[0025] 2) This invention analyzes and identifies significant energy consumption influencing factors for commercial office buildings in cold climate zones, commercial office buildings in hot summer and cold winter climate zones, administrative office buildings in cold climate zones, administrative office buildings in hot summer and cold winter climate zones, hotels in cold climate zones, hotels in hot summer and cold winter climate zones, hospitals in cold climate zones, hospitals in hot summer and cold winter climate zones, shopping centers in cold climate zones, shopping centers in hot summer and cold winter climate zones, cultural buildings in cold climate zones, and cultural buildings in hot summer and cold winter climate zones. It then constructs personalized energy consumption regression equations for each type of public building and determines the corresponding parameters. This personalized approach accounts for the differences between buildings, improving the targetedness and accuracy of energy consumption predictions.

[0026] 3) This invention uses the XGBoost model to establish a revised energy consumption prediction model based on the energy consumption calculation regression model. This model generates revised building energy consumption forecasts, significantly improving the accuracy, adaptability, and flexibility of forecasts, enhancing the ability to process large-scale data, and preventing overfitting. This dual prediction and correction mechanism effectively reduces prediction errors and improves the reliability of prediction results.

[0027] 4) This invention utilizes an integrated framework to fuse the building energy consumption prediction values obtained by the energy consumption prediction correction model with those obtained by the AI energy consumption prediction module to obtain the final building energy consumption prediction results. This reduces the variance of the building energy consumption prediction values, effectively improves the accuracy of building energy consumption predictions, and enhances the generalization capability of the method. This integrated prediction strategy fully leverages the advantages of different models, improves the stability and accuracy of predictions, and reduces the bias and risks that may exist in a single model.

[0028] 5) The building energy consumption prediction system of the present invention realizes the automatic calculation of energy consumption of large public buildings. By integrating the energy consumption prediction values of heterogeneous AI models, it improves the stability and accuracy of building energy consumption prediction, increases the generalization ability of the prediction model of the present invention, and reduces the dependence on building sample data and expert evaluation conclusions. It is particularly suitable for smart building scenarios with high requirements for prediction stability, and provides strong support for energy conservation and consumption reduction and sustainable development of public buildings. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The present invention will be further described below with reference to the accompanying drawings and examples.

[0030] Figure 1 Schematic diagram of the flow of a method for predicting energy consumption of large public buildings according to an embodiment of the present invention.

[0031] Figure 2 Schematic diagram of the stepwise regression method according to an embodiment of the present invention.

[0032] Figure 3 Schematic diagram of the process of building an XGBoost model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] Example 1 like Figure 1 As shown in FIG, the energy consumption prediction method for large public buildings based on the XGBoost model includes the following steps: Step 1: Classify public buildings by climate zone and purpose, collect energy consumption data of various types of public buildings, and construct a building energy consumption sample dataset; In the embodiment, public buildings are classified according to climate zones and uses, and the specific classifications include: commercial office buildings in cold climate zones, commercial office buildings in hot summer and cold winter climate zones, administrative unit office buildings in cold climate zones, administrative unit office buildings in hot summer and cold winter climate zones, hotel buildings in cold climate zones, hotel buildings in hot summer and cold winter climate zones, hospital buildings in cold climate zones, hospital buildings in hot summer and cold winter climate zones, shopping center-type buildings in cold climate zones, shopping center-type buildings in hot summer and cold winter climate zones, cultural buildings in cold climate zones, and cultural buildings in hot summer and cold winter climate zones.

[0034] Step 2: Analyze the factors affecting energy consumption of public buildings; Step 3: Use the stepwise regression method to analyze the significance of each influencing factor on building energy consumption, and screen the influencing factors of various types of public buildings; The stepwise regression method is used to analyze the significance of the impact of various factors on building energy consumption, including: 1) Use various energy consumption influencing factors to conduct linear regression statistics on energy consumption indicators and perform F value test; 2) Based on the effect of each influencing factor on the energy consumption index, introduce them into the regression equation in descending order; the effect of each influencing factor on the energy consumption index is determined by the size of the partial correlation coefficient between each factor and the energy consumption index, or by the F test value of the regression between each factor and the energy consumption index; 3) When the introduced factors become insignificant due to the subsequent introduction of other factors, they are eliminated; 4) Repeat steps 2) and 3) until no insignificant factors are eliminated from the regression equation and no significant factors can be introduced into the regression equation.

[0035] Step 4: Based on the results of step 3, construct energy consumption calculation regression models for various types of public buildings; The regression model for calculating energy consumption of commercial office buildings in cold climate zones is: y = 2.08530 + 0.07332 x 1+ 0.45082 x 2+2.31729 x 4+1.67997 x 5+1.4367 x 6-2.07138 x 7; Where y represents the building energy consumption per unit area; x 1 is the personnel density, that is, the number of people per thousand square meters; x 2 is the daily operation time of the building; x 4 is the area ratio of the dining area; x 5 is the proportion of commercial area; x 6 is the area ratio of the data center; x 7 is the percentage of garage area; The regression model for calculating energy consumption of commercial office buildings in hot summer and cold winter climate zones is: y =1.54 +0.00026226 x 0+1.8367 x 1+0.02375 x 2+1.755 x 6+100.75 x 12 ; Where y represents the building energy consumption per unit area; x 0 is the building area; x 1 is the population density; x 2 represents the daily operation time of the building; x 6 represents the area ratio of the data center; x 12 It is the product of the air-conditioning area ratio and the natural logarithm of the number of days the air-conditioning is in operation.

[0036] The regression model for calculating energy consumption of administrative office buildings in cold climate zones is: y =2.0853 +0.07332 x 1+0.45082 x 2+2.31729 x 4+1.43670 x 6-2.07138 x 7; Where y represents the building energy consumption per unit area; x 1 is the staff density, that is, the number of employees per thousand square meters; x 2 is the daily operation time of the building; x 4 is the area ratio of the dining area; x 6 is the area ratio of the data center; x 7 is the proportion of underground garage area.

[0037] The regression model for calculating energy consumption of administrative office buildings in hot summer and cold winter climate zones is: ; In the formula is the building area, x 1 is the population density; is the number of operating hours per day, It is the product of the air-conditioned area ratio and the natural logarithm of the air-conditioned degree days.

[0038] The regression model for calculating energy consumption of hotel buildings in cold climate zones is: ; Where y represents the building energy consumption per unit area; x 5 is the commercial area, i.e. the proportion of shop area; The number of cold storages per 100 square meters; It represents the product of the number of guest rooms per 100 square meters and the occupancy rate.

[0039] The regression model for calculating energy consumption of hotel buildings in hot summer and cold winter climate zones is: ; Where y represents the building energy consumption per unit area; x 1 is the staff density, that is, the number of employees per thousand square meters of building area; Indicates the annual heating days of the building; Indicates the annual cooling days of the building; It is the product of the number of rooms per 100 square meters of building area and the annual occupancy rate; Indicates whether the hotel provides on-site laundry service. The hotel provides on-site laundry service. Indicates that the hotel does not provide laundry service.

[0040] The regression model for calculating energy consumption of hospital buildings in cold climate zones is: ; Where y represents the building energy consumption per unit area; is the building area, The number of beds provided is the number of inpatient beds; The number of surgeries per year; The number of hospital employees.

[0041] The regression model for calculating energy consumption of hospital buildings in hot summer and cold winter climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the number of beds provided by the hospital; represents the number of outpatient visits per year; It is the heating degree days HDD; Indicates the number of air-conditioning degree days CDD.

[0042] The regression model for calculating energy consumption of shopping mall-type buildings in cold climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the daily running time; Indicates the proportion of air-conditioned area, that is, the cooling area per 10,000 square meters; The area of the dining area; It is the ratio of garage area, i.e. parking area per thousand square meters; Indicates warehouse area per 10,000 square meters; The energy consumption calculation regression model for shopping mall buildings in hot summer and cold winter climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the personnel density, i.e. the number of employees per thousand square meters; Indicates the daily running time; It represents the sum of the number of freezers and refrigerators per thousand square meters; It is the product of the number of days the air conditioner is in operation and the air conditioning area ratio.

[0043] The regression model for calculating energy consumption of cultural buildings in cold climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the personnel density, i.e. the number of employees per thousand square meters; It is the product of the number of heating days and the proportion of heating area; It represents the product of the number of days the air conditioner is in operation and the proportion of the cooling area; Indicates the proportion of constant temperature and humidity area; The energy consumption calculation regression model for cultural buildings in hot summer and cold winter climate zones is: ; Where y represents the building energy consumption per unit area; Indicates the daily opening hours.

[0044] Step 5: Based on the energy consumption calculation regression model, an energy consumption prediction correction model is established; using the energy consumption calculation regression model of step 4 and the trained energy consumption prediction correction model, the energy consumption prediction value of the public building is calculated.

[0045] like Figure 3 As shown in the figure, the XGBoost model is used to establish an energy consumption prediction correction model. The input data items of the energy consumption prediction correction model include the energy consumption regression calculation value y, the climate type C t 、Use Classification c , building area X0, population density X1, daily operation time of the building X2, air-conditioning area ratio X3, catering area ratio X4, commercial area ratio X5, data room area ratio X6, garage area ratio X7, heating days per year X8, air-conditioning operation days per year X9, number of beds provided X 10 The output data of the energy consumption prediction and correction model are the building energy consumption prediction value and energy efficiency level.

[0046] Example 2 The difference between the building energy consumption prediction method of this embodiment and the building energy consumption prediction method of Example 1 is step 5. In addition to the XGBoost model, step 5 of this embodiment also uses a multi-layer perceptron (MLP) to predict the energy consumption of large public buildings based on the original sample data of public buildings. The prediction results of the XGBoost model and the multi-layer perceptron (MLP) are integrated using a bagging strategy to obtain the final building energy consumption prediction result. Steps 1-4 of the building energy consumption prediction method of this embodiment are the same as those of Example 1.

[0047] The method for predicting energy consumption of a large public building according to an embodiment includes the following steps: Step 1: Classify public buildings by climate zone and purpose, collect energy consumption data of various types of public buildings, and construct a building energy consumption sample dataset; Step 2: Analyze the factors affecting energy consumption of public buildings; Step 3: Use the stepwise regression method to analyze the significance of each influencing factor on building energy consumption, and screen the influencing factors of various types of public buildings; Step 4: Based on the results of step 3, construct energy consumption calculation regression models for various types of public buildings; Step 5: XGBoost and multi-layer perceptron (MLP) are used to construct base models for energy consumption prediction of large public buildings. The input data of the XGBoost model includes the original sample data of public buildings and the energy consumption regression calculation value, and the input of the multi-layer perceptron (MLP) is the original sample data of public buildings. The prediction results of the base models are integrated using the Bagging integration strategy to obtain the final building energy consumption prediction result.

[0048] Example 3 The difference between the building energy consumption prediction method of this embodiment and the building energy consumption prediction method of Example 2 is step 5. In addition to the XGBoost model and the multi-layer perceptron (MLP), step 5 of this embodiment also uses an energy consumption calculation regression model as the base model of the bagging integration framework. The bagging integration strategy is used to fuse the energy consumption prediction results of XGBoost and the multi-layer perceptron (MLP) with the building energy consumption prediction value obtained by the energy consumption calculation regression model to obtain the final building energy consumption prediction result. Steps 1-4 of the building energy consumption prediction method of this embodiment are the same as those of Example 1 or Example 2.

[0049] In an embodiment, in the case where the original sample data of the building is lacking, the energy consumption prediction value obtained by combining the energy consumption calculation regression model with the energy consumption prediction correction model is used to construct a training sample set of the multi-layer perceptron MLP to train the multi-layer perceptron MLP, so as to reduce the dependence on the original sample data of the building and improve the generalization ability and accuracy of the AI model.

[0050] Example 4 This embodiment provides a public building energy consumption prediction system based on the method of the second or third embodiment, including the following modules: Building data collection module: used to collect sample data of various public buildings, including the climate type C of public buildings t 、Use Classification c , building area X0, population density X1, daily operating hours of the building X2, air-conditioned area ratio X3, catering area ratio X4, commercial area ratio X5, data room area ratio X6, garage area ratio X7, heating days per year X8, air-conditioning operation days per year X9, number of beds provided and actual energy consumption of the building; Energy consumption influencing factors screening module: Use stepwise regression method to screen and determine the significant influencing factors of various types of public buildings; construct energy consumption regression equations for various types of public buildings and determine the parameters of the energy consumption regression equations; Building energy consumption regression calculation module: used to calculate the building energy consumption regression prediction value based on the significant influencing factors of public buildings obtained by the energy consumption influencing factor screening module and the parameters of the corresponding energy consumption regression equation; Energy consumption prediction and correction module: Use XGBoost to build an energy consumption prediction and correction model. The input data of the energy consumption prediction and correction model includes the original sample data of public buildings and the building energy consumption regression prediction value obtained by the building energy consumption regression calculation module. The energy consumption prediction and correction model is used to obtain the corrected building energy consumption prediction value. AI energy consumption prediction module: uses AI models to calculate the energy consumption prediction value of public buildings based on the original sample data of public buildings; Integrated prediction module: The Bagging integration strategy is used to integrate the building energy consumption prediction value obtained by the energy consumption prediction correction module with the energy consumption prediction value obtained by the AI energy consumption prediction module to obtain the final building energy consumption prediction result.

[0051] In another embodiment, the integrated prediction module uses a Stacking integration framework to obtain the final building energy consumption prediction result.

Claims

1. A large-scale public building energy consumption prediction method based on the XGBoost model is characterized by: The following steps are involved: Step 1: Classify public buildings by climate zone and purpose, collect energy consumption data of various types of public buildings, and construct a building energy consumption sample dataset; Step 2: Analyze the factors affecting energy consumption of public buildings; Step 3: Use the stepwise regression method to analyze the significance of each influencing factor on building energy consumption, and screen the influencing factors of various types of public buildings; Step 4: Based on the results of step 3, construct energy consumption calculation regression models for various types of public buildings; Step 5: Based on the energy consumption calculation regression model, an energy consumption prediction correction model is established; and the energy consumption prediction value of the public building is calculated using the energy consumption calculation regression model and the energy consumption prediction correction model.

2. The method for predicting energy consumption of large public buildings according to claim 1, characterized in that: In step 1, the public buildings are classified according to climate zones and uses, and the specific classifications include: commercial office buildings in cold climate zones, commercial office buildings in hot summer and cold winter climate zones, administrative office buildings in cold climate zones, administrative office buildings in hot summer and cold winter climate zones, hotel buildings in cold climate zones, hotel buildings in hot summer and cold winter climate zones, hospital buildings in cold climate zones, hospital buildings in hot summer and cold winter climate zones, shopping center-type buildings in cold climate zones, shopping center-type buildings in hot summer and cold winter climate zones, cultural buildings in cold climate zones, and cultural buildings in hot summer and cold winter climate zones.

3. The method for predicting energy consumption of large public buildings according to claim 2, characterized in that: In step 3, the stepwise regression method is used to analyze the significance of the impact of each influencing factor on building energy consumption, specifically including: 1) Use various energy consumption influencing factors to conduct linear regression statistics on energy consumption indicators and perform F value test; 2) Based on the effect of each influencing factor on the energy consumption index, introduce them into the regression equation in descending order; the effect of each influencing factor on the energy consumption index is determined by the size of the partial correlation coefficient between each factor and the energy consumption index, or by the F test value of the regression between each factor and the energy consumption index; 3) When the introduced factors become insignificant due to the subsequent introduction of other factors, they are eliminated; 4) Repeat steps 2) and 3) until no insignificant factors are eliminated from the regression equation and no significant factors can be introduced into the regression equation.

4. The method for predicting energy consumption of large public buildings according to claim 3, characterized in that: In step 4, the regression model for calculating energy consumption of commercial office buildings in cold climate zones is: y = 2.08530+ 0.07332 x 1 + 0.45082 x 2 +2.31729 x 4+1.67997 x 5+1.4367 x 6-2.07138 x 7； Where y represents the building energy consumption per unit area; x 1 is the population density; x 2 is the daily operation time of the building; x 4 is the area ratio of the dining area; x 5 is the proportion of commercial area; x 6 is the area ratio of the data center; x 7 is the percentage of garage area; The regression model for calculating energy consumption of commercial office buildings in hot summer and cold winter climate zones is: y =1.54 +0.00026226 x 0 +1.8367 x 1 +0.02375 x 2 +1.755 x 6 +100.75 x 12 ； Where y represents the building energy consumption per unit area; x 0 is the building area; x 12 It is the product of the air-conditioning area ratio and the natural logarithm of the number of days the air-conditioning is in operation.

5. The method for predicting energy consumption of large public buildings according to claim 3, characterized in that: In step 4, the regression model for calculating energy consumption of administrative office buildings in cold climate zones is: y =2.0853 +0.07332 x 1+0.45082 x 2+2.31729 x 4+1.43670 x 6-2.07138 x 7； Where y represents the building energy consumption per unit area; x 1 is the population density; x 2 is the daily operation time of the building; x 4 is the area ratio of the dining area; x 6 is the area ratio of the data center; x 7 is the proportion of underground garage area; The regression model for calculating energy consumption of administrative office buildings in hot summer and cold winter climate zones is: ； In the formula is the building area, x 1 is the population density; is the number of operating hours per day, It is the product of the air-conditioned area ratio and the natural logarithm of the air-conditioned degree days.

6. The method for predicting energy consumption of large public buildings according to claim 3, characterized in that: In step 4, the regression model for calculating energy consumption of hotel buildings in cold climate zones is: ； Where y represents the building energy consumption per unit area; x 5 is the commercial area, i.e. the proportion of shop area; The number of cold storages per 100 square meters; It represents the product of the number of guest rooms per 100 square meters and the occupancy rate; The regression model for calculating energy consumption of hotel buildings in hot summer and cold winter climate zones is: ； Where y represents the building energy consumption per unit area; x 1 is the population density; Indicates the annual heating days of the building; Indicates the annual cooling days of the building; It is the product of the number of rooms per 100 square meters of building area and the annual occupancy rate; Indicates whether the hotel offers on-site laundry services.

7. The method for predicting energy consumption of large public buildings according to claim 3, characterized in that: In step 4, the regression model for calculating energy consumption of hospital buildings in cold climate zones is: ； Where y represents the building energy consumption per unit area; is the building area, The number of beds provided; The number of surgeries per year; is the number of hospital staff; The regression model for calculating energy consumption of hospital buildings in hot summer and cold winter climate zones is: ； Where y represents the building energy consumption per unit area; Indicates the number of beds provided; represents the number of outpatient visits per year; It is the heating degree days HDD; Indicates the number of air-conditioning degree days CDD.

8. The method for predicting energy consumption of large public buildings according to claim 3, characterized in that: In step 4, the energy consumption calculation regression model for shopping mall buildings in cold climate zones is: ； Where y represents the building energy consumption per unit area; Indicates the daily running time; Indicates the proportion of air-conditioned area; The area of the dining area; is the percentage of garage area; Indicates warehouse area per 10,000 square meters; The energy consumption calculation regression model for shopping mall buildings in hot summer and cold winter climate zones is: ； Where y represents the building energy consumption per unit area; Indicates the density of people; Indicates the daily running time; It represents the sum of the number of freezers and refrigerators per thousand square meters; It is the product of the number of days the air conditioner is in operation and the air conditioning area ratio.

9. The method for predicting energy consumption of large public buildings according to claim 3, characterized in that: In step 4, the regression model for calculating energy consumption of cultural buildings in cold climate zones is: ； Where y represents the building energy consumption per unit area; Indicates the density of people; It is the product of the number of heating days and the proportion of heating area; It represents the product of the number of days the air conditioner is in operation and the proportion of the cooling area; Indicates the proportion of constant temperature and humidity area; The energy consumption calculation regression model for cultural buildings in hot summer and cold winter climate zones is: ； Where y represents the building energy consumption per unit area; Indicates the daily opening hours.

10. The method for predicting energy consumption of large public buildings according to any one of claims 3 to 9, characterized in that: In step 4, the step 5 uses the XGBoost model to establish an energy consumption prediction correction model, and the input data items of the energy consumption prediction correction model include the energy consumption regression calculation value y, the climate type C t 、Use Classification c , building area X0, population density X1, daily operation time of the building X2, air-conditioning area ratio X3, catering area ratio X4, commercial area ratio X5, data room area ratio X6, garage area ratio X7, heating days per year X8, air-conditioning operation days per year X9, number of beds provided X 10 .

Images