Building power load peak clipping potential estimation method and device

Abstract

The invention discloses a building power load peak clipping potential estimation method and device, and relates to the technical field of building energy management, and the method comprises the following steps: obtaining historical data of a reference building and a target building, and correspondingly constructing a data set A and a data set B; constructing a machine learning peak clipping potential prediction model based on an XGboost algorithm, and training and optimizing the machine learning peak clipping potential prediction model based on the data set A; using a K-means clustering method to extract a load characteristic curve of the data set A during a historical demand response event; and analyzing the building power load peak clipping potential of the target building corresponding to the data set B based on the demand response load characteristics corresponding to the load characteristic curve. According to the invention, under the condition of lack of detailed building and equipment information, the power demand peak clipping potential of a large-scale building is estimated, and the operation efficiency and reliability of a power system can be improved.

Description

Technical Field

[0001] The invention relates to the technical field of building energy management, and in particular to a method and a device for estimating the peak shaving potential of power load of a building. Background Art

[0002] With population growth and urbanization, the demand for electricity in buildings is increasing, which puts great pressure on the national grid and may lead to insufficient power supply. Demand response (DR) programs provide a more economical alternative to building more power plants by incentivizing end users to reduce their electricity demand during specific time periods. However, it is still challenging to accurately estimate the peak demand shaving potential of buildings on a large scale, mainly due to the lack of detailed building and equipment information, so the traditional data-driven method has low assessment accuracy.

[0003] In the current Internet environment, user activity is one of the important indicators to measure the health of a platform. In order to improve user activity and enhance user stickiness, a technical means is urgently needed to push accurate information to users who have not been active for a long time, so as to stimulate their interest in re-using the platform.

[0004] Therefore, in order to meet actual needs, a technology for estimating the peak shaving potential of building power load is provided. Summary of the invention

[0005] In view of the defects existing in the prior art, the purpose of the present invention is to provide a method and device for estimating the peak shaving potential of building power load, which can estimate the peak shaving potential of power demand of large-scale buildings in the absence of detailed building and equipment information, thereby helping to improve the operating efficiency and reliability of the power system.

[0006] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0007] In a first aspect, the present application provides a method for estimating the peak shaving potential of power loads in a building, the method comprising the following steps:

[0008] Obtain historical data of the reference building and the target building, and construct dataset A and dataset B accordingly;

[0009] Building a machine learning peak shaving potential prediction model based on the XGboost algorithm, and training and optimizing the machine learning peak shaving potential prediction model based on the data set A;

[0010] Extracting the load characteristic curve of the data set A during the historical demand response event using the K-means clustering method;

[0011] Based on the demand response load characteristics corresponding to the load characteristic curve, the building power load peak shaving potential of the target building corresponding to the data set B is analyzed.

[0012] On the basis of the above technical solution, the acquisition of historical data of the reference building and the target building, corresponding to constructing data set A and data set B, includes the following steps:

[0013] Acquire the number of floors, the year of completion, the building area, the main purpose of the building, the historical demand response plan, the power reduction capacity of the cooling system, the power reduction capacity of the ventilation system, and the time of occurrence of historical demand response events in the historical data of the reference building to construct the data set A;

[0014] The number of floors of the building, the year of building completion, the building area, and the annual energy consumption information without historical demand response events in the historical data of the target building are obtained to construct the data set B.

[0015] On the basis of the above technical solution, the method further comprises the following steps:

[0016] Calculating building DR characteristics corresponding to different reference buildings in the data set A;

[0017] Based on the building DR characteristics corresponding to different reference buildings in the data set A, a cluster DR characteristic score index is calculated.

[0018] On the basis of the above technical solution, the method is configured with a building DR characteristic coefficient calculation formula, and the building DR characteristic coefficient calculation formula is:

[0019] in,

[0020] P bl is the baseline power of the building when not participating in demand response, P DR is the power consumption of the building during the demand response event, and DRprofile is the DR characteristic coefficient of the building.

[0021] On the basis of the above technical solution, the method is configured with a DR characteristic calculation formula, which is:

[0022] in,

[0023] x represents a data point, C represents a cluster, μ j Represents cluster C i The average value of the points in the cluster, K represents the number of clusters.

[0024] On the basis of the above technical solution, the method is configured with a cluster DR characteristic score index calculation formula, the cluster DR characteristic score index calculation formula:

[0025] in,

[0026] d in (x) represents the average distance between point x and other points in the same cluster, d out (x) represents the minimum average distance between point x and points in other clusters, and s(x) is the clustering DR feature score indicator.

[0027] In a second aspect, the present application provides a device for estimating the peak shaving potential of power loads in a building, the device comprising:

[0028] The data set acquisition module is used to acquire the historical data of the reference building and the target building, and construct the data set A and the data set B accordingly;

[0029] A model optimization module, which is used to build a machine learning peak shaving potential prediction model based on the XGboost algorithm, and train and optimize the machine learning peak shaving potential prediction model based on the data set A;

[0030] A curve acquisition module, which is used to extract the load characteristic curve of the data set A during the historical demand response event by using a K-means clustering method;

[0031] A potential estimation module is used to analyze the building power load peak shaving potential of the target building corresponding to the data set B based on the demand response load characteristics corresponding to the load characteristic curve.

[0032] On the basis of the above technical solution, the data set acquisition module is further used to acquire the number of building floors, the year of building completion, the building area, the main purpose of the building, the historical demand response plan, the power reduction capacity of the cooling system, the power reduction capacity of the ventilation system, and the time of occurrence of historical demand response events in the historical data of the reference building to construct the data set A;

[0033] The data set acquisition module is also used to acquire the number of building floors, the year of building completion, the building area, and the annual energy consumption information without historical demand response events from the historical data of the target building to construct the data set B.

[0034] On the basis of the above technical solution, the potential estimation module is also used to calculate the building DR characteristics corresponding to different reference buildings in the data set A;

[0035] The potential estimation module is also used to calculate and obtain a cluster DR characteristic score index based on the building DR characteristics corresponding to different reference buildings in the data set A.

[0036] On the basis of the above technical solution, the device is equipped with a building DR characteristic coefficient calculation formula, and the building DR characteristic coefficient calculation formula is:

[0037] in,

[0038] P bl is the baseline power of the building when not participating in demand response, P DR is the power consumption of the building during the demand response event, and DRprofile is the DR characteristic coefficient of the building.

[0039] On the basis of the above technical solution, the device is equipped with a DR characteristic calculation formula, which is:

[0040] in,

[0041] x represents a data point, C represents a cluster, μ j Represents cluster C i The average value of the points in the cluster, K represents the number of clusters.

[0042] On the basis of the above technical solution, the device is equipped with a cluster DR characteristic score index calculation formula, the cluster DR characteristic score index calculation formula is:

[0043] in,

[0044] d in (x) represents the average distance between point x and other points in the same cluster, d out (x) represents the minimum average distance between point x and points in other clusters, and s(x) is the clustering DR feature score indicator.

[0045] Compared with the prior art, the advantages of the present invention are:

[0046] The present invention estimates the peak shaving potential of power demand of large-scale buildings in the absence of detailed building and equipment information, which helps to improve the operating efficiency and reliability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0048] Figure 1A flowchart of the steps of a method for estimating the peak shaving potential of power loads of a building according to an embodiment of the present invention;

[0049] Figure 2 A schematic diagram of the steps of a method for estimating the peak shaving potential of power loads of a building according to an embodiment of the present invention;

[0050] Figure 3 A building power load curve diagram of the building power load peak shaving potential estimation method according to an embodiment of the present invention;

[0051] Figure 4 It is a correlation curve diagram of the number of clusters K and the cluster DR characteristic score of the building power load peak shaving potential estimation method according to an embodiment of the present invention;

[0052] Figure 5 A building DR characteristic curve diagram of a building power load peak shaving potential estimation method according to an embodiment of the present invention;

[0053] Figure 6 It is a structural block diagram of a device for estimating the peak shaving potential of power load of a building according to an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0055] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings.

[0056] The embodiments of the present application provide a method and device for estimating the peak shaving potential of power load in a building, which can estimate the peak shaving potential of power demand in a large-scale building in the absence of detailed building and equipment information, thereby helping to improve the operating efficiency and reliability of the power system.

[0057] In order to achieve the above technical effects, the overall idea of ​​this application is as follows:

[0058] A method for estimating the peak shaving potential of power load in a building, the method comprising the following steps:

[0059] S1. Obtain historical data of reference buildings and target buildings, and construct dataset A and dataset B accordingly;

[0060] S2. Build a machine learning peak shaving potential prediction model based on the XGboost algorithm, and train and optimize the machine learning peak shaving potential prediction model based on data set A;

[0061] S3, extracting the load characteristic curve of dataset A during the historical demand response event using K-means clustering method;

[0062] S4. Based on the demand response load characteristics corresponding to the load characteristic curve, the building power load peak shaving potential of the target building corresponding to the data set B is analyzed.

[0063] The embodiments of the present application are further described in detail below in conjunction with the accompanying drawings.

[0064] First, see Figures 1 to 5 As shown, an embodiment of the present application provides a method for estimating the peak shaving potential of power load of a building, the method comprising the following steps:

[0065] S1. Obtain historical data of reference buildings and target buildings, and construct dataset A and dataset B accordingly;

[0066] S2. Build a machine learning peak shaving potential prediction model based on the XGboost algorithm, and train and optimize the machine learning peak shaving potential prediction model based on data set A;

[0067] S3, extracting the load characteristic curve of dataset A during the historical demand response event using K-means clustering method;

[0068] S4. Based on the demand response load characteristics corresponding to the load characteristic curve, the building power load peak shaving potential of the target building corresponding to the data set B is analyzed.

[0069] The embodiments of the present application estimate the peak shaving potential of power demand of large-scale buildings in the absence of detailed building and equipment information, which helps to improve the operating efficiency and reliability of the power system.

[0070] Furthermore, the acquisition of historical data of the reference building and the target building, corresponding to constructing the data set A and the data set B, includes the following steps:

[0071] Acquire the number of floors, the year of completion, the building area, the main purpose of the building, the historical demand response plan, the power reduction capacity of the cooling system, the power reduction capacity of the ventilation system, and the time of occurrence of historical demand response events in the historical data of the reference building to construct the data set A;

[0072] The number of floors of the building, the year of building completion, the building area, and the annual energy consumption information without historical demand response events in the historical data of the target building are obtained to construct the data set B.

[0073] Furthermore, the method further comprises the following steps:

[0074] Calculating building DR characteristics corresponding to different reference buildings in the data set A;

[0075] Based on the building DR characteristics corresponding to different reference buildings in the data set A, a cluster DR characteristic score index is calculated.

[0076] Furthermore, the method is configured with a building DR characteristic coefficient calculation formula, and the building DR characteristic coefficient calculation formula is:

[0077] in,

[0078] P bl is the baseline power of the building when not participating in demand response, P DR is the power consumption of the building during the demand response event, and DRprofile is the DR characteristic coefficient of the building.

[0079] Furthermore, the method is configured with a DR characteristic calculation formula, and the DR characteristic calculation formula is:

[0080] in,

[0081] x represents a data point, C represents a cluster, μ j Represents cluster C i The average value of the points in the cluster, K represents the number of clusters.

[0082] Furthermore, the method is configured with a cluster DR characteristic score index calculation formula, the cluster DR characteristic score index calculation formula is:

[0083] in,

[0084] d in (x) represents the average distance between point x and other points in the same cluster, d out (x) represents the minimum average distance between point x and points in other clusters, and s(x) is the clustering DR feature score indicator.

[0085] Based on the technical solution of the embodiment of the present application, during the specific implementation, the situation is as follows:

[0086] The method for estimating the peak shaving potential of the building power load comprises the following steps:

[0087] (1) Collect detailed building and system information data and historical response data, and construct dataset A and dataset B based on whether the building data is rich;

[0088] (2) A machine learning peak shaving potential prediction model was constructed based on the XGboost algorithm, and the training set of dataset A was used for training and optimization;

[0089] (3) The K-means clustering method is used to extract the load characteristic curve of the test set buildings in dataset A during the DR event;

[0090] (4) Based on the demand response load characteristics extracted in step (3), infer the peak shaving capacity of the data-scarce buildings in dataset B.

[0091] Specifically, step (1) is as follows:

[0092] (11) Detailed building and system information data and building power load data including historical demand response (DR) events are collected. According to the number of characteristics of the factors affecting the building peak shaving capacity that can be collected, a data-rich building dataset A and a data-scarce building dataset B are constructed. The characteristics of the factors affecting the building peak shaving capacity in dataset A include ten main characteristics, including the number of building floors, the year the building was built, the building area, the main use, the DR plan, the power reduction capacity of the cooling system and ventilation system, the power reduction capacity that other equipment may obtain, and the day and time of the day when the DR event occurs. Dataset B only contains building variables (number of building floors, year the building was built, and building area) and annual energy consumption information without DR events.

[0093] (12) The two data sets A and B are divided into training set and test set in a ratio of 7:3, and the outlier data beyond the range of building variables are excluded, and the missing data items are filled by the mean filling method.

[0094] Specifically, step (2) is as follows:

[0095] (21) Because the sample size is small and the input variables are discrete, the current SOTA tree ensemble regression model XGboost algorithm model is used to construct a regression prediction model for building peak shaving capacity.

[0096] (22) During the model training process, four evaluation indicators, namely, squared deviation, variance, mean squared error, and R2, are visualized to prune the model. The squared deviation captures the systematic error of the model, and the variance captures the fluctuation of the training data. Higher squared deviation values ​​indicate that the model is prone to underfitting, while higher variance values ​​indicate that the model is overfitting. The mean squared error measures the average of the squared errors (composed of the squared deviation and the variance). Finally, R2 is used to verify the ability of the trained model to explain the dataset. Find the best hyperparameters so that the model has high performance and is not overfitting. Specifically, the model's tree estimator and the number of samples at the leaf node are iterated from one to ten. Finally, the average R2 value of the optimal model (i.e., the model with the optimal hyperparameters) is evaluated and verified through cross-validation.

[0097] Specifically, step (3) is as follows:

[0098] (31) The building DR characteristic coefficient is defined as the percentage of the building power consumption during the DR event to the baseline power. The specific calculation formula is as follows:

[0099] in,

[0100] P bl is the baseline power of the building when not participating in demand response, P DR is the power consumption of the building during the demand response event, and DRprofile is the DR characteristic coefficient of the building.

[0101] (32) Considering the independence of the DR characteristics of different buildings, the k-means clustering algorithm is used to extract the DR characteristics of different types of buildings in data set A. The specific calculation formula is as follows:

[0102] in,

[0103] The purpose of K-means clustering is to find the cluster center that minimizes the distance within each cluster. x represents the data point, C represents the cluster, and μ j Represents cluster C i K represents the number of clusters. K is a hyperparameter that needs to be fine-tuned to better extract DR features.

[0104] (33) In order to better measure the performance of clustering and further fine-tune the hyperparameter K, a new clustering DR feature score indicator is defined, and its calculation formula is as follows:

[0105] in,

[0106] d in (x) represents the average distance between point x and other points in the same cluster, i.e., the internal distance, d out (x) represents the minimum average distance between point x and points in other clusters, i.e., the inter-cluster distance, and s(x) is the cluster DR feature score index;

[0107] Unlike other metrics that only consider internal distances (e.g., the sum of squared distances), the cluster DR feature score varies between -1 and 1, taking into account both internal and external cluster distances, and smaller internal cluster distances and larger external cluster distances represent better clustering performance (i.e., s(x) is closer to 1).

[0108] Furthermore, step (4) is specifically as follows:

[0109] (41) Based on the clustering results of step (3), the XGBoost classification model is trained to learn the correlation between building characteristics (i.e., four building variables: number of floors, building year, building area, and building type) and the DR characteristic coefficients identified in dataset A. The overall accuracy used in the performance evaluation of the classification model is as follows, where N represents the number of all samples:

[0110]

[0111] (42) The trained XGBoost classification model is fitted to data set B, and the architectural features of buildings without historical response data are matched with the DR characteristics. The peak shaving potential of buildings without historical DR information is further inferred and calculated. The specific calculation formula is as follows:

[0112] PSC=(1-DRprofile)*kWh on-peak *R_t DR / on-peak ;

[0113]

[0114] It should be noted that the technical solution of the embodiment of the present application assumes that DR events only occur during peak power consumption periods, where: Represents the ratio of building peak load to annual power load, R_t DR / on-peak Indicates the ratio of DR duration to peak period

[0115] Figure 3 This is the building power load curve of the test case data set A in a certain week. Figure 4 It is the correlation curve between the number of clusters K of the test case and the cluster DR feature score. Figure 5 The DR characteristic curve of the building extracted by the test case is shown in Table 1 below, which shows the estimated value of peak shaving capacity calculated by inference on the data-scarce building by the test case.

[0116] Table 1

[0117]

[0118] Second, see Figure 6 As shown, an embodiment of the present application provides a building power load peak shaving potential estimation device based on the building power load peak shaving potential estimation method mentioned in the first aspect, and the device includes:

[0119] The data set acquisition module is used to acquire the historical data of the reference building and the target building, and construct the data set A and the data set B accordingly;

[0120] A model optimization module, which is used to build a machine learning peak shaving potential prediction model based on the XGboost algorithm, and train and optimize the machine learning peak shaving potential prediction model based on the data set A;

[0121] A curve acquisition module, which is used to extract the load characteristic curve of the data set A during the historical demand response event by using a K-means clustering method;

[0122] A potential estimation module is used to analyze the building power load peak shaving potential of the target building corresponding to the data set B based on the demand response load characteristics corresponding to the load characteristic curve.

[0123] The embodiments of the present application estimate the peak shaving potential of power demand of large-scale buildings in the absence of detailed building and equipment information, which helps to improve the operating efficiency and reliability of the power system.

[0124] Furthermore, the data set acquisition module is further used to acquire the number of building floors, the year of building completion, the building area, the main purpose of the building, the historical demand response plan, the power reduction capacity of the cooling system, the power reduction capacity of the ventilation system, and the time of occurrence of historical demand response events in the historical data of the reference building to construct the data set A;

[0125] The data set acquisition module is also used to acquire the number of building floors, the year of building completion, the building area, and the annual energy consumption information without historical demand response events from the historical data of the target building to construct the data set B.

[0126] Furthermore, the potential estimation module is also used to calculate the building DR characteristics corresponding to different reference buildings in the data set A;

[0127] The potential estimation module is also used to calculate and obtain a cluster DR characteristic score index based on the building DR characteristics corresponding to different reference buildings in the data set A.

[0128] Furthermore, the device is provided with a calculation formula for the building DR characteristic coefficient, and the calculation formula for the building DR characteristic coefficient is:

[0129] in,

[0130] P bl is the baseline power of the building when not participating in demand response, P DR is the power consumption of the building during the demand response event, and DRprofile is the DR characteristic coefficient of the building.

[0131] Furthermore, the device is configured with a DR characteristic calculation formula, and the DR characteristic calculation formula is:

[0132] in,

[0133] x represents a data point, C represents a cluster, μ j Represents cluster C i The average value of the points in the cluster, K represents the number of clusters.

[0134] Furthermore, the device is configured with a cluster DR characteristic score index calculation formula, the cluster DR characteristic score index calculation formula:

[0135] in,

[0136] d in (x) represents the average distance between point x and other points in the same cluster, d out (x) represents the minimum average distance between point x and points in other clusters, and s(x) is the clustering DR feature score indicator.

[0137] It should be noted that the device for estimating the peak-shaving potential of power loads of buildings mentioned in the second aspect is similar to the technical principles of the method for estimating the peak-shaving potential of power loads of buildings mentioned in the first aspect in terms of technical issues, technical means and technical effects, and will not be elaborated here.

[0138] In the description of the present application, it should be noted that the terms "upper", "lower", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present application. Unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0139] It should be noted that, in this application, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0140] The above description is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest range consistent with the principles and novel features applied for herein.

Claims

1. A method for estimating the peak load shaving potential of buildings. It is characterized in that The method comprises the following steps: Obtain historical data of the reference building and the target building, and construct dataset A and dataset B accordingly; Building a machine learning peak shaving potential prediction model based on the XGboost algorithm, and training and optimizing the machine learning peak shaving potential prediction model based on the data set A; Extracting the load characteristic curve of the data set A during the historical demand response event using the K-means clustering method; Based on the demand response load characteristics corresponding to the load characteristic curve, the building power load peak shaving potential of the target building corresponding to the data set B is analyzed.

2. The method for estimating the peak shaving potential of power load of a building according to claim 1, It is characterized in that The acquisition of historical data of the reference building and the target building, corresponding to constructing the data set A and the data set B, includes the following steps: Acquire the number of floors, the year of completion, the building area, the main purpose of the building, the historical demand response plan, the power reduction capacity of the cooling system, the power reduction capacity of the ventilation system, and the time of occurrence of historical demand response events in the historical data of the reference building to construct the data set A; The number of floors of the building, the year of building completion, the building area, and the annual energy consumption information without historical demand response events in the historical data of the target building are obtained to construct the data set B.

3. The method for estimating the peak shaving potential of power load of a building according to claim 1, It is characterized in that The method further comprises the following steps: Calculating building DR characteristics corresponding to different reference buildings in the data set A; Based on the building DR characteristics corresponding to different reference buildings in the data set A, a cluster DR characteristic score index is calculated.

4. The method for estimating the peak shaving potential of building power load according to claim 3, It is characterized in that The method is configured with a building DR characteristic coefficient calculation formula, and the building DR characteristic coefficient calculation formula is: in, P bl is the baseline power of the building when not participating in demand response, P DR is the power consumption of the building during the demand response event, and DRprofile is the DR characteristic coefficient of the building.

5. The method for estimating the peak shaving potential of building power load according to claim 4, It is characterized in that The method is configured with a DR characteristic calculation formula, which is: in, x represents a data point, C represents a cluster, μ j Represents cluster C i The average value of the points in the cluster, K represents the number of clusters.

6. The method for estimating the peak shaving potential of power load of a building according to claim 5, It is characterized in that The method is configured with a cluster DR characteristic score index calculation formula, the cluster DR characteristic score index calculation formula is: in, d in (x) represents the average distance between point x and other points in the same cluster, d out (x) represents the minimum average distance between point x and points in other clusters, and s(x) is the clustering DR feature score indicator.

7. A device for estimating the peak load shaving potential of a building, It is characterized in that The device comprises: The data set acquisition module is used to acquire the historical data of the reference building and the target building, and construct the data set A and the data set B accordingly; A model optimization module, which is used to build a machine learning peak shaving potential prediction model based on the XGboost algorithm, and train and optimize the machine learning peak shaving potential prediction model based on the data set A; A curve acquisition module, which is used to extract the load characteristic curve of the data set A during the historical demand response event by using a K-means clustering method; A potential estimation module is used to analyze the building power load peak shaving potential of the target building corresponding to the data set B based on the demand response load characteristics corresponding to the load characteristic curve.

8. The device for estimating the peak shaving potential of power load of a building according to claim 7, Features: The data set acquisition module is further used to acquire the number of building floors, the year of building completion, the building area, the main purpose of the building, the historical demand response plan, the power reduction capacity of the cooling system, the power reduction capacity of the ventilation system, and the time of occurrence of historical demand response events from the historical data of the reference building to construct the data set A; The data set acquisition module is also used to acquire the number of building floors, the year of building completion, the building area, and the annual energy consumption information without historical demand response events from the historical data of the target building to construct the data set B.

9. The device for estimating the peak shaving potential of power load of a building according to claim 7, Features: The potential estimation module is also used to calculate the building DR characteristics corresponding to different reference buildings in the data set A; The potential estimation module is also used to calculate and obtain a cluster DR characteristic score index based on the building DR characteristics corresponding to different reference buildings in the data set A.

10. The device for estimating the peak shaving potential of power load of a building according to claim 7, It is characterized in that The device is equipped with a building DR characteristic coefficient calculation formula, and the building DR characteristic coefficient calculation formula is: in, P bl is the baseline power of the building when not participating in demand response, P DR is the power consumption of the building during the demand response event, and DRprofile is the DR characteristic coefficient of the building.

Images