A User Electricity Consumption Behavior Classification Method and System Based on Hybrid Clustering Framework

Abstract

This invention discloses a user electricity consumption behavior classification method and system based on a hybrid clustering framework. The hybrid clustering framework is used to deeply mine user electricity consumption behavior. The first layer categorizes users into two classes based on whether they participate in electricity demand response: Class A and Class B. The second layer is a deep clustering layer, which further clusters the electricity consumption behavior of Class B users participating in demand response, resulting in four subcategories: Class B includes users participating in electricity demand response whose air conditioning operation events occur during the execution of the electricity demand response event; Class C includes users participating in electricity demand response whose electricity demand response events occur during the execution of the air conditioning operation event; Class D includes users participating in electricity demand response whose electricity demand response events occur first and end before the air conditioning operation event; and Class E includes users participating in electricity demand response whose air conditioning operation events occur first and end before the electricity demand response event. Finally, air conditioning operation control strategies are provided for each type of user participating in electricity demand response.

Description

Technical Field

[0001] This invention belongs to the field of residential air conditioning load participation in electricity demand response technology, specifically involving a user electricity behavior classification method and system based on a hybrid clustering framework. Background Technology

[0002] As one of the main electrical loads in buildings, air conditioning loads possess a certain degree of flexibility and adjustability. Currently, there is considerable research on the operation strategies of air conditioning systems in large buildings, but research on air conditioning systems in residential buildings is relatively less mature. Precise segmentation of user behavior can help power suppliers better predict and manage electricity demand. By analyzing users' electricity consumption patterns, habits, and preferences, peak and off-peak electricity demand can be accurately predicted for different time periods and regions, helping suppliers to rationally allocate and optimize energy resources and improve power supply efficiency.

[0003] Correspondingly, by predicting and analyzing similar users, more accurate forecasts of peak and off-peak electricity loads and fluctuations in electricity demand can be achieved, thus laying the foundation for providing more precise control strategies and energy allocation schemes. Currently, the classification of user electricity consumption behavior is usually limited to a superficial level, and further exploration of the degree of similarity has not received sufficient attention. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a user electricity consumption behavior classification method and system based on a hybrid clustering framework to address the shortcomings of the existing technology, thereby solving the technical problem of inaccurate clustering of existing user electricity consumption behavior.

[0005] The present invention adopts the following technical solution:

[0006] A user electricity consumption behavior classification method based on a hybrid clustering framework includes the following steps:

[0007] S1. Obtain data on air conditioning power consumption, indoor temperature, and outdoor temperature of residents in residential buildings to obtain a user dataset;

[0008] S2. Extract the air conditioner runtime and air conditioner start time for each user from the user dataset obtained in step S1.

[0009] S3. Construct a hybrid clustering framework, considering the demand response time, execution time, and execution duration. Whether or not a user participates in the demand response is passed into the hybrid clustering framework as the first-level clustering criterion. The air conditioner runtime and air conditioner start-up time of each user obtained in step S2 are passed into the first level of the hybrid clustering framework to perform the first-level clustering of electricity consumption behavior, resulting in Group A, which does not participate in the electricity demand response event, and Group B, which participates in the electricity demand response event.

[0010] S4. Considering the degree of overlap and start time between the demand response execution process and the air conditioning operation period, the degree of overlap and start time of the two events are passed into the hybrid clustering framework as the second-level clustering criteria. The data of B-type users participating in the power demand response event obtained in step S3 are passed into the second level of the hybrid clustering framework for second-level - deep clustering. Through analysis, the B-type group participating in the power demand response event is divided into four categories.

[0011] S5. Based on the results obtained in step S4, provide control strategies for various types of users participating in power demand response.

[0012] Specifically, in step S1, the electricity consumption of air conditioning, indoor temperature and outdoor temperature data of residents in residential buildings are preprocessed. If the difference between two adjacent points exceeds the set error threshold range, it is considered abnormal load data. The preprocessed user electricity consumption data is then saved in the corresponding list for each user.

[0013] Furthermore, the user dataset used in the corresponding list for each user is:

[0014] users = {user1[t open ,t operate ,p t_use ],user2[t open ,t operate ,p t_use ],…,usern[t open ,t operate ,p t_use ]}

[0015] Among them, t open For the time the user turns on the air conditioner, t operate For the user's air conditioner running time, p t_use This shows the distribution of users' electricity consumption time.

[0016] Specifically, in step S3, the second derivative evaluation index and the silhouette coefficient are introduced as evaluation indexes in the hybrid clustering framework.

[0017] Furthermore, the evaluation index for the second derivative is:

[0018] SSE k_best =min(SSE) j∈(2,k) ”)

[0019] Among them, SSE k_best "To determine the optimal number of clusters, SSE" j Let be the sum of squared errors for the j-th iteration.

[0020] Furthermore, the contour coefficient is:

[0021]

[0022] Where S(i) is the contour coefficient value of the i-th sample point, a(i) is the cohesion of the sample point, and b(i) is the inter-class distance.

[0023] Specifically, in step S3, the first layer of the hybrid clustering framework is a shallow clustering layer, which takes into account the user's air conditioning start time, air conditioning running time, and the time of the demand response event released by the power grid. The maximum number of clusters is 5. The curves of the sum of squared errors, the profile coefficient, and the second derivative evaluation index are plotted for each instance, and the optimal number of clusters is 2. Category A is defined as the category that does not participate in power demand response, and category B is defined as the category that participates in power demand response.

[0024] Specifically, in step S4, the second layer of the hybrid clustering framework is a deep clustering layer, which takes into account the start time of user air conditioning operation, the duration of air conditioning operation, and the time of the demand response event issued by the power grid. The maximum number of clusters is 10. During this process, the curves of the sum of squared errors, the curves of the profile coefficient, and the curves of the second derivative evaluation index are plotted each time. The optimal number of clusters is 4. By analyzing the average daily operation of air conditioning for various types of users and the execution period of the power demand response event, the following categories are identified: the first category is air conditioning operation during the execution period of the power demand response event; the second category is power demand response during the execution period of the air conditioning operation event; the third category is where the implementation time of the air conditioning operation event and the power demand response event overlaps, with the power demand response event preceding the air conditioning operation event; and the fourth category is where the implementation time of the air conditioning operation event and the power demand response event overlaps, with the air conditioning operation event preceding the power demand response event.

[0025] Specifically, in step S5, the power demand response control strategies under various user electricity consumption categories are as follows:

[0026] Based on the results obtained using the hybrid clustering framework, users' electricity consumption behavior is divided into four categories: Category A, users who do not participate in electricity demand response; Category B, users who participate in electricity demand response, and whose air conditioning operation event occurs during the execution of the electricity demand response event; Category C, users who participate in electricity demand response, and whose electricity demand response event occurs during the execution of the air conditioning operation event; Category D, users who participate in electricity demand response, and whose electricity demand response event precedes the air conditioning operation event; and Category E, users who participate in electricity demand response, and whose air conditioning operation event precedes the electricity demand response event.

[0027] Secondly, embodiments of the present invention provide a user electricity consumption behavior classification system based on a hybrid clustering framework, comprising:

[0028] The data module acquires data on air conditioning power consumption, indoor temperature, and outdoor temperature of residents in residential buildings to obtain user electricity consumption data.

[0029] The extraction module extracts the air conditioner runtime and air conditioner start time for each user from the user electricity consumption data obtained from the data module.

[0030] The first clustering module constructs a hybrid clustering framework, considering the demand response time, execution time, and execution duration. Whether or not a user participates in the demand response is passed into the hybrid clustering framework as the first-level clustering criterion. The air conditioning runtime and air conditioning start-up time of each user obtained by the extraction module are passed into the first level of the hybrid clustering framework to perform the first-level clustering of electricity consumption behavior, resulting in Group A, which does not participate in the electricity demand response event, and Group B, which participates in the electricity demand response event.

[0031] The second clustering module considers the degree of overlap and start time between the demand response execution process and the air conditioning operation process. It inputs the degree of overlap and the start time of the two events into the hybrid clustering framework as the second-level clustering criteria. The user data of the B class participating in the power demand response event obtained from the first clustering module is input into the second layer of the hybrid clustering framework for second-level - deep clustering, dividing the B class participating in the power demand response event into four categories.

[0032] The output module provides control strategies for various types of users participating in power demand response based on the results obtained from the second clustering module.

[0033] Compared with the prior art, the present invention has at least the following beneficial effects:

[0034] A user electricity consumption behavior classification method based on a hybrid clustering framework is proposed. In situations of power resource shortage, this method mobilizes more user groups to participate in electricity demand response, thereby alleviating grid pressure. During this period, to obtain sufficient electricity response, it is necessary to send response invitations to the target group. By dividing users into two categories, those who participate in electricity demand response and those who do not, the target group can be better understood. Then, the electricity consumption behavior of users who participate in electricity demand response is further subdivided to better implement subsequent control strategies.

[0035] Furthermore, the collected data on air conditioning power consumption, indoor temperature, and outdoor temperature of residential buildings are preprocessed, primarily to check for missing and outlier values. Specifically, if the difference between two consecutive values ​​in the power consumption data exceeds a set error threshold, it is considered abnormal load data. This preprocessing ensures high-quality data for verification.

[0036] Furthermore, in clustering methods, there are usually three evaluation metrics to judge the quality of clustering. To examine the cluster density and separation, this invention uses the silhouette coefficient metric for evaluation. In clustering algorithms, the sum of squared errors refers to the error of each data point, that is, the Euclidean distance from it to the centroid of its nearest cluster. The sum of squared errors is then obtained by summing these values. Curves showing the variation of the sum of squared errors under different numbers of clusters were plotted. Generally, the elbow rule is used to observe the variation curve of the sum of squared errors to find the optimal number of clusters. A second derivative evaluation criterion is introduced. By calculating the second derivative of the sum of squared errors curve before and after each number of clusters, the point where the slope of the curve changes abruptly is obtained, and this point is the optimal cluster point.

[0037] Furthermore, in order to ensure the maximum response volume during times of power shortage, a shallow clustering layer of a hybrid clustering framework is used to first screen out users who can participate in power demand response, and then issue response requests to these users; the air conditioning running time and air conditioning start time of these users are recorded.

[0038] Furthermore, the first step of clustering yields two groups: those that do not participate in electricity demand response and those that do. However, the classification obtained from the first step is only a superficial one. To further explore the similarity of electricity consumption behavior among users of the same category, deeper clustering is needed. Through the deep clustering layer of the hybrid clustering framework proposed in this invention, the similarity of user electricity consumption behavior can be further explored, thus providing a foundation for the implementation of subsequent control strategies.

[0039] Furthermore, in order to obtain the maximum response to alleviate grid strain while ensuring the lowest possible electricity costs for users, and taking into full account the different types of user electricity consumption, the set temperature and the operating time at that temperature during each user's air conditioning event were calculated.

[0040] It is understandable that the beneficial effects of the second aspect mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0041] In summary, this invention performs deeper clustering of user electricity consumption behavior. Through a hybrid clustering framework, it effectively distinguishes different groups and provides clear clustering divisions. By observing the clustering results, we can see that different data points are grouped into clusters with similar characteristics or behavioral patterns, thus providing a better understanding of the data's inherent structure and features. In this process, a second-order derivative evaluation criterion is introduced to select the number of clusters, which improves the accuracy of cluster selection to a certain extent. Furthermore, relevant departments can implement response control strategies based on the different user categories obtained, in order to achieve the maximum electricity consumption response during the electricity demand response period.

[0042] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0043] Figure 1 This is a framework diagram for hybrid clustering;

[0044] Figure 2 This is a flowchart illustrating the clustering process within each layer of the hybrid clustering framework.

[0045] Figure 3 A graph showing the average operating status of user air conditioners and the execution status of power demand response events;

[0046] Figure 4 This is a graph showing the variation of the sum of squared errors in the second layer.

[0047] Figure 5 This is a graph showing the variation of the contour coefficients in the second layer.

[0048] Figure 6 This is a graph showing the changes in the evaluation index of the second-order derivative at the second level.

[0049] Figure 7 A graph showing the overlap between the average operating status of user air conditioners and the execution of power demand response events;

[0050] Figure 8 This is a clustering diagram of residential users in a smart grid pilot zone in Australia.

[0051] Figure 9 Diagram illustrating control strategies for various user air conditioning systems participating in electricity demand response;

[0052] Figure 10 This is a graph showing the average hourly electricity consumption of four user groups obtained from a deep clustering layer of users in a certain community. Detailed Implementation

[0053] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0055] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0056] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this invention generally indicates that the preceding and following objects have an "or" relationship.

[0057] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0058] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0059] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0060] This invention provides a method for classifying user electricity consumption behavior based on a hybrid clustering framework. The first layer is a preliminary clustering layer, which divides users into category A and category B based on whether they participate in demand response events. Category A consists of users who do not participate in electricity demand response, while category B consists of users who do participate in electricity demand response. The second layer is a deep clustering layer, which further clusters the electricity consumption behavior of category B users obtained in the first layer. The evaluation at this layer is based on the degree of overlap between the start time of air conditioning operation and the execution period of demand response events.

[0061] Please see Figure 1 The present invention discloses a user electricity consumption behavior classification method based on a hybrid clustering framework, comprising the following steps:

[0062] S1. Collect data such as air conditioning power consumption, indoor temperature and outdoor temperature of residents in residential buildings to obtain historical datasets. First, preprocess the data to remove abnormal data and ensure the integrity and stability of the data.

[0063] Based on existing residential user data from the Australian smart grid pilot zone, and after appropriate data preprocessing, this study selected air conditioning power consumption and corresponding hourly meteorological data from January 2014 for analysis, totaling 721 data points. Based on this, a hybrid clustering framework was employed to conduct in-depth analysis of user electricity consumption behavior.

[0064] The data for this study comes from the Smart Grid Smart City (SGSC) pilot project, jointly funded by the Australian government and an industry consortium led by Ausgrid. The dataset collects household appliance data from 808 residential users between 2012 and 2014, collected every half hour, including interval readings of electricity consumption; home network plug readings; peak events; peak event responses; and offer and accept behaviors. The study focuses on users' air conditioning power consumption and corresponding hourly meteorological data for January 2014, totaling 721 data points.

[0065] Before data analysis, the hourly load data to be analyzed is first preprocessed to remove abnormal data. Normally, the data from two adjacent load points are relatively similar, meaning the historical load curve is smooth and continuous. During data preprocessing, if the difference between two adjacent points is too large, exceeding the set error threshold, it is considered abnormal load data. For the filtered abnormal or missing data, the average value of adjacent days at the same time is used to replace it, ensuring data integrity and stability.

[0066] By analyzing electricity meter data, the on / off status and runtime of users' air conditioners were obtained. Since users' air conditioner event behavior can occur multiple times a day, this study only considers the air conditioner operation event that is closest to the demand response event. The processed data is stored in the corresponding list for each user.

[0067] users = {user1[t open ,t operate ,p t_use ],user2[t open ,t operate ,p t_use ],…,usern[t open ,t operate ,p t_use ]} (1)

[0068] Where userss is the user dataset used in this study, user1 is the user with the label 1, and t open For the time the user turns on the air conditioner, t operate For the user's air conditioner running time, p t_use This shows the distribution of users' electricity consumption time.

[0069] S2. Based on the user dataset processed in step S1, extract the air conditioner running time and air conditioner start time for each user.

[0070] S3. Considering the execution time and duration of the demand response time, whether or not to participate in the power demand response is passed into the hybrid clustering framework as the first-level clustering criterion. The user dataset processed in step S2 is passed into the first level of the hybrid clustering framework to perform the first-level clustering of electricity consumption behavior, resulting in class A (not participating in the power demand response event) and class B (participating in the power demand response event).

[0071] Hybrid clustering framework

[0072] Clustering strategy for each layer

[0073] The "elbow rule" is a commonly used method to determine optimal parameters, especially in data analysis, to determine the optimal number of clusters. This method determines the optimal parameter values ​​by plotting a graph and observing the position of the "elbow." In cluster analysis, it is used to find the optimal number of clusters.

[0074] Specifically, the elbow rule is based on the idea that as the number of clusters increases, the sum of squared errors (SSE) gradually decreases, as shown in equation (2). A graph is plotted showing the relationship between the number of clusters and SSE, and a decreasing trend in SSE is observed. When this trend shows a significant and slow change, the current number of clusters is considered the optimal number of clusters.

[0075] To accurately determine the number of clusters when the change is slow, a second derivative evaluation metric (SDEM) is introduced, as shown in equation (5).

[0076] There are three evaluation metrics for the quality of clustering results:

[0077] Silhouette Coefficient: The value ranges from [-1, 1], and the closer it is to 1, the better the clustering result.

[0078] Davies-Bouldin index: The smaller the value, the better the clustering result;

[0079] Calinski-Harabasz index: A higher value indicates a better clustering result.

[0080] Since the silhouette coefficient does not require real labels and does not need to know the real labels or category information of the data in advance, this study chose the silhouette coefficient as another evaluation index, as shown in Equation (6).

[0081]

[0082] Among them, SSE j Let x be the sum of squared errors for the j-th iteration, i be the i-th cluster data, j be the current number of clusters, k be the defined maximum number of clusters, and x be the sum of squared errors for the j-th iteration. ij For the i-th data in the j-th cluster, c j It is the cluster center of the j-th cluster.

[0083] SSE j_ =SSE j -SSE j-1 (3)

[0084] SSE j+ =SSE j+1 -SSE j (4)

[0085] SSE k_best =min(SSE) j∈(2,k) (5)

[0086] Among them, SSE j_ ' is the left-hand guide for the current number of clusters, SSE j+ ' is the right-hand derivative of the current number of clusters. For continuously changing functions, the left-hand derivative is the same as the right-hand derivative, SSE. k_best " is the optimal number of clusters, and the second derivative represents the concavity or convexity of the function. For concave functions, it takes the minimum value.

[0087]

[0088]

[0089] Simplifying, we get equation (8):

[0090]

[0091] Where S(i) is the contour coefficient value of the i-th sample point, a(i) is the cohesion of the sample point, and b(i) is the inter-class distance.

[0092] First layer - shallow clustering

[0093] This layer is a shallow clustering layer, taking into account the user's air conditioner start time, air conditioner running duration, and the time of the demand response event published by the power grid. Based on whether users participate in electricity demand response, they are divided into group A (non-participating) and group B (participating). The analysis is further enhanced by examining the average operating conditions of users' air conditioners and the execution status of electricity demand response events. Figure 3 .

[0094] Second layer - deep clustering

[0095] This layer is a deep clustering layer, taking into account the user's air conditioning start time, air conditioning runtime, and the time of the demand response event published by the power grid. The maximum number of clusters is specified as 9. During this process, the curves showing the changes in the sum of squared errors, the profile coefficient, and the second-derivative evaluation index are plotted for each iteration. Figure 4 , Figure 5 and Figure 6 As shown. The optimal number of clusters was found to be 4. This was achieved by analyzing the average daily operating conditions of air conditioners for various user types and the execution period of power demand response events, such as... Figure 7 As shown in the diagram. The first category after analysis is when the air conditioning operation event occurs during the execution of the power demand response event; the second category is when the power demand response event occurs during the execution of the air conditioning operation event; the third category is when the two events overlap, with the power demand response event preceding the air conditioning operation event; and the fourth category is when the two events overlap, with the air conditioning operation event preceding the power demand response event.

[0096] S4. Considering the degree of overlap between the demand response execution process and the air conditioning operation period, as well as their start times, the degree of overlap and the start time of the two events are passed into the hybrid clustering framework as the second-level clustering criteria. The relevant data of B-class users obtained in step S3 are passed into the second layer of the hybrid clustering framework to perform second-level - deep clustering, and finally the B-class users are divided into four categories.

[0097] Analysis of Clustering Results Using a Hybrid Clustering Framework

[0098] Hybrid clustering frameworks can effectively distinguish different groups and provide clear clustering partitions. By observing the clustering results, we can see that different data points are divided into clusters with similar characteristics or behavioral patterns, thus gaining a better understanding of the inherent structure and characteristics of the data.

[0099] The final user electricity behavior classification results obtained from the application of a hybrid clustering framework in the Australian smart city pilot project. Figure 8 Among them, Category A is a user category that does not participate in electricity demand response; Category B is a user category that participates in electricity demand response, and the air conditioning operation event occurs during the execution of the electricity demand response event; Category C is a user category that participates in electricity demand response, and the electricity demand response event occurs during the execution of the air conditioning operation event; Category D is a user category that participates in electricity demand response, and the electricity demand response event precedes the air conditioning operation event; Category E is a user category that participates in electricity demand response, and the air conditioning operation event precedes the electricity demand response event.

[0100] Electricity demand response control strategies under various user electricity consumption categories

[0101] Based on the results obtained using a hybrid clustering framework, users' electricity consumption behavior was divided into five categories, such as... Figure 8 As shown, corresponding control strategies are proposed for each category. Through reasonable, low-energy-consumption, and efficient control strategies, we can constrain users' electricity consumption behavior and provide practical solutions for responding to peak power grid stress events. This will contribute to the sustainable use of energy and the stable operation of the power system, thus laying the foundation for building a smart and reliable power supply network.

[0102] S5. Based on the user classification obtained using the hybrid clustering framework in step S4, provide control strategies for each type of user participating in electricity demand response, such as... Figure 9 As shown.

[0103] In another embodiment of the present invention, a user electricity consumption behavior classification system based on a hybrid clustering framework is provided. This system can be used to implement the above-mentioned user electricity consumption behavior classification method based on a hybrid clustering framework. Specifically, the user electricity consumption behavior classification system based on a hybrid clustering framework includes a data module, an extraction module, a first clustering module, a second clustering module, and an output module.

[0104] The data module acquires data on air conditioning electricity consumption, indoor temperature, and outdoor temperature of residents in residential buildings to obtain user datasets.

[0105] The extraction module extracts the air conditioner runtime and air conditioner start time for each user from the user dataset obtained by the data module.

[0106] The first clustering module constructs a hybrid clustering framework, considering the demand response time, execution time, and execution duration. Whether or not a user participates in the demand response is passed into the hybrid clustering framework as the first-level clustering criterion. The air conditioning runtime and air conditioning start-up time of each user obtained by the extraction module are passed into the first level of the hybrid clustering framework to perform the first-level clustering of electricity consumption behavior, resulting in Group A, which does not participate in the electricity demand response event, and Group B, which participates in the electricity demand response event.

[0107] The second clustering module considers the degree of overlap and start time between the demand response execution process and the air conditioning operation process. It inputs the degree of overlap and the start time of the two events into the hybrid clustering framework as the second-level clustering criteria. The user data of the B class participating in the power demand response event obtained from the first clustering module is input into the second layer of the hybrid clustering framework for second-level - deep clustering, dividing the B class participating in the power demand response event into four categories.

[0108] The output module provides control strategies for various types of users participating in power demand response based on the results obtained from the second clustering module.

[0109] In another embodiment of the present invention, a terminal device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to achieve corresponding method flows or corresponding functions. The processor described in this embodiment of the present invention can be used for the operation of a user electricity behavior classification method based on a hybrid clustering framework, including:

[0110] Data on air conditioning power consumption, indoor temperature, and outdoor temperature for residents of residential buildings are collected to obtain a user dataset. The air conditioning runtime and start-up time for each user are extracted from the user dataset. A hybrid clustering framework is constructed, considering demand response time, execution time, and execution duration. Whether or not a user participates in demand response is used as the first-level clustering criterion. The obtained air conditioning runtime and start-up time for each user are then used as the first-level clustering criterion for electricity consumption behavior, resulting in group A (non-participating in demand response events) and group B (participating in demand response events). The overlap between the demand response execution process and the air conditioning operation period, as well as the start time, are used as the second-level clustering criterion. Data from group B users participating in demand response events are then used as the second-level clustering criterion for deeper clustering. Through analysis, group B users participating in demand response events are divided into four categories. Control strategies are then proposed for each category of users participating in demand response events.

[0111] In another embodiment of the present invention, a storage medium is also provided, specifically a computer-readable storage medium (memory). This computer-readable storage medium is a memory device in a terminal device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and extended storage media supported by the terminal device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device.

[0112] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the user electricity consumption behavior classification method based on a hybrid clustering framework in the above embodiments; one or more instructions in the computer-readable storage medium are loaded by the processor and executed as follows:

[0113] Data on air conditioning power consumption, indoor temperature, and outdoor temperature for residents of residential buildings are collected to obtain a user dataset. The air conditioning runtime and start-up time for each user are extracted from the user dataset. A hybrid clustering framework is constructed, considering demand response time, execution time, and execution duration. Whether or not a user participates in demand response is used as the first-level clustering criterion. The obtained air conditioning runtime and start-up time for each user are then used as the first-level clustering criterion for electricity consumption behavior, resulting in group A (non-participating in demand response events) and group B (participating in demand response events). The overlap between the demand response execution process and the air conditioning operation period, as well as the start time, are used as the second-level clustering criterion. Data from group B users participating in demand response events are then used as the second-level clustering criterion for deeper clustering. Through analysis, group B users participating in demand response events are divided into four categories. Control strategies are then proposed for each category of users participating in demand response events.

[0114] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0115] The validation was performed using a hybrid clustering framework applied to user data from an Australian smart city pilot project. The variation curve of the sum of squared errors from the deep clustering layer is shown below. Figure 4 The profile coefficient variation curve is as follows: Figure 5 And the second derivative evaluation criterion curve, such as Figure 6 The study found that the optimal number of clusters is 4. The final user electricity behavior classification results are as follows: Figure 8 As shown.

[0116] Among them, by applying a hybrid clustering framework to the electricity consumption behavior data of residents in the Australian pilot area, the user groups were ultimately divided into five categories: A, B, C, D, and E.

[0117] Group A consists of groups that do not participate in electricity demand response, while groups B, C, D, and E are groups that do participate in demand response.

[0118] Group B is characterized by air conditioning operation events occurring during the execution of electricity demand response events;

[0119] Group C is characterized by electricity demand response events occurring during the execution of air conditioning operation events;

[0120] Group D is characterized by electricity demand response events occurring first, and ending before air conditioning operation events;

[0121] Group E is characterized by air conditioning operation events occurring first, and ending before the electricity demand response event.

[0122] To further verify the reliability and universality of the segmentation, data on the electricity consumption behavior of users in a certain residential community were used for validation. During the validation process, the second-level clustering was directly defined as four categories of electricity consumption behavior. Finally, the average hourly electricity consumption of the four user groups across the entire dataset was obtained, as follows: Figure 10 As shown.

[0123] As the amount of data increases, the uncertainty of user electricity consumption behavior will also increase. For user groups of categories B and C, the boundary remains clear and can be determined based on the range of the maximum continuous operating time of the air conditioner and the range of the power demand response execution time.

[0124] For users in categories D and E, the curves are steeper when calculating the average hourly electricity consumption for each category, and a peak occurs at a certain point in time.

[0125] Based on the location of the peak value, we verified whether the user group participated in electricity demand response. Based on the peak value's position at the time point of the electricity demand response execution phase, we determined whether the group belonged to one of the four pre-defined groups. Ultimately, the diagram shows that type 3 is class B users, type 2 is class C users, type 1 is class D users, and type 4 is class E users. This conclusion is consistent with the classification results of the user dataset from the Australian experimental area.

[0126] In summary, this invention presents a user electricity consumption behavior classification method and system based on a hybrid clustering framework. This method performs deeper clustering of user electricity consumption behavior, effectively distinguishing different groups and providing clear clustering divisions. By observing the clustering results, we can see that different data points are grouped into clusters with similar characteristics or behavioral patterns, thus providing a better understanding of the data's internal structure and features. In this process, a second-order derivative evaluation criterion is introduced to select the number of clusters, improving the accuracy of cluster selection to some extent. The hybrid clustering framework strategy proposed in this invention provides guidance for power management departments to formulate optimized control strategies for different user groups. Simultaneously, through the hybrid clustering framework, power management departments can more accurately plan the allocation of power resources. Based on the characteristics of user groups, the optimal operating time window and appropriate power allocation strategies can be determined. At different times, power management departments can adjust the power supply according to the needs of user groups to meet user demands and reduce energy waste.

[0127] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0128] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0129] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0130] In the embodiments provided by this invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0131] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0132] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0133] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0134] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0135] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0136] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0137] The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.

Claims

1. A user electricity consumption behavior classification method based on a hybrid clustering framework, characterized in that, Includes the following steps: S1. Obtain data on air conditioning power consumption, indoor temperature, and outdoor temperature of residents in residential buildings to obtain a user dataset; S2. Extract the air conditioner runtime and air conditioner start time for each user from the user dataset obtained in step S1. S3. Construct a hybrid clustering framework, considering the demand response time, execution time, and execution duration. Whether or not a user participates in the demand response is passed into the hybrid clustering framework as the first-level clustering criterion. The air conditioner runtime and air conditioner start-up time of each user obtained in step S2 are passed into the first level of the hybrid clustering framework to perform the first-level clustering of electricity consumption behavior, resulting in Group A, which does not participate in the electricity demand response event, and Group B, which participates in the electricity demand response event. S4. Consider the degree of overlap and start time between the demand response execution process and the air conditioning operation process. Input the degree of overlap and start time of the two events into the hybrid clustering framework as the second-level clustering criteria. The data of B-type users participating in the electricity demand response event obtained in step S3 are fed into the second layer of the hybrid clustering framework for second-level - deep clustering. Through analysis, the B-type group participating in the electricity demand response event is divided into four categories. S5. Based on the results obtained in step S4, provide control strategies for various types of users participating in power demand response.

2. The user electricity consumption behavior classification method based on a hybrid clustering framework according to claim 1, characterized in that, In step S1, the electricity consumption of air conditioning, indoor temperature and outdoor temperature data of residents in residential buildings are preprocessed. If the difference between two adjacent points exceeds the set error threshold range, it is considered abnormal load data. The preprocessed user electricity consumption data is saved in the corresponding list of each user.

3. The user electricity consumption behavior classification method based on a hybrid clustering framework according to claim 2, characterized in that, The user dataset used in the corresponding list for each user is: users={user1[t open ,t operate ,p t_use ],user2[t open ,t operate ,p t_use ],…,usern[t open ,t operate ,p t_use ]} Among them, t open For the time the user turns on the air conditioner, t operate For the user's air conditioner running time, p t_use This shows the distribution of users' electricity consumption time.

4. The user electricity consumption behavior classification method based on a hybrid clustering framework according to claim 1, characterized in that, In step S3, the second derivative evaluation index and the silhouette coefficient are introduced as evaluation indexes in the hybrid clustering framework.

5. The user electricity consumption behavior classification method based on a hybrid clustering framework according to claim 4, characterized in that, The evaluation index for the second derivative is: SSE k_best "=min(SSE j∈(2,k) ") Among them, SSE k_best "To determine the optimal number of clusters, SSE" j Let be the sum of squared errors for the j-th iteration.

6. The user electricity consumption behavior classification method based on a hybrid clustering framework according to claim 4, characterized in that, The profile coefficient is: Where S(i) is the contour coefficient value of the i-th sample point, a(i) is the cohesion of the sample point, and b(i) is the inter-class distance.

7. The user electricity consumption behavior classification method based on a hybrid clustering framework according to claim 1, characterized in that, In step S3, the first layer of the hybrid clustering framework is a shallow clustering layer. The user's air conditioning start time, air conditioning running time, and the time of the demand response event released by the power grid are input. The maximum number of clusters is 5. The error sum of squares change curve, profile coefficient change curve, and second derivative evaluation index change curve are plotted for each time. The optimal number of clusters is 2. Class A is defined as the category that does not participate in power demand response, and Class B is defined as the category that participates in power demand response.

8. The user electricity consumption behavior classification method based on a hybrid clustering framework according to claim 1, characterized in that, In step S4, the second layer of the hybrid clustering framework is a deep clustering layer. The inputs include the start time of the user's air conditioning operation, the air conditioning runtime, and the time of the demand response event issued by the power grid. The maximum number of clusters is 10. During this process, the curves showing the changes in the sum of squared errors, the profile coefficient, and the second-order derivative evaluation index are plotted for each instance. The optimal number of clusters is found to be 4. By analyzing the average daily operation of air conditioning for various user types and the execution period of the power demand response event, the following categories are identified: the first category is air conditioning operation events occurring during the execution period of the power demand response event; the second category is power demand response events occurring during the execution period of the air conditioning operation event; the third category is air conditioning operation events and power demand response events coinciding in time, with the power demand response event preceding and ending before the air conditioning operation event; and the fourth category is air conditioning operation events and power demand response events coinciding in time, with the air conditioning operation event preceding and ending before the power demand response event.

9. The user electricity consumption behavior classification method based on a hybrid clustering framework according to claim 1, characterized in that, In step S5, the specific power demand response control strategies under various user electricity consumption categories are as follows: Based on the results obtained using the hybrid clustering framework, users' electricity consumption behavior is divided into four categories: Category A, users who do not participate in electricity demand response; Category B, users who participate in electricity demand response, and whose air conditioning operation event occurs during the execution of the electricity demand response event; Category C, users who participate in electricity demand response, and whose electricity demand response event occurs during the execution of the air conditioning operation event; Category D, users who participate in electricity demand response, and whose electricity demand response event precedes the air conditioning operation event; and Category E, users who participate in electricity demand response, and whose air conditioning operation event precedes the electricity demand response event.

10. A user electricity consumption behavior classification system based on a hybrid clustering framework, characterized in that, include: The data module acquires data on air conditioning electricity consumption, indoor temperature, and outdoor temperature of residents in residential buildings to obtain user datasets; The extraction module extracts the air conditioner runtime and air conditioner start time for each user from the user dataset obtained by the data module. The first clustering module constructs a hybrid clustering framework, considering the demand response time, execution time, and execution duration. Whether or not a user participates in the demand response is passed into the hybrid clustering framework as the first-level clustering criterion. The air conditioning runtime and air conditioning start-up time of each user obtained by the extraction module are passed into the first level of the hybrid clustering framework to perform the first-level clustering of electricity consumption behavior, resulting in Group A, which does not participate in the electricity demand response event, and Group B, which participates in the electricity demand response event. The second clustering module considers the degree of overlap and start time between the demand response execution process and the air conditioning operation process. The degree of overlap and the start time of the two events are passed into the hybrid clustering framework as the second-level clustering criteria. The user data of category B participating in electricity demand response events obtained from the first clustering module is passed into the second layer of the hybrid clustering framework for second-level - deep clustering, which divides the category B group participating in electricity demand response events into four categories. The output module provides control strategies for various types of users participating in power demand response based on the results obtained from the second clustering module.

Images