Multi-level clustering method for micro-energy network users and load data acquisition system

Abstract

The invention relates to a multi-level clustering method for micro-energy network users and a load data acquisition system, the load data acquisition system comprises a side end distributed acquisition unit, a storage unit and a cloud end centralized processing unit, the load data acquired by the side end distributed acquisition unit is stored in the storage unit, and the cloud end centralized processing unit is used for processing the load data acquired by the side end distributed acquisition unit. The cloud centralized processing unit preprocesses the stored power load data, and then constructs a basic feature representation hierarchy unit, a user power consumption behavior portrait hierarchy unit and a schedulable potential mining hierarchy unit of the micro energy network user through a clustering algorithm, a user power consumption behavior portrait method and a schedulable potential evaluation method. The method has the characteristics of comprehensive user load data feature analysis and high reference value of the multi-level clustering result, can reflect the behavior rule of the micro-energy network users participating in demand response, deeply excavates the response potential of the micro-energy network comprehensive energy users, remarkably improves the system operation economy, and promotes the consumption of renewable energy sources.

Description

Technical Field

[0001] This invention relates to the field of clustering methods for electricity consumption behavior of microgrid users, and in particular to a multi-level clustering method and load data acquisition system for microgrid users. Technical Background

[0002] The demand for carbon reduction and efficiency improvement in the energy sector continues to grow, and the energy field is showing a new development trend of multi-energy coexistence, complementarity, and green transformation. As an important access mode for integrated energy, microgrids, with their flexible load characteristics and distributed energy features, help promote the absorption of renewable energy by the main grid and improve the operation and dispatch level of the entire integrated energy system.

[0003] Currently, my country's carbon market is becoming increasingly sophisticated, and the carbon tax policy is poised for implementation. Research on clustering methods for typical electricity consumption behaviors in microgrids can help to reasonably define carbon emission responsibilities and accurately control corporate carbon gains and losses in carbon reduction-related policies, representing a new path for the energy industry to achieve low-carbon and sustainable development.

[0004] A microgrid is a small-scale distributed integrated energy system that manages, transmits, and distributes energy, consisting of distributed energy sources, energy storage devices, energy conversion devices, loads, monitoring and protection devices, etc. It has flexible energy use methods and can make reasonable use of users' energy demand characteristics and local resource conditions through various energy conversions and load aggregation to achieve cascade utilization of energy. It is the physical carrier of the load side of the energy internet and also a high-quality dispatch resource.

[0005] With the development of science and technology, microgrids are being used more widely, the number of electricity users is constantly increasing, and the demand for electricity is also growing. It is essential to monitor users' electricity consumption, collect data on different users' electricity habits, electricity information, and electricity load, analyze this data to determine the electricity consumption characteristics of different users, classify users based on these characteristics, and adjust the power system in a timely manner according to user needs. Summary of the Invention

[0006] To address the problems and shortcomings of the existing technologies, the purpose of this invention is to provide a multi-level clustering method for microgrid users and a load data acquisition system. The technical solution adopted is as follows:

[0007] A multi-level clustering method for microgrid users and a load data acquisition system, comprising:

[0008] Collect user power load data and perform data preprocessing;

[0009] The data preprocessing involves removing abnormal user power load data and supplementing missing data values.

[0010] Based on the preprocessed microgrid user power load data, typical daily load data is selected, and the typical daily power load data of microgrid users is clustered to obtain the electricity consumption behavior categories of microgrid users. Basic feature representation hierarchical units are established, and the typical daily power load data of each unit is the cluster center load data.

[0011] Based on the basic feature representation hierarchy, an optimal user electricity consumption behavior feature index system is established. According to the load data of the cluster center of each basic feature representation hierarchy unit and the optimal user electricity consumption behavior feature index system, each unit in the basic feature representation hierarchy is scored and quantified to obtain the microgrid user electricity consumption behavior profile and establish user electricity consumption behavior profile hierarchy units.

[0012] Based on the hierarchical profile of user electricity consumption behavior, the dispatchable potential attribute indicators are evaluated. Microgrid users are clustered according to categorical and numerical indicators to obtain the results of microgrid user dispatchable potential attribute similarity clustering. Energy supply main unit, consumption main unit and supply-consumption balance unit at the dispatchable potential evaluation level are established to characterize the clustering characteristics of microgrid.

[0013] Furthermore, the process of collecting user power load data and performing data preprocessing includes:

[0014] Obtain microgrid user power load data from user power load data acquisition devices;

[0015] Based on the microgrid user power load data, calculate the microgrid user... exist Median of daily total load data ;

[0016] Step 1.3. Calculate each data point With median The absolute difference is expressed as follows:

[0017] ;

[0018] In the above formula, For users exist The j-th data point and the median The absolute difference For users exist The j-th data point of the day is for the user. exist The median of all load data during the day;

[0019] Calculate the median of all absolute differences ;

[0020] If data collection point absolute difference Greater than or less than of times or This multiple allows us to determine the user's identity. Electricity load data collection in Collection point in the middle of the day The system is constantly affected by noise, which can cause outliers to appear. These outliers are then removed.

[0021] Mark MicroNet users Missing data in power load data and the date of the missing data And determine the start time of the power load data loss period. and end time ;

[0022] The microgrid user power load data is segmented into weekly segments based on the date label of the load data, as expressed by: ,in This indicates the load data for week 1, and the date labels for the microgrid user power load data are aligned from Monday to Sunday.

[0023] Extracting the dates of missing data from microgrid user power load data Corresponding week date All load data Further extraction of load data for the corresponding time period with missing load data yielded the following results. ,in The corresponding collection points are Calculate the average value for each data point, using the following expression:

[0024] ,in Indicates user exist Total number of missing data points per day's load. The date of missing data within the load data collection period. Corresponding week date The first of all daily load data Load data at time t during the week;

[0025] In the above formula, For users exist The j-th data point and the median The absolute difference For users exist The j-th data point of the day is for the user. exist The median of all load data during the day;

[0026] The average value of each collected data point is used as the missing value to fill in the gaps, as shown in the expression:

[0027] ;

[0028] in, To determine the corresponding week date within the load data collection period The average value at time j across all daily load data. To determine the corresponding week date within the load data collection period The average value at time j+1 in all daily load data. To determine the corresponding week date within the load data collection period The average value at time j+n-1 in all daily load data.

[0029] Furthermore, based on the preprocessed microgrid user power load data, typical daily load data is selected, and clustering of the typical daily power load data of microgrid users is performed. This is done by measuring the distance between the microgrid user power load data vector and the cluster center using an improved K-means clustering algorithm based on time-domain alignment, including the following steps:

[0030] Construct the distance matrix D of typical daily electricity load data vectors of microgrid users;

[0031] For microgrid users Typical daily electricity load data With MicroNetwork users Typical daily electricity load data Construct a distance matrix D, where the number of rows in distance matrix D is the number of microgrid user i. Data from 1 measurement point, the number of columns is that of microgrid user j. The distance matrix D, representing the data from the measurement points, is expressed as follows:

[0032] ;

[0033] Construct a cumulative distance matrix D based on the distance matrix of typical daily electricity load data vectors of microgrid users. w The expression is:

[0034]

[0035] in, For indexing;

[0036] Based on the cumulative distance matrix D of the microgrid user power load vector w Able to obtain micro-energy network users Typical daily electricity load data With MicroNetwork users Typical daily electricity load data The dynamic time warp distance between them is .

[0037] Furthermore, the establishment of the optimal user electricity consumption behavior characteristic index system includes:

[0038] Calculate the mutual information and correlation between the initial feature index system and user electricity consumption behavior categories;

[0039] Based on the correlation between the initial feature index system and the user's electricity consumption behavior category, the optimal user electricity consumption behavior feature index system is selected through maximum correlation and minimum redundancy criterion.

[0040] Furthermore, step 3.1, which involves calculating the mutual information and correlation between the initial feature index system and the user's electricity consumption behavior categories, includes the following steps:

[0041] Step 3.1.1. For continuous numerical type indicators: Discretize continuous numerical type indicators, that is, the first one in the initial feature indicator system. Individual electricity consumption behavior characteristics indicators For continuous numerical indicators, Divided into Calculate the interval in the initial feature index system. Individual electricity consumption behavior characteristics indicators Information entropy, expressed as:

[0042]

[0043] in, Indicates that there is Electricity consumption behavior characteristics of individual users Falling in Each interval Indicates the total number of users;

[0044] Step 3.1.2. For discrete numerical type indices: the first in the initial feature index system Individual electricity consumption behavior characteristics indicators For discrete numerical indexes, calculate the first value in the initial feature index system. Individual electricity consumption behavior characteristics indicators Information entropy, expressed as:

[0045]

[0046] in, Indicates that there is Electricity consumption behavior characteristics of individual users For the first A discrete value;

[0047] Step 3.1.3. Calculate the user's electricity consumption behavior category Information entropy, expressed as:

[0048]

[0049] Among them, C k Indicates the Cth k User electricity consumption behavior categories This represents the total number of user electricity consumption behavior categories. Indicates the Cth k Number of users in each electricity consumption behavior category;

[0050] Step 3.1.4. Calculation and The joint information entropy between them is expressed as:

[0051]

[0052] in, This indicates that it belongs to user electricity consumption behavior category C. k And electricity consumption behavior characteristics indicators Falling in Number of users in each interval;

[0053] Step 3.1.5. Calculate the first... Individual electricity consumption behavior characteristics indicators Related to user electricity consumption behavior categories mutual information and based on mutual information Calculate correlation The expression is:

[0054]

[0055] Based on the Individual electricity consumption behavior characteristics indicators Related to user electricity consumption behavior categories mutual information Calculate the correlation of mutual information The expression is:

[0056] .

[0057] Furthermore, the step of characterizing the load data of the cluster centers of each basic feature-based hierarchical unit and the optimal user electricity consumption behavior characteristic index system includes the following steps:

[0058] Step a. Calculate user electricity consumption behavior category C kThe first of all users Optimal electricity consumption behavior characteristic indicators average ;

[0059] Step b. Record user electricity consumption behavior category C k All users in the middle Optimal electricity consumption behavior characteristic indicators maximum value and minimum value ;

[0060] Step c. Calculation .

[0061] Furthermore, the evaluation of the schedulable potential attribute indicators includes the following steps:

[0062] Step 4.1. Count and record the number of microgrid users. Typical daily photovoltaic output and electric vehicles Distributed flexible resources Percentage in typical daily electricity load data ;

[0063] Step 4.2. Based on the microgrid users Typical daily electricity load data, calculating microgrid users Typical daily schedulable potential index The expression is:

[0064] ,

[0065] in, A larger value indicates greater volatility and higher dispatchability of typical daily power load data for the user, while a smaller value indicates lower dispatchability. Indicates MicroNet users Typical daily maximum power load; Indicates MicroNet users Typical daily minimum power load; Indicates MicroNet users Typical daily average power load; based on dispatchable potential indicators Based on experience, scheduling experts categorize users into those with strong scheduling potential and those with weak scheduling potential.

[0066] Step 4.3. Use vectors to describe the electricity consumption attributes of users in the microgrid. The expression is:

[0067] .

[0068] Furthermore, the categorical indicators are: user electricity consumption behavior category indicators and dispatchability potential strength indicators. The clustering of microgrid users based on categorical and numerical indicators includes the following steps:

[0069] Step a. Standardize numerical indicators and select the number of clusters according to the clustering objective. Random selection 1 data point is used as the initial cluster center C s Initialize the number of iterations Z s =1;

[0070] Step b. Calculate the combined distance between each user attribute vector and the cluster center;

[0071] Step c. Assign users to cluster centers based on the combined distance between user attribute vectors and cluster centers. In each cluster;

[0072] Step d. Calculation The average distance between the power load vectors in each cluster is used as the new cluster center, and the cluster center C is updated. s+1 Number of iterations Z s +1;

[0073] Step e. Calculate C s With C s+1 The distance between new cluster centers and old cluster centers is considered when the distance between them is greater than a minimum distance threshold. If the new cluster centers are less than or equal to the minimum distance threshold, replace the old cluster centers with the new ones and return to step b for iteration; The iteration ends when the number of iterations exceeds the maximum number of iterations. At this point, clustering terminates.

[0074] A system for collecting user load data in microgrids includes:

[0075] The data acquisition module is used to collect user power load data and perform data preprocessing.

[0076] The clustering module is used to select typical daily load data based on the preprocessed microgrid user power load data, cluster the typical daily power load data of microgrid users, obtain the electricity consumption behavior categories of microgrid users, establish basic feature representation hierarchical units, and the typical daily power load data of each unit is the cluster center load data.

[0077] The hierarchical unit establishment module is used to establish an optimal user electricity consumption behavior characteristic index system based on the basic feature representation hierarchy. According to the load data of the cluster center of each basic feature representation hierarchy unit and the optimal user electricity consumption behavior characteristic index system, each unit in the basic feature representation hierarchy is scored and quantified to obtain the microgrid user electricity consumption behavior profile and establish the user electricity consumption behavior profile hierarchical unit.

[0078] The module for establishing a scheduling potential assessment hierarchical unit is used to assess scheduling potential attribute indicators based on user electricity consumption behavior profiles. It clusters microgrid users according to categorical and numerical indicators to obtain the results of microgrid user scheduling potential attribute similarity clustering, establishes scheduling potential assessment hierarchical units, and characterizes the clustering features of microgrids.

[0079] Furthermore, the distributed sensor data acquisition module is used to collect power load data, meaning that it is used every [time period]. Data is collected once, daily. Data points, continuously collected Daily power load data.

[0080] A computer device includes a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor, when executing the computer program, implements the steps of any one of the methods for multi-level clustering of microgrid users.

[0081] A computer program is stored on a computer storage medium, and when the computer program is executed by a processor, it implements any of the steps of a multi-level clustering method for microgrid users.

[0082] Compared with existing technologies, its advantages are as follows:

[0083] This invention provides a multi-level clustering method and load data acquisition system for microgrid users. The load data acquisition system includes an edge-distributed acquisition unit, a storage unit, and a cloud-based centralized processing unit. The load data collected by the edge-distributed acquisition unit is stored in the storage unit. The cloud-based centralized processing unit preprocesses the stored power load data. After preprocessing, the user power load data is used to construct basic feature representation units, user power consumption behavior profiling units, and dispatchable potential mining units for microgrid users through an improved K-means clustering algorithm based on time-domain alignment, a user electricity consumption behavior profiling method, a dispatchable potential assessment method, and a clustering algorithm that calculates comprehensive similarity across index types. This method features comprehensive analysis of microgrid user load data characteristics and high reference value for multi-level clustering results. It can realistically and comprehensively reflect the behavioral patterns of microgrid users participating in demand response, deeply explore the response potential of integrated energy users in microgrids, significantly improve the economic efficiency of system operation, and promote the consumption of renewable energy. Attached Figure Description

[0084] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0085] Figure 1 This is a block diagram of the multi-level clustering method for microgrid users and the load data acquisition system of the present invention.

[0086] Figure 2 This is a structural diagram of the microgrid user power load data acquisition system of the present invention;

[0087] Figure 3 This is a flowchart of the microgrid user power load data preprocessing process of the present invention;

[0088] Figure 4 This is a flowchart of the user electricity consumption behavior profiling method of the present invention;

[0089] Figure 5 This is a flowchart of the dynamic time warping process of the present invention;

[0090] Figure 6 This is a flowchart illustrating the mutual information and correlation between the initial feature index system of this invention and user electricity consumption behavior categories;

[0091] Figure 7 Flowchart for selecting the optimal user electricity consumption behavior characteristic index system for this invention;

[0092] Figure 8 A flowchart for establishing a hierarchical unit for user electricity consumption behavior profiling in this invention;

[0093] Figure 9 This is a flowchart illustrating how the present invention clusters microgrid users based on categorical and numerical indicators. Detailed Implementation

[0094] 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 embodiments of the present invention, and not all embodiments. 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.

[0095] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0096] Example 1

[0097] This invention provides an embodiment, such as Figure 2 As shown, a microgrid user load data acquisition system includes an edge-distributed acquisition unit, a storage unit, and a cloud-based centralized processing unit. The edge-distributed acquisition unit is connected to the storage unit, and the storage unit is connected to the cloud-based centralized processing unit. The edge-distributed acquisition unit consists of a distributed sensor data acquisition module, a distributed sensor data conversion module, and a distributed sensor data processing module. The distributed sensor data acquisition module is used to acquire power load data. The acquired power load data is processed by the distributed sensor data conversion module and the distributed sensor data processing module, and then stored in the storage unit. The cloud-based centralized processing unit preprocesses the stored power load data. The distributed sensor data acquisition module is used to acquire power load data, meaning it acquires data every [number of data points]. Data is collected once, daily. Data points, continuously collected Daily power load data.

[0098] A system for collecting user load data in microgrids includes:

[0099] The data acquisition module is used to collect user power load data and perform data preprocessing.

[0100] The clustering module is used to select typical daily load data based on the preprocessed microgrid user power load data, cluster the typical daily power load data of microgrid users, obtain the electricity consumption behavior categories of microgrid users, establish basic feature representation hierarchical units, and the typical daily power load data of each unit is the cluster center load data.

[0101] The hierarchical unit establishment module is used to establish an optimal user electricity consumption behavior characteristic index system based on the basic feature representation hierarchy. According to the load data of the cluster center of each basic feature representation hierarchy unit and the optimal user electricity consumption behavior characteristic index system, each unit in the basic feature representation hierarchy is scored and quantified to obtain the microgrid user electricity consumption behavior profile and establish the user electricity consumption behavior profile hierarchical unit.

[0102] The module for establishing a scheduling potential assessment hierarchical unit is used to assess scheduling potential attribute indicators based on user electricity consumption behavior profiles. It clusters microgrid users according to categorical and numerical indicators to obtain the results of microgrid user scheduling potential attribute similarity clustering, establishes scheduling potential assessment hierarchical units, and characterizes the clustering features of microgrids.

[0103] Example 2

[0104] The present invention provides another embodiment, such as... Figure 3 As shown, user power load data is collected through a microgrid user power load data acquisition device, and the data is preprocessed. Data preprocessing includes removing abnormal user power load data and supplementing missing data. Data removal utilizes the characteristic that microgrid user power load data exhibits an approximately weekly cycle; if there are abnormal user power load data... On a certain day A certain collection point If the power load value is greater than / less than the maximum / minimum load for the same period, then the user will be... exist All daily load data were marked as potentially abnormal and further analyzed using the median absolute value difference method. Missing data was imputed by calculating the average value of each data point, with the specific steps as follows:

[0105] Step 1.1. Obtain microgrid user power load data from the user power load data acquisition device;

[0106] Step 1.2. Based on the obtained microgrid user power load data, calculate the microgrid user... exist Median of daily total load data ;

[0107] Step 1.3. Calculate each data point With median The absolute difference is expressed as follows:

[0108] ;

[0109] In the above formula, To determine the corresponding week date within the load data collection period The average value at time j across all daily load data. To determine the corresponding week date within the load data collection period The average value at time j+1 in all daily load data. To determine the corresponding week date within the load data collection period The average value at time j+n-1 in all daily load data.

[0110] If the missing data is October 21st, corresponding to a Tuesday, the average of the load data at time j from all Tuesdays within the collection period will be used as the starting time for the missing data. The same applies to time j+1. Here, time j corresponds to the start time of the missing power load data period. The end time is at time j+n-1. .

[0111] Step 1.4. Calculate the median of all absolute differences. ;

[0112] Step 1.5. If the data collection point absolute difference Greater than or less than of times or This multiple allows us to determine the user's identity. Electricity load data collection in Collection point in the middle of the day The system is constantly affected by noise, which can cause outliers to appear. These outliers are then removed.

[0113] Step 1.6. Mark Microgrid users Missing data in power load data and the date of the missing data And determine the start time of the power load data loss period. and end time ;

[0114] Step 1.7. Divide the microgrid user power load data into segments based on the date label of the load data, using weeks as the scale. The expression is: ,in: This indicates the load data for week 1, and the date labels for the microgrid user power load data are aligned from Monday to Sunday.

[0115] Step 1.8. Extract the dates containing missing data from the microgrid user power load data. Corresponding week date All load data Further extraction of load data for the corresponding time period with missing load data yielded the following results. ,in The corresponding collection points are Calculate the average value for each data point, using the following expression: ,in Indicates user exist Total number of missing data points per day's load. The date of missing data within the load data collection period. Corresponding week date The first of all daily load data Load data at time t during the week;

[0116] For example, if the missing data is October 21st, the corresponding weekday is Tuesday. This indicates all weekdays within the data collection period. Load data at time t during the week.

[0117] Step 1.9. Use the average value of each collected data point as the missing value to fill in the missing values. The expression is:

[0118] ;

[0119] in To determine the corresponding week date within the load data collection period The average value at time j across all daily load data. To determine the corresponding week date within the load data collection period The average value at time j+1 in all daily load data. To determine the corresponding week date within the load data collection period The average value at time j+n-1 in all daily load data.

[0120] Example 3

[0121] The present invention provides another embodiment, such as... Figure 4 As shown, based on the preprocessed microgrid user power load data, typical daily load data is selected, and the typical daily power load data of microgrid users is clustered to obtain the electricity consumption behavior categories of microgrid users. Basic feature representation hierarchical units are established, and the typical daily power load data of each unit is the cluster center load data.

[0122] Based on the basic feature representation hierarchy, an optimal user electricity consumption behavior feature index system is established. According to the load data of the cluster center of each basic feature representation hierarchy unit and the optimal user electricity consumption behavior feature index system, each unit in the basic feature representation hierarchy is scored and quantified to obtain the microgrid user electricity consumption behavior profile and establish user electricity consumption behavior profile hierarchy units.

[0123] Example 4

[0124] The present invention provides another embodiment, such as... Figure 5 As shown, clustering of typical daily electricity load data of microgrid users is performed by measuring the distance between the microgrid user electricity load data vector and the cluster center using an improved K-means clustering algorithm based on time-domain alignment. The specific steps are as follows:

[0125] Step 2.1. Construct the distance matrix D of the typical daily electricity load data vector of microgrid users;

[0126] For microgrid users Typical daily electricity load data With MicroNetwork users Typical daily electricity load data Construct a distance matrix D, where the number of rows in distance matrix D is the number of microgrid user i. Data from 1 measurement point, the number of columns is that of microgrid user j. The distance matrix D, representing the data from the measurement points, is expressed as follows:

[0127] ;

[0128] Step 2.2. Construct the cumulative distance matrix D based on the distance matrix of typical daily electricity load data vectors of microgrid users. w The expression is:

[0129]

[0130] in, For indexing.

[0131] Step 2.3. Based on the cumulative distance matrix D of the microgrid user power load vector... w We can find micro-energy network users Typical daily electricity load data With MicroNetwork users Typical daily electricity load data The dynamic time warp distance between them is .

[0132] Example 5

[0133] The present invention provides another embodiment, such as... Figure 6 As shown, the mutual information and correlation between the initial feature index system and user electricity consumption behavior categories are calculated using the following steps:

[0134] Step 3.1.1. For continuous numerical type indicators: Discretize continuous numerical type indicators, that is, the first one in the initial feature indicator system. Individual electricity consumption behavior characteristics indicators For continuous numerical indicators, Divided into Calculate the interval in the initial feature index system. Individual electricity consumption behavior characteristics indicators Information entropy, expressed as:

[0135]

[0136] Where, N v Indicates that there are N v Electricity consumption behavior characteristics of individual users Falling in There are several intervals, where N represents the total number of users;

[0137] Step 3.1.2. For discrete numerical type indices: the first in the initial feature index system Individual electricity consumption behavior characteristics indicators For discrete numerical indexes, calculate the first value in the initial feature index system. Individual electricity consumption behavior characteristics indicators Information entropy, expressed as:

[0138]

[0139] Where, N v Indicates that there are N v Electricity consumption behavior characteristics of individual users For the first A discrete value;

[0140] Step 3.1.3. Calculate the information entropy of user electricity consumption behavior category C, expressed as:

[0141]

[0142] Among them, C k Indicates the Cth k User electricity consumption behavior categories This represents the total number of user electricity consumption behavior categories. Indicates the Cth k Number of users in each electricity consumption behavior category;

[0143] Step 3.1.4. Calculation The joint information entropy between C and α is expressed as:

[0144]

[0145] in, This indicates that it belongs to user electricity consumption behavior category C. k And electricity consumption behavior characteristics indicators Falling in Number of users in each interval;

[0146] Step 3.1.5. Calculate the first... Individual electricity consumption behavior characteristics indicators Mutual information with user electricity consumption behavior category C and based on mutual information Calculate correlation The expression is:

[0147]

[0148] Based on the Individual electricity consumption behavior characteristics indicators Mutual information with user electricity consumption behavior category C Calculate the correlation of mutual information The expression is:

[0149] .

[0150] Example 6

[0151] The present invention provides another embodiment, such as... Figure 7 As shown, based on the correlation between the initial feature index system and user electricity consumption behavior categories, the optimal user electricity consumption behavior feature index system is selected through maximum correlation and minimum redundancy criterion measurement, and D is adopted. e (T,C) represents the correlation between the initial characteristic indicators and the user's electricity consumption behavior category. ,in: Represents the total number of features; using Indicates redundancy among initial feature indices To select the optimal index, define... Using the traversal method, let the optimal set of electricity consumption behavior characteristic indicators be T. best , making T best Traverse from [0,0,0,0,…,0] to [1,1,1,1,…,1], that is, traverse all combinations of the initial feature indicators and select one of them. The largest initial combination of characteristic indicators is taken as the optimal set of electricity consumption behavior characteristic indicators T. best .

[0152] Example 7

[0153] The present invention provides another embodiment, which is a multi-level clustering method for microgrid users.

[0154] The specific steps for calculating the correlation between feature indicators are as follows:

[0155] Step a. Calculate the first Individual characteristic indicators Information entropy , No. Individual characteristic indicators Information entropy , No. Individual characteristic indicators With the Individual characteristic indicators Inter-information ;

[0156] Step b. Calculate the mutual information after correlation correction based on this. , representing characteristic indicators and The correlation between them.

[0157] .

[0158] Example 8

[0159] The present invention provides another embodiment, which is a multi-level clustering method for microgrid users.

[0160] like Figure 8 As shown, based on the optimal user electricity consumption behavior characteristic index, each unit in the basic feature representation layer is scored and quantified. The specific steps are as follows:

[0161] Step a. Calculate user electricity consumption behavior category C k The first of all users Optimal electricity consumption behavior characteristic indicators average ;

[0162] Step b. Record user electricity consumption behavior category C k All users in the middle Optimal electricity consumption behavior characteristic indicators maximum value and minimum value ;

[0163] Step c. Calculation .

[0164] Example 9

[0165] The present invention provides another embodiment, which is a multi-level clustering method for microgrid users.

[0166] Each unit in the basic feature representation hierarchy is scored and quantified to obtain a microgrid user electricity consumption behavior profile, and a user electricity consumption behavior profile hierarchy unit is established. The specific steps are as follows:

[0167] Calculate user electricity consumption behavior category C k The The scores of the optimal electricity consumption behavior characteristic indicators are used to normalize the user electricity consumption behavior category C using the Softmax function. k The optimal electricity consumption behavior characteristic index vector is expressed as:

[0168]

[0169] The user electricity consumption behavior category C after normalization by the Softmax function. k The optimal electricity consumption behavior feature vector is used as the scoring result and as the category label for the electricity consumption behavior of this type of user, thereby obtaining the electricity consumption behavior profile of the microgrid user and establishing a hierarchical unit for the user electricity consumption behavior profile.

[0170] Example 10

[0171] The present invention provides another embodiment, which is a multi-level clustering method for microgrid users.

[0172] like Figure 9 As shown, a clustering method that calculates comprehensive similarity across indicator types is used to cluster all users of the microgrid according to categorical and numerical indicators. The specific steps are as follows:

[0173] Step a. Standardize numerical indicators and select the number of clusters according to the clustering objective. Random selection 1 data point is used as the initial cluster center C s Initialize the number of iterations Z s =1;

[0174] Step b. Calculate the combined distance between each user attribute vector and the cluster center;

[0175] In calculating the composite distance, the squared Euclidean distance is used for numerical feature indicators, and the Hamming distance is used for categorical feature indicators. The expression is as follows:

[0176]

[0177] in , This indicates that when categorical indicators belong to the same category... Conversely, when categorical indicators belong to different categories... , This represents the user's electricity consumption attributes after maximum and minimum standardization. This represents the electricity consumption attributes of users in the cluster centers after maximal and minimum standardization.

[0178] Step c. Assign users to cluster centers based on the combined distance between user attribute vectors and cluster centers. In each cluster;

[0179] Step d. Calculation The average distance between the power load vectors in each cluster is used as the new cluster center, and the cluster center C is updated. s+1 Number of iterations Z s +1;

[0180] Step e. Calculate C s With C s+1 The distance between new cluster centers and old cluster centers is considered when the distance between them is greater than a minimum distance threshold. If the new cluster centers are less than or equal to the minimum distance threshold, replace the old cluster centers with the new ones and return to step b for iteration; The iteration ends when the number of iterations exceeds the maximum number of iterations. At this point, clustering terminates.

[0181] Example 11

[0182] Based on the same inventive concept, embodiments of the present invention also provide a computer device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor. When the processor executes the computer program, it implements the steps of the multi-level clustering method for microgrid users described in embodiments 1-10.

[0183] Example 12

[0184] Based on the same inventive concept, this embodiment of the invention also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the steps of the multi-level clustering method for microgrid users described in embodiments 1-10.

[0185] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0186] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. 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 illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0187] 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.

[0188] 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.

[0189] In the description of this specification, the terms "one embodiment," "some embodiments," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0190] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A multi-level clustering method for microgrid users and a load data acquisition system, characterized in that, include: Collect user power load data and perform data preprocessing; Based on the preprocessed microgrid user power load data, typical daily load data is selected, and the typical daily power load data of microgrid users is clustered to obtain the electricity consumption behavior categories of microgrid users. Basic feature representation hierarchical units are established, and the typical daily power load data of each unit is the cluster center load data. Based on the basic feature representation hierarchy, an optimal user electricity consumption behavior feature index system is established. According to the load data of the cluster center of each basic feature representation hierarchy unit and the optimal user electricity consumption behavior feature index system, each unit in the basic feature representation hierarchy is scored and quantified to obtain the microgrid user electricity consumption behavior profile and establish user electricity consumption behavior profile hierarchy units. Based on the hierarchical profile of user electricity consumption behavior, the dispatchable potential attribute indicators are evaluated. Microgrid users are clustered according to categorical and numerical indicators to obtain the results of microgrid user dispatchable potential attribute similarity clustering. Dispatchable potential evaluation hierarchical units are established to characterize the clustering characteristics of microgrids.

2. The multi-level clustering method for microgrid users according to claim 1, characterized in that, Collect user power load data and perform data preprocessing, including: Obtain microgrid user power load data from user power load data acquisition devices; Based on the microgrid user power load data, calculate the microgrid user... exist Median of daily total load data ; Calculate each data point With median The absolute difference is expressed as follows: ； In the above formula, For users exist The j-th data point and the median The absolute difference For users exist The j-th data point of the day is for the user. exist The median of all load data during the day; Calculate the median of all absolute differences ; If data collection point absolute difference Greater than or less than of times or This multiple allows us to determine the user's identity. Electricity load data collection in Collection point in the middle of the day The system is constantly affected by noise, which can cause outliers to appear. These outliers are then removed. Mark MicroNet users Missing data in power load data and the date of the missing data And determine the start time of the power load data loss period. and end time ; The microgrid user power load data is segmented into weekly segments based on the date label of the load data, as expressed by: ,in This indicates the load data for the first week, and the date labels for the microgrid user power load data are aligned from Monday to Sunday. Extracting the dates of missing data from microgrid user power load data Corresponding week date All load data Further load data was extracted for the corresponding time periods when load data was missing. ,in The corresponding collection points are Calculate the average value for each data point, using the following expression: ,in Indicates user exist Total number of missing data points per day's load. The date of missing data within the load data collection period. Corresponding week date The first of all daily load data Load data at time t during the week; The average value of each collected data point is used as the missing value to fill in the gaps, as shown in the expression: ； in, To determine the corresponding week date within the load data collection period The average value at time j across all daily load data. To determine the corresponding week date within the load data collection period The average value at time j+1 in all daily load data. To determine the corresponding week date within the load data collection period The average value at time j+n-1 in all daily load data.

3. The multi-level clustering method for microgrid users according to claim 1, characterized in that, The process involves selecting typical daily load data from the preprocessed microgrid user electricity load data and clustering this data. This is achieved using an improved K-means clustering algorithm based on time-domain alignment to measure the distance between the microgrid user electricity load data vector and the cluster centers. The methods include: Construct the distance matrix D of typical daily electricity load data vectors of microgrid users; For microgrid users Typical daily power load data With MicroNetwork users Typical daily power load data Construct a distance matrix D, where the number of rows in distance matrix D is the number of microgrid user i. Data from 1 measurement point, the number of columns is that of microgrid user j. The distance matrix D, representing the data from the measurement points, is expressed as follows: ； Construct a cumulative distance matrix D based on the distance matrix of typical daily electricity load data vectors of microgrid users. w The expression is: ； in, For indexing; Based on the cumulative distance matrix D of the microgrid user power load vector w Able to obtain micro-energy network users Typical daily power load data With MicroNetwork users Typical daily power load data The dynamic time warp distance between them is .

4. The multi-level clustering method for microgrid users according to claim 1, characterized in that, The establishment of the optimal user electricity consumption behavior characteristic index system includes: Calculate the mutual information and correlation between the initial feature index system and user electricity consumption behavior categories; Based on the correlation between the initial feature index system and the user's electricity consumption behavior category, the optimal user electricity consumption behavior feature index system is selected through maximum correlation and minimum redundancy criterion.

5. A multi-level clustering method for microgrid users according to claim 1, characterized in that, The system of load data from cluster centers of hierarchical units and the optimal user electricity consumption behavior characteristic index system based on each basic feature characterization level includes the following steps: Step a. Calculate user electricity consumption behavior category C k The first of all users Optimal electricity consumption behavior characteristic indicators average ; Step b. Record user electricity consumption behavior category C k All users in the middle Optimal electricity consumption behavior characteristic indicators maximum value and minimum value ; Step c. Calculation .

6. The multi-level clustering method for microgrid users according to claim 1, characterized in that, The evaluation of schedulable potential attributes includes the following steps: Step 4.

1. Count and record the number of microgrid users. Typical daily photovoltaic output and electric vehicles Distributed flexible resources Percentage in typical daily electricity load data ; Step 4.

2. Based on the microgrid users Typical daily electricity load data, calculating microgrid users Typical daily schedulable potential index The expression is: ， in, A larger value indicates greater volatility and higher dispatchability of typical daily power load data for the user, while a smaller value indicates lower dispatchability. Indicates MicroNet users Typical daily maximum power load; Indicates MicroNet users Typical daily minimum power load; Indicates MicroNet users Typical daily average power load; based on dispatchable potential indicators Based on experience, scheduling experts categorize users into those with strong scheduling potential and those with weak scheduling potential. Step 4.

3. Use vectors to describe the electricity consumption attributes of users in the microgrid. The expression is: 。 7. The multi-level clustering method for microgrid users according to claim 1, characterized in that, The clustering of microgrid users based on categorical and numerical indicators includes the following steps: Step a. Standardize numerical indicators and select the number of clusters according to the clustering objective. Random selection 1 data point as the initial cluster center C s Initialize the number of iterations Z s =1; Step b. Calculate the combined distance between each user attribute vector and the cluster center; Step c. Assign users to cluster centers based on the combined distance between user attribute vectors and cluster centers. In each cluster; Step d. Calculation The average distance between the power load vectors in each cluster is used as the new cluster center, and the cluster center C is updated. s+1 Number of iterations Z s +1; Step e. Calculate C s With C s+1 The distance between new cluster centers and old cluster centers is considered when the distance between them is greater than a minimum distance threshold. If the new cluster centers are less than or equal to the minimum distance threshold, replace the old cluster centers with the new ones and return to step b for iteration; When the iteration ends; When the number of iterations is greater than the maximum number of iterations At this point, clustering terminates.

8. A system for acquiring user load data in a microgrid, characterized in that, include: The data acquisition module is used to collect user power load data and perform data preprocessing. The clustering module is used to select typical daily load data based on the preprocessed microgrid user power load data, cluster the typical daily power load data of microgrid users, obtain the electricity consumption behavior categories of microgrid users, establish basic feature representation hierarchical units, and the typical daily power load data of each unit is the cluster center load data. The hierarchical unit establishment module is used to establish an optimal user electricity consumption behavior characteristic index system based on the basic feature representation hierarchy. According to the load data of the cluster center of each basic feature representation hierarchy unit and the optimal user electricity consumption behavior characteristic index system, each unit in the basic feature representation hierarchy is scored and quantified to obtain the microgrid user electricity consumption behavior profile and establish the user electricity consumption behavior profile hierarchical unit. The module for establishing a scheduling potential assessment hierarchical unit is used to assess scheduling potential attribute indicators based on user electricity consumption behavior profiles. It clusters microgrid users according to categorical and numerical indicators to obtain the results of microgrid user scheduling potential attribute similarity clustering, establishes scheduling potential assessment hierarchical units, and characterizes the clustering features of microgrids.

9. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of a multi-level clustering method for microgrid users as described in any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium contains a computer program, which, when executed by a processor, implements the steps of a multi-level clustering method for microgrid users as described in any one of claims 1-7.

Images