A method, device and terminal device for estimating user cluster baseline load
Through clustering and graph convolutional neural network technology, the problem of low accuracy of cluster baseline load estimation is solved, and higher estimation accuracy and transaction fairness are achieved.
Patent Information
- Application Number
- CN202110780395.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-07-09
- Publication Date
- 2025-03-25
- Estimated Expiration
- 2041-07-09
AI Technical Summary
The accuracy of estimation of cluster baseline load in the prior art has affected the fairness of transactions between power system operators and aggregators.
By obtaining the historical load data of the user cluster to be estimated, performing cluster analysis, computing the spatial correlation between multiple users, constructing an undirected graph of spatial correlation, and estimating the cluster baseline load using graph convolutional neural network.
It significantly improves the estimation accuracy of the cluster baseline load, ensures the fairness of demand response settlement, and is conducive to the implementation and promotion of demand response projects.
Smart Images
Figure CN113705957B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power grid demand response, and in particular to a user cluster baseline load estimation method, device and terminal equipment. Background Art
[0002] With the advancement of the marketization of the power system, demand response technology has received more and more attention. Aggregators gather a large number of users to participate in demand response by paying participation compensation, thereby obtaining a larger overall capacity, and then sell these capacities to power system operators for profit. The bonus paid by the power system operator to the aggregator is based on the cluster load reduction, which is equal to "the difference between the load that should be consumed if all users do not participate in demand response and the load actually consumed after participating in demand response", where the latter is the actual measurement data and the former is defined as the baseline load of the cluster users, that is, the load that should be consumed if all users do not participate in demand response.
[0003] The cluster user baseline load is the guarantee for arranging reasonable trading strategies in the market environment. Once the cluster baseline load is overestimated, the corresponding cluster reduction will be overestimated, and the interests of the power system operator will be damaged; once the cluster baseline load is underestimated, the corresponding cluster reduction will also be underestimated, and the interests of the aggregator will be damaged. In order to ensure the fairness of transactions between power system operators and aggregators, it is necessary to accurately estimate the cluster baseline load.
[0004] In the prior art, the cluster baseline load estimation result is usually a simple sum of the individual user baseline load estimation results. However, due to the large fluctuations in the electricity consumption behavior of individual users, the baseline load estimation for individual users often has large errors, so the accuracy of the cluster baseline estimation result obtained by this method is not ideal. Summary of the invention
[0005] In view of this, an embodiment of the present invention provides a method for estimating a user cluster baseline load to solve the problem of low cluster baseline load estimation accuracy in the prior art.
[0006] In order to achieve the above object, the present invention adopts the following technical scheme:
[0007] In a first aspect, an embodiment of the present invention provides a method for estimating a baseline load of a user cluster, comprising: obtaining historical load data of a user cluster to be estimated; clustering the user cluster to be estimated according to the historical load data to obtain multiple categories of users and load data of the multiple categories of users; calculating the spatial correlations between the multiple categories of users according to the load data of the multiple categories of users, and constructing a spatial correlation matrix according to the spatial correlations; taking the multiple categories of users as nodes of a spatial correlation undirected graph, and the load data corresponding to the multiple categories of users as node features; constructing the edges of the spatial correlation undirected graph according to the spatial correlation matrix; and estimating the baseline load of the user cluster to be estimated based on the spatial correlation undirected graph.
[0008] Based on the first aspect, in some embodiments, obtaining historical load data of the user cluster to be estimated includes: obtaining historical load data of the user cluster to be estimated on K non-DR days before the DR day d, represents the load data of user m on the kth non-DR day before DR day d; where, represents the mth user, M target users representing the cluster baseline load estimate; It means there are T time periods in a day, d represents a DR day, and k represents the kth non-DR day before the DR day.
[0009] Based on the first aspect, in some embodiments, clustering the user cluster to be estimated according to the historical load data to obtain multiple types of users and load data of the multiple types of users includes: calculating the typical load pattern of user m on the K non-DR days according to the historical load data; the typical load pattern The calculation formula is:
[0010]
[0011] Calculate the typical load patterns of all M users on the K non-DR days, determine the number of clusters according to the evaluation index of clustering performance, and cluster all M users into N clusters;
[0012] Determine the clustering center of each type of user based on the objective function min f:
[0013]
[0014] In the formula is the cluster center of the nth user, Indicates the N types of users obtained, Represents the nth type of user.
[0015] Based on the first aspect, in some embodiments, calculating the spatial correlation between the multiple types of users according to the load data of the multiple types of users, and forming a spatial correlation matrix according to the spatial correlation, includes: calculating the spatial correlation of the typical load patterns of the i-th user and the j-th user at the K non-DR day aggregation levels, as follows:
[0016]
[0017] in, is the spatial correlation between the i-th user and the j-th user, l i,d-k represents the aggregated load data of the i-th user on the k-th non-DR day before the DR day d, l j,d-k Similarly, the spatial correlation indicators between N types of users are used to form an N-order square matrix H, where the square matrix H is a spatial correlation matrix. The elements in the square matrix H are
[0018] Based on the first aspect, in some embodiments, the multiple types of users are used as nodes of a spatial correlation undirected graph, and the load data corresponding to the multiple types of users are node features, including: L n,d =[l n,d-K ,…,l n,d-1 ] T represents the load data of the n-th user at the aggregate level on the K non-DR event days, and the n-th user is regarded as a node V on the spatial correlation undirected graph n ; The load data of the nth user at the aggregate level on the K non-DR event days and the load data of the aggregate level on the DR event day F n =[L n,d ,l n,d ] as node V n The node characteristics of the undirected graph are The node feature set is
[0019] Based on the first aspect, in some embodiments, constructing the edge of the spatial correlation undirected graph according to the spatial correlation matrix includes: i and V j The spatial correlation between them is the edge of the spatial correlation undirected graph, and the edge is defined as e i,j , e i,j The length of is numerically equal to the value of the spatial correlation of the typical load patterns of the i-th user and the j-th user at the K non-DR day aggregation level, that is, The edge between N types of users i,j The data constitutes an N-order symmetric matrix E, because It is concluded that E=H, and the square matrix H is the spatial correlation matrix.
[0020] Based on the first aspect, in some embodiments, estimating the baseline load of the user cluster to be estimated based on the spatial correlation undirected graph includes: calculating the Laplacian matrix L of the spatial correlation undirected graph G G , extract the time information from the historical load data of the user cluster and define is the time label matrix corresponding to the historical load data; select some node features in the spatial correlation undirected graph, and calculate the cluster load data of all types of users in the K non-DR days and non-DR periods before the DR date. and the corresponding time label T LBL As feature X train ,as follows:
[0021]
[0022] Select some node features in the spatial correlation undirected graph, and use the cluster load data of all types of users in the non-DR period on the DR day d as the label Y train ,as follows:
[0023]
[0024] Label Y train With feature X train Put it into the graph convolutional neural network for training, as follows:
[0025] Y train =Ug(Λ)U T X train
[0026] in is the Laplace matrix L G The eigenvalue matrix, U T is the characteristic function matrix corresponding to the eigenvalue, g() is the functional relationship between the feature and the label, and the cluster baseline load estimation model that fully considers the change information of the user load pattern is trained; the cluster load data of all types of users in the DR period of K non-DR days before the DR day and the corresponding time label features are selected and put into the trained cluster baseline load estimation model to obtain the estimation result of the user cluster baseline load.
[0027] In a second aspect, an embodiment of the present invention provides a user cluster baseline load estimation device, comprising: an information acquisition module, used to obtain historical load data of a user cluster to be estimated; a clustering module, used to cluster the user cluster to be estimated according to the historical load data, and obtain multiple categories of users and load data of the multiple categories of users; an undirected graph construction module, used to calculate the spatial correlation between the multiple categories of users according to the load data of the multiple categories of users, and construct a spatial correlation matrix according to the spatial correlation; the multiple categories of users are used as nodes of a spatial correlation undirected graph, and the load data corresponding to the multiple categories of users are node features; according to the spatial correlation matrix, the edges of the spatial correlation undirected graph are constructed; a load estimation module, used to estimate the baseline load of the user cluster to be estimated based on the spatial correlation undirected graph.
[0028] In a third aspect, an embodiment of the present invention provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the user cluster baseline load estimation method as described in any one of the above-mentioned first aspects are implemented.
[0029] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the user cluster baseline load estimation method as described in any one of the above-mentioned first aspects are implemented.
[0030] In an embodiment of the present invention, historical load data of a user cluster to be estimated is obtained; the user cluster to be estimated is clustered according to the historical load data to obtain multiple categories of users and load data of the multiple categories of users; the spatial correlation between the multiple categories of users is calculated according to the load data of the multiple categories of users, and a spatial correlation matrix is constructed according to the spatial correlation; the multiple categories of users are used as nodes of a spatial correlation undirected graph, and the load data corresponding to the multiple categories of users are node features; the edges of the spatial correlation undirected graph are constructed according to the spatial correlation matrix; the baseline load of the user cluster to be estimated is estimated based on the spatial correlation undirected graph, which can greatly improve the accuracy of cluster baseline load estimation. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0032] Figure 1 It is a flowchart of a method for estimating a user cluster baseline load provided by an embodiment of the present invention;
[0033] Figure 2 It is a schematic diagram of the structure of a user cluster baseline load estimation device provided by an embodiment of the present invention;
[0034] Figure 3 It is a schematic diagram of a terminal device provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0035] The present invention will be described more clearly below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the effects of the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several variations and improvements may be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.
[0036] In order to make the purpose, technical solutions and advantages of the present invention more clear, specific embodiments will be described below in conjunction with the accompanying drawings.
[0037] The embodiment of the present invention provides a method for estimating a user cluster baseline load. Figure 1 As shown, the user cluster baseline load estimation method may include steps 101 to 105 .
[0038] Step 101: Obtain historical load data of the user cluster to be estimated.
[0039] The estimation of the aggregated baseline load (ABL) on the demand response (DR) event day is based on the historical load data of the user cluster to be estimated before the DR event day.
[0040] Based on the load data of K non-DR days before the DR event day d. First, the typical load pattern of all users on these K non-DR days should be calculated, where: represents the mth user, M target users representing the cluster baseline load estimate; Indicates that there are T moments in a day. On this basis, use represents the load data of user m on the kth non-DR day before DR day d.
[0041] Step 102: Cluster the user clusters to be estimated according to the historical load data to obtain multiple types of users and load data of multiple types of users.
[0042] The calculation formula for the typical load pattern of user m on the above K non-DR days is:
[0043]
[0044] According to the above formula, the typical load patterns of all M users on the above K non-DR days are calculated, and based on this, the users are clustered into N categories. The selection of the number of clusters N is determined based on the two clustering performance evaluation indicators, the ratio of the within cluster sum of squares to the between cluster variation (Ratio of within Cluster Sum of Squares to Between Cluster Variation, WCBCR) and the Davies-Bouldin index (DBI). Different cluster numbers are selected, and the clustering performance DBI and WCBCR under each cluster number are calculated. The requirements of DBI and WCBCR and the complexity of clustering are comprehensively considered to determine the optimal number of clusters, and clustering is performed using the K-means clustering algorithm.
[0045] Indicates the N types of users obtained, Represents the nth type of users, and determines the clustering center of each type of users according to the objective function minf:
[0046]
[0047] In the formula is the cluster center of the nth type of user.
[0048] Step 103: Calculate the spatial correlations between the multiple types of users according to the load data of the multiple types of users, and construct a spatial correlation matrix according to the spatial correlations.
[0049] represents the aggregated load data of the nth user on the kth non-DR day before the DR day d, l i,d-k represents the aggregated load data of the i-th user on the k-th non-DR day before the DR day d, l j,d-k Same reason.
[0050] The spatial correlation of the typical load patterns of the i-th user and the j-th user at the above K non-DR day aggregation level is calculated as follows:
[0051]
[0052] in is the spatial correlation between the i-th user and the j-th user, and its essence is the difference between the i-th user and the j-th user.
[0053] The spatial correlation indicators between N types of users form an N-order square matrix It is called the spatial correlation matrix, and the elements of the H matrix are defined as follows:
[0054]
[0055] Step 104: constructing a spatial correlation undirected graph according to the load data corresponding to the multiple types of users and the spatial correlation matrix between the multiple types of users;
[0056] The spatial association undirected graph contains nodes and edges. First, the nodes of the node undirected graph are constructed. n,d =[l n,d-K ,…,l n,d-1 ] T represents the load data of the n-th user at the aggregate level of the above K non-DR days. In the construction of the undirected graph, the n-th user is the node V on the undirected graph. n , the load data of the nth user at the aggregate level on the K non-DR event days and the load data at the aggregate level on the DR event day F n =[L n,d ,l n,d ] as node V n The node characteristics of the undirected graph are: The node feature set is
[0057] Then construct the edges of the spatial correlation undirected graph and define the connection nodes V i and V j The side is e i,j , e i,j The length of the edge of is numerically equal to the difference between the typical load patterns of the i-th user and the j-th user at the above K non-DR day aggregation levels, that is, On this basis, the edge data between N types of users can form an N-order symmetric matrix because It turns out that E=H.
[0058] Step 105: Estimate the baseline load of the user cluster to be estimated based on the spatial correlation undirected graph.
[0059] After the nodes and edges of the undirected graph are constructed, an undirected graph G containing N types of user load information and the spatial correlation between them can be obtained. For the above undirected graph G, the adjacency matrix describing its topological structure is first obtained: As follows:
[0060] A G =H (5)
[0061] From the adjacency matrix A G The corresponding degree matrix can be obtained and the Laplacian matrix Degree matrix D G It is an N-order diagonal square matrix, and the values on the diagonal are the number of edges connecting the corresponding nodes. Gis the difference between the degree matrix and the adjacency matrix, as follows:
[0062] L G =D G -A G (6)
[0063] Extract the time information from the historical load data of the user cluster and define It is the time label matrix corresponding to the historical load data.
[0064] Select some node features in the figure, that is, the cluster load data and corresponding time labels of all types of users in the K non-DR days and non-DR periods before the DR day as feature X train ,as follows:
[0065]
[0066] Select some node features in the figure, that is, the cluster load data of all types of users in the non-DR period on DR day d as the label Y train ,as follows:
[0067]
[0068] Label Y train With feature X train Put it into the graph convolutional neural network for training, as follows:
[0069] Y train =Ug(Λ)U T X train (9)
[0070] in is the Laplace matrix L G The eigenvalue matrix, U T is the characteristic function matrix corresponding to the eigenvalue, g() is the functional relationship between the feature and the label, and the cluster baseline load estimation model that fully considers the change information of user load pattern is trained.
[0071] The cluster load data and corresponding time labels of all types of users in the DR period of K non-DR days before the DR day are selected as new features, and input into the trained cluster baseline load estimation model. According to the functional relationship g() between the features and labels in the model, the estimation result of the cluster baseline load can be obtained.
[0072] When applying the above-mentioned user cluster load estimation method, the undirected graph needs to be updated synchronously, that is, when estimating the cluster baseline load on each DR event day, an undirected graph needs to be rebuilt for training.
[0073] The present invention utilizes a novel graph convolutional neural network to estimate cluster baseline load, which greatly improves the estimation accuracy of cluster baseline load, ensures fairness in demand response settlement, and is conducive to the implementation and promotion of demand response projects.
[0074] Example 1, taking a cluster of thousands of residential users in London as the research object, the load data of these residents were obtained through the official website of the London government. The data set contains the electricity consumption data of 5567 London households from November 2011 to February 2014, with a time resolution of 30 minutes. Among them, 4411 users did not participate in the demand response plan, and their electricity prices remained unchanged. After removing users with serious load data loss, there were 3588 users who did not participate in demand response and had relatively complete data. The present invention selected 2000 users from them, and based on their electricity consumption data, conducted an ABL estimation experiment based on a graph convolutional neural network, verifying the estimation effect of the cluster baseline load estimation method provided by the present invention.
[0075] The cluster baseline load estimation method based on graph convolutional neural network includes the following steps:
[0076] Step 1: Randomly select 2000 users from 3588 users, and use the load data of the 5 non-DR days before the DR event date as the basis, that is, K = 5. According to formula (1), calculate the typical load pattern of these 2000 users in the 5 non-DR days before the DR event date. Select different cluster numbers and calculate the clustering performance DBI and WCBCR under each cluster number. On this basis, comprehensively consider the requirements of DBI and WCBCR and the complexity of clustering to determine the optimal cluster number and perform K-means clustering.
[0077] Step 2: Calculate the cluster load data of each cluster user and find the typical cluster load pattern of each cluster user in the five non-DR days before the DR day. According to formula (3), calculate the differences between different types of cluster users and form a matrix, which is the spatial correlation matrix of users, which can describe the spatial distribution of different types of users.
[0078] Step 3: Build an undirected graph. In this graph, each user cluster is regarded as a point, and the difference in historical typical load patterns between different types of users is regarded as the length of the edge connecting the corresponding nodes. At the same time, the load data of the cluster users corresponding to the node on the DR day and 5 days before the DR day are selected as the node features of the node. This undirected graph contains both the load information of all cluster users and the spatial correlation of these cluster users.
[0079] Step 4: Based on step 3, the model is trained. During the training process, the node features in the historical data, that is, the load data of all class users in the non-DR period before the DR day and the time labels of these data are selected as the features in the training process. Accordingly, the load data of all class users in the non-DR period on the DR day are selected as the labels in the training process. They are input into the graph convolutional neural network model for training.
[0080] Step 5: After the model training is completed, the load data of all types of users in the DR period of the five non-DR days before the DR day are input into the above-trained model as features in the training process to obtain the cluster baseline load estimation result.
[0081] Step 6: The above are the steps for estimating the cluster baseline load for a certain DR day. For subsequent DR days, repeat the above 5 steps to estimate them separately.
[0082] Step 7: The above are the steps for estimating the cluster baseline load for a specific 2,000 users on multiple DR days. For other cluster users, repeat the above 6 steps to estimate.
[0083] Step 8: The present invention selects 100 groups of users, each with a user size of 2,000 households. The present invention first estimates the cluster baseline load of these 100 groups of users, and then selects three indicators, namely, mean absolute error MAE, mean absolute percentage error MAPE, and bias rate Bias, to evaluate the estimation performance, and calculates the average estimation performance of the 100 groups of users. The results are as follows: the average MAE is 19.3091, the average MAPE is 2.5243, and the average Bias is -1.8071. Compared with traditional methods, the present invention has significantly improved accuracy.
[0084] Step 9: On November 25, 2013, the load data at 8 time points from 17:00 to 20:00 estimated by the present invention are: 664.852kW, 754.328kW, 860.593kW, 934.964kW, 978.65kW, 1022.9kW, 1035.07kW, 1027.9kW. Correspondingly, the actual loads at these 8 time points are: 688.984kW, 785.694kW, 860.217kW, 937.158kW, 1007.19, 1021.11kW, 1032.91kW. It can be seen that the estimated results of the present invention are close to the actual results and have high accuracy.
[0085] See also Figure 2The embodiment of the present invention provides a user cluster baseline load estimation device 30, including: an information acquisition module 310, a clustering module 320, an undirected graph construction module 330 and a load estimation module 340.
[0086] The information acquisition module 310 is used to acquire historical load data of the user cluster to be estimated.
[0087] The clustering module 320 is used to cluster the user clusters to be estimated according to the historical load data to obtain multiple types of users and load data of the multiple types of users.
[0088] The undirected graph construction module 330 is used to calculate the spatial correlation between multiple types of users based on the load data of multiple types of users, and to construct a spatial correlation matrix based on the spatial correlation; the multiple types of users are used as nodes of the spatial correlation undirected graph, and the load data corresponding to the multiple types of users are used as node features; and the edges of the spatial correlation undirected graph are constructed based on the spatial correlation matrix.
[0089] The load estimation module 340 is used to estimate the baseline load of the user cluster to be estimated based on the spatial correlation undirected graph.
[0090] Figure 3 Schematic diagram of a terminal device provided by an embodiment of the present invention. Figure 3 As shown, the terminal device 4 of this embodiment includes: a processor 40, a memory 41, and a computer program 42 stored in the memory 41 and executable on the processor 40, such as a user cluster baseline load estimation program. When the processor 40 executes the computer program 42, the steps in the above-mentioned user cluster baseline load estimation method embodiment are implemented, such as Figure 1 Alternatively, when the processor 40 executes the computer program 42, the functions of each module / unit in the above-mentioned device embodiments are realized, for example, Figure 2 The functions of modules 310 to 340 are shown.
[0091] Exemplarily, the computer program 42 may be divided into one or more modules / units, which are stored in the memory 41 and executed by the processor 40 to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program 42 in the terminal device 4. For example, the computer program 42 may be divided into an information acquisition module, a clustering module, an undirected graph construction module, and a load estimation module.
[0092] The terminal device 4 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The terminal device may include, but is not limited to, a processor 40 and a memory 41. Those skilled in the art will appreciate that Figure 3 It is only an example of the terminal device 4 and does not constitute a limitation on the terminal device 4. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal device may also include input and output devices, network access devices, buses, etc.
[0093] The processor 40 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.
[0094] The memory 41 may be an internal storage unit of the terminal device 4, such as a hard disk or memory of the terminal device 4. The memory 41 may also be an external storage device of the terminal device 4, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device 4. Further, the memory 41 may also include both an internal storage unit and an external storage device of the terminal device 4. The memory 41 is used to store the computer program and other programs and data required by the terminal device. The memory 41 may also be used to temporarily store data that has been output or is to be output.
[0095] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by 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 embodiment can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.
[0096] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0097] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.
[0098] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0099] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0100] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0101] If the integrated module / unit is implemented in the form of 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, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased 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 electric carrier signals and telecommunication signals.
[0102] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.
Claims
1. A user cluster baseline load estimation method, characterized in that: include: Obtain historical load data of the user cluster to be estimated; Clustering the user clusters to be estimated according to the historical load data to obtain multiple categories of users and load data of the multiple categories of users; Calculating the spatial correlations between the multiple types of users according to the load data of the multiple types of users, and constructing a spatial correlation matrix according to the spatial correlations; The multiple types of users are used as nodes of a spatially associated undirected graph, and the load data corresponding to the multiple types of users are node features; Constructing edges of the spatial correlation undirected graph according to the spatial correlation matrix; estimating a baseline load of the user cluster to be estimated based on the spatial correlation undirected graph; The step of calculating the spatial correlations between the multiple types of users according to the load data of the multiple types of users, and forming a spatial correlation matrix according to the spatial correlations includes: Calculate the spatial correlation of the typical load patterns of the i-th user and the j-th user at the aggregation level of K non-DR days: in, is the spatial correlation between the i-th user and the j-th user, l i,d-k represents the aggregated load data of the i-th user on the k-th non-DR day before the DR day d, l j,d-k Similarly; The spatial correlation indicators between N types of users are used to form an N-order square matrix H, where the square matrix H is a spatial correlation matrix. The elements in the square matrix H are The estimating the baseline load of the to-be-estimated user cluster based on the spatial correlation undirected graph includes: Calculate the Laplacian matrix L of the spatial correlation undirected graph G G , extract the time information from the historical load data of the user cluster and define is the time label matrix corresponding to the historical load data; Select some node features in the spatial correlation undirected graph, and calculate the cluster load data of all types of users in the K non-DR days and non-DR periods before the DR day. and the corresponding time label T LBL As feature X train ,as follows: Select some node features in the spatial correlation undirected graph, and use the cluster load data of all types of users in the non-DR period on the DR day d as the label Y train ,as follows: Label Y train With feature X train Put it into the graph convolutional neural network for training, as follows: Y train =Ug(Λ)U T X train in is the Laplace matrix L G The eigenvalue matrix of is the characteristic function matrix corresponding to the eigenvalue, g() is the functional relationship between the feature and the label, and the cluster baseline load estimation model that fully considers the change information of the user load pattern is trained; The cluster load data of all types of users in the DR period of K non-DR days before the DR day and the corresponding time label features are selected and put into the trained cluster baseline load estimation model to obtain the estimation result of the user cluster baseline load.
2. The user cluster baseline load estimation method according to claim 1, characterized in that: The step of obtaining historical load data of the user cluster to be estimated includes: Obtain the historical load data of the user cluster to be estimated for K non-DR days before the DR day d. represents the load data of user m on the kth non-DR day before DR day d; in, represents the mth user, M target users representing the cluster baseline load estimate; represents T moments in a day, d represents a DR day, and k represents the kth non-DR day before the DR day.
3. The user cluster baseline load estimation method according to claim 1, characterized in that: The clustering of the user cluster to be estimated according to the historical load data to obtain multiple types of users and load data of the multiple types of users includes: According to the historical load data, a typical load pattern of user m on K non-DR days is calculated; Typical load mode The calculation formula is: Calculate the typical load patterns of M users on the K non-DR days, determine the number of clusters according to the evaluation index of clustering performance, and cluster the M users into N clusters; Determine the clustering centers of various types of users based on the objective function minf: In the formula is the cluster center of the nth type of user, Indicates the N types of users obtained, Represents the nth type of user.
4. The user cluster baseline load estimation method according to claim 1, characterized in that: The method of using the multiple types of users as nodes of a spatially correlated undirected graph and using the load data corresponding to the multiple types of users as node features includes: The nth user is regarded as a node V on the spatial correlation undirected graph n , represents the load data of the nth user at the aggregate level on K non-DR event days; The load data of the nth user at the aggregate level on the K non-DR event days and the load data of the aggregate level on the DR event day are calculated as follows: n =[L n,d ,l n,d ] as node V n Node characteristics; Among them, the node set on the undirected graph is The node feature set is 5. The user cluster baseline load estimation method according to claim 1, characterized in that: The step of constructing the edges of the spatial correlation undirected graph according to the spatial correlation matrix includes: Set the node V i and V j The spatial correlation between them is the edge of the spatial correlation undirected graph, and the edge is defined as e i,j , e i,j The length of is numerically equal to the value of the spatial correlation of the typical load patterns of the i-th user and the j-th user at the aggregation level of K non-DR days, that is, The edge between N types of users i,j The data constitutes an N-order symmetric matrix E, because It is concluded that E=H, and the square matrix H is the spatial correlation matrix.
6. A user cluster baseline load estimation device, characterized in that: include: An information acquisition module, used to acquire historical load data of the user cluster to be estimated; A clustering module, used for clustering the user clusters to be estimated according to the historical load data to obtain multiple types of users and load data of the multiple types of users; An undirected graph construction module is used to calculate the spatial correlation between the multiple types of users according to the load data of the multiple types of users, and to construct a spatial correlation matrix according to the spatial correlation; the multiple types of users are used as nodes of the spatial correlation undirected graph, and the load data corresponding to the multiple types of users are node features; Constructing edges of the spatial correlation undirected graph according to the spatial correlation matrix; A load estimation module, used for estimating the baseline load of the user cluster to be estimated based on the spatial correlation undirected graph; In the undirected graph construction module, the spatial correlations between the multiple types of users are calculated according to the load data of the multiple types of users, and a spatial correlation matrix is constructed according to the spatial correlations, including: Calculate the spatial correlation of the typical load patterns of the i-th user and the j-th user at the aggregation level of K non-DR days: in, is the spatial correlation between the i-th user and the j-th user, l i,d-k represents the aggregated load data of the i-th user on the k-th non-DR day before the DR day d, l j,d-k Similarly; The spatial correlation indicators between N types of users are used to form an N-order square matrix H, where the square matrix H is a spatial correlation matrix. The elements in the square matrix H are The load estimation module is specifically used for: Calculate the Laplacian matrix L of the spatial correlation undirected graph G G , extract the time information from the historical load data of the user cluster and define is the time label matrix corresponding to the historical load data; Select some node features in the spatial correlation undirected graph, and calculate the cluster load data of all types of users in the K non-DR days and non-DR periods before the DR day. and the corresponding time label T LBL As feature X train ,as follows: Select some node features in the spatial correlation undirected graph, and use the cluster load data of all types of users in the non-DR period on the DR day d as the label Y train ,as follows: Label Y train With feature X train Put it into the graph convolutional neural network for training, as follows: Y train =Ug(Λ)U T X train in is the Laplace matrix L G The eigenvalue matrix of is the characteristic function matrix corresponding to the eigenvalue, g() is the functional relationship between the feature and the label, and the cluster baseline load estimation model that fully considers the change information of the user load pattern is trained; The cluster load data of all types of users in the DR period of K non-DR days before the DR day and the corresponding time label features are selected and put into the trained cluster baseline load estimation model to obtain the estimation result of the user cluster baseline load.
7. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for estimating a baseline load of a user cluster as claimed in any one of claims 1 to 5 are implemented.
8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for estimating a baseline load of a user cluster as claimed in any one of claims 1 to 5 are implemented.