A residential user cluster load forecasting method, system, device and medium

Abstract

The application discloses a kind of resident user cluster load prediction method, system, equipment and medium, the method includes the following steps: based on the resident user grouping result of the resident user cluster of pre-acquired, current time space graph data for obtaining facing resident user cluster load prediction is constructed;Based on the current time space graph data, using the self-adapting space-time synchronous graph convolutional neural network model that is pre-trained is predicted, obtains the next time load prediction value of each resident user grouping in the resident user cluster;The next time load prediction value of each resident user grouping is aggregated, and the total load prediction value of the resident user cluster is obtained.The application specifically proposes a kind of resident user cluster short-term load prediction method based on graph neural network and clustering, can significantly improve resident user cluster load prediction precision.

Description

Technical Field

[0001] This invention belongs to the field of power system load forecasting technology, and specifically relates to a method, system, equipment and medium for forecasting clustered loads of residential users. Background Technology

[0002] With the widespread deployment and use of smart meters and other measuring devices on the user side, massive amounts of heterogeneous residential user data from multiple sources have been collected and stored, providing a solid data foundation for user-level load forecasting. Accurate residential user cluster load forecasting is a crucial foundation for promoting demand-side management of smart distribution networks and assisting power grid companies in achieving peak shaving and valley filling. However, due to the influence of multiple complex factors such as user electricity consumption patterns, weather conditions, market electricity prices, and policies, user electricity load exhibits strong randomness and high uncertainty. Therefore, residential user cluster load forecasting still faces significant challenges and difficulties.

[0003] Currently, existing methods for predicting residential user cluster loads mainly include random forests, support vector machines, artificial neural networks, recurrent neural networks, and long short-term memory neural networks. However, these existing methods focus on mining the temporal correlation of user electricity load sequences, while ignoring the potential spatial correlations between user electricity consumption behaviors (specifically, within the same region, residential users share the same geographical space, meteorological conditions, holiday information, electricity pricing policies, and other comprehensive factors). Therefore, a new method for predicting residential user cluster loads is urgently needed to further improve prediction accuracy. Summary of the Invention

[0004] The purpose of this invention is to provide a method, system, device, and medium for predicting residential user cluster loads, in order to solve one or more of the aforementioned technical problems. Specifically, this invention proposes a short-term load prediction method for residential user clusters based on graph neural networks and clustering, which can significantly improve the accuracy of residential user cluster load prediction.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] The first aspect of this invention provides a method for predicting clustered loads of residential users, comprising the following steps:

[0007] Based on the pre-acquired resident user grouping results of the resident user cluster, a spatiotemporal map data for the current time moment is constructed to predict the load of the resident user cluster.

[0008] Based on the spatiotemporal graph data at the current moment, a pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model is used to predict the load forecast value of each resident user group in the resident user cluster at the next moment.

[0009] The next-time load forecast values ​​of each residential user group are aggregated to obtain the total load forecast value of the residential user cluster.

[0010] A further improvement of the method of the present invention is that the step of obtaining the residential user grouping results of the pre-acquired residential user cluster includes: based on the similarity of users' electricity consumption behavior, grouping the historical load sequences of all residential users in the residential user cluster by clustering to obtain the residential user grouping results.

[0011] A further improvement to the method of the present invention is that the step of constructing the spatiotemporal map data for obtaining the current time-space graph for predicting the cluster load of residential users includes:

[0012] The aggregated load sequence of each residential user group is used as the feature sequence of each node in the graph structure data, and the association between nodes is represented by an adjacency matrix. Based on the aggregated load sequence of each residential user group and the adjacency matrix, the spatiotemporal graph data for the current time of residential user cluster load prediction is constructed.

[0013] The adjacency matrix construction methods include:

[0014] If the correlation coefficient between the corresponding historical load sequences of two nodes is greater than or equal to a certain preset threshold, then it is considered that there is a connection between the two nodes, and the corresponding element of the adjacency matrix is ​​set to 1; otherwise, it is considered that there is no connection between the two nodes, and the corresponding element of the adjacency matrix is ​​set to 0.

[0015] Alternatively, the correlation coefficient between the corresponding historical load sequences of the two nodes can be used as the corresponding element of the adjacency matrix;

[0016] The expression for calculating the correlation coefficient between the aggregated load sequences of each residential user group is as follows:

[0017]

[0018] In the formula, ρ ij Represents the correlation coefficient. All represent aggregated load sequences X i and X j Pearson correlation coefficient, COV(X) i ,X j ) represents X i and X j covariance, and They represent X respectively i and X j The standard deviation.

[0019] A further improvement to the method of the present invention is that the step of obtaining the pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model includes:

[0020] Constructing graph data samples for cluster load forecasting of residential users in, Input to the model, For model output, Let A be the graph signal matrix composed of node features at time t, where A is the adjacency matrix and H is the length of the history sequence.

[0021] An adaptive spatiotemporal synchronous graph convolutional neural network model is trained using the constructed graph data samples. After reaching the preset convergence condition, the pre-trained adaptive spatiotemporal synchronous graph convolutional neural network model is obtained.

[0022] A second aspect of the present invention provides a residential user cluster load forecasting system, comprising:

[0023] The current time spatiotemporal map data acquisition module is used to construct and obtain current time spatiotemporal map data for load prediction of the residential user cluster based on the pre-acquired residential user cluster resident user grouping results;

[0024] The prediction module is used to make predictions based on the current spatiotemporal graph data and a pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model to obtain the next time load prediction value for each resident user group in the resident user cluster.

[0025] The aggregation module is used to aggregate the next-time load forecast values ​​of each residential user group to obtain the total load forecast value of the residential user cluster.

[0026] A further improvement of the system of the present invention is that, in the current time spatiotemporal map data acquisition module, the step of acquiring the pre-acquired residential user grouping results of the residential user cluster includes: based on the similarity of users' electricity consumption behavior, grouping the historical load sequences of all residential users in the residential user cluster by clustering to obtain the residential user grouping results.

[0027] A further improvement of the system of the present invention is that, in the current time spatiotemporal map data acquisition module, the step of constructing and obtaining current time spatiotemporal map data for residential user cluster load prediction includes:

[0028] The aggregated load sequence of each residential user group is used as the feature sequence of each node in the graph structure data, and the association between nodes is represented by an adjacency matrix. Based on the aggregated load sequence of each residential user group and the adjacency matrix, the spatiotemporal graph data for the current time of residential user cluster load prediction is constructed.

[0029] The adjacency matrix construction methods include:

[0030] If the correlation coefficient between the corresponding historical load sequences of two nodes is greater than or equal to a certain preset threshold, then it is considered that there is a connection between the two nodes, and the corresponding element of the adjacency matrix is ​​set to 1; otherwise, it is considered that there is no connection between the two nodes, and the corresponding element of the adjacency matrix is ​​set to 0.

[0031] Alternatively, the correlation coefficient between the corresponding historical load sequences of the two nodes can be used as the corresponding element of the adjacency matrix;

[0032] The expression for calculating the correlation coefficient between the aggregated load sequences of each residential user group is as follows:

[0033]

[0034] In the formula, ρ ij Represents the correlation coefficient. All represent aggregated load sequences X i and X j Pearson correlation coefficient, COV(X) i ,X j ) represents X i and X j covariance, and They represent X respectively i and X j The standard deviation.

[0035] A further improvement of the system of the present invention is that the step of obtaining the pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model in the prediction module includes:

[0036] Constructing graph data samples for cluster load forecasting of residential users in, Input to the model, For model output, Let A be the graph signal matrix composed of node features at time t, where A is the adjacency matrix and H is the length of the history sequence.

[0037] An adaptive spatiotemporal synchronous graph convolutional neural network model is trained using the constructed graph data samples. After reaching the preset convergence condition, the pre-trained adaptive spatiotemporal synchronous graph convolutional neural network model is obtained.

[0038] A third aspect of the present invention provides an electronic device comprising:

[0039] At least one processor; and,

[0040] A memory communicatively connected to the at least one processor; wherein,

[0041] The memory stores instructions that can be executed by the at least one processor, which, when executed, enable the at least one processor to perform any of the above-described residential user cluster load forecasting methods of the present invention.

[0042] The fourth aspect of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the residential user cluster load prediction method described in any one of the present invention.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] To address the technical shortcomings of existing traditional short-term load forecasting methods for residential user clusters that ignore the potential spatial correlations between the electricity consumption behaviors of adjacent users, this invention proposes a short-term load forecasting method for residential user clusters based on graph neural networks and clustering, which can significantly improve the accuracy of load forecasting for residential user clusters. More specifically, this invention utilizes graph neural networks to not only mine the temporal correlation of historical load sequences of residential users but also to mine the spatial correlation between users' electricity consumption behaviors, thereby significantly improving the accuracy of short-term load forecasting for residential user clusters. This invention is applicable to short-term load forecasting of residential user clusters with strong spatiotemporal correlations and has significant practical application value and promising prospects for promoting demand-side management of smart distribution networks and assisting power grid companies in achieving peak shaving and valley filling. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art are briefly introduced below; obviously, the drawings described below are some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart illustrating a method for predicting clustered loads of residential users provided in an embodiment of the present invention.

[0047] Figure 2 This is a schematic diagram of the adjacency matrix and spatiotemporal graph structure in an embodiment of the present invention;

[0048] Figure 3 This is a schematic diagram of the adaptive spatiotemporal synchronization graph convolutional neural network structure in an embodiment of the present invention;

[0049] Figure 4 This is a schematic diagram of a short-term load forecasting method for residential user clusters based on graph neural networks and clustering, provided in a specific embodiment of the present invention.

[0050] Figure 5 This is a schematic block diagram of a residential user cluster load prediction system provided in an embodiment of the present invention. Detailed Implementation

[0051] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0052] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0053] The present invention will now be described in further detail with reference to the accompanying drawings:

[0054] Please see Figure 1 The present invention provides a method for predicting clustered loads of residential users, comprising the following steps:

[0055] Step 1: Based on the pre-acquired resident user grouping results of the resident user cluster, construct the current time spatiotemporal map data for load prediction of the resident user cluster;

[0056] Step 2: Based on the spatiotemporal graph data at the current time, use a pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model to make predictions and obtain the next time load prediction values ​​for each resident user group in the resident user cluster.

[0057] Step 3: Aggregate the next-time load forecast values ​​of each resident user group in the resident user cluster to obtain the total load forecast value of the resident user cluster.

[0058] In this embodiment of the invention, the step of obtaining the pre-acquired residential user grouping results of the residential user cluster may be as follows: based on the similarity of users' electricity consumption behavior, the historical load sequences of all residential users in the residential user cluster are grouped by clustering to obtain the residential user grouping results.

[0059] In this embodiment of the invention, based on the similarity of users' electricity consumption behavior, the K-means clustering method is used to group the historical load sequences of all residential users. Specifically:

[0060] (1) Construct a two-dimensional matrix X from the historical load sequences of all residential users. N×T Where N is the number of residential users and T is the length of the historical load sequence;

[0061] (2) Set the value of parameter K for the K-means clustering method, and choose Euclidean distance as the similarity measure for K-means clustering. The Euclidean distance between two points x and y in n-dimensional space is calculated as follows:

[0062]

[0063] In the formula: n is the spatial dimension; x k and y k These are the k-th attribute feature values ​​of x and y, respectively;

[0064] (3) K resident user groups S are obtained by K-means clustering. i , i = 1, 2, ..., K;

[0065] (4) Sum the historical load sequences of all residents in each resident user group to obtain the resident user group S. i The aggregate load sequence X i , i = 1, 2, ..., K.

[0066] In this embodiment of the invention, the specific steps for constructing spatiotemporal map data for load prediction of residential user clusters based on the pre-acquired resident user grouping results are as follows:

[0067] The spatiotemporal graph data includes node features and the relationships between nodes;

[0068] The aggregated load sequence of each residential user group is used as the feature sequence of each node in the graph structure data;

[0069] The relationships between nodes are represented by an adjacency matrix. The first method for constructing the adjacency matrix is ​​as follows: if the correlation coefficient between the corresponding historical load sequences of two nodes is not less than a certain threshold, then a connection is considered to exist between the two nodes, and the corresponding element in adjacency matrix A is set to 1; otherwise, no connection is considered to exist between the two nodes, and the corresponding element in adjacency matrix A is set to 0. The specific calculation formula is as follows: In the formula, A ij Let ξ be the element in the i-th row and j-th column of the adjacency matrix A, and ξ be the threshold. The second method for constructing the adjacency matrix is ​​as follows: The correlation coefficient between the corresponding historical load sequences of two nodes is used as the corresponding element of the adjacency matrix. The specific calculation formula is as follows: A ij =ρ ij The two construction methods described above are exemplified as follows: Figure 2 As shown.

[0070] The correlation coefficient between the aggregated load sequences of each residential user group is calculated as follows:

[0071]

[0072] In the formula, ρ ij Represents the correlation coefficient. All represent aggregated load sequences X i and X j Pearson correlation coefficient, COV(X) i ,X j ) represents X i and X j covariance, and They represent X respectively i and X j The standard deviation.

[0073] In summary, based on the aggregated load sequences and adjacency matrices of each residential user group, a spatiotemporal graph data for residential user cluster load forecasting is jointly constructed.

[0074] In this embodiment of the invention, the step of obtaining the pre-trained adaptive spatiotemporal synchronous graph convolutional neural network model includes: constructing graph data samples for residential user cluster load prediction. in, Input to the model, For model output, Let A be the graph signal matrix composed of node features at time t, A be the adjacency matrix, and H be the length of the historical sequence. Then, the ASTSGCN model is trained using the constructed graph data samples. An exemplary model structure is shown below. Figure 3 As shown.

[0075] In this embodiment of the invention, the trained ASTSGCN model is used to predict the future cluster load of residential users, resulting in... in, The load forecast value for residential user group k is given for the next future time step. Then, by aggregating the load forecast values ​​for each residential user group, the total load forecast value for the residential user cluster for the next future time step is obtained. The specific calculation formula is as follows: In the formula, P next This is the predicted total load value for the residential user cluster.

[0076] In summary, addressing the technical shortcomings of existing traditional short-term load forecasting methods for residential user clusters that ignore the potential spatial correlation between the electricity consumption behaviors of adjacent users, this invention specifically proposes a short-term load forecasting method for residential user clusters based on graph neural networks and clustering. By constructing spatiotemporal graph data, an Adaptive Spatial-Temporal Synchronous Graph Convolutional Network (ASTSGCN) is trained to mine the electricity consumption patterns of residential user clusters, predict the load value of each residential user group at the next moment, and then aggregate the load forecast values ​​of each residential user group to obtain the total load forecast value of the entire residential user cluster. This significantly improves the accuracy of load forecasting for residential user clusters. Specifically, the method provided by this invention first groups residential user clusters using K-means clustering based on the similarity of electricity consumption patterns; then, based on the Pearson correlation coefficient, a graph structure data for load forecasting of residential user clusters is established; finally, using graph data samples of residential user clusters, the Adaptive Spatial-Temporal Synchronous Graph Convolutional Network (ASTSGCN) model is used to achieve short-term load forecasting for residential user clusters. The technical solution provided by this invention takes into account the potential spatial correlation between the electricity consumption behavior of adjacent users, which can significantly improve the accuracy of load forecasting for residential user clusters. It is of great significance for refined demand-side management, assisting in peak shaving and valley filling, and supporting the consumption of new energy sources.

[0077] Please see Figure 4 In this embodiment of the invention, a method for predicting clustered loads of residential users uses a publicly available dataset from the Commission for Energy Regulation (CER) of Ireland. The complete dataset contains electricity load data from over 5,000 residential users from August 2009 to December 2010, with a data sampling frequency of once every 30 minutes. This embodiment selects 3,639 residential users with complete sampling data to form the embodiment dataset. The entire embodiment dataset is divided into a training set, a validation set, and a test set in a ratio of 0.7:0.2:0.1.

[0078] Step 1: Based on the similarity of user electricity consumption behavior, the historical load sequences of all residential users are grouped using the K-means clustering method. To overcome the limitation of manually selecting the K value in the K-means clustering method, this embodiment selects the number of clusters K one by one within the range of [3, 17], and performs the following operations for each K value, as follows:

[0079] (1) Construct a two-dimensional matrix with a dimension of 3639×25730 from the historical load sequences of all 3639 residential users;

[0080] (2) Euclidean distance is chosen as the similarity measure for the K-means clustering method;

[0081] (3) K resident user groups are obtained by K-means clustering method;

[0082] (4) Sum the historical load sequences of all residents in each resident user group to obtain the aggregated load sequence of each resident user group.

[0083] Step 2: Based on the clustering results, construct spatiotemporal map data for residential user cluster load prediction. Similarly, this embodiment of the invention performs the following operations for each K value, specifically as follows:

[0084] (1) Calculate the Pearson correlation coefficient between the aggregated load sequences of each resident user group under the cluster size K value;

[0085] (2) Construct the corresponding adjacency matrices using the first adjacency matrix construction method and the second adjacency matrix construction method respectively, where the threshold ξ is 0.9;

[0086] (3) Based on the aggregated load sequence and adjacency matrix of each residential user group, a spatiotemporal graph data for residential user cluster load prediction is jointly constructed.

[0087] Step 3: Using the constructed spatiotemporal graph data, train an Adaptive Spatiotemporal Synchronous Graph Convolutional Neural Network (ASTSGCN) to mine the electricity consumption patterns of residential user clusters, predict the load value of each residential user group at the next moment, and then aggregate the load prediction values ​​of each residential user group to obtain the total load prediction value of the entire residential user cluster. Similarly, this embodiment performs the following operations for each K value, as follows:

[0088] (1) Construct graph data samples for cluster load prediction of residential users, and then train the ASTSGCN model. The model training parameters are set as shown in Table 1.

[0089] Table 1. Parameters of the ASTSGCN Model

[0090]

[0091]

[0092] (2) Apply the trained ASTSGCN model to predict the future load of the residential user cluster, and obtain the load prediction values ​​of K residential user groups at the next time step. Then, by aggregating the load prediction values ​​of the K residential user groups, obtain the total load prediction value of the residential user cluster composed of 3639 residential users at the next time step.

[0093] (3) The mean absolute percentage error (MAPE), mean absolute error (MAE), and root mean squared error (RMSE) were used as prediction performance evaluation indicators to evaluate the prediction accuracy of the ASTSGCN model under different numbers of clusters and different adjacency matrices. The results of specific implementation examples are shown in Table 2.

[0094] Table 2. Comparison of prediction performance of the ASTSGCN model under different numbers of clusters and different adjacency matrices.

[0095]

[0096] (Note: 0-1 represents the first method of constructing the adjacency matrix, and ρ represents the second method of constructing the adjacency matrix.)

[0097] Comparison of the results from the examples shows that, regardless of the value of the number of clusters K, the second adjacency matrix construction method can generally improve the prediction accuracy of the ASTSGCN model compared to the first adjacency matrix construction method.

[0098] (4) To verify the superiority of the technical solution of this invention, this embodiment selects Long Short-Term Memory (LSTM), Random Forest (RF), Support Vector Regression (SVR), and Decision Tree (DT) as benchmark prediction methods, and compares their prediction accuracy with that of the ASTSGCN model. It should be noted that the input to all benchmark prediction methods is the historical load sequence aggregated from the residential user clusters, while the output is the total load of the residential user clusters at the next future time. Specific results are shown in Table 3.

[0099] Table 3. Comparison of results from different prediction methods

[0100] Prediction methods MAE(kW) MAPE (%) RMSE (kW) LSTM 58.4 2.8 79.6 RF 70.1 3.1 102.6 SVR 158.9 6.2 280.5 DT 69.7 3.1 99.6 ASSGCN 44.8 2.0 63.0

[0101] Table 3 shows the prediction performance evaluation results of the ASTSGCN model, referring to the results based on the second adjacency matrix construction method when the optimal cluster size K is 12. The comparative analysis in Table 3 shows that the ASTSGCN model achieved the best prediction accuracy in all three evaluation metrics: MAE, MAPE, and RMSE. In other words, compared to all benchmark prediction methods, the ASTSGCN model shows varying degrees of accuracy improvement in all three different prediction performance evaluation metrics.

[0102] The following are embodiments of the apparatus of the present invention, which can be used to execute embodiments of the method of the present invention. For details not omitted in the apparatus embodiments, please refer to the embodiments of the method of the present invention.

[0103] Please see Figure 5 In another embodiment of the present invention, a residential user cluster load prediction system is provided, comprising:

[0104] The current time spatiotemporal map data acquisition module is used to construct and obtain current time spatiotemporal map data for load prediction of the residential user cluster based on the pre-acquired residential user cluster resident user grouping results;

[0105] The prediction module is used to make predictions based on the current spatiotemporal graph data and a pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model to obtain the next time load prediction value for each resident user group in the resident user cluster.

[0106] The aggregation module is used to aggregate the next-time load forecast values ​​of each residential user group to obtain the total load forecast value of the residential user cluster.

[0107] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions from the computer storage medium to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used for the operation of a residential user cluster load forecasting method.

[0108] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the residential user cluster load forecasting method in the above embodiments.

[0109] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application 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.

[0110] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0111] 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.

[0112] 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.

[0113] 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 method for predicting clustered loads of residential users, characterized in that, Includes the following steps: Based on the pre-acquired resident user grouping results of the resident user cluster, a spatiotemporal map data for the current time moment is constructed to predict the load of the resident user cluster. Based on the spatiotemporal graph data at the current moment, a pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model is used to predict the load forecast value of each resident user group in the resident user cluster at the next moment. By aggregating the next-time load forecast values ​​of each residential user group, the total load forecast value of the residential user cluster is obtained; in, The steps for obtaining the pre-acquired residential user grouping results of the residential user cluster include: based on the similarity of users' electricity consumption behavior, grouping the historical load sequences of all residential users in the residential user cluster by clustering to obtain the residential user grouping results; The steps for constructing the current-time spatiotemporal graph data for residential user cluster load prediction include: using the aggregated load sequence of each residential user group as the feature sequence of each node in the graph structure data, and using an adjacency matrix to represent the association between nodes; and constructing the current-time spatiotemporal graph data for residential user cluster load prediction based on the aggregated load sequence of each residential user group and the adjacency matrix. The adjacency matrix construction methods include: If the correlation coefficient between the corresponding historical load sequences of two nodes is greater than or equal to a certain preset threshold, then it is considered that there is a connection between the two nodes, and the corresponding element of the adjacency matrix is ​​set to 1; otherwise, it is considered that there is no connection between the two nodes, and the corresponding element of the adjacency matrix is ​​set to 0. Alternatively, the correlation coefficient between the corresponding historical load sequences of the two nodes can be used as the corresponding element of the adjacency matrix; The expression for calculating the correlation coefficient between the aggregated load sequences of each residential user group is as follows: ； In the formula, Represents the correlation coefficient. All represent aggregated load sequences and Pearson correlation coefficient, express and covariance, and They represent and The standard deviation.

2. The method for predicting clustered loads of residential users according to claim 1, characterized in that, The steps for obtaining the pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model include: Constructing graph data samples for cluster load forecasting of residential users ;in, Input to the model, For model output, for t A is a graph signal matrix composed of node features at any given time, where A is the adjacency matrix. H The length of the historical sequence; An adaptive spatiotemporal synchronous graph convolutional neural network model is trained using the constructed graph data samples. After reaching the preset convergence condition, the pre-trained adaptive spatiotemporal synchronous graph convolutional neural network model is obtained.

3. A residential user cluster load forecasting system, characterized in that, include: The current time spatiotemporal map data acquisition module is used to construct and obtain current time spatiotemporal map data for load prediction of the residential user cluster based on the pre-acquired residential user cluster resident user grouping results; The prediction module is used to make predictions based on the current spatiotemporal graph data and a pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model to obtain the next time load prediction value for each resident user group in the resident user cluster. The aggregation module is used to aggregate the next-time load forecast values ​​of each residential user group to obtain the total load forecast value of the residential user cluster. in, In the current time-space map data acquisition module, the steps for acquiring the pre-acquired residential user grouping results of the residential user cluster include: based on the similarity of users' electricity consumption behavior, grouping the historical load sequences of all residential users in the residential user cluster by clustering to obtain the residential user grouping results; The steps in the current-moment spatiotemporal map data acquisition module to construct and obtain current-moment spatiotemporal map data for residential user cluster load prediction include: The aggregated load sequence of each residential user group is used as the feature sequence of each node in the graph structure data, and the association between nodes is represented by an adjacency matrix. Based on the aggregated load sequence of each residential user group and the adjacency matrix, the spatiotemporal graph data for the current time of residential user cluster load prediction is constructed. The adjacency matrix construction methods include: If the correlation coefficient between the corresponding historical load sequences of two nodes is greater than or equal to a certain preset threshold, then it is considered that there is a connection between the two nodes, and the corresponding element of the adjacency matrix is ​​set to 1; otherwise, it is considered that there is no connection between the two nodes, and the corresponding element of the adjacency matrix is ​​set to 0. Alternatively, the correlation coefficient between the corresponding historical load sequences of the two nodes can be used as the corresponding element of the adjacency matrix; The expression for calculating the correlation coefficient between the aggregated load sequences of each residential user group is as follows: ； In the formula, Represents the correlation coefficient. All represent aggregated load sequences and Pearson correlation coefficient, express and covariance, and They represent and The standard deviation.

4. A residential user cluster load forecasting system according to claim 3, characterized in that, In the prediction module, the steps for obtaining the pre-trained adaptive spatiotemporal synchronization graph convolutional neural network model include: Constructing graph data samples for cluster load forecasting of residential users ;in, Input to the model, For model output, for t A is a graph signal matrix composed of node features at any given time, where A is the adjacency matrix. H The length of the historical sequence; An adaptive spatiotemporal synchronous graph convolutional neural network model is trained using the constructed graph data samples. After reaching the preset convergence condition, the pre-trained adaptive spatiotemporal synchronous graph convolutional neural network model is obtained.

5. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the residential user cluster load forecasting method as described in claim 1 or 2.

6. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the residential user cluster load forecasting method as described in claim 1 or 2.

Images