Power load prediction method based on industrial user group dynamic correlation

Abstract

The invention belongs to the technical field of power system operation and control, and provides an industrial user group dynamic correlation-based power load prediction method, which comprises the steps of obtaining an optimal feature set of each industrial user load based on a maximum correlation and minimum redundancy criterion; obtaining a correlation coefficient matrix of each industrial user load based on the optimal feature set; according to the correlation coefficient matrix, utilizing a graph convolutional neural network to realize accurate division of industrial user groups; based on the industrial user group, capturing a sequential relationship of the load of the industrial user group; constructing a new graph structure, inputting the sequential relationship of the loads of the industrial user groups into the dynamic graph learning model, and performing adaptive learning to generate a dynamic graph adjacency matrix; and training a new graph structure data-based graph convolutional neural network by using the dynamic graph adjacency matrix, and predicting a load value of each industrial user group. According to the invention, dynamic monitoring and prediction of power consumption behaviors can be realized.

Description

Technical Field

[0001] The present disclosure belongs to the technical field of power system operation and control, and in particular relates to a power load forecasting method based on dynamic correlation of industrial user groups. Background Art

[0002] With the continuous advancement of new power system construction and power market reform, new roles and technologies such as virtual power plants and demand response are constantly emerging, energy consumption patterns are gradually diversified, industrial users' electricity consumption behavior is more complex, and their loads are showing more complex and changeable new characteristics and new forms. Traditional extensive system-level load forecasting methods are gradually unable to meet the management needs of the power system.

[0003] However, due to multiple factors, including industry type, meteorological conditions, and market electricity prices, industrial user-level loads are highly random and volatile. Accurate load forecasting for industrial user groups (collections of multiple users) remains difficult and presents significant challenges. Furthermore, existing methods for segmenting industrial user groups often employ traditional methods such as K-means clustering and hierarchical clustering. These methods are often sensitive to initial values ​​and outliers, exhibit high computational complexity, and exhibit poor model generalization. Furthermore, they fail to fully account for the varying impacts of different characteristics on the loads of different industrial users, reducing the accuracy and stability of industrial user group segmentation results, further impacting the accuracy of industrial user group load forecasting.

[0004] Furthermore, most existing load forecasting methods for industrial user groups focus solely on exploring the temporal correlations within a group's own load series, while ignoring the potential relationships between different industrial user groups. Few industrial user group load forecasting methods consider the potential correlations between the electricity consumption behaviors of different industrial user groups, but these methods typically only consider the overall static correlations between different industrial user groups. However, the correlations between the loads of different industrial user groups often change over time, limiting further improvements in the accuracy of industrial user group load forecasting. Summary of the Invention

[0005] In order to solve the above problems, the present invention provides an electric power load forecasting method based on the dynamic correlation of industrial user groups, which captures the electricity consumption correlation between industrial user groups in real time and further carries out refined load forecasting for industrial user groups. It is of great significance to reduce the difficulty of power grid planning and scheduling and ensure the safe and stable operation of the power system.

[0006] In a first aspect, an embodiment of the present disclosure provides a method for power load forecasting based on dynamic correlation of industrial user groups, comprising: Obtaining the optimal feature set for each industrial user load based on the maximum correlation minimum redundancy criterion, including: constructing a feature set based on original feature data, and calculating the maximum correlation index between other related features and the industrial user load using the feature set; selecting normalized data of two other features with the same sequence length in the feature set, discretely distributing data points in the normalized data of the two other features in pairs in a two-dimensional space, constructing a data set, and calculating the minimum redundancy index between other related features and the industrial user load based on the mutual information of the data set; obtaining a maximum correlation minimum redundancy (mRMR) index based on the maximum correlation index and the minimum redundancy index; the data set that maximizes the maximum correlation minimum redundancy (mRMR) index is the optimal feature set for the industrial user; Based on the optimal feature set, the correlation coefficient matrix of each industrial user load is obtained; Based on the correlation coefficient matrix, the graph convolutional neural network is used to accurately divide industrial user groups; Based on industrial user groups, capture the temporal relationship of industrial user group loads; Construct a new graph structure, input the temporal relationship of the load of each industrial user group into the dynamic graph learning model, and adaptively learn to generate the dynamic graph adjacency matrix; The dynamic graph adjacency matrix is ​​used to train a graph convolutional neural network based on the new graph structure data to predict the load values ​​of various industrial user groups.

[0007] Further, Normalize the original feature data to the interval [0, 1] according to the following formula; All normalized data are constructed into feature sets according to each industrial user and time series.

[0008] Further, Based on the optimal feature set, the correlation coefficient matrix of each industrial user load is obtained, including: Filter out the non-repeated features contained in the optimal feature set of all industrial user loads, form a new feature set, and determine the correlation coefficient vector; The new correlation coefficient vectors of all industrial user loads are concatenated to obtain the correlation coefficient matrix.

[0009] Further, Based on the correlation coefficient matrix, the graph convolutional neural network is used to accurately segment industrial user groups, including: Use the K-means algorithm to initially cluster industrial user loads, and combine it with the silhouette coefficient formula to determine the initial number of clusters, and obtain the initial cluster center and nearby samples; According to the initial cluster center and nearby samples, mark the correlation coefficient matrix; Construct graph data structures and graph adjacency matrices; The labeled correlation coefficient matrix is ​​input into the graph convolutional neural network model based on the graph structure to achieve accurate division of all industrial user groups.

[0010] Further, Based on industrial user groups, the temporal relationship of industrial user group loads is captured, including: At each data collection moment, the normalized data of the load of each industrial user in each industrial user group is added together to obtain the load of each industrial user group; Multi-layer dilated causal convolution is used to capture the temporal relationship of the load of each industrial user group.

[0011] Further, Construct a new graph structure, input the temporal relationship of the load of each industrial user group into the dynamic graph learning model, and adaptively learn to generate a dynamic graph adjacency matrix, including: The load of each industrial user group is used as a node in the new graph structure, and the temporal relationship of the load of each industrial user group is used as the relationship between the nodes in the new graph structure; In the dynamic graph learning layer, the correlation between the nodes of each new graph structure is measured by the Mahalanobis distance, and the covariance matrix in the Mahalanobis distance is set as a trainable weight parameter to ensure the feasibility of real-time updating of the dynamic graph adjacency matrix.

[0012] In a second aspect, the embodiments of the present disclosure further provide a power load forecasting system based on the dynamic correlation of industrial user groups, including an optimal feature set determination module, a correlation coefficient matrix construction module, an industrial user group division module, a time series relationship calculation module, a dynamic graph adjacency matrix generation module, and a load value prediction module; An optimal feature set determination module is used to obtain the optimal feature set for each industrial user load based on the maximum correlation minimum redundancy criterion, including: constructing a feature set based on original feature data, and using the feature set to calculate the maximum correlation index between other related features and the industrial user load; selecting normalized data of two other features with the same sequence length in the feature set, discretely distributing the data points in the normalized data of the two other features in pairs in a two-dimensional space to construct a data set, and calculating the minimum redundancy index between other related features and the industrial user load based on the mutual information of the data set; obtaining the maximum correlation minimum redundancy (mRMR) index based on the maximum correlation index and the minimum redundancy index; the data set that maximizes the maximum correlation minimum redundancy (mRMR) index is the optimal feature set for the industrial user; A correlation coefficient matrix construction module is used to obtain the correlation coefficient matrix of each industrial user load based on the optimal feature set; The industrial user group segmentation module is used to accurately segment industrial user groups using a graph convolutional neural network based on the correlation coefficient matrix; A time series relationship calculation module is used to capture the time series relationship of the load of industrial user groups based on industrial user groups; The dynamic graph adjacency matrix generation module is used to construct a new graph structure, input the temporal relationship of the load of each industrial user group into the dynamic graph learning model, and adaptively learn to generate the dynamic graph adjacency matrix; The load value prediction module is used to use the dynamic graph adjacency matrix to train a graph convolutional neural network based on new graph structure data to predict the load value of each industrial user group.

[0013] In a third aspect, based on the same inventive concept, an embodiment of the present disclosure further provides an electronic device, comprising at least one processor and at least one memory electrically connected; The memory is electrically connected to the processor, wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the power load forecasting method based on dynamic correlation of industrial user groups as described in any of the preceding items.

[0014] In a fourth aspect, based on the same inventive concept, an embodiment of the present disclosure further provides a computer storage medium having a computer program stored thereon; When the computer program is executed by a processor, the power load forecasting method based on dynamic correlation of industrial user groups as described in any of the preceding items is implemented.

[0015] In a fifth aspect, based on the same inventive concept, an embodiment of the present disclosure further provides a computer program product, wherein the computer program product is stored in at least one storage medium; The computer program product includes several instructions for causing at least one computer device to execute the power load forecasting method based on dynamic correlation of industrial user groups as described in any of the above items.

[0016] Compared with the existing technology, the present invention provides a power load forecasting method based on the dynamic correlation of industrial user groups, which has the following beneficial effects: The embodiments of the present disclosure can adapt to the continuous changes in prediction scenarios, fully consider the differences in the impact of different characteristics on the load of each industrial user and the topological structure information between different industrial users, and provide more real-time and flexible decision support for industrial controllable loads to participate in demand response. It has important practical significance for improving the ability of industrial users to participate in demand response and optimizing energy management.

[0017] Other features and advantages of the present disclosure will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present disclosure. The purposes and other advantages of the present disclosure can be realized and obtained by the structures indicated in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A schematic diagram of a power load forecasting method based on dynamic correlation of industrial user groups according to an embodiment of the present disclosure is shown; Figure 2 A schematic structural diagram of an electronic device according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure more clear, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present disclosure without making any creative efforts shall fall within the scope of protection of the present disclosure.

[0021] Attachment Figure 1 FIG. 1 shows a flow chart of a method for predicting power load based on dynamic correlation of industrial user groups according to an embodiment of the present disclosure. Figure 1 As shown, the power load forecasting method based on the dynamic correlation of industrial user groups according to an embodiment of the present disclosure includes: S1, based on the maximum correlation and minimum redundancy criterion, the optimal feature set of each industrial user load is obtained.

[0022] S11, collect original feature data.

[0023] The original feature data includes historical data on industrial user loads and historical data on other features related to industrial user loads. The historical data on other features related to industrial user loads include weather data, electricity prices, etc.

[0024] S12, normalize the original feature data and construct a feature set.

[0025] First, the original feature data is normalized to the interval [0, 1] according to the following formula: (1) Where, X o is the original feature data; X min and X max They are the original feature data and X o The minimum and maximum values ​​of characteristic data of the same type; X yes X o Normalized data, that is, 0≤ X ≤1.

[0026] Secondly, all normalized data are constructed into feature sets according to each industrial user and time series.

[0027] S13, using the feature set, calculating the maximum correlation index between other related features and the industrial user load.

[0028] For each industrial user, the Pearson correlation coefficient is used to represent the correlation between other related features and the industrial user load. The closer the absolute value is to 1, the greater the correlation is; the closer it is to 0, the smaller the correlation is. The formula for calculating the Pearson correlation coefficient is as follows: (2) Where, X i The first feature set i Normalized data of other features; Y is the normalized data of industrial user load in the feature set; f cov ( X i ,Y ) is the covariance of the normalized data of other features and the normalized data of industrial user load, σ ( X i )and σ ( Y ) are the normalized data of other features respectively X i Normalized data with industrial user load Y The standard deviation of r ( X i ,Y ) is the Pearson correlation coefficient.

[0029] We can further obtain the maximum correlation index D ( S,Y )for: (3) Where, S and N S are the optimal feature set of industrial user load and the number of other features it contains, respectively.

[0030] S14, determining a minimum redundancy index.

[0031] The Maximum Information Coefficient (MIC) is used to measure the redundancy between different features. It adds normalization to the mutual information and represents the amount of information contained in one random variable in another random variable. The calculation process is as follows: First, assume X i and X j They are the first i The first and j Normalized data of other features, and the sequence length is n .Will X i and X j The data points in the pair are discretely distributed in 2D space, which is recorded as data set D ( X i ,X j ). Grid the 2D space and make the x and y The directions are divided into A and B intervals.

[0032] Data Collection D ( X i ,X j ) in the grid G ( A,B ) is calculated as follows: (4) Where, p ( D a,b ) is the data set D ( X i ,X j ) are located in the region ( a,b ) probability; p( D a )and p ( D b ) are the data points located in the region a and b The probability of each region is proportional to the number of data points located in the region.

[0033] Next, take the data set D ( X i ,X j ) in the grid G ( A,B ) Mutual information under different partitioning methods M ( D,G ), and normalize it to get: (5) Thus we get X i and X j The MIC calculation formula is: (6) In the formula, the MIC value range is [0,1]. The closer the value is to 1, the greater the correlation between the two data; the closer it is to 0, the smaller the correlation.

[0034] The minimum redundancy index can be further obtained as: (7).

[0035] S15, obtaining the maximum relevance minimum redundancy (mRMR) index and determining the optimal feature set.

[0036] Combining equations (3) and (7), the final maximum correlation minimum redundancy mRMR index is: (8) Make I Get the maximum value of the data set D ( X i ,X j ) is the optimal feature set for industrial users S .

[0037] Through the above process, the optimal feature set can be screened for each industrial user load. S .

[0038] S2, based on the optimal feature set, obtains the correlation coefficient matrix of each industrial user load.

[0039] S21, filter out the optimal feature set of all industrial user loads S Contained in N non-repeating features to form a new feature set F , determine the correlation coefficient vector.

[0040] For each industrial user, combine formula (2) to obtain the industrial user load and their respective optimal feature sets S The Pearson correlation coefficient between the features included in the is used to obtain the correlation coefficient vector of each industrial user load, and the obtained correlation coefficient vector is padded. The specific operation is as follows: For a certain industrial user, if the new feature set F A certain feature in the optimal feature set of the industrial user load S In the correlation coefficient vector, the corresponding correlation coefficient is filled in. If the new feature set F A feature in the load is not in the optimal feature set of the industrial user S In the correlation coefficient vector, 0 is filled in the corresponding position, so that the dimension of each industrial user load is obtained. N The new correlation coefficient vector of : , F For the new feature set, Y' For the new feature set F Industrial user load in.

[0041] S22, concatenate the new correlation coefficient vectors of all industrial user loads to obtain a dimension of ( M,N )'s correlation coefficient matrix H : (9).

[0042] Where, M , N represents the dimension of matrix H.

[0043] S3, based on the correlation coefficient matrix, uses graph convolutional neural network to achieve accurate division of industrial user groups.

[0044] S31, use the K-means algorithm to initially cluster the industrial user load, and combine the silhouette coefficient formula to determine the initial number of clusters, and obtain the initial cluster center and nearby samples.

[0045] Specifically, the K-means algorithm is used to divide industrial users based on their load data, and the optimal cluster number k is determined according to the silhouette coefficient, and finally the industrial users are divided into k industrial user groups.

[0046] S32: Label the correlation coefficient matrix based on the initial cluster center and nearby samples.

[0047] In the correlation coefficient matrix, the data related to the initial cluster center and nearby samples are regarded as labeled data, and the data unrelated to the initial cluster center and nearby samples are regarded as unlabeled data.

[0048] S33, construct graph data structure and graph adjacency matrix.

[0049] Specifically, each industrial user is regarded as a node (1-dimensional element) of the graph data structure, and the Pearson correlation coefficient between industrial user loads is used as the corresponding 2-dimensional element of the graph adjacency matrix. The association relationship between the nodes of the graph data structure is finely quantified, and the topological structure information between nodes of different graph data structures is considered.

[0050] The Pearson correlation coefficient between industrial user loads is calculated based on the normalized data of industrial user loads at each data collection moment.

[0051] In S34, the annotated correlation coefficient matrix is ​​input into the Graph Convolution Neural Networks (GCNNs) model based on the graph structure in S33 to achieve accurate segmentation of all industrial user groups.

[0052] S4, based on industrial user groups, captures the temporal relationship of the load of industrial user groups.

[0053] S41 , at each data collection moment, adding the normalized data of the load of each industrial user in each industrial user group to obtain the load of each industrial user group.

[0054] S42, uses multi-layer dilated causal convolution to capture the temporal relationship of the load of each industrial user group.

[0055] First, dilated causal convolution makes t The output at time +1 is only t This is related to the input before the moment, which is in line with the real logic of industrial user group load data analysis containing time series. On the other hand, the receptive field of the convolution operation is increased by injecting holes into the standard convolution kernel. The receptive field size calculation formula of the convolution is: (10) Where, z is the size of the original convolution kernel, d is the expansion factor.

[0056] Then, skip connections are used to concatenate the outputs of each expanded causal convolutional layer to obtain the temporal relationship (causal relationship and long-term dependency) of the industrial user group load, avoiding degradation problems caused by excessive stacking. S5: Construct a new graph structure and input the temporal relationship of the load of each industrial user group into the dynamic graph learning model, and adaptively learn to generate a dynamic graph adjacency matrix.

[0057] S51 , taking the load of each industrial user group as a node of a new graph structure, and taking the temporal relationship of the load of each industrial user group as the association between the nodes of the new graph structure.

[0058] S52, in the dynamic graph learning layer, the correlation between the nodes of each new graph structure is measured by the Mahalanobis distance, and the covariance matrix in the Mahalanobis distance is set as a trainable weight parameter to ensure the feasibility of real-time updating of the dynamic graph adjacency matrix.

[0059] First, the Mahalanobis distance is used to measure the correlation between the nodes of each new graph structure. The calculation formula is: (11) Where, x p t and x q t are the nodes of the new graph structure p and q exist t The temporal feature sequence of the moment, W is the covariance matrix.

[0060] Secondly, the covariance matrix in the Mahalanobis distance is set as a trainable weight parameter to ensure the feasibility of real-time updating of the dynamic graph adjacency matrix. Then, the Mahalanobis distance is used to calculate the correlation between nodes in the new graph structure. The calculation formula is: (12) (13) Where, G t ( x i t , x j t )for t time x p t and x q trelevance; σ D is the standard deviation; for t The normalized correlation matrix between the nodes of each new graph structure at the moment is t The dynamic graph adjacency matrix at time.

[0061] S6. Finally, the dynamic graph adjacency matrix is ​​used to train a graph convolutional neural network based on the new graph structure data to predict the load value of each industrial user group.

[0062] The embodiment of the present disclosure performs a fine division of industrial user groups based on a graph convolutional neural network model, which can fully consider the differences in the impact of different characteristics on the load of each industrial user and the topological structure information between different industrial users. Through an end-to-end modeling approach, a fine and stable division of industrial user groups is achieved; the dynamic graph learning method for the correlation of industrial user groups can capture the electricity consumption correlation between groups in real time, thereby realizing dynamic monitoring and prediction of electricity consumption behavior. This method can adapt to the continuous changes in prediction scenarios and provide more real-time and flexible decision support for industrial controllable loads to participate in demand response. It has important practical significance for improving the ability of industrial users to participate in demand response and optimizing energy management.

[0063] Based on the above method, the embodiment of the present disclosure also provides an electric power load forecasting system based on the dynamic correlation of industrial user groups corresponding to the above method, including an optimal feature set determination module, a correlation coefficient matrix construction module, an industrial user group division module, a time series relationship calculation module, a dynamic graph adjacency matrix generation module, and a load value prediction module; An optimal feature set determination module is used to obtain the optimal feature set of each industrial user load based on the maximum correlation and minimum redundancy criterion; A correlation coefficient matrix construction module is used to obtain the correlation coefficient matrix of each industrial user load based on the optimal feature set; The industrial user group segmentation module is used to accurately segment industrial user groups using a graph convolutional neural network based on the correlation coefficient matrix; A time series relationship calculation module is used to capture the time series relationship of the load of industrial user groups based on industrial user groups; The dynamic graph adjacency matrix generation module is used to construct a new graph structure, input the temporal relationship of the load of each industrial user group into the dynamic graph learning model, and adaptively learn to generate the dynamic graph adjacency matrix; The load value prediction module is used to use the dynamic graph adjacency matrix to train a graph convolutional neural network based on new graph structure data to predict the load value of each industrial user group.

[0064] Based on the same inventive concept as the above disclosure, the present disclosure also provides an electronic device. Figure 2As shown, the electronic device of an embodiment of the present disclosure includes at least one electrically connected processor and at least one memory, wherein the memory is electrically connected to the processor, wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the power load forecasting method based on dynamic correlation of industrial user groups as described above.

[0065] It should be noted that the electrical connection between the above-mentioned units does not necessarily mean the connection between lines. An indirect connection method can be applied to the embodiments of the present disclosure as long as the purpose of the present disclosure is achieved.

[0066] Based on the same inventive concept, the present disclosure also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the power load forecasting method based on dynamic correlation of industrial user groups as described above.

[0067] Based on the same inventive concept, the present disclosure also provides a computer program product, which is stored in at least one storage medium; the computer program product includes several instructions for enabling at least one computer device to execute the above-mentioned power load forecasting method based on dynamic correlation of industrial user groups.

[0068] Although the present disclosure has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure.

Claims

1. A method for predicting power load based on dynamic correlation of industrial user groups, characterized in that: The method comprises: Obtaining the optimal feature set for each industrial user load based on the maximum correlation minimum redundancy criterion, including: constructing a feature set based on original feature data, and calculating the maximum correlation index between other related features and the industrial user load using the feature set; selecting normalized data of two other features with the same sequence length in the feature set, discretely distributing data points in the normalized data of the two other features in pairs in a two-dimensional space, constructing a data set, and calculating the minimum redundancy index between other related features and the industrial user load based on the mutual information of the data set; obtaining a maximum correlation minimum redundancy (mRMR) index based on the maximum correlation index and the minimum redundancy index; the data set that maximizes the maximum correlation minimum redundancy (mRMR) index is the optimal feature set for the industrial user; Based on the optimal feature set, the correlation coefficient matrix of each industrial user load is obtained; Based on the correlation coefficient matrix, the graph convolutional neural network is used to accurately divide industrial user groups; Based on industrial user groups, capture the temporal relationship of industrial user group loads; Construct a new graph structure, input the temporal relationship of the load of each industrial user group into the dynamic graph learning model, and adaptively learn to generate the dynamic graph adjacency matrix; The dynamic graph adjacency matrix is ​​used to train a graph convolutional neural network based on the new graph structure data to predict the load values ​​of various industrial user groups.

2. The power load forecasting method based on dynamic correlation of industrial user groups according to claim 1 is characterized in that: Normalize the original feature data and construct a feature set, including: Normalize the original feature data to the interval [0, 1] according to the following formula; All normalized data are constructed into feature sets according to each industrial user and time series.

3. The power load forecasting method based on dynamic correlation of industrial user groups according to claim 1 is characterized in that: Based on the optimal feature set, the correlation coefficient matrix of each industrial user load is obtained, including: Filter out the non-repeated features contained in the optimal feature set of all industrial user loads, form a new feature set, and determine the correlation coefficient vector; The new correlation coefficient vectors of all industrial user loads are concatenated to obtain the correlation coefficient matrix.

4. The power load forecasting method based on dynamic correlation of industrial user groups according to claim 1 is characterized in that: Based on the correlation coefficient matrix, the graph convolutional neural network is used to accurately segment industrial user groups, including: Use the K-means algorithm to initially cluster industrial user loads, and combine it with the silhouette coefficient formula to determine the initial number of clusters, and obtain the initial cluster center and nearby samples; According to the initial cluster center and nearby samples, mark the correlation coefficient matrix; Construct graph data structures and graph adjacency matrices; The labeled correlation coefficient matrix is ​​input into the graph convolutional neural network model based on the graph structure to achieve accurate division of all industrial user groups.

5. The power load forecasting method based on dynamic correlation of industrial user groups according to claim 1 is characterized in that: Based on industrial user groups, the temporal relationship of industrial user group loads is captured, including: At each data collection moment, the normalized data of the load of each industrial user in each industrial user group is added together to obtain the load of each industrial user group; Multi-layer dilated causal convolution is used to capture the temporal relationship of the load of each industrial user group.

6. The power load forecasting method based on dynamic correlation of industrial user groups according to claim 1 is characterized in that: Construct a new graph structure, input the temporal relationship of the load of each industrial user group into the dynamic graph learning model, and adaptively learn to generate a dynamic graph adjacency matrix, including: The load of each industrial user group is used as a node in the new graph structure, and the temporal relationship of the load of each industrial user group is used as the relationship between the nodes in the new graph structure; In the dynamic graph learning layer, the correlation between the nodes of each new graph structure is measured by the Mahalanobis distance, and the covariance matrix in the Mahalanobis distance is set as a trainable weight parameter to ensure the feasibility of real-time updating of the dynamic graph adjacency matrix.

7. A power load forecasting system based on dynamic correlation of industrial user groups, characterized by: The system includes an optimal feature set determination module, a correlation coefficient matrix construction module, an industrial user group division module, a time series relationship calculation module, a dynamic graph adjacency matrix generation module, and a load value prediction module; An optimal feature set determination module is used to obtain the optimal feature set for each industrial user load based on the maximum correlation minimum redundancy criterion, including: constructing a feature set based on the original feature data, and using the feature set to calculate the maximum correlation index between other related features and the industrial user load; selecting normalized data of two other features with the same sequence length in the feature set, discretely distributing the data points in the normalized data of the two other features in pairs in a two-dimensional space to construct a data set, and calculating the minimum redundancy index between other related features and the industrial user load based on the mutual information of the data set; obtaining the maximum correlation minimum redundancy (mRMR) index based on the maximum correlation index and the minimum redundancy index; the data set that maximizes the maximum correlation minimum redundancy (mRMR) index is the optimal feature set for the industrial user; A correlation coefficient matrix construction module is used to obtain the correlation coefficient matrix of each industrial user load based on the optimal feature set; The industrial user group segmentation module is used to accurately segment industrial user groups using a graph convolutional neural network based on the correlation coefficient matrix; A time series relationship calculation module is used to capture the time series relationship of the load of industrial user groups based on industrial user groups; The dynamic graph adjacency matrix generation module is used to construct a new graph structure, input the temporal relationship of the load of each industrial user group into the dynamic graph learning model, and adaptively learn to generate the dynamic graph adjacency matrix; The load value prediction module is used to use the dynamic graph adjacency matrix to train a graph convolutional neural network based on new graph structure data to predict the load value of each industrial user group.

8. An electronic device, characterized in that: comprising at least one processor and at least one memory electrically connected; The memory is electrically connected to the processor, wherein the memory stores instructions that can be executed by at least one of the processors, and the instructions are executed by at least one of the processors to enable at least one of the processors to execute the power load forecasting method based on dynamic correlation of industrial user groups as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer readable storage medium stores a computer program; When the computer program is executed by a processor, the method for predicting power load based on dynamic correlation of industrial user groups according to any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that The computer program product is stored in at least one storage medium; The computer program product includes several instructions for causing at least one electronic device to execute the power load forecasting method based on dynamic correlation of industrial user groups as described in any one of claims 1 to 6.

Images