Multi-region Load Probability Prediction Method and System Considering Spatiotemporal Characteristics

Abstract

The present invention belongs to the technical field of electric load forecasting, and discloses a multi-substation area load probability forecasting method and system considering spatio-temporal characteristics. Geographical location data, historical load data and external data related to the load of the substation area are collected and preprocessed; the substation area is taken as a node, and the electrical connection between the substation areas is taken as an edge to construct a graph structure; feature extraction is performed on the graph structure to obtain spatial features between multiple substation areas, and operations are performed on the spatial features to obtain temporal features; dynamic fusion of the spatial and temporal features is performed to obtain the fused features, and load probability forecasting is performed based on the fused features; a loss function is defined, and the model parameters are optimized to obtain a trained load probability forecasting model. The present invention realizes the forecasting of the load of multiple substation areas by considering the actual physical space information and temporal feature information, reflects the uncertainty of the load through the way of probability forecasting, and provides more comprehensive and reliable load information for the scheduling, planning and operation management of the power system.

Description

Technical Field

[0001] The present invention relates to the technical field of electric load forecasting, and particularly to a multi-substation area load probability forecasting method and system considering spatio-temporal characteristics. Background Art

[0002] The statements in this part merely provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] With the continuous expansion of the scale of the power system and the widespread popularization of electric vehicles, the load characteristics of the multi-substation area power grid have become increasingly complex. There are electrical connections and power interactions between multiple substations. Their loads are not only affected by the electrical equipment within their own substations but also by the associated effects of other substations.

[0004] At the same time, load data has obvious time series characteristics, including periodicity, trend, and uncertainty. Traditional load forecasting methods, such as those based on statistical models, are difficult to effectively capture the complex spatial correlation and time series characteristics between multiple substations. Machine learning methods, although they can handle complex data to a certain extent, do not dig deep enough into the spatial structure of multi-substation area data. The recurrent neural network (RNN) and its variant LSTM in deep learning perform well in processing time series data and can learn the long-term dependence relationship of load data, but they do not adequately consider the spatial relationship between substations in multi-substation area data. The graph convolutional network (GCN) is good at processing data with graph structures and can mine the actual spatial associations between multiple substations, but most do not fully integrate the actual spatial information, resulting in inaccurate capture of spatial correlation by the model.

[0005] Currently, some load forecasting methods based on deep learning have been proposed. However, some methods cannot fully integrate multi-substation area data and actual spatial and time characteristics in the model structure, resulting in limited prediction accuracy, and rarely consider the uncertainty of load forecasting, and cannot provide sufficient risk information for the decision-making of the power system. Therefore, a probability forecasting method that can comprehensively consider the actual spatial structure of multiple substations, the differences in load characteristics in different regions, and load uncertainty is needed. Summary of the Invention

[0006] To solve the above problems, the present invention proposes a multi-substation area load probability forecasting method and system considering spatio-temporal characteristics, which integrates the advantages of multiple models, considers the actual physical spatial information and time feature information to achieve the forecasting of multi-substation area loads, and reflects the uncertainty of loads through probability forecasting, providing more comprehensive and reliable load information for the scheduling, planning, and operation management of the power system.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] In a first aspect, the present invention provides a multi-district load probability prediction method considering spatio-temporal characteristics, including the following steps:

[0009] Collect the geographical location data, historical load data, and external data related to the load of each district, and preprocess the collected data;

[0010] Based on the preprocessed data, construct a graph structure with each district as a node and the electrical connection between districts as an edge;

[0011] Extract features from the graph structure data to obtain the spatial features between multiple districts, and input the spatial features into the LSTM layer in chronological order to obtain temporal features;

[0012] Dynamically fuse the spatial features and temporal features to obtain the fused features, and perform load probability prediction based on the fused features;

[0013] Define a loss function, optimize the model parameters, and obtain a trained multi-district load probability prediction model.

[0014] As an alternative implementation, preprocessing the collected data specifically includes:

[0015] Clean the collected data, remove outliers and missing values, normalize the data, divide the normalized data into multiple time windows in chronological order, and use the data within each time window as a sample.

[0016] As an alternative implementation, extracting features from the graph structure data specifically includes: using multiple stacked GCN layers to perform linear transformation and ReLU processing of the activation function on the node features and adjacency matrix of the graph structure to obtain the spatial features between multiple districts.

[0017] As an alternative implementation, the adjacency matrix is constructed according to the physical connection relationship and power transmission characteristics between districts, and the element values in the adjacency matrix are set according to the line impedance and power transmission capacity between districts.

[0018] As an alternative implementation, the node features of the graph structure are the geographical location information, load, and related data of each district.

[0019] As an alternative implementation, use a cross-gating block to dynamically fuse the spatial features and temporal features. The cross-gating block consists of two gating units, which are respectively used to fuse the spatial features extracted by GCN and the time series features learned by LSTM, generate a gating signal by calculating the weighted sum of the input features, and dynamically fuse the spatial and temporal features according to the gating signal.

[0020] Second aspect, the present invention provides a multi - substation area load probability prediction system considering spatio - temporal characteristics, including:

[0021] A data collection and pre - processing module, configured to: collect geographical location data, historical load data, and external data related to the load of each substation area, and pre - process the collected data;

[0022] A graph structure construction module, configured to: based on the pre - processed data, construct a graph structure with each substation area as a node and the electrical connection between substation areas as an edge;

[0023] A feature extraction module, configured to: extract features from the graph structure data to obtain spatial features between multiple substation areas, and input the spatial features into an LSTM layer in chronological order to obtain temporal features;

[0024] A load probability prediction module, configured to: dynamically fuse the spatial features and temporal features to obtain fused features, and perform load probability prediction based on the fused features;

[0025] A model training module, configured to: define a loss function, optimize model parameters, and obtain a trained multi - substation area load probability prediction model.

[0026] Third aspect, the present invention provides an electronic device, including a memory and a processor, as well as computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the method described in the first aspect is completed.

[0027] Fourth aspect, the present invention provides a computer - readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the method described in the first aspect is completed.

[0028] Fifth aspect, the present invention provides a computer program product, including a computer program. When the computer program is executed by a processor, the method described in the first aspect is implemented.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] The present disclosure proposes a multi-district load probability prediction method and system considering spatio-temporal characteristics, constructs a CG-GCN-LSTM model, gives full play to the advantages of GCN in mining the spatial correlation of multiple districts, LSTM in learning time series characteristics, and the cross-gating block in dynamically fusing spatio-temporal characteristics, and can capture the complex characteristics of the actual space and load data of multiple districts more comprehensively and accurately. Compared with a single model or a simple combined model, the prediction accuracy of the load is significantly improved. By constructing reasonable graph structure data and adjacency matrix, the physical connection and power transmission relationship of the multi-district power grid are effectively reflected, enabling the model to better learn the actual spatial dependence and data spatial dependence between districts, and providing more accurate spatial information support for load prediction. The introduction of the cross-gating block realizes the adaptive fusion of spatial and temporal features, can flexibly adjust the feature fusion strategy according to different load data characteristics and prediction tasks, improves the adaptability and robustness of the model, and can still maintain good prediction performance in the face of data noise and distribution changes. The model finally performs probability prediction, and can stably and reliably consider different load differences and load uncertainties for multi-district load prediction in practical applications, providing a strong guarantee for the scientific dispatching and efficient operation of the power system.

[0031] Advantages of additional aspects of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The specification drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0033] Figure 1 It is a flowchart of the multi-district load probability prediction method considering spatio-temporal characteristics provided in Embodiment 1 of the present invention;

[0034] Figure 2 It is a schematic structural diagram of the spatio-temporal model of multi-district charging piles;

[0035] Figure 3 It is a schematic structural diagram of the CG-GCN-LSTM model. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] The present invention will be further described below in conjunction with the drawings and embodiments.

[0037] It should be noted that the following detailed description is exemplary and is intended to provide further description of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0038] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units need not be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0039] In the case of no conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other.

[0040] Embodiment 1

[0041] As Figure 1 shown, this embodiment provides a multi-substation area load probability prediction method considering spatio-temporal characteristics, including the following steps:

[0042] S1. Collect the geographical location data, historical load data and external data related to the load of each substation area, and preprocess the collected data;

[0043] S2. Based on the preprocessed data, take each substation area as a node and the electrical connection between the substation areas as an edge to construct a graph structure;

[0044] S3. Extract features from the graph structure data to obtain the spatial features between multiple substation areas, and input the spatial features into the LSTM layer in chronological order to obtain the time features;

[0045] S4. Dynamically fuse the spatial features and time features to obtain the fused features, and perform load probability prediction based on the fused features;

[0046] S5. Define a loss function, optimize the model parameters, and obtain a trained multi-substation area load probability prediction model.

[0047] Collect the geographical location (latitude and longitude) information of each substation area, and use devices such as smart meters and sensors to collect the historical load data of each substation area, including the active power, reactive power, voltage and other information of each substation area. At the same time, collect external data related to the load, such as meteorological data (temperature, humidity, wind speed), date type (weekday, holiday). Clean the collected data to remove outliers and missing values. For outliers, use statistical methods (3σ principle) for identification and correction; for missing values, use linear interpolation method for filling according to the time series characteristics and correlation of the data. Then, perform normalization processing on the data to map all data to the interval [0, 1].

[0048] All data are normalized using the normalization method:

[0049] ;

[0050] where x is the original data, and are the minimum and maximum values of the data features respectively, is the normalized data.

[0051] The normalized data are divided into multiple time windows in chronological order. The data within each time window are used as a sample. Each sample contains the load data and relevant external data of multiple substations within that time window. These samples are constructed into graph-structured data and used as the input for the subsequent model. The construction of the graph structure is as follows:

[0052] The multi-substation power grid is regarded as a graph structure, where each substation is a node of the graph, and the electrical connections between substations are edges. According to the physical connection relationship and power transmission characteristics between substations, an adjacency matrix is constructed to describe the spatial correlation strength between substations. The element values in the adjacency matrix can be set according to factors such as the line impedance and power transmission capacity between substations. If there is a direct and tight electrical connection between substation i and substation j, the corresponding element value in the adjacency matrix is relatively large; if the connection is weak or there is no direct connection, the value is small or 0.

[0053] Node information of the graph structure: Each substation is used as a node of the graph. In addition to the traditional load and related data (such as active power, reactive power, voltage, etc.) as node features, the geographical location information of the substation, the longitude and latitude coordinates, is also incorporated. These information are combined into a feature vector to represent each node, and the feature vector of node is [load, longitude, latitude].

[0054] Edge information and adjacency matrix: The construction of the edge not only considers the electrical connection relationship between substations, but also combines their geographical distances. For two substations with electrical connections, the closer the geographical distance, the greater the weight of the edge; otherwise, the smaller the weight.

[0055] Adjacency matrix is used to describe the connection relationship and weight between nodes. If there is an electrical connection between substation and , and the geographical distance is , the adjacency matrix element can be defined as:

[0056] ;

[0057] Among them and are adjustable parameters used to control the influence degree of geographical distance on edge weights.

[0058] Use GCN to process the constructed graph structure data. Through convolutional operations on the graph, GCN can automatically learn the spatial feature representation between multiple substations. The input of GCN is the node features of the graph (the load and related data of each substation) and the adjacency matrix. After stacking multiple GCN layers, higher-level spatial features are gradually extracted. In each layer of GCN, through linear transformation and non-linear activation function ReLU processing on the node features and the adjacency matrix, the output features of this layer are obtained. These output features integrate the information of the substation itself and the correlation information of adjacent substations, thus effectively capturing the spatial correlation between multiple substations.

[0059] The input of the GCN layer includes the node feature matrix , the adjacency matrix and the node position information matrix (constituted by the longitude and latitude information of the nodes). The calculation of the first layer of GCN is extended on the basis of the traditional GCN, considering the actual spatial position information of the nodes. The calculation of the first layer of GCN can be expressed as:

[0060] ;

[0061] Among them,

[0062] ;

[0063] ;

[0064] In the formula, is the normalized adjacency matrix, is the degree matrix (diagonal matrix), is the learnable weight matrix of the first layer, is the activation function.

[0065] Add the node feature matrix and the node position information matrix so that the model can consider both the load characteristics and spatial position characteristics of the nodes during feature propagation. In order to extract higher-level spatial features, stack multiple GCN layers. The calculation of the th layer of GCN is:

[0066] ;

[0067] Among them is the output of the previous layer, is the The learnable weight matrix of the layer. By stacking multiple layers of GCNs, the model can gradually capture the synergy of load changes between multiple regions and the load distribution pattern in space.

[0068] To enhance the model's capture of spatial features, a spatial attention mechanism is introduced. Calculate the attention weights for each node Define the query vector , the key vector .

[0069] ;

[0070] The dot product similarity is: ;

[0071] The updated feature of node is:

[0072] ;

[0073] Among them, and are the initial feature vectors of nodes and respectively, and are learnable weight matrices.

[0074] Input the spatial features output by the GCN layer into the LSTM layer in chronological order. The LSTM has memory cells and gating mechanisms, and can effectively learn the long-term temporal dependencies of load data. The LSTM layer processes the input sequence data step by step in time, and through the coordinated action of the forget gate, input gate, and output gate, decides which information to retain, which information to update, and which information to output. During the processing, the LSTM can remember the trends, periodicity, etc. of the load data, so as to effectively predict future load changes. By stacking multiple layers of LSTM, the model's learning ability for complex time series features can be further enhanced.

[0075] Arrange the spatial features output by the GCN layer in chronological order and input them into a network containing 3 layers of LSTM. The number of hidden units in each layer of LSTM is set to 64. In the first layer of LSTM, the input sequence passes through the forget gate , input gate , output gate and memory cell calculations to obtain the output :

[0076] ;

[0077] ;

[0078] ;

[0079] ;

[0080] ;

[0081] wherein, is the weight matrix, is the bias vector, is the Sigmoid function, represents element-wise multiplication. The second and third layer LSTMs process the output of the previous layer in the same way to further learn the time series features.

[0082] After the GCN layer and the LSTM layer, a cross-gating block is introduced. This module consists of two gating units, which are respectively used to fuse the spatial features extracted by the GCN and the time series features learned by the LSTM. One gating unit is responsible for controlling the information flow of the spatial features, and the other gating unit is responsible for controlling the information flow of the time features. The gating unit generates a gating signal by calculating the weighted sum of the input features, and dynamically fuses the spatial and time features according to the gating signal. The gating signal is obtained through operations such as linear transformation, non-linear activation, and weighted summation of the spatial and time features. In this way, the cross-gating block can adaptively adjust the fusion ratio of the spatial and time features according to different input data and prediction task requirements, so as to better utilize the advantages of the two types of features and improve the prediction accuracy.

[0083] The cross-gating block contains two gating units, which respectively process the spatial features output by the GCN and the time features output by the LSTM ;

[0084] The fused spatial features: ;

[0085] For the spatial feature gating unit, calculate the gating signal : ; The fused time features: ;

[0086] The finally fused features: ;

[0087] After the feature output after cross-gating block fusion, a deep-AR base is used for probability prediction. The probability prediction layer adopts the Gaussian mixture model (GMM) method to map the fused features to the probability distribution of the load. For the Gaussian mixture model, the negative log-likelihood loss function is adopted, and the root mean square error (RMSE), quantile loss normalization sum ( ), and the continuous ranked probability score (CRPS) are used to measure the difference between the probability distribution predicted by the model and the true load value.

[0088] By inputting a Gaussian distribution and using the negative log-likelihood as the loss function:

[0089] ;

[0090] Adopt a DeepAR infrastructure with 3-layer LSTM. To evaluate the performance, the root mean square error (RMSE), quantile loss normalization sum ( ), and the continuous ranked probability score (CRPS) (the lower the score, the better the performance) are reported. The calculation formula of

[0091] is as follows:

[0092] Among them, ， represents the predicted quantile value.

[0093] For the predicted Gaussian distribution , the score is defined as:

[0094] ;

[0095] Among them, are the probability density function (PDF) and cumulative distribution function (CDF) of the Gaussian distribution respectively. For the prediction time domain , the CRPS score is defined as the average value of the score: .

[0096] Embodiment 2

[0097] This embodiment provides a multi-substation area load probability prediction system considering spatio-temporal characteristics, including:

[0098] A data acquisition and preprocessing module, configured to: acquire the geographical location data, historical load data, and external data related to the load of each substation area, and preprocess the acquired data;

[0099] A graph structure construction module, configured to: based on the preprocessed data, construct a graph structure with each substation area as a node and the electrical connections between substation areas as edges;

[0100] A feature extraction module, configured to: extract features from the graph structure data to obtain the spatial features between multiple substation areas, and input the spatial features into the LSTM layer in chronological order to obtain temporal features;

[0101] A load probability prediction module, configured to: dynamically fuse the spatial features and temporal features to obtain the fused features, and perform load probability prediction based on the fused features;

[0102] A model training module, configured to: define a loss function, optimize the model parameters, and obtain a trained multi-substation area load probability prediction model.

[0103] It should be noted here that the above modules correspond to the steps described in Embodiment 1. The examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the content disclosed in the above Embodiment 1. It should be noted that the above modules can be executed in a computer system such as a set of computer-executable instructions as part of the system.

[0104] In more embodiments, there is also provided:

[0105] An electronic device, including a memory and a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the method described in Embodiment 1 is completed. For the sake of brevity, it will not be elaborated here.

[0106] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be 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. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0107] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0108] A computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the method described in Embodiment 1.

[0109] The method in Embodiment 1 can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the art such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, registers, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0110] A computer program product includes a computer program, which, when executed by a processor, implements the method described in Embodiment 1.

[0111] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which are executed in a device on a target real or virtual processor to perform the process / method as described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules can be combined or divided as needed. The machine-executable instructions for program modules can be executed locally or within a distributed device. In a distributed device, program modules can be located in local and remote storage media.

[0112] The computer program code for implementing the method of the present invention can be written in one or more programming languages. These computer program codes can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing devices, such that when the program code is executed by the computer or other programmable data processing devices, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the computer, partially on the computer, as an independent software package, partially on the computer and partially on a remote computer, or entirely on a remote computer or server.

[0113] In the context of the present invention, the computer program code or related data can be carried by any suitable carrier to enable a device, apparatus, or processor to perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, etc. Examples of signals can include electrical, optical, radio, acoustic, or other forms of propagated signals, such as carrier waves, infrared signals, etc.

[0114] Those of ordinary skill in the art will realize that the units and algorithm steps of the examples described in conjunction with this embodiment can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0115] Although the specific implementation manners of the present invention have been described above in conjunction with the accompanying drawings, it is not a limitation on the protection scope of the present invention. Those skilled in the art should understand that, based on the technical solution of the present invention, various modifications or deformations that can be made by those skilled in the art without creative efforts are still within the protection scope of the present invention.

Claims

1. A multi-substation area load probability prediction method considering spatio-temporal characteristics, characterized in that It includes the following steps: Collect the geographical location data, historical load data, and external data related to the load of each substation area, and preprocess the collected data; Based on the preprocessed data, construct a graph structure with each substation area as a node and the electrical connections between substation areas as edges; Extract features from the graph structure data to obtain the spatial features between multiple substation areas, and input the spatial features into the LSTM layer in chronological order to obtain the temporal features; Dynamically fuse the spatial features and temporal features to obtain the fused features, and perform load probability prediction based on the fused features; Define a loss function, optimize the model parameters, and obtain a trained multi-substation area load probability prediction model; For the dynamic fusion of the spatial features and temporal features, a cross-gating block is used to dynamically fuse the spatial features and temporal features. The cross-gating block consists of two gating units, which are respectively used to fuse the spatial features extracted by the GCN and the time series features learned by the LSTM, generate a gating signal by calculating the weighted sum of the input features, and dynamically fuse the spatial and temporal features according to the gating signal; For the load probability prediction based on the fused features, after the features output by the cross-gating block are fused, a deep-AR basis is used for probability prediction. The probability prediction layer adopts the Gaussian mixture model method to map the fused features to the probability distribution of the load.

2. The multi-region load probability prediction method considering spatio-temporal characteristics according to claim 1, wherein Preprocess the collected data, specifically: Clean the collected data, remove outliers and missing values, normalize the data, divide the normalized data into multiple time windows in chronological order, and use the data within each time window as a sample.

3. The multi-region load probability prediction method considering spatio-temporal characteristics according to claim 1, characterized in that, Extract features from the graph structure data, specifically: use multiple stacked GCN layers to perform linear transformation and ReLU processing of the activation function on the node features and adjacency matrix of the graph structure to obtain the spatial features between multiple substation areas.

4. The multi-region load probability prediction method considering spatio-temporal characteristics according to claim 3, characterized in that, The adjacency matrix is constructed according to the physical connection relationship and power transmission characteristics between substation areas, and the element values in the adjacency matrix are set according to the line impedance and power transmission capacity between substation areas.

5. The multi-region load probability prediction method considering spatio-temporal characteristics according to claim 3, characterized in that The node features of the graph structure are the geographical location information, load, and related data of each substation area.

6. A multi-substation area load probability prediction system considering spatio-temporal characteristics, characterized in that, It includes: A data collection and preprocessing module, configured to: collect the geographical location data, historical load data, and external data related to the load of each substation area, and preprocess the collected data; A graph structure construction module, configured to: based on the preprocessed data, construct a graph structure with each substation area as a node and the electrical connections between substation areas as edges; A feature extraction module, configured to: extract features from the graph structure data to obtain the spatial features between multiple substation areas, and input the spatial features into the LSTM layer in chronological order to obtain the temporal features; A load probability prediction module, configured to: dynamically fuse the spatial features and temporal features to obtain the fused features, and perform load probability prediction based on the fused features; A model training module, configured to: define a loss function, optimize the model parameters, and obtain a trained multi-substation area load probability prediction model; The dynamic fusion of spatial features and temporal features is carried out by using a cross-gating block to fuse the spatial features and temporal features. The cross-gating block consists of two gating units, which are respectively used to fuse the spatial features extracted by GCN and the time series features learned by LSTM, generate a gating signal by calculating the weighted sum of the input features, and dynamically fuse the spatial and temporal features according to the gating signal; For the load probability prediction based on the fused features, after the feature output fused by the cross-gating block, a deep-AR basis is used for probability prediction. The probability prediction layer adopts the Gaussian mixture model (GMM) method to map the fused features to the probability distribution of the load.

7. An electronic device, characterized in that, It includes a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the method described in any one of claims 1-5 is completed.

8. A computer-readable storage medium, characterized in that, For storing computer instructions, when the computer instructions are executed by the processor, the method described in any one of claims 1-5 is completed.

9. A computer program product, characterized in that, It includes a computer program, and when the computer program is executed by the processor, the method described in any one of claims 1-5 is implemented.

Images