Regional photovoltaic prediction method and device, server and storage medium

Abstract

The invention discloses a regional photovoltaic prediction method and device, a server and a storage medium, and the method comprises the steps: carrying out the clustering of the historical data of each station in a region, and obtaining a station clustering matrix; inputting the station clustering matrix and the historical data of each station into a trained graph convolutional neural network; the graph convolutional neural network comprises a self-attention unit, the self-attention unit is used for calculating an attention score matrix between each node and all other nodes in the same category to capture the relation strength between the nodes, and the attention score matrix and a matrix output by a convolutional layer are fused to obtain the relation strength between the nodes. The method comprises the following steps: taking a plurality of stations as input of a normalization layer, and outputting a directed topological graph of the stations by using the normalization layer, nodes in the directed topological graph comprising feature dimensions of historical meteorological information and power generation information; and inputting the directed topological graph into the trained graph neural network, and obtaining a regional photovoltaic prediction result according to a result output by the graph neural network.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic prediction technology, and in particular to a regional photovoltaic prediction method, device, server and storage medium. Background Art

[0002] With the profound changes in the global energy structure, solar photovoltaic technology, as a key driving force for renewable clean energy, has become increasingly important. The output power of photovoltaic power stations is directly affected by solar irradiance, which in turn is affected by a variety of meteorological conditions. Therefore, accurate prediction of solar irradiance is crucial for optimizing the operation strategy of photovoltaic power stations, improving the efficiency of grid dispatching, and ensuring the stability and economy of power supply.

[0003] The existing regional photovoltaic power prediction method mainly predicts the power of a single station based on the total predicted irradiance of the horizontal plane, and then expands it to the entire region. However, this method has low prediction efficiency. The existing method first predicts a single station and then expands it to the entire region. When there are many stations in the region, this method has a large amount of calculation and takes a long time, making the overall prediction efficiency low. Summary of the invention

[0004] The embodiments of the present invention provide a regional photovoltaic prediction method, device, server and storage medium to solve the technical problem of low efficiency in regional photovoltaic power prediction in the prior art.

[0005] In a first aspect, an embodiment of the present invention provides a regional photovoltaic prediction method, comprising:

[0006] Cluster the historical data of each site in the region to obtain a site clustering matrix;

[0007] Inputting the site clustering matrix and the historical data of each site into the trained graph convolutional neural network;

[0008] The graph convolutional neural network includes: a self-attention unit, which uses the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and fuses the attention score matrix with the matrix output by the convolution layer as the input of a normalization layer, and uses the normalization layer to output a site directed topology map, and the nodes in the directed topology map include: feature dimensions of historical meteorological information and power generation information;

[0009] The directed topological graph is input into a trained graph neural network, and a regional photovoltaic prediction result is obtained according to the output of the graph neural network.

[0010] In a second aspect, an embodiment of the present invention further provides a regional photovoltaic prediction device, comprising:

[0011] The clustering module is used to cluster the historical data of each site in the region to obtain a site clustering matrix;

[0012] An input module, used for inputting the site clustering matrix and the historical data of each site into the trained graph convolutional neural network;

[0013] A directed topology map output module, used to use the self-attention unit included in the graph convolutional neural network, use the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and fuse the attention score matrix with the matrix output by the convolution layer as the input of a normalization layer, and use the normalization layer to output a site directed topology map, the nodes in the directed topology map include: feature dimensions of historical meteorological information and power generation information;

[0014] The directed topological graph input module is used to input the directed topological graph into the trained graph neural network, and obtain the regional photovoltaic prediction result according to the result output by the graph neural network.

[0015] In a third aspect, an embodiment of the present invention further provides a server, the server comprising:

[0016] one or more processors;

[0017] a storage device for storing one or more programs,

[0018] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the regional photovoltaic prediction methods provided in the above embodiments.

[0019] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer executable instructions, wherein the computer executable instructions, when executed by a computer processor, are used to execute the regional photovoltaic prediction method provided in the above embodiment.

[0020] The regional photovoltaic prediction method, device, server and storage medium provided by the embodiment of the present invention cluster the historical data of each site in the region to obtain a site clustering matrix; the site clustering matrix and the historical data of each site are input into the trained graph convolutional neural network; the graph convolutional neural network includes: a self-attention unit, which uses the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and fuses the attention score matrix with the matrix output by the convolution layer, and uses the normalization layer to output a site directed topological graph, and the nodes in the directed topological graph include: the characteristic dimensions of historical meteorological information and power generation information; the directed topological graph is input into the trained graph neural network, and the regional photovoltaic prediction result is obtained according to the output of the graph neural network. Using the above method, the association relationship between each site can be fully utilized, and by processing its relationship, its spatial characteristics and the correlation between the meteorological impact on it can be fully utilized, and the self-attention mechanism is introduced so that each node can selectively focus on certain specific nodes when aggregating neighbor information. Multiple layers of graph convolutional layers are stacked, and each layer captures the relationship between nodes of different scales. The output of each layer is used as the input of the next layer to build a graph neural network model for regional photovoltaic power prediction. This not only improves the efficiency of regional photovoltaic power prediction, but also improves the accuracy of prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0022] Figure 1 is a flow chart of a regional photovoltaic prediction method provided in Embodiment 1 of the present invention;

[0023] Figure 2 is a flow chart of a regional photovoltaic prediction method provided by Embodiment 2 of the present invention;

[0024] Figure 3 is a schematic diagram of the structure of a regional photovoltaic prediction device provided in Embodiment 3 of the present invention;

[0025] Figure 4 It is a schematic diagram of the structure of the server provided in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0026] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0027] Embodiment 1

[0028] Figure 1 : is a flow chart of a regional photovoltaic prediction method provided in Embodiment 1 of the present invention. This embodiment is applicable to the case of efficient prediction of regional photovoltaics. The method can be executed by a regional photovoltaic prediction device, and specifically includes the following steps:

[0029] Step 110, clustering the historical data of each site in the area to obtain a site clustering matrix.

[0030] For example, the fuzzy C-means clustering algorithm can be used to cluster the historical actual data of each station in the region, obtain the membership of each station to each category, and form a station-category membership matrix. Through the above clustering, different single stations are divided into different categories by weight. The station-category membership matrix is ​​used to assist in the later connection between the stations and improve the accuracy of the model prediction.

[0031] Step 120: input the site clustering matrix and the historical data of each site into the trained graph convolutional neural network.

[0032] Graph convolutional neural networks design a method to extract features from graph data, so that we can use these features to perform node classification, graph classification, and edge prediction on graph data.

[0033] In this embodiment, the trained graph convolutional neural network can be used to generate a regional photovoltaic site map. For example, a corresponding regional site attribute matrix can be established for historical data, and it can be input into the trained graph convolutional neural network together with the site clustering matrix.

[0034] Step 130, the graph convolutional neural network includes: a self-attention unit, using the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and the attention score matrix is ​​fused with the matrix output by the convolution layer, and the two are used as the input of the normalization layer, and the normalization layer is used to output a site directed topology map, and the nodes in the directed topology map include: feature dimensions of historical meteorological information and power generation information;

[0035] In this embodiment, a self-attention mechanism is introduced in constructing a graph structure. Based on the clustering results, each site can be regarded as a node in the graph. The connection weights between nodes are based on the self-attention mechanism, and the strength of the relationship between nodes is captured by calculating the attention score between each node and all other nodes in the same category. There is no need to calculate the attention score of each node with all nodes, thereby reducing the amount of calculation and improving the model capability. Exemplarily, the graph convolutional neural network includes: a self-attention unit, which uses the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the strength of the relationship between nodes. In the graph convolutional neural network, the features of the nodes of the initial input are continuously extracted through multiple layers of convolution, and the matrix output in the last convolutional layer is fused with the attention score matrix output by the self-attention unit, and together they are used as the input of a normalized layer, and a directed topological graph of all sites in the output area is used by a normalized layer.

[0036] Step 140: input the directed topological graph into the trained graph neural network, and obtain the regional photovoltaic prediction result according to the output result of the graph neural network.

[0037] Graph neural networks use neural networks to learn graph structure data, extract and discover features and patterns in graph structure data, and achieve prediction results. Using the trained graph neural network, by inputting the directed topological graph obtained in the above steps, the corresponding prediction results are obtained.

[0038] Exemplarily, an optimization algorithm (Adam) can be used for model training; mean square error (MSE) is used as the loss function; the best hyperparameters (including learning rate, number of hidden layers, number of neurons) are selected through Bayesian optimization; to prevent overfitting, the data is divided into training set, validation set and test set, and EarlyStopping is used on the validation set to ensure the generalization ability of the model.

[0039] Furthermore, the above model can be corrected and optimized in real time. A real-time data stream system is established to continuously receive satellite cloud image data. Combined with the irradiance data estimated by real-time satellite cloud images, the weights of the neural network are gradually updated using online learning technology to ensure the adaptability of the prediction model over time and environmental changes.

[0040] In this embodiment, the historical data of each site in the region is clustered to obtain a site clustering matrix; the site clustering matrix and the historical data of each site are input into the trained graph convolutional neural network; the graph convolutional neural network includes: a self-attention unit, which uses the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and the attention score matrix is ​​fused with the matrix output by the convolution layer, which are used as the input of the normalization layer, and the normalization layer is used to output a site directed topological graph, and the nodes in the directed topological graph include: the characteristic dimensions of historical meteorological information and power generation information; the directed topological graph is input into the trained graph neural network, and the regional photovoltaic prediction result is obtained according to the output of the graph neural network. Using the above method, the association relationship between each site can be fully utilized, and by processing its relationship, its spatial characteristics and the correlation between the impact of meteorology on it can be fully utilized, and the self-attention mechanism is introduced so that each node can selectively focus on certain specific nodes when aggregating neighbor information. Multiple layers of graph convolutional layers are stacked, and each layer captures the relationship between nodes of different scales. The output of each layer is used as the input of the next layer to construct a graph neural network model for regional photovoltaic power prediction. It not only improves the efficiency of regional photovoltaic power forecasting, but also improves the accuracy of the forecast.

[0041] In a preferred implementation of this embodiment, before clustering the historical data of each site in the region, the method may further add the following steps: standardize the historical data collected at each site to ensure that each feature is on the same scale. Further, the following steps may be added: collect the longitude, latitude and altitude of each photovoltaic site; collect the historical actual data of each site; collect meteorological forecast data from different meteorological sources that are closely related to the actual photovoltaic power; use K nearest neighbor interpolation to process missing values ​​in the data set, and use the boundary value method to replace abnormal values ​​with boundary values. Using the above method, the data can be preprocessed to further improve the accuracy of photovoltaic predictions at regional sites.

[0042] Embodiment 2

[0043] Figure 2It is a flow chart of the regional photovoltaic prediction method provided by the second embodiment of the present invention. This embodiment is optimized based on the above embodiment, and the historical data of each site is specifically optimized as the total irradiance of the slope. And the following steps are added: according to the geographical location and meteorological parameters of the photovoltaic power station, the solar zenith angle and azimuth within the time range of the data set are calculated; based on the calculated zenith angle, the total irradiance, direct irradiance and scattered irradiance of the horizontal surface under the clear sky state are estimated; the total irradiance on the inclined surface of the photovoltaic panel is estimated by combining the tilt angle, orientation and clear sky total irradiance, direct irradiance, scattered irradiance at the location of the photovoltaic panel, and the zenith angle and azimuth of the sun. The predicted irradiance on the photovoltaic panel is obtained by using the total irradiance of the clear sky on the horizontal plane and the total irradiance of the inclined surface of the photovoltaic panel, and combining the horizontal total irradiance of the meteorological forecast in the data set; determine the tilt angle and orientation of the photovoltaic panel when the correlation between the actual photovoltaic power and the predicted irradiance is the highest; weighted processing is performed on the predicted irradiance of the photovoltaic panel provided by different meteorological sources to obtain the total irradiance of the slope.

[0044] See also Figure 2 , the regional photovoltaic prediction method comprises:

[0045] Step 210, according to the geographical location and meteorological parameters of the photovoltaic power station, calculate the solar zenith angle and azimuth within the time range of the data set, and based on the calculated zenith angle, estimate the total irradiance, direct irradiance and diffuse irradiance of the horizontal surface under clear sky conditions.

[0046] In the existing photovoltaic prediction technology, the influence of the tilt angle is not fully considered. Since photovoltaic panels are usually installed at an angle, the actual received irradiance is significantly different from the total irradiance on the horizontal plane. This leads to insufficient accuracy when the total irradiance on the horizontal plane is used directly for prediction, especially when the amount of data is less than one year, the annual trend cannot be reflected, resulting in greater deviation. In addition, the existing slope conversion method does not fully consider meteorological factors such as temperature and air pressure. The above reasons are superimposed, resulting in a decrease in the accuracy of photovoltaic prediction. Therefore, it is necessary to accurately calculate the irradiance. For example, it is necessary to calculate the solar zenith angle and azimuth within a certain period of time for each site's geographical location and meteorological parameters, and based on the zenith angle, the total irradiance, direct irradiance and scattered irradiance of the horizontal plane under clear sky conditions are calculated. Optionally, the NREL SPA algorithm can be used to calculate the solar zenith angle and azimuth within the time range of the data set based on the geographical location (longitude, latitude, altitude) and meteorological parameters (pressure, temperature) of the photovoltaic power station.

[0047] Step 220, estimate the total irradiance on the inclined surface of the photovoltaic panel by combining the tilt angle, orientation and clear sky total irradiance, direct irradiance, diffuse irradiance at its location, and the zenith angle and azimuth angle of the sun.

[0048] Based on the specific configuration of the photovoltaic panel, i.e., the orientation, tilt angle and other parameters, the clear sky total irradiance, direct irradiance, and diffuse irradiance are calculated, and the total irradiance on the inclined surface of the photovoltaic panel is estimated based on the zenith angle and azimuth of the sun. Optionally, the Haurwitz model can be used to infer the total irradiance, direct irradiance and diffuse irradiance of the horizontal surface under clear sky conditions based on the calculated zenith angle.

[0049] Step 230, using the total irradiance of the clear sky on the horizontal plane and the total irradiance of the inclined surface of the photovoltaic panel, combined with the total horizontal irradiance of the weather forecast in the data set, the predicted irradiance on the photovoltaic panel is obtained.

[0050] Various meteorological parameters related to irradiance in meteorological data are used, such as air pressure, temperature, and weather. The total horizontal irradiance is obtained by combining various meteorological forecasts, and the predicted irradiance on the photovoltaic panel is obtained based on the total horizontal irradiance combined with the position, angle and other characteristics of the photovoltaic panel. For example, in order to solve the problem of unknown tilt angle and orientation of photovoltaic panels in practical applications, a particle swarm optimization algorithm can be used to determine the tilt angle and orientation of photovoltaic panels when the correlation between actual photovoltaic power and predicted irradiance is the highest. For example, the Klucher model can be used to estimate the total irradiance on the inclined surface of the photovoltaic panel.

[0051] Step 240, determining the photovoltaic panel tilt angle and orientation when the correlation between the actual photovoltaic power and the predicted irradiance is the highest, performing weighted processing on the photovoltaic panel predicted irradiance provided by different meteorological sources, and obtaining the total irradiance of the slope.

[0052] Furthermore, the predicted irradiance of photovoltaic panels provided by different meteorological sources can be weighted to reduce the impact of data fluctuations from a single meteorological source.

[0053] Step 250, clustering the historical data of each site in the area to obtain a site clustering matrix, and inputting the site clustering matrix and the historical data of each site into the trained graph convolutional neural network.

[0054] Step 260, the graph convolutional neural network includes: a self-attention unit, which uses the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and fuses the attention score matrix with the matrix output by the convolution layer as the input of a normalization layer, and uses the normalization layer to output a site directed topology map, and the nodes in the directed topology map include: feature dimensions of historical meteorological information and power generation information.

[0055] Step 270: input the directed topological graph into the trained graph neural network, and obtain the regional photovoltaic prediction result according to the result output by the graph neural network.

[0056] This embodiment optimizes the historical data of each site into the total irradiance of the slope. And add the following steps: according to the geographical location and meteorological parameters of the photovoltaic power station, calculate the solar zenith angle and azimuth within the time range of the data set; based on the calculated zenith angle, calculate the total irradiance, direct irradiance and scattered irradiance of the horizontal surface under clear sky conditions; combine the tilt angle, orientation of the photovoltaic panel and the total irradiance, direct irradiance, scattered irradiance of the clear sky at its position, as well as the zenith angle and azimuth of the sun to estimate the total irradiance on the inclined surface of the photovoltaic panel. Using the total irradiance of the clear sky on the horizontal plane and the total irradiance of the inclined surface of the photovoltaic panel, combined with the total horizontal irradiance of the meteorological forecast in the data set, the predicted irradiance on the photovoltaic panel is obtained; determine the tilt angle and orientation of the photovoltaic panel when the correlation between the actual photovoltaic power and the predicted irradiance is the highest; weight the predicted irradiance of the photovoltaic panel provided by different meteorological sources to obtain the total irradiance of the slope. The total irradiance on the inclined surface of the photovoltaic panel can be accurately calculated through the solar irradiance slope conversion technology, which solves the irradiance difference problem caused by the tilt angle of the photovoltaic panel and improves the accuracy of the prediction. In addition, the predicted irradiance of the photovoltaic panel provided by different meteorological sources can be weighted to reduce the impact of fluctuations in data from a single meteorological source and improve the stability of the prediction results.

[0057] Embodiment 3

[0058] Figure 3 is a schematic diagram of the structure of the regional photovoltaic prediction device provided by the third embodiment of the present invention, see Figure 3 , the regional photovoltaic prediction device comprises:

[0059] A clustering module 310 is used to cluster the historical data of each site in the region to obtain a site clustering matrix;

[0060] An input module 320, used to input the site clustering matrix and the historical data of each site into the trained graph convolutional neural network;

[0061] A directed topology map output module 330 is used to use the self-attention unit included in the graph convolutional neural network to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and fuse the attention score matrix with the matrix output by the convolution layer as the input of a normalization layer, and use the normalization layer to output a site directed topology map, where the nodes in the directed topology map include: feature dimensions of historical meteorological information and power generation information;

[0062] The directed topological graph input module 340 is used to input the directed topological graph into the trained graph neural network, and obtain the regional photovoltaic prediction result according to the result output by the graph neural network.

[0063] The regional photovoltaic prediction device provided in this embodiment obtains a site clustering matrix by clustering the historical data of each site in the region; the site clustering matrix and the historical data of each site are input into the trained graph convolutional neural network; the graph convolutional neural network includes: a self-attention unit, which uses the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and fuses the attention score matrix with the matrix output by the convolution layer, and uses the normalization layer to output a site directed topological graph, and the nodes in the directed topological graph include: the characteristic dimensions of historical meteorological information and power generation information; the directed topological graph is input into the trained graph neural network, and the regional photovoltaic prediction result is obtained according to the output of the graph neural network. Using the above method, the association relationship between each site can be fully utilized, and by processing its relationship, its spatial characteristics and the correlation between the impact of meteorology on it can be fully utilized, and the self-attention mechanism is introduced so that each node can selectively focus on certain specific nodes when aggregating neighbor information. Multiple layers of graph convolutional layers are stacked, and each layer captures the relationship between nodes of different scales. The output of each layer is used as the input of the next layer to build a graph neural network model for regional photovoltaic power prediction. This not only improves the efficiency of regional photovoltaic power prediction, but also improves the accuracy of prediction.

[0064] Based on the above embodiments, the clustering module includes:

[0065] The use unit is used to cluster the historical actual data of each station in the region by using the fuzzy C-means clustering algorithm, obtain the membership degree of each station to each category, and form a station clustering matrix.

[0066] Based on the above embodiments, the device further includes:

[0067] The standardization processing module is used to standardize the historical data collected at each site to ensure that all features are at the same scale.

[0068] Based on the above embodiments, the device further includes:

[0069] The first collection module is used to collect the longitude, latitude and altitude of each photovoltaic site;

[0070] The second collection module is used to collect the historical actual data of each site;

[0071] The third collection module is used to collect meteorological forecast data from different meteorological sources that are closely related to the actual photovoltaic power generation;

[0072] The replacement module is used to process the missing values ​​in the data set using K nearest neighbor interpolation and replace the outliers with boundary values ​​using the boundary value method.

[0073] Based on the above embodiments, the historical data of each site includes: total irradiance of the slope.

[0074] Based on the above embodiments, the device further includes:

[0075] The zenith angle and azimuth angle calculation module is used to calculate the solar zenith angle and azimuth angle within the time range of the data set according to the geographical location and meteorological parameters of the photovoltaic power station;

[0076] The irradiance estimation module is used to estimate the total irradiance, direct irradiance and diffuse irradiance of the horizontal surface under clear sky conditions based on the calculated zenith angle;

[0077] The total irradiance estimation module is used to estimate the total irradiance on the inclined surface of the photovoltaic panel by combining the tilt angle, orientation and total clear sky irradiance, direct irradiance, diffuse irradiance at its location, as well as the zenith angle and azimuth angle of the sun.

[0078] Based on the above embodiments, the device further includes:

[0079] The module is used to obtain the predicted irradiance on the photovoltaic panel by using the total irradiance of the clear sky on the horizontal plane and the total irradiance on the inclined surface of the photovoltaic panel, and combining the total horizontal irradiance of the meteorological forecast in the data set;

[0080] A determination module, used to determine the tilt angle and orientation of the photovoltaic panel when the correlation between the actual photovoltaic power and the predicted irradiance is the highest;

[0081] The weighted processing module is used to perform weighted processing on the predicted irradiance of photovoltaic panels provided by different meteorological sources to obtain the total irradiance of the slope.

[0082] The regional photovoltaic prediction device provided in the embodiment of the present invention can execute the regional photovoltaic prediction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0083] Embodiment 4

[0084] Figure 4 A schematic diagram of the structure of a server provided in Embodiment 4 of the present invention. Figure 4 A block diagram of an exemplary server 12 suitable for use in implementing embodiments of the present invention is shown. Figure 4 The server 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0085] like Figure 4As shown, the server 12 is in the form of a general purpose computing device. The components of the server 12 may include, but are not limited to: one or more processors or processing units 16, server memory 28, and a bus 18 connecting different server components (including server memory 28 and processing unit 16).

[0086] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0087] The server 12 typically includes a variety of computer server readable media. These media can be any available media that can be accessed by the server 12, including volatile and non-volatile media, removable and non-removable media.

[0088] Server memory 28 may include computer server readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache 32. Server 12 may further include other removable / non-removable, volatile / non-volatile computer server storage media. By way of example only, storage server 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 4 not shown, usually called a "hard drive"). Although Figure 4 Not shown in the figure, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The server memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present invention.

[0089] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the server memory 28, such program modules 42 including, but not limited to, operating a server, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described herein.

[0090] The server 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), may communicate with one or more devices that enable a user to interact with the server 12, and / or may communicate with any device that enables the server 12 to communicate with one or more other computing devices (e.g., network cards, modems, etc.). Such communication may be performed via an input / output (I / O) interface 22. In addition, the server 12 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the server 12 via a bus 18. It should be understood that, although not shown in the figure, the server 12 may be assembled to use other hardware and / or software modules, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID servers, tape drives, and data backup storage servers, etc.

[0091] The processing unit 16 executes various functional applications and data processing by running the programs stored in the server memory 28 , such as implementing the regional photovoltaic prediction method provided in the embodiment of the present invention.

[0092] Embodiment 5

[0093] Embodiment 5 of the present invention further provides a storage medium comprising computer executable instructions, wherein the computer executable instructions, when executed by a computer processor, are used to execute any of the regional photovoltaic prediction methods provided in the above embodiments.

[0094] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0095] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0096] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0097] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or device. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0098] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A regional photovoltaic prediction method, characterized in that: include: Cluster the historical data of each site in the region to obtain a site clustering matrix; Inputting the site clustering matrix and the historical data of each site into the trained graph convolutional neural network; The graph convolutional neural network includes: a self-attention unit, which uses the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and fuses the attention score matrix with the matrix output by the convolution layer as the input of a normalization layer, and uses the normalization layer to output a site directed topology map, and the nodes in the directed topology map include: feature dimensions of historical meteorological information and power generation information; The directed topological graph is input into a trained graph neural network, and a regional photovoltaic prediction result is obtained according to the output of the graph neural network.

2. The method according to claim 1, characterized in that The historical data of each site in the region is clustered to obtain a site clustering matrix, including: The fuzzy C-means clustering algorithm is used to cluster the historical actual data of each station in the region, and the membership degree of each station to each category is obtained to form a station clustering matrix.

3. The method according to claim 1, characterized in that: Before clustering the historical data of each site in the region, the method further includes: The historical data collected at each site were standardized to ensure that all features were at the same scale.

4. The method according to claim 3, characterized in that: Before the standardization process of the collected historical data of each site, the method further includes: Collect the longitude, latitude and altitude of each PV site; Collect historical actual data of each site; Collect weather forecast data from different meteorological sources that are closely related to the actual PV power generation; K nearest neighbor interpolation is used to process missing values ​​in the data set, and the boundary value method is used to replace outliers with boundary values.

5. The method according to claim 1, characterized in that The historical data of each site includes: total irradiance on the slope.

6. The method according to claim 5, characterized in that The method further comprises: According to the geographical location of the photovoltaic power plant and meteorological parameters, the solar zenith angle and azimuth angle within the time range of the data set are calculated; Based on the calculated zenith angle, the total irradiance, direct irradiance and diffuse irradiance on the horizontal surface under clear sky conditions are estimated; The total irradiance on the inclined surface of the photovoltaic panel is estimated by combining the tilt angle and orientation of the photovoltaic panel, the total clear sky irradiance, direct irradiance, diffuse irradiance at its location, and the zenith angle and azimuth of the sun.

7. The method according to claim 6, characterized in that The method further comprises: The predicted irradiance on the photovoltaic panel is obtained by using the total irradiance of the clear sky on the horizontal plane and the total irradiance of the inclined surface of the photovoltaic panel, combined with the horizontal total irradiance of the meteorological forecast in the dataset; Determine the tilt angle and orientation of the photovoltaic panels when the actual photovoltaic power is most correlated with the predicted irradiance; The predicted irradiance of photovoltaic panels provided by different meteorological sources is weighted to obtain the total irradiance of the slope.

8. A regional photovoltaic prediction device, characterized in that: include: The clustering module is used to cluster the historical data of each site in the region to obtain a site clustering matrix; An input module, used for inputting the site clustering matrix and the historical data of each site into the trained graph convolutional neural network; A directed topology map output module, used to use the self-attention unit included in the graph convolutional neural network, use the self-attention unit to calculate the attention score matrix between each node and all other nodes in the same category to capture the relationship strength between nodes, and fuse the attention score matrix with the matrix output by the convolution layer, and use the normalization layer to output a site directed topology map, and the nodes in the directed topology map include: feature dimensions of historical meteorological information and power generation information; The directed topological graph input module is used to input the directed topological graph into the trained graph neural network, and obtain the regional photovoltaic prediction result according to the result output by the graph neural network.

9. A server, characterized in that: The server comprises: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the regional photovoltaic prediction method as described in any one of claims 1-8.

10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to execute the regional photovoltaic prediction method as described in any one of claims 1-8 when executed by a computer processor.

Images