A method, storage medium and device for predicting ultra-short-term power of regional distributed photovoltaic aggregation

Abstract

The present invention discloses a method, storage medium and device for predicting ultra-short-term power of regional distributed photovoltaic aggregation. The method comprises: obtaining historical power data of each distributed photovoltaic power station in a distributed photovoltaic cluster and meteorological data of the area where the distributed photovoltaic cluster is located; using a correlation analysis method to determine photovoltaic output influencing factors from the meteorological data; inputting the meteorological data and historical power data corresponding to the photovoltaic output influencing factors into a pre-built prediction model to obtain power prediction values ​​of each distributed photovoltaic power station; based on the loss coefficient and the power prediction value of each distributed photovoltaic power station, obtaining the ultra-short-term power prediction value of regional distributed photovoltaic aggregation; the loss coefficient is obtained according to the topological relationship of each distributed photovoltaic power station in the distributed photovoltaic cluster. The present invention takes into account the spatial correlation between each distributed photovoltaic power station, and can improve the accuracy of distributed photovoltaic aggregation ultra-short-term prediction.

Description

Technical Field

[0001] The invention relates to a method, storage medium and equipment for predicting ultra-short-term power of regional distributed photovoltaic aggregation, and belongs to the technical field of photovoltaic power prediction. Background Art

[0002] In recent years, under the background of energy transformation, distributed photovoltaic power generation has become an important new energy power generation method. Distributed photovoltaic power generation has the characteristics of high randomness, strong volatility and obvious intermittency. After large-scale access to the power grid, it may lead to a decline in power quality, an increase in harmonics and power supply reliability problems, posing severe challenges to the safe and stable operation of the power grid. Therefore, accurately predicting the power of distributed photovoltaic power generation systems and providing accurate photovoltaic power generation power prediction data for dispatching can achieve optimal dispatch planning and regulation.

[0003] Most of the existing distributed photovoltaic cluster power prediction methods use the prediction results of a single distributed photovoltaic as the smallest unit, and calculate the regional prediction results through the accumulation method or statistical upscaling method. They face the following problems that need to be solved urgently: Problem 1. Since the output characteristics of distributed photovoltaic power stations under the same geographical and climatic environment are similar, the historical power of each site will have a certain correlation with the future output of the adjacent sites, that is, there is a certain spatial coupling relationship, which leads to the existing method of predicting and accumulating a single photovoltaic power station or sub-region separately, which will increase the prediction error; Problem 2. The existing method does not consider the distributed photovoltaic cluster as an aggregated whole. The line loss of the aggregated power when it is outputting power to the outside under the actual network constraints leads to a decrease in the accuracy of the prediction results. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method, device, storage medium and equipment for regional distributed photovoltaic aggregation ultra-short-term power prediction, which takes into account the spatial correlation between distributed photovoltaic power stations and can improve the accuracy of distributed photovoltaic aggregation ultra-short-term prediction. To achieve the above purpose, the present invention is implemented by the following technical solutions:

[0005] In a first aspect, the present invention provides a method for predicting ultra-short-term power of regional distributed photovoltaic aggregation, comprising:

[0006] Obtain the historical power data of each distributed photovoltaic power station in the distributed photovoltaic cluster and the meteorological data of the area where the distributed photovoltaic cluster is located;

[0007] Adopt correlation analysis method to determine the factors affecting photovoltaic output from the acquired meteorological data;

[0008] The meteorological data and historical power data corresponding to the factors affecting photovoltaic output are input into the pre-built prediction model to obtain the power prediction value of each distributed photovoltaic power station; based on the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster and the power prediction value of each distributed photovoltaic power station, the regional distributed photovoltaic aggregation ultra-short-term power prediction value is obtained;

[0009] The loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster is obtained by the following steps:

[0010] According to the topological relationship of each distributed photovoltaic power station in the distributed photovoltaic cluster, a topological structure diagram is obtained; according to the topological structure diagram, an aggregation point of the distributed photovoltaic cluster is determined;

[0011] Based on the grid parameters of each distributed photovoltaic power station in the distributed photovoltaic cluster, the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster is calculated.

[0012] In combination with the first aspect, optionally, the using of a correlation analysis method to determine photovoltaic output influencing factors from the acquired meteorological data includes:

[0013] Initialize various meteorological factors in meteorological data;

[0014] Calculate the The correlation coefficient between each meteorological factor and photovoltaic power is calculated by the following formula:

[0015] ,

[0016] In the formula, Indicates The correlation coefficient between meteorological factors and photovoltaic power is Indicates The meteorological factor Sample values, Indicates The sample mean of the meteorological factors, Indicates The power value of each sample, represents the power mean of all samples, Indicates the number of meteorological factors;

[0017] The calculated correlation coefficients are sorted by correlation degree to obtain a correlation degree sequence, and the meteorological factors corresponding to the top n correlation coefficients in the correlation degree sequence are determined as the factors affecting photovoltaic output.

[0018] In combination with the first aspect, optionally, the step of inputting meteorological data and historical power data corresponding to photovoltaic output influencing factors into a pre-built prediction model to obtain power prediction values ​​of each distributed photovoltaic power station includes:

[0019] Through the feature matrix Summarize distributed photovoltaic clusters The meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of each distributed photovoltaic power station, the characteristic matrix It is expressed as:

[0020] ,

[0021] In the formula, Indicates the meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of the first distributed photovoltaic power station, The meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of the second distributed photovoltaic power station are shown. Indicates Meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of each distributed photovoltaic power station;

[0022] The feature matrix Input the pre-built prediction model to obtain the power prediction value of each distributed photovoltaic power station, which is expressed by the following formula:

[0023] ,

[0024] In the formula, Indicates the distributed photovoltaic cluster The topological structure diagram of a distributed photovoltaic power station. Indicates The characteristic matrix of the step , Indicates The characteristic matrix of the step , is the spatiotemporal graph convolutional neural network function; Indicates The power prediction value of each distributed photovoltaic power station is Indicates The power prediction value of each distributed photovoltaic power station is calculated.

[0025] In combination with the first aspect, optionally, based on the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster and the power prediction value of each distributed photovoltaic power station, the regional distributed photovoltaic aggregation ultra-short-term power prediction value is obtained, which is expressed by the following formula:

[0026] ,

[0027] In the formula, represents the regional distributed photovoltaic aggregate ultra-short-term power forecast value, Indicates The power prediction value of a distributed photovoltaic power station, Represents the total number of distributed photovoltaic power stations in the distributed photovoltaic cluster, Indicates The loss coefficient between each distributed photovoltaic power station and the aggregation point.

[0028] In combination with the first aspect, optionally, the pre-built prediction model includes a feature input layer, a spatial feature extraction layer, a temporal feature extraction layer and a prediction output layer;

[0029] The feature matrix Input the feature input layer;

[0030] The spatial feature extraction layer uses the graph convolutional network GCN to perform spatial correlation modeling to extract the spatial features of the topological structure. The propagation formula of the graph convolutional network GCN is as follows:

[0031] ,

[0032] In the formula, Indicates The graph convolutional network features of the layer, Indicates The graph convolutional network features of the layer, represents the adjacency matrix in the topological graph, for Added to the identity matrix I, for The degree matrix of Indicates The graph convolutional network parameters of the layer, Represents the Sigmoid activation function;

[0033] The spatial feature extraction layer uses a two-layer graph convolutional network GCN to capture spatial dependencies, which is expressed as follows:

[0034] ,

[0035] In the formula, Represents the feature matrix , represents the adjacency matrix in the topological graph, Represents the spatiotemporal graph convolutional neural network function used by the graph convolutional network GCN, Represents the feature matrix Adjacency matrix in topological graph The spatial dependence of Represents the preprocessing result. ; represents the weight matrix of the first layer, represents the weight matrix of the second layer; ReLU is the ReLU activation function, which is used to map all negative values ​​to 0, while positive values ​​remain unchanged. Represents the Sigmoid activation function;

[0036] The temporal feature extraction layer uses a recurrent neural network GRU to model temporal correlation, which is expressed by the following formula:

[0037] ,

[0038] ,

[0039] ,

[0040] ,

[0041] In the formula, express The door of renewal at all times, represents the weight of the update gate, represents the bias of the update gate, represents the graph convolution process, express The feature matrix at time, express Output at the moment;

[0042] express The reset door of the moment, represents the weight of the reset gate, Indicates the deviation of the reset gate;

[0043] express The candidate state at the moment, represents the weight of the candidate state, represents the deviation of the candidate state;

[0044] express Output at the moment;

[0045] The prediction output layer outputs the power prediction value of each distributed photovoltaic power station, which is expressed by the following formula:

[0046] ,

[0047] In the formula, Indicates that a single distributed photovoltaic power station The power prediction value at the time, represents the weight of the prediction output layer, Represents the bias of the predicted output layer.

[0048] In combination with the first aspect, optionally, the topological structure diagram is obtained according to the topological relationship of each distributed photovoltaic power station in the distributed photovoltaic cluster, which is represented by the following formula:

[0049] ,

[0050] In the formula, Represents a topological structure diagram; Represents a node set on the topological structure diagram, and the node represents each distributed photovoltaic power station in the distributed photovoltaic cluster. , Indicates the first distributed photovoltaic power station, Indicates the second distributed photovoltaic power station, Indicates Distributed photovoltaic power stations, Represents the total number of distributed photovoltaic power stations in the distributed photovoltaic cluster; It represents the set of edges connecting the nodes on the graph, where the edges represent the power lines connecting the distributed photovoltaic power stations; Represents the adjacency matrix in the topological structure graph, which is expressed by the following formula:

[0051] ,

[0052] In the formula, express The elements in .

[0053] In combination with the first aspect, optionally, the aggregation point of the distributed photovoltaic cluster is determined as follows: a node to which the power flow ultimately flows and transmits power to an external power grid is used as the aggregation point.

[0054] In combination with the first aspect, optionally, the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster is calculated based on the grid parameters of each distributed photovoltaic power station in the distributed photovoltaic cluster, including:

[0055] Calculate the active power of each distributed photovoltaic power station using the following formula:

[0056] ,

[0057] In the formula, Indicates The active power of a distributed photovoltaic power station, For the The voltage amplitude of a distributed photovoltaic power station, Indicates The voltage amplitude of a distributed photovoltaic power station, Represents the total number of distributed photovoltaic power stations in the distributed photovoltaic cluster; It is A distributed photovoltaic power station and The real part of the admittance matrix between distributed photovoltaic power stations represents the conductance, It is A distributed photovoltaic power station and The imaginary part of the admittance matrix between distributed photovoltaic power stations represents the electrical susceptance; Indicates A distributed photovoltaic power station and The voltage phase angle difference between distributed photovoltaic power stations;

[0058] Calculate the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster by the following formula:

[0059] ,

[0060] In the formula, Represents the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster, Indicates The active power of a distributed photovoltaic power station, Indicates the first The power of a distributed photovoltaic power station.

[0061] In a second aspect, the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon. When the computer program / instruction is executed by a processor, the steps of the regional distributed photovoltaic aggregation ultra-short-term power prediction method described in the first aspect are implemented.

[0062] In a third aspect, the present invention provides a computer device, characterized in that it includes:

[0063] Memory, for storing computer programs / instructions;

[0064] A processor is used to execute the computer program / instructions to implement the steps of the regional distributed photovoltaic aggregation ultra-short-term power prediction method described in the first aspect.

[0065] Compared with the prior art, the method, storage medium and device for predicting ultra-short-term power of regional distributed photovoltaic aggregation provided by the embodiment of the present invention have the following beneficial effects:

[0066] The present invention inputs meteorological data and historical power data corresponding to factors affecting photovoltaic output into a pre-built prediction model to obtain power prediction values ​​of each distributed photovoltaic power station; based on the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster and the power prediction value of each distributed photovoltaic power station, the regional distributed photovoltaic aggregation ultra-short-term power prediction value is obtained; based on the grid parameters of each distributed photovoltaic power station in the distributed photovoltaic cluster, the present invention calculates the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster; the present invention takes into account the line loss of the aggregation prediction under the actual network constraints, and can improve the accuracy of the distributed photovoltaic aggregation ultra-short-term prediction;

[0067] The present invention obtains a topological structure diagram according to the topological relationship of each distributed photovoltaic power station in a distributed photovoltaic cluster; according to the topological structure diagram, the aggregation point of the distributed photovoltaic cluster is determined; through the topological structure diagram, the present invention can simultaneously predict the power of multiple distributed photovoltaic power stations and obtain the aggregated power. Compared with the existing research method of modeling a single photovoltaic power station or sub-region separately and then accumulating them, the spatial coupling relationship of the output of regional distributed photovoltaic power stations is considered, and the spatial correlation between each distributed photovoltaic power station is considered, thereby reducing the error of cluster prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a schematic flow chart of a method for predicting ultra-short-term power of regional distributed photovoltaic aggregation in Embodiment 1 of the present invention;

[0069] Figure 2 It is a structural schematic diagram of a prediction model in a method for regional distributed photovoltaic aggregation ultra-short-term power prediction in Example 1 of the present invention. DETAILED DESCRIPTION

[0070] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0071] Embodiment 1:

[0072] like Figure 1 As shown, this embodiment provides a method for predicting ultra-short-term power of regional distributed photovoltaic aggregation, including:

[0073] Obtain the historical power data of each distributed photovoltaic power station in the distributed photovoltaic cluster and the meteorological data of the area where the distributed photovoltaic cluster is located;

[0074] Adopt correlation analysis method to determine the factors affecting photovoltaic output from the acquired meteorological data;

[0075] The meteorological data and historical power data corresponding to the factors affecting photovoltaic output are input into the pre-built prediction model to obtain the power prediction value of each distributed photovoltaic power station; based on the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster and the power prediction value of each distributed photovoltaic power station, the regional distributed photovoltaic aggregation ultra-short-term power prediction value is obtained.

[0076] The loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster is obtained by the following steps:

[0077] According to the topological relationship of each distributed photovoltaic power station in the distributed photovoltaic cluster, a topological structure diagram is obtained;

[0078] According to the topological structure diagram, determining the aggregation point of the distributed photovoltaic cluster;

[0079] Based on the grid parameters of each distributed photovoltaic power station in the distributed photovoltaic cluster, the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster is calculated.

[0080] The specific implementation steps are as follows.

[0081] Step 1: Obtain the topological relationship of each distributed photovoltaic power station in the distributed photovoltaic cluster and the grid parameters of each distributed photovoltaic power station.

[0082] Step 2: Obtain the topological structure diagram and determine the aggregation point of the distributed photovoltaic cluster.

[0083] Step 2.1: According to the topological relationship of each distributed photovoltaic power station in the distributed photovoltaic cluster, a topological structure diagram is obtained.

[0084] It is expressed by the following formula:

[0085] ,

[0086] In the formula, Represents a topology diagram.

[0087] In the formula, It represents a set of nodes on the topological structure diagram, and the nodes represent each distributed photovoltaic power station in the distributed photovoltaic cluster. , Indicates the first distributed photovoltaic power station, Indicates the second distributed photovoltaic power station, Indicates Distributed photovoltaic power stations, Represents the total number of distributed photovoltaic power stations in the distributed photovoltaic cluster.

[0088] In the formula, It represents the set of edges connecting the nodes on the graph, and the edges represent the power lines connecting the distributed photovoltaic power stations.

[0089] In the formula, Represents the adjacency matrix in the topological structure graph, which is expressed by the following formula:

[0090] ,

[0091] In the formula, express The elements in .

[0092] Step 2.2: According to the topological structure diagram, determine the aggregation point of the distributed photovoltaic cluster.

[0093] The aggregation point of the distributed photovoltaic cluster is determined as the node where the power flow finally flows and transmits power to the external power grid.

[0094] Step 3: Based on the grid parameters of each distributed photovoltaic power station in the distributed photovoltaic cluster, the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster is calculated.

[0095] Step 3.1: Calculate the active power of each distributed photovoltaic power station using the following formula:

[0096] ,

[0097] In the formula, Indicates The active power of a distributed photovoltaic power station, For the The voltage amplitude of a distributed photovoltaic power station, Indicates The voltage amplitude of a distributed photovoltaic power station, Represents the total number of distributed photovoltaic power stations in the distributed photovoltaic cluster; It is A distributed photovoltaic power station and The real part of the admittance matrix between distributed photovoltaic power stations represents the conductance, It is A distributed photovoltaic power station and The imaginary part of the admittance matrix between distributed photovoltaic power stations represents the electrical susceptance; Indicates A distributed photovoltaic power station and The voltage phase angle difference between distributed photovoltaic power stations.

[0098] Step 3.2: Calculate the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster by the following formula:

[0099] ,

[0100] In the formula, Represents the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster, Indicates The active power of a distributed photovoltaic power station, Indicates the first The power of a distributed photovoltaic power station.

[0101] Step 4: Obtain the historical power data of each distributed photovoltaic power station in the distributed photovoltaic cluster and the meteorological data of the area where the distributed photovoltaic cluster is located.

[0102] In this embodiment, the meteorological data of the area where the distributed photovoltaic cluster is located includes temperature, air pressure, wind speed, humidity, and irradiance.

[0103] In this embodiment, meteorological data is collected through a distributed photovoltaic cluster or provided by a meteorological bureau.

[0104] Step 5: Use correlation analysis method to determine the factors affecting photovoltaic output from the acquired meteorological data.

[0105] Step 5.1: Initialize various meteorological factors in the meteorological data.

[0106] Step 5.2: Calculate the The correlation coefficient between each meteorological factor and photovoltaic power is calculated by the following formula:

[0107] ,

[0108] In the formula, Indicates The correlation coefficient between meteorological factors and photovoltaic power is Indicates The meteorological factor Sample values, Indicates The sample mean of the meteorological factors, Indicates The power value of each sample, represents the power mean of all samples, Indicates the number of meteorological factors.

[0109] Step 5.3: Sort the calculated correlation coefficients by correlation degree to obtain a correlation degree sequence, and determine the meteorological factors corresponding to the top n correlation coefficients in the correlation degree sequence as the factors affecting photovoltaic output.

[0110] Photovoltaic output is mainly affected by meteorological factors. Selecting too few meteorological factors will lead to large fluctuations in the prediction results and excessive reliance on a single indicator for accuracy. Selecting too many meteorological factors will make model training complicated and invalid indicators will interfere with the prediction results. Therefore, selecting appropriate influencing indicators will increase prediction accuracy and model training efficiency.

[0111] It should be noted that the value of n is determined according to actual conditions. In this embodiment, the value of n is 4.

[0112] In this embodiment, historical output, temperature, humidity, and irradiance are selected as photovoltaic output influencing factors.

[0113] This embodiment can select photovoltaic output influencing factors with a high degree of correlation by calculating the correlation coefficient.

[0114] Step 6: Obtain the regional distributed photovoltaic aggregation ultra-short-term power forecast value.

[0115] Step 6.1: Input the meteorological data and historical power data corresponding to the factors affecting photovoltaic output into the pre-built prediction model to obtain the power prediction value of each distributed photovoltaic power station.

[0116] Step 6.1.1: Through the feature matrix Summarize distributed photovoltaic clusters The meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of each distributed photovoltaic power station, the characteristic matrix It is expressed as:

[0117] ,

[0118] In the formula, Indicates the meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of the first distributed photovoltaic power station, The meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of the second distributed photovoltaic power station are shown. Indicates The meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of each distributed photovoltaic power station.

[0119] In this embodiment, , is the historical power data, is the temperature data, is the humidity data, The irradiance data.

[0120] Step 6.1.2: Convert the feature matrix Input the pre-built prediction model to obtain the power prediction value of each distributed photovoltaic power station, which is expressed by the following formula:

[0121] ,

[0122] In the formula, Indicates the distributed photovoltaic cluster The topological structure diagram of a distributed photovoltaic power station. Indicates The characteristic matrix of the step , Indicates The characteristic matrix of the step , is the spatiotemporal graph convolutional neural network function; Indicates The power prediction value of each distributed photovoltaic power station is Indicates The power prediction value of each distributed photovoltaic power station is calculated.

[0123] It should be noted that this embodiment studies the ultra-short-term power prediction of regional distributed photovoltaic power stations, and predicts the power for the next hour every 5 minutes. Therefore, the historical time step is set to 48 and the output time step is set to 12.

[0124] like Figure 2 As shown, the pre-built prediction model includes a feature input layer, a spatial feature extraction layer, a temporal feature extraction layer and a prediction output layer.

[0125] The feature input layer is used to input the feature matrix .

[0126] In this embodiment, the feature matrix include .

[0127] like Figure 2 As shown, the spatial feature extraction layer uses the graph convolutional network GCN to model spatial correlation to extract the spatial features of the topological structure. The propagation formula of the graph convolutional network GCN is as follows:

[0128] ,

[0129] In the formula, Indicates The graph convolutional network features of the layer, Indicates The graph convolutional network features of the layer, represents the adjacency matrix in the topological graph, for Added to the identity matrix I, for The degree matrix of Indicates The graph convolutional network parameters of the layer, Represents the Sigmoid activation function.

[0130] Specifically, the spatial feature extraction layer uses a two-layer graph convolutional network GCN to capture spatial dependencies, which is expressed as follows:

[0131] ,

[0132] In the formula, Represents the feature matrix , represents the adjacency matrix in the topological graph, Represents the spatiotemporal graph convolutional neural network function used by the graph convolutional network GCN, Represents the feature matrix Adjacency matrix in topological graph The spatial dependence of Represents the preprocessing result. ; represents the weight matrix of the first layer, represents the weight matrix of the second layer; ReLU is the ReLU activation function, which is used to map all negative values ​​to 0, while positive values ​​remain unchanged. Represents the Sigmoid activation function.

[0133] like Figure 2 As shown, the time feature extraction layer uses the recurrent neural network GRU to model the time correlation, which is expressed by the following formula:

[0134] ,

[0135] ,

[0136] ,

[0137] ,

[0138] In the formula, express The door of renewal at all times, represents the weight of the update gate, represents the bias of the update gate, represents the graph convolution process, express The feature matrix at time, express Output at the moment;

[0139] In the formula, express The reset door of the moment, represents the weight of the reset gate, Indicates the deviation of the reset gate;

[0140] In the formula, express The candidate state at the moment, represents the weight of the candidate state, represents the deviation of the candidate state;

[0141] In the formula, express Output at the moment.

[0142] The prediction output layer outputs the power prediction value of each distributed photovoltaic power station, which is expressed by the following formula:

[0143] ,

[0144] In the formula, Indicates that a single distributed photovoltaic power station The power prediction value at the time, represents the weight of the prediction output layer, Represents the bias of the predicted output layer.

[0145] Step 6.2: Based on the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster and the power prediction value of each distributed photovoltaic power station, the regional distributed photovoltaic aggregation ultra-short-term power prediction value is obtained.

[0146] It is expressed by the following formula:

[0147] ,

[0148] In the formula, represents the regional distributed photovoltaic aggregate ultra-short-term power forecast value, Indicates The power prediction value of a distributed photovoltaic power station, Represents the total number of distributed photovoltaic power stations in the distributed photovoltaic cluster, Indicates The loss coefficient between each distributed photovoltaic power station and the aggregation point.

[0149] This embodiment can simultaneously predict the power of multiple distributed photovoltaic power stations and obtain the aggregated power through the topological structure diagram. Compared with the existing research method of modeling a single photovoltaic power station or sub-region separately and then accumulating them, it takes into account the spatial coupling relationship of the output of regional distributed photovoltaic power stations and the spatial correlation between each distributed photovoltaic power station, thereby reducing the error of cluster prediction.

[0150] This embodiment takes into account the line loss of the aggregation prediction under the actual network constraints, which can improve the accuracy of the ultra-short-term prediction of distributed photovoltaic aggregation.

[0151] Embodiment 2:

[0152] This embodiment provides a computer-readable storage medium having a computer program / instruction stored thereon, characterized in that when the computer program / instruction is executed by a processor, the steps of the regional distributed photovoltaic aggregation ultra-short-term power prediction method described in Example 1 are implemented.

[0153] Embodiment 3:

[0154] This embodiment provides a computer device, characterized by comprising:

[0155] Memory, for storing computer programs / instructions;

[0156] A processor is used to execute the computer program / instructions to implement the steps of the regional distributed photovoltaic aggregation ultra-short-term power prediction method described in Example 1.

[0157] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0158] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0159] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0160] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0161] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.

Claims

1. A method for predicting ultra-short-term power of regional distributed photovoltaic aggregation, characterized in that: include: Obtain the historical power data of each distributed photovoltaic power station in the distributed photovoltaic cluster and the meteorological data of the area where the distributed photovoltaic cluster is located; Adopt correlation analysis method to determine the factors affecting photovoltaic output from the acquired meteorological data; The meteorological data and historical power data corresponding to the factors affecting photovoltaic output are input into the pre-built prediction model to obtain the power prediction value of each distributed photovoltaic power station; based on the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster and the power prediction value of each distributed photovoltaic power station, the regional distributed photovoltaic aggregation ultra-short-term power prediction value is obtained; The loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster is obtained by the following steps: According to the topological relationship of each distributed photovoltaic power station in the distributed photovoltaic cluster, a topological structure diagram is obtained; according to the topological structure diagram, an aggregation point of the distributed photovoltaic cluster is determined; Based on the grid parameters of each distributed photovoltaic power station in the distributed photovoltaic cluster, the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster is calculated, including: Calculate the active power of each distributed photovoltaic power station using the following formula: ， In the formula, Indicates The active power of a distributed photovoltaic power station, For the The voltage amplitude of a distributed photovoltaic power station, Indicates The voltage amplitude of a distributed photovoltaic power station, Represents the total number of distributed photovoltaic power stations in the distributed photovoltaic cluster; It is A distributed photovoltaic power station and The real part of the admittance matrix between distributed photovoltaic power stations represents the conductance, It is A distributed photovoltaic power station and The imaginary part of the admittance matrix between distributed photovoltaic power stations represents the electrical susceptance; Indicates A distributed photovoltaic power station and The voltage phase angle difference between distributed photovoltaic power stations; Calculate the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster by the following formula: ， In the formula, Represents the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster, Indicates The active power of a distributed photovoltaic power station, Indicates the first The power of a distributed photovoltaic power station.

2. The method for regional distributed photovoltaic aggregation ultra-short-term power prediction according to claim 1 is characterized in that: The correlation analysis method is used to determine the factors affecting photovoltaic output from the acquired meteorological data, including: Initialize various meteorological factors in meteorological data; Calculate the The correlation coefficient between each meteorological factor and photovoltaic power is calculated by the following formula: ， In the formula, Indicates The correlation coefficient between meteorological factors and photovoltaic power is Indicates The meteorological factor Sample values, Indicates The sample mean of the meteorological factors, Indicates The power value of each sample, represents the power mean of all samples, Indicates the number of meteorological factors; The calculated correlation coefficients are sorted by correlation degree to obtain a correlation degree sequence, and the meteorological factors corresponding to the top n correlation coefficients in the correlation degree sequence are determined as the factors affecting photovoltaic output.

3. The method for regional distributed photovoltaic aggregation ultra-short-term power prediction according to claim 1 is characterized in that: The meteorological data and historical power data corresponding to the factors affecting photovoltaic output are input into a pre-built prediction model to obtain the power prediction value of each distributed photovoltaic power station, including: Through the feature matrix Summarize distributed photovoltaic clusters The meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of each distributed photovoltaic power station, the characteristic matrix It is expressed as: ， In the formula, Indicates the meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of the first distributed photovoltaic power station, The meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of the second distributed photovoltaic power station are shown. Indicates Meteorological data and historical power data corresponding to the factors affecting the photovoltaic output of each distributed photovoltaic power station; The feature matrix Input the pre-built prediction model to obtain the power prediction value of each distributed photovoltaic power station, which is expressed by the following formula: ， In the formula, Indicates the distributed photovoltaic cluster The topological structure diagram of a distributed photovoltaic power station. Indicates The characteristic matrix of the step , Indicates The characteristic matrix of the step , is the spatiotemporal graph convolutional neural network function; Indicates The power prediction value of each distributed photovoltaic power station is Indicates The power prediction value of each distributed photovoltaic power station is calculated.

4. The method for regional distributed photovoltaic aggregation ultra-short-term power prediction according to claim 1 is characterized in that: Based on the loss coefficient between each distributed photovoltaic power station and the aggregation point in the distributed photovoltaic cluster and the power prediction value of each distributed photovoltaic power station, the regional distributed photovoltaic aggregation ultra-short-term power prediction value is obtained, which is expressed by the following formula: ， In the formula, represents the regional distributed photovoltaic aggregate ultra-short-term power forecast value, Indicates The power prediction value of a distributed photovoltaic power station, Represents the total number of distributed photovoltaic power stations in the distributed photovoltaic cluster, Indicates The loss coefficient between each distributed photovoltaic power station and the aggregation point.

5. The method for regional distributed photovoltaic aggregation ultra-short-term power prediction according to claim 3 is characterized in that: The pre-built prediction model includes a feature input layer, a spatial feature extraction layer, a temporal feature extraction layer and a prediction output layer; The feature matrix Input the feature input layer; The spatial feature extraction layer uses the graph convolutional network GCN to perform spatial correlation modeling to extract the spatial features of the topological structure. The propagation formula of the graph convolutional network GCN is as follows: ， In the formula, Indicates The graph convolutional network features of the layer, Indicates The graph convolutional network features of the layer, represents the adjacency matrix in the topological graph, for Added to the identity matrix I, for The degree matrix of Indicates The graph convolutional network parameters of the layer, Represents the Sigmoid activation function; The spatial feature extraction layer uses a two-layer graph convolutional network GCN to capture spatial dependencies, which is expressed as follows: ， In the formula, Represents the feature matrix , represents the adjacency matrix in the topological graph, Represents the spatiotemporal graph convolutional neural network function used by the graph convolutional network GCN, Represents the feature matrix Adjacency matrix in topological graph The spatial dependence of Represents the preprocessing result. ; represents the weight matrix of the first layer, represents the weight matrix of the second layer; ReLU is the ReLU activation function, which is used to map all negative values ​​to 0, while positive values ​​remain unchanged. Represents the Sigmoid activation function; The temporal feature extraction layer uses a recurrent neural network GRU to model temporal correlation, which is expressed by the following formula: ， ， ， ， In the formula, express The door of renewal at all times, represents the weight of the update gate, represents the bias of the update gate, represents the graph convolution process, express The feature matrix at time, express Output at the moment; express The reset door of the moment, represents the weight of the reset gate, Indicates the deviation of the reset gate; express The candidate status at the moment, represents the weight of the candidate state, represents the deviation of the candidate state; express Output at the moment; The prediction output layer outputs the power prediction value of each distributed photovoltaic power station, which is expressed by the following formula: ， In the formula, Indicates that a single distributed photovoltaic power station The power prediction value at the time, represents the weight of the prediction output layer, Represents the bias of the predicted output layer.

6. The method for regional distributed photovoltaic aggregation ultra-short-term power prediction according to claim 1 is characterized in that: According to the topological relationship of each distributed photovoltaic power station in the distributed photovoltaic cluster, a topological structure diagram is obtained, which is represented by the following formula: ， In the formula, Represents a topological structure diagram; Represents a node set on the topological structure diagram, and the node represents each distributed photovoltaic power station in the distributed photovoltaic cluster. , Indicates the first distributed photovoltaic power station, Indicates the second distributed photovoltaic power station, Indicates Distributed photovoltaic power stations, Represents the total number of distributed photovoltaic power stations in the distributed photovoltaic cluster; It represents the set of edges connecting the nodes on the graph, where the edges represent the power lines connecting the distributed photovoltaic power stations; Represents the adjacency matrix in the topological structure graph, which is expressed by the following formula: ， In the formula, express The elements in .

7. The method for regional distributed photovoltaic aggregation ultra-short-term power prediction according to claim 1 is characterized in that: The aggregation point of the distributed photovoltaic cluster is determined as follows: the node to which the power flow finally flows and transmits power to the external power grid is used as the aggregation point.

8. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the steps of the method for regional distributed photovoltaic aggregation ultra-short-term power prediction described in any one of claims 1-7 are implemented.

9. A computer device, characterized in that: include: Memory, for storing computer programs / instructions; A processor for executing the computer program / instructions to implement the steps of the regional distributed photovoltaic aggregation ultra-short-term power prediction method described in any one of claims 1-7.

Images