Power distribution network multi-cascade distributed photovoltaic output prediction method and device

Abstract

The invention relates to a power distribution network multi-cascade distributed photovoltaic output prediction method and device, and the method comprises the steps: building a power distribution network multi-cascade distributed photovoltaic connection distribution relation data sample library; based on distributed photovoltaic historical related data, in combination with a distributed photovoltaic connection distribution relation data sample library, establishing a power distribution network multi-cascade distributed photovoltaic data sample library; based on the multi-cascade distributed photovoltaic data sample library, analyzing multi-cascade distributed photovoltaic output influence characteristics, and establishing an influence characteristic library for different cascade distributed photovoltaic; an LSTM model is constructed based on the influence feature library, data in the multi-cascade distributed photovoltaic data sample library is adopted for training, and a distribution network distributed photovoltaic multi-cascade output prediction model is obtained; and inputting the input parameter data of the to-be-predicted time period into the power distribution network distributed photovoltaic multi-cascade output prediction model to obtain a power distribution network multi-cascade distributed photovoltaic output prediction result. According to the invention, the distributed photovoltaic output condition of each step of the power distribution network can be accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed photovoltaic power station output prediction, and in particular to a method and device for predicting multi-level distributed photovoltaic output in a distribution network. Background Art

[0002] As energy depletion and energy conservation and environmental protection issues become increasingly serious, distributed photovoltaic power generation has been used more and more widely. With the increasing penetration rate of distributed photovoltaics in the distribution network, while alleviating energy tension and environmental deterioration, the intermittent and volatile nature of photovoltaic power generation also poses great challenges to the safe, reliable and economical operation of the distribution network. Accurate distributed photovoltaic output forecasting can predict future photovoltaic power generation, provide scientific decision-making basis for automatic power generation control and grid dispatching of distribution networks, effectively improve the utilization efficiency of distributed photovoltaics, effectively reduce the impact of high penetration of distributed photovoltaics on distribution networks, improve power supply reliability, and ensure safe, reliable and economical operation of distribution networks.

[0003] At present, distributed photovoltaic output prediction methods can be divided into physical methods and statistical prediction methods according to the prediction principle. The physical method is to use physical models such as solar radiation transfer equation, solar position model, photovoltaic cell model and inverter efficiency model to establish a prediction algorithm to predict power generation. The statistical prediction method conducts statistical analysis on historical operation data to find out the inherent laws between photovoltaic power generation and various influencing factors, establish a prediction algorithm to predict power generation. As a new school of statistical prediction, artificial intelligence prediction methods have outstanding performance in distributed photovoltaic prediction. With the development of various artificial intelligence algorithms, they have been initially applied in the field of distributed photovoltaic prediction.

[0004] Distributed photovoltaics have an impact on multiple levels of the distribution network. In order to avoid the multi-faceted impact of high-penetration distributed photovoltaics on the distribution network, it is necessary to predict the output of multi-level distributed photovoltaics in the distribution network. This is a technical problem that needs to be urgently solved by technical personnel in this field. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a method and device for predicting the output of multi-level distributed photovoltaic power in a distribution network, which can accurately predict the output of distributed photovoltaic power at each level of the distribution network and predict the output characteristics of distributed photovoltaic power in advance.

[0006] The technical solution adopted by the present invention to solve the technical problem is: to provide a method for predicting the output of multi-level distributed photovoltaic power in a distribution network, comprising the following steps:

[0007] Establish a sample database of the distribution relationship data of multi-level distributed photovoltaic connections in the distribution network;

[0008] Based on the historical data related to distributed photovoltaics, combined with the data sample library of multi-level distributed photovoltaic connection distribution relationship of the distribution network, a data sample library of multi-level distributed photovoltaics in the distribution network is established;

[0009] Based on the multi-level distributed photovoltaic data sample library of the distribution network, the impact characteristics of the multi-level distributed photovoltaic output are analyzed, and an impact characteristic library for distributed photovoltaics of different levels is established;

[0010] Building an LSTM model based on the impact feature library, and training the LSTM model using data in the distribution network multi-level distributed photovoltaic data sample library to obtain a distribution network distributed photovoltaic multi-level output prediction model;

[0011] The input parameter data of the time period to be predicted is input into the distribution network distributed photovoltaic multi-level output prediction model to obtain the distribution network multi-level distributed photovoltaic output prediction result.

[0012] The establishment of a distribution network multi-level distributed photovoltaic connection distribution relationship data sample library specifically includes:

[0013] Obtain distribution network topology data and distributed photovoltaic user account information;

[0014] Based on the grid topology data and distributed photovoltaic user ledger information, the connection relationship between photovoltaic users, substations, feeders, busbars, substations and regions is analyzed and verified;

[0015] Based on the obtained connection relationships among photovoltaic users, substations, feeders, busbars, substations and regions, a distribution network multi-level distributed photovoltaic connection distribution relationship data sample library is established.

[0016] The distributed photovoltaic user ledger information includes: photovoltaic user ID, longitude and latitude location, altitude, installed capacity, grid connection type, meter type, installation inclination, component area and voltage level.

[0017] The distributed photovoltaic historical data is combined with the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library to establish a distribution network multi-level distributed photovoltaic data sample library, specifically including:

[0018] Obtain the historical distributed photovoltaic power generation active power data and the corresponding historical micro-meteorological element real-time data;

[0019] According to the correlation between photovoltaic users in the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library, the missing values ​​of the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data are completed;

[0020] A multi-level distributed photovoltaic data sample library for distribution network is established based on the completed distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data.

[0021] The method of completing missing values ​​of the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data according to the correlation between photovoltaic users in the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library specifically includes:

[0022] Screening the missing values ​​of the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data;

[0023] Calculate the correlation between photovoltaic users in the same period according to the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library;

[0024] The data samples of photovoltaic users that are most relevant to the photovoltaic users with missing data are selected, and the data of the photovoltaic users with missing data are completed by using the correlation.

[0025] The method of analyzing the impact characteristics of multi-level distributed photovoltaic output based on the distribution network multi-level distributed photovoltaic data sample library and establishing an impact characteristic library for distributed photovoltaics at different levels specifically includes:

[0026] For distributed photovoltaics at different levels, based on different time scales and different spatial scales in the multi-level distributed photovoltaic data sample library of the distribution network, the influence characteristics of distributed photovoltaic output are extracted by correlation analysis method, and multiple common influence characteristics are eliminated;

[0027] The importance of influencing features after eliminating multiple common influencing features is evaluated through the random forest algorithm, and an influencing feature set is selected for distributed photovoltaics at different levels to build an influencing feature library for distributed photovoltaics at different levels.

[0028] The technical solution adopted by the present invention to solve the technical problem is: to provide a multi-level distributed photovoltaic output prediction device for a distribution network, comprising:

[0029] The first establishment module is used to establish a data sample library of multi-level distributed photovoltaic connection distribution relationship of the distribution network;

[0030] The second establishment module is used to establish a distribution network multi-level distributed photovoltaic data sample library based on the distributed photovoltaic historical related data and in combination with the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library;

[0031] The third establishment module is used to analyze the impact characteristics of multi-level distributed photovoltaic output based on the multi-level distributed photovoltaic data sample library of the distribution network, and establish an impact characteristic library for distributed photovoltaics of different levels;

[0032] Constructing a training module for constructing an LSTM model based on the impact feature library, and using the data in the distribution network multi-level distributed photovoltaic data sample library to train the LSTM model to obtain a distribution network distributed photovoltaic multi-level output prediction model;

[0033] The prediction module is used to input the input parameter data of the time period to be predicted into the distribution network distributed photovoltaic multi-level output prediction model to obtain the distribution network multi-level distributed photovoltaic output prediction result.

[0034] The technical solution adopted by the present invention to solve its technical problem is: to provide an electronic device, including a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for predicting the multi-level distributed photovoltaic output of the distribution network when executing the computer program.

[0035] The technical solution adopted by the present invention to solve its technical problem is: providing a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned distribution network multi-level distributed photovoltaic output prediction method are implemented.

[0036] Beneficial Effects

[0037] Due to the adoption of the above-mentioned technical scheme, compared with the prior art, the present invention has the following advantages and positive effects: the present invention constructs a multi-level distributed photovoltaic impact feature based on the multi-level distributed photovoltaic connection distribution relationship and the multi-level distributed photovoltaic data, and then constructs a distribution network multi-level distributed photovoltaic output prediction model based on the multi-level distributed photovoltaic impact feature. The obtained prediction model can accurately predict the distributed photovoltaic output of each level of the distribution network, thereby predicting the output characteristics of distributed photovoltaics in advance, so as to support the optimized scheduling of the multi-level operation of the distribution network and improve the intelligent operation level of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flow chart of a method for predicting multi-level distributed photovoltaic output in a distribution network according to a first embodiment of the present invention;

[0039] Figure 2 It is a schematic diagram of the LSTM model structure in the first embodiment of the present invention. DETAILED DESCRIPTION

[0040] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0041] The first embodiment of the present invention relates to a method for predicting multi-level distributed photovoltaic output in a distribution network, wherein the multiple levels in the method include user level, substation level, feeder level, bus level, substation level and regional level, wherein the regional level may be a county, city, province or a specific identification area. The method first establishes a sample library of data on the distribution relationship of multi-level distributed photovoltaic connection in the distribution network; then, the missing values ​​of the historical data of power generation power of all distributed photovoltaic users in the distribution network and the corresponding historical data of micro-meteorological elements are cleaned and screened, and the missing values ​​are processed to establish a photovoltaic data sample library; then, based on a large amount of sample data in the photovoltaic data sample library, the impact characteristics of multi-level distributed photovoltaic output are analyzed, and an impact feature library for distributed photovoltaics of different levels is established; thereafter, based on a deep learning algorithm, a multi-level distributed photovoltaic output prediction model is established through training with a large amount of sample data for the impact characteristics of multi-level distributed photovoltaic output; finally, the input feature data of the demand prediction period is input to obtain the prediction results of multi-level distributed photovoltaic output of users, substations, feeders, busbars, substations and regions in the distribution network. Figure 1 As shown, the specific steps include:

[0042] Step 1: Establish a sample database of distribution relationship data of multi-level distributed photovoltaic connections in distribution networks.

[0043] This step specifically includes: obtaining the distribution network topology data and distributed photovoltaic user ledger information; wherein the distributed photovoltaic user ledger information includes photovoltaic user ID, longitude and latitude location, altitude, installed capacity, grid connection type, meter type, installation inclination, component area and voltage level, etc. Based on the grid topology data and distributed photovoltaic user ledger information, the connection relationship between photovoltaic users, substations, feeders, buses, substations and regions is analyzed and verified. According to the obtained connection relationship between photovoltaic users, substations, feeders, buses, substations and regions, a distribution network multi-level distributed photovoltaic connection distribution relationship data sample library is established.

[0044] Step 2: Based on the distributed photovoltaic historical related data and in combination with the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library, a distribution network multi-level distributed photovoltaic data sample library is established.

[0045] This step specifically includes: obtaining distributed photovoltaic historical power generation active power data and corresponding historical micro-meteorological element real-time data; wherein the micro-meteorological element real-time data includes total radiation irradiance, temperature, relative humidity, wind speed, wind direction, air pressure, rainfall, cloud cover, weather type, etc. According to the correlation between photovoltaic users in the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library, the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data are supplemented with missing values. According to the supplemented distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data, a distribution network multi-level distributed photovoltaic data sample library is established.

[0046] When filling missing values, if the data gap period is short, the interpolation method can be used directly to fill the gap. When the data gap period is long, the correlation method can be used to fill the gap. Among them, the correlation method specifically includes:

[0047] Screening the missing values ​​of the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data;

[0048] Calculate the correlation between photovoltaic users in the same period according to the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library;

[0049] The data samples of photovoltaic users that are most relevant to the photovoltaic users with missing data are selected, and the data of the photovoltaic users with missing data are completed by using the correlation.

[0050] After completing the missing data, this implementation can also normalize all the data to unify the basic units of measurement so that the subsequent training of the neural network model can converge faster. After the normalization is completed, the normalized data is used to establish a multi-level distributed photovoltaic data sample library for the distribution network.

[0051] Step 3: Based on the multi-level distributed photovoltaic data sample library of the distribution network, analyze the impact characteristics of the multi-level distributed photovoltaic output and establish an impact characteristic library for distributed photovoltaics of different levels. This step specifically includes:

[0052] For distributed photovoltaics of different levels, based on the different time scales and different spatial scales in the multi-level distributed photovoltaic data sample library of the distribution network, the influence characteristics of distributed photovoltaic output are extracted by correlation analysis, and multiple common influence characteristics are eliminated. By eliminating multiple common influence characteristics, the dimension of the influence characteristics is reduced to speed up the training speed of the subsequent neural network model.

[0053] The importance of influencing features after eliminating multiple common influencing features is evaluated through the random forest algorithm, and an influencing feature set is selected for distributed photovoltaics at different levels to build an influencing feature library for distributed photovoltaics at different levels.

[0054] Step 4: construct an LSTM model based on the impact feature library, and use the data in the distribution network multi-level distributed photovoltaic data sample library to train the LSTM model to obtain a distribution network distributed photovoltaic multi-level output prediction model.

[0055] In this step, the LSTM model is an advanced version of the recurrent neural network (RNN). Unlike the RNN, the RNN receives all data, while the LSTM model receives data selectively and can selectively store information.

[0056] like Figure 2 As shown in the figure, the LSTM model has three gate units: (1) Input gate: determines whether there will be information input to the memory unit at this moment; (2) Output gate: determines whether there will be information output from the memory unit at each moment. (3) Forget gate: determines whether the value in the memory unit is forgotten at each moment. If it is opened, the value in the memory unit will be cleared, that is, forgotten.

[0057] The LSTM model has three main stages:

[0058] (1) Forgetting stage: Selectively forget the input passed by the previous node. The specific calculation is as follows:

[0059] Forget gate output f t :f t =σ(W hf H t-1 +W xf x t +b f )

[0060] f stands for forget, which acts as a forget gate to control whether the value of the previous state needs to be forgotten or remembered for the next operation.

[0061] (2) Selective memory stage: The input of this stage is selectively memorized.

[0062] Input gate output i t :i t =σ(W hi H t-1 +W xi x t +b i )

[0063] Temporary status

[0064] Update temporary status Get the new unit state C t :

[0065] (3) Output stage: Determine the current output. The new unit in the previous stage is transformed through a tanh activation function to obtain the output h t .h t-1 and x t Are the two inputs of the output gate, and the calculation formula is as follows:

[0066] o t =σ(W xo H t-1 +W xo x t +b0)

[0067] Where σ() is the activation function, which is usually a sigmoid function. The values ​​obtained after the activation function are between [0, 1], 1 means the door is fully open, and 0 means it is closed; W f , W i , W C , W o The weights corresponding to the forget gate, input gate, memory unit, and output gate; b f , b i , b C , b o is the corresponding bias vector, Represents the product operation; Represents addition operation.

[0068] The LSTM model is an improvement on RNN, with the goal of improving the long-term data dependency problem of RNN. In response to the serious gradient vanishing problem of RNN, the long-term memory functional unit of the LSTM model enables better screening of data and information, and its gate structure facilitates the network to reasonably output information when a large amount of data flows into the network.

[0069] This implementation method selects an influencing feature set for different levels of distributed photovoltaics from the influencing feature library, uses a large amount of sample data to train and test the LSTM model, and obtains suitable parameters. After the training is completed, a distribution network distributed photovoltaic multi-level output prediction algorithm model can be obtained.

[0070] Step 5, input the input parameter data of the time period to be predicted into the distribution network distributed photovoltaic multi-level output prediction model to obtain the distribution network multi-level distributed photovoltaic output prediction result, that is, obtain the distribution network user-station-feeder-bus-substation-region multi-level distributed photovoltaic output prediction result.

[0071] It is not difficult to find that the present invention constructs a multi-level distributed photovoltaic impact feature based on the multi-level distributed photovoltaic connection distribution relationship and multi-level distributed photovoltaic data, and then constructs a distribution network multi-level distributed photovoltaic output prediction model based on the multi-level distributed photovoltaic impact feature. The obtained prediction model can accurately predict the distributed photovoltaic output of each level of the distribution network, thereby predicting the output characteristics of distributed photovoltaics in advance, so as to support the optimized scheduling of multi-level operation of the distribution network and improve the intelligent operation level of the distribution network.

[0072] The second embodiment of the present invention relates to a multi-level distributed photovoltaic output prediction device for a distribution network, comprising:

[0073] The first establishment module is used to establish a data sample library of multi-level distributed photovoltaic connection distribution relationship of the distribution network;

[0074] The second establishment module is used to establish a distribution network multi-level distributed photovoltaic data sample library based on the distributed photovoltaic historical related data and in combination with the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library;

[0075] The third establishment module is used to analyze the impact characteristics of multi-level distributed photovoltaic output based on the multi-level distributed photovoltaic data sample library of the distribution network, and establish an impact characteristic library for distributed photovoltaics of different levels;

[0076] Constructing a training module for constructing an LSTM model based on the impact feature library, and using the data in the distribution network multi-level distributed photovoltaic data sample library to train the LSTM model to obtain a distribution network distributed photovoltaic multi-level output prediction model;

[0077] The prediction module is used to input the input parameter data of the time period to be predicted into the distribution network distributed photovoltaic multi-level output prediction model to obtain the distribution network multi-level distributed photovoltaic output prediction result.

[0078] The first establishment module includes:

[0079] The first acquisition unit is used to acquire the distribution network topology data and distributed photovoltaic user account information;

[0080] An analysis unit, used for analyzing and verifying the connection relationship between photovoltaic users, substations, feeders, buses, substations and regions based on the grid topology data and distributed photovoltaic user ledger information;

[0081] The first establishing unit is used to establish a distribution network multi-level distributed photovoltaic connection distribution relationship data sample library according to the obtained connection relationship between photovoltaic users, substations, feeders, busbars, substations and regions.

[0082] The distributed photovoltaic user ledger information includes: photovoltaic user ID, longitude and latitude location, altitude, installed capacity, grid connection type, meter type, installation inclination, component area and voltage level.

[0083] The second establishing module comprises:

[0084] The second acquisition unit is used to acquire the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data;

[0085] A completion unit, configured to complete missing values ​​of the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data according to the correlation between photovoltaic users in the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library;

[0086] The second establishing unit is used to establish a multi-level distributed photovoltaic data sample library for the distribution network according to the completed distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data.

[0087] The completion unit comprises:

[0088] A screening subunit, used for screening the missing values ​​of the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data;

[0089] A calculation subunit, used for calculating the correlation between photovoltaic users in the same period according to the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library;

[0090] The completion subunit is used to select data samples of photovoltaic users that are most relevant to the photovoltaic users with missing data, and use the correlation to complete the data of the photovoltaic users with missing data.

[0091] The third building module comprises:

[0092] An extraction unit is used to extract the influencing features of distributed photovoltaic output by a correlation analysis method for distributed photovoltaics of different levels based on different time scales and different spatial scales in the multi-level distributed photovoltaic data sample library of the distribution network, and eliminate multiple common influencing features;

[0093] A construction unit is used to evaluate the importance of influencing features after eliminating multiple common influencing features through a random forest algorithm, select influencing feature sets for distributed photovoltaics at different levels, and construct an influencing feature library for distributed photovoltaics at different levels.

[0094] A third embodiment of the present invention relates to an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for predicting multi-level distributed photovoltaic output in a distribution network of the first embodiment when executing the computer program.

[0095] A fourth embodiment of the present invention relates to a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for predicting multi-level distributed photovoltaic output in a distribution network of the first embodiment.

[0096] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0097] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate 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.

[0098] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction method, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0099] 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 1The steps for the functions specified in one or more boxes.

[0100] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for predicting the output of multi-level distributed photovoltaic power in a distribution network, characterized in that: The following steps are involved: Establish a sample database of the distribution relationship data of multi-level distributed photovoltaic connections in the distribution network; Based on the historical data related to distributed photovoltaics, combined with the data sample library of multi-level distributed photovoltaic connection distribution relationship of the distribution network, a data sample library of multi-level distributed photovoltaics in the distribution network is established; Based on the multi-level distributed photovoltaic data sample library of the distribution network, the impact characteristics of the multi-level distributed photovoltaic output are analyzed, and an impact characteristic library for distributed photovoltaics of different levels is established; An LSTM model is constructed based on the impact feature library, and the LSTM model is trained using the data in the distribution network multi-level distributed photovoltaic data sample library to obtain a distribution network distributed photovoltaic multi-level output prediction model; the input parameter data of the time period to be predicted is input into the distribution network distributed photovoltaic multi-level output prediction model to obtain a distribution network multi-level distributed photovoltaic output prediction result.

2. The method for predicting the output of multi-level distributed photovoltaic power in a distribution network according to claim 1, characterized in that: The establishment of a distribution network multi-level distributed photovoltaic connection distribution relationship data sample library specifically includes: Obtain distribution network topology data and distributed photovoltaic user account information; Based on the grid topology data and distributed photovoltaic user ledger information, the connection relationship between photovoltaic users, substations, feeders, busbars, substations and regions is analyzed and verified; Based on the obtained connection relationships among photovoltaic users, substations, feeders, busbars, substations and regions, a distribution network multi-level distributed photovoltaic connection distribution relationship data sample library is established.

3. The method for predicting the output of multi-level distributed photovoltaic power in a distribution network according to claim 2, characterized in that: The distributed photovoltaic user ledger information includes: photovoltaic user ID, longitude and latitude location, altitude, installed capacity, grid connection type, meter type, installation inclination, component area and voltage level.

4. The method for predicting the output of multi-level distributed photovoltaic power in a distribution network according to claim 1, characterized in that: The distributed photovoltaic historical data is combined with the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library to establish a distribution network multi-level distributed photovoltaic data sample library, specifically including: Obtain the historical distributed photovoltaic power generation active power data and the corresponding historical micro-meteorological element real-time data; According to the correlation between photovoltaic users in the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library, the missing values ​​of the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data are completed; A multi-level distributed photovoltaic data sample library for distribution network is established based on the completed distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data.

5. The method for predicting the output of multi-level distributed photovoltaic power in a distribution network according to claim 4, characterized in that: The method of completing missing values ​​of the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data according to the correlation between photovoltaic users in the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library specifically includes: Screening the missing values ​​of the distributed photovoltaic historical power generation active power data and the corresponding historical micro-meteorological element real-time data; Calculate the correlation between photovoltaic users in the same period according to the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library; The data samples of photovoltaic users that are most relevant to the photovoltaic users with missing data are selected, and the data of the photovoltaic users with missing data are completed by using the correlation.

6. The method for predicting the output of multi-level distributed photovoltaic power in a distribution network according to claim 1, characterized in that: The method of analyzing the impact characteristics of multi-level distributed photovoltaic output based on the distribution network multi-level distributed photovoltaic data sample library and establishing an impact characteristic library for distributed photovoltaics at different levels specifically includes: For distributed photovoltaics at different levels, based on different time scales and different spatial scales in the multi-level distributed photovoltaic data sample library of the distribution network, the influence characteristics of distributed photovoltaic output are extracted by correlation analysis method, and multiple common influence characteristics are eliminated; The importance of influencing features after eliminating multiple common influencing features is evaluated through the random forest algorithm, and an influencing feature set is selected for distributed photovoltaics at different levels to build an influencing feature library for distributed photovoltaics at different levels.

7. A multi-level distributed photovoltaic output prediction device for a distribution network, characterized in that: include: The first establishment module is used to establish a data sample library of multi-level distributed photovoltaic connection distribution relationship of the distribution network; The second establishment module is used to establish a distribution network multi-level distributed photovoltaic data sample library based on the distributed photovoltaic historical related data and in combination with the distribution network multi-level distributed photovoltaic connection distribution relationship data sample library; The third establishment module is used to analyze the impact characteristics of multi-level distributed photovoltaic output based on the multi-level distributed photovoltaic data sample library of the distribution network, and establish an impact characteristic library for distributed photovoltaics of different levels; Constructing a training module for constructing an LSTM model based on the impact feature library, and using the data in the distribution network multi-level distributed photovoltaic data sample library to train the LSTM model to obtain a distribution network distributed photovoltaic multi-level output prediction model; The prediction module is used to input the input parameter data of the time period to be predicted into the distribution network distributed photovoltaic multi-level output prediction model to obtain the distribution network multi-level distributed photovoltaic output prediction result.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method for predicting multi-level distributed photovoltaic output in a distribution network as described in any one of claims 1-7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting multi-level distributed photovoltaic output in a distribution network as claimed in any one of claims 1 to 7 are implemented.

Images