Distributed photovoltaic power prediction method and device for micro-grid in heavy overload area

Abstract

The invention relates to the technical field of distributed photovoltaic prediction, and particularly provides a distributed photovoltaic power prediction method and device for a heavy overload area microgrid, and the method comprises the steps: obtaining a modal key value label corresponding to a distributed photovoltaic output scene of the heavy overload area microgrid; and taking the modal key value label as the input of a pre-trained distributed photovoltaic power prediction model of the heavy overload area micro-grid, and obtaining a photovoltaic power prediction result corresponding to the distributed photovoltaic output scene output by the pre-trained distributed photovoltaic power prediction model of the heavy overload area micro-grid. The technical scheme provided by the invention can be used as a distributed photovoltaic prediction method for a micro-grid or a power distribution area in a heavy and overload area, and compared with a traditional time sequence prediction method, the method has higher pertinence.

Description

Technical Field

[0001] The present invention relates to the field of distributed photovoltaic prediction technology, and in particular to a distributed photovoltaic power prediction method and device for a microgrid in a heavily overloaded area. Background Art

[0002] In recent years, the construction of new power systems has continued to advance, and distributed power sources, electric vehicles, energy storage, and microgrids have developed rapidly. However, with the large-scale access of a high proportion of distributed photovoltaics to microgrids and distribution network areas, the probability of overvoltage, power quality degradation, and reverse heavy overload in local areas has greatly increased. Therefore, it is necessary to further improve the accuracy of local distributed photovoltaic predictions, alleviate the problem of local heavy overloads, enable demand-side distributed access to the power system in a safer, more environmentally friendly, efficient, and economical manner, improve the overall energy utilization efficiency of the region, and ensure the stable operation of local power supply.

[0003] Compared with the new energy forecast of traditional large power grids, the distributed photovoltaic power forecast of microgrids in heavily overloaded areas has complex influencing factors, such as a large number of short-term, local unexpected shadow obstructions, which are reflected in the presence of a large number of abnormal glitches and spikes in the data. This makes the traditional photovoltaic forecast based on meteorological factors have significant lack of applicability in heavily overloaded areas. Summary of the invention

[0004] In order to overcome the above-mentioned defects, the present invention proposes a distributed photovoltaic power prediction method and device for a microgrid in a heavily overloaded area.

[0005] In a first aspect, a distributed photovoltaic power prediction method for a microgrid in a heavily overloaded area is provided, and the distributed photovoltaic power prediction method for a microgrid in a heavily overloaded area includes:

[0006] Get the modal key-value tag corresponding to the distributed photovoltaic output scenario of the microgrid in the heavily overloaded area;

[0007] The modal key-value label is used as the input of a pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area, and a photovoltaic power prediction result corresponding to the distributed photovoltaic output scenario output by the pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area is obtained.

[0008] Preferably, the training process of the pre-trained distributed photovoltaic power prediction model for the heavily overloaded regional microgrid includes:

[0009] The sliding window algorithm is used to obtain the historical data of distributed photovoltaic power in the microgrid in the heavily overloaded area;

[0010] Clustering the distributed photovoltaic power history data of the microgrid in the heavily overloaded area, and taking each type of distributed photovoltaic power history data as distributed photovoltaic power history data corresponding to a distributed photovoltaic output scenario;

[0011] Set the modal key-value labels of the distributed photovoltaic power history data corresponding to different distributed photovoltaic output scenarios, and use the distributed photovoltaic power history data corresponding to different distributed photovoltaic output scenarios to construct training data;

[0012] The training data is used to train the distributed photovoltaic power prediction model of the initial heavily overloaded regional microgrid to obtain the pre-trained distributed photovoltaic power prediction model of the heavily overloaded regional microgrid.

[0013] Furthermore, the distributed photovoltaic power prediction model of the microgrid in the initial heavily overloaded area includes a long short-term memory network, an attention mechanism network and a fully connected neural network connected in sequence.

[0014] Furthermore, the mathematical model of the long short-term memory network is as follows:

[0015] h t =O t *tanh(C t )

[0016] In the above formula, h t is the output feature of the long short-term memory network at time t, O t is the output gate output feature of the LSTM network at time t, C t is the output feature of the memory unit of the long short-term memory network at time t.

[0017] Furthermore, the output characteristics of the output gate of the long short-term memory network at time t are as follows:

[0018] O t =σ( <W o ,[h t-1 ,x t ]>+b o )

[0019] The output characteristics of the memory unit of the long short-term memory network at time t are as follows:

[0020]

[0021] In the above formula, σ is the sigmoid function, W o is the output weight, h t-1 is the output feature of the long short-term memory network at time t-1, x t is the input sequence at time t, b o is the output bias, C t-1is the output feature of the memory unit of the long short-term memory network at time t-1, f t is the output sequence of the forget gate at time t, is the candidate memory state, r t To preserve information weight.

[0022] Furthermore, the output sequence of the forget gate at time t is as follows:

[0023] f t =σ( <W f ,[h t-1 ,x t ]>+b f )

[0024] The candidate memory states are as follows:

[0025]

[0026] The retained information weights are as follows:

[0027] r t =σ( <W r ,[h t-1 ,x t ]>+b r )

[0028] In the above formula, W f is the forget weight matrix, b f is the forget bias, W c is the candidate memory weight, b c is the candidate bias, W r To preserve information weight, b r To retain the bias.

[0029] Furthermore, the mathematical model of the attention mechanism network is as follows:

[0030]

[0031] In the above formula, Attention is the output feature of the mathematical model of the attention mechanism network, α i is the normalized similarity value corresponding to the distributed photovoltaic output scenario i, Value i is the output feature of the long short-term memory network in distributed photovoltaic output scenario i, i∈[1,n], n is the number of distributed photovoltaic output scenarios.

[0032] Furthermore, the normalized similarity value corresponding to the distributed photovoltaic output scene i is as follows:

[0033]

[0034] In the above formula, e is a natural constant, SML i is the similarity between the target sequence and the modal key-value label corresponding to distributed photovoltaic output scenario i.

[0035] Furthermore, the similarity between the target sequence and the modal key-value label corresponding to the distributed photovoltaic output scenario i is as follows:

[0036]

[0037] In the above formula, Array is the target sequence, Key i is the modal key-value label corresponding to the distributed photovoltaic output scenario i.

[0038] In a second aspect, a distributed photovoltaic power prediction device for a microgrid in a heavily overloaded area is provided, wherein the distributed photovoltaic power prediction device for the microgrid in a heavily overloaded area comprises:

[0039] An acquisition module is used to obtain the modal key value tag corresponding to the distributed photovoltaic output scenario of the microgrid in the heavily overloaded area;

[0040] A prediction module is used to use the modal key-value label as the input of a pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area, and obtain a photovoltaic power prediction result corresponding to the distributed photovoltaic output scenario output by the pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area.

[0041] In a third aspect, a computer device is provided, comprising: one or more processors;

[0042] The processor is configured to execute one or more programs;

[0043] When the one or more programs are executed by the one or more processors, the distributed photovoltaic power prediction method for the heavily overloaded area microgrid is implemented.

[0044] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed, the distributed photovoltaic power prediction method for the microgrid in the heavily overloaded area is implemented.

[0045] The above one or more technical solutions of the present invention have at least one or more of the following beneficial effects:

[0046] The present invention provides a distributed photovoltaic power prediction method and device for a heavily overloaded regional microgrid, comprising: obtaining a modal key-value tag corresponding to a distributed photovoltaic output scenario of a heavily overloaded regional microgrid; using the modal key-value tag as an input of a pre-trained distributed photovoltaic power prediction model for a heavily overloaded regional microgrid, and obtaining a photovoltaic power prediction result corresponding to the distributed photovoltaic output scenario output by the pre-trained distributed photovoltaic power prediction model for a heavily overloaded regional microgrid. The technical solution provided by the present invention can be used as a distributed photovoltaic prediction method for a heavily overloaded regional microgrid or a distribution station area, and is more targeted than traditional time series prediction methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a schematic flow chart of the main steps of the distributed photovoltaic power prediction method for a microgrid in a heavily overloaded area according to an embodiment of the present invention;

[0048] Figure 2 It is a schematic diagram of the training process of a distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area according to an embodiment of the present invention;

[0049] Figure 3 This is a comparison diagram of distributed photovoltaic prediction results in a non-interference area according to an embodiment of the present invention;

[0050] Figure 4 This is a test set result diagram of the distributed photovoltaic power prediction model of the heavily overloaded area microgrid according to an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The specific implementation modes of the present invention will be further described in detail below in conjunction with the accompanying drawings.

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] Example 1

[0054] See attached Figure 1 , Figure 1 FIG. 1 is a flow chart showing the main steps of a distributed photovoltaic power prediction method for a microgrid in a heavily overloaded area according to an embodiment of the present invention. Figure 1 As shown, the distributed photovoltaic power prediction method of the microgrid in the heavily overloaded area in the embodiment of the present invention mainly includes the following steps:

[0055] Step S101: Obtaining a modal key-value tag corresponding to a distributed photovoltaic output scenario of a microgrid in a heavily overloaded area;

[0056] The modal key-value label is used as the input of a pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area, and a photovoltaic power prediction result corresponding to the distributed photovoltaic output scenario output by the pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area is obtained.

[0057] In this embodiment, the training process of the pre-trained distributed photovoltaic power prediction model of the heavily overloaded regional microgrid is as follows: Figure 2 As shown, including:

[0058] The sliding window algorithm is used to obtain the historical data of distributed photovoltaic power in the microgrid in the heavily overloaded area;

[0059] Clustering the distributed photovoltaic power history data of the microgrid in the heavily overloaded area, and taking each type of distributed photovoltaic power history data as distributed photovoltaic power history data corresponding to a distributed photovoltaic output scenario;

[0060] Set the modal key-value labels of the distributed photovoltaic power history data corresponding to different distributed photovoltaic output scenarios, and use the distributed photovoltaic power history data corresponding to different distributed photovoltaic output scenarios to construct training data;

[0061] The training data is used to train the distributed photovoltaic power prediction model of the initial heavily overloaded regional microgrid to obtain the pre-trained distributed photovoltaic power prediction model of the heavily overloaded regional microgrid.

[0062] In a specific implementation, the gradient descent method is used to train each weight matrix and bias matrix in long short-term memory learning, and the training stop condition is set to that the 2-norm of the training set error is less than 1.0*10-3.

[0063] In one embodiment, the distributed photovoltaic power prediction model of the initial heavily overloaded regional microgrid includes a long short-term memory network, an attention mechanism network, and a fully connected neural network connected in sequence.

[0064] In one embodiment, the mathematical model of the long short-term memory network is as follows:

[0065] h t =O t *tanh(C t )

[0066] In the above formula, h t is the output feature of the long short-term memory network at time t, O t is the output gate output feature of the LSTM network at time t, Ct is the output feature of the memory unit of the long short-term memory network at time t.

[0067] In one embodiment, the output gate output characteristics of the long short-term memory network at time t are as follows:

[0068] O t =σ( <W o ,[h t-1 , x t ]>+b o )

[0069] The output characteristics of the memory unit of the long short-term memory network at time t are as follows:

[0070]

[0071] In the above formula, σ is the sigmoid function, W o is the output weight, h t-1 is the output feature of the long short-term memory network at time t-1, x t is the input sequence at time t, b o is the output bias, C t-1 is the output feature of the memory unit of the long short-term memory network at time t-1, f t is the output sequence of the forget gate at time t, is the candidate memory state, r t To preserve information weight.

[0072] In one embodiment, the forget gate output sequence at time t is as follows:

[0073] f t =σ( <W f ,[h t-1 , x t ]>+b f )

[0074] The candidate memory states are as follows:

[0075]

[0076] The retained information weights are as follows:

[0077] r t =σ( <W r ,[h t-1 , x t ]>+b r )

[0078] In the above formula, W f is the forget weight matrix, b f is the forget bias, W cis the candidate memory weight, b c is the candidate bias, W r To preserve information weight, b r To retain the bias.

[0079] In one embodiment, the mathematical model of the attention mechanism network is as follows:

[0080]

[0081] In the above formula, Attention is the output feature of the mathematical model of the attention mechanism network, α i is the normalized similarity value corresponding to the distributed photovoltaic output scenario i, Value i is the output feature of the long short-term memory network in distributed photovoltaic output scenario i, i∈[1,n], n is the number of distributed photovoltaic output scenarios.

[0082] In one embodiment, the normalized similarity value corresponding to the distributed photovoltaic output scene i is as follows:

[0083]

[0084] In the above formula, e is a natural constant, SML i is the similarity between the target sequence and the modal key-value label corresponding to distributed photovoltaic output scenario i.

[0085] In one embodiment, the similarity between the target sequence and the modal key-value tag corresponding to the distributed photovoltaic output scenario i is as follows:

[0086]

[0087] In the above formula, Array is the target sequence, Key i is the modal key-value label corresponding to the distributed photovoltaic output scenario i.

[0088] The above algorithm is verified by taking the distributed photovoltaic output data of a certain heavily overloaded area as an example. First, a data set of a pure distributed photovoltaic heavily overloaded area is tested. The data set contains 7 days of 15-minute distributed photovoltaic data of microgrid areas. 6 days are divided into training sets and 1 day is divided into test sets for algorithm testing and verification. The results are as follows Figure 3 shown.

[0089] Figure 3(a) is the distributed photovoltaic prediction curve on day T0, and (b) is the actual distributed photovoltaic output curve on day T0. The photovoltaic load involved in the calculation is predicted to reach a maximum load of about 118kW at around 2 pm. The mean absolute error MAE between the predicted data and the actual data is 0.027515, the root mean square error MSE is 0.00075708, and the root mean square error RMSE is 0.027515.

[0090] Then, load forecasting was carried out for a photovoltaic integrated system in a heavily overloaded area. Since the distributed photovoltaic system was equipped with post-meter energy storage, it was impossible to decouple the distributed photovoltaic output and energy storage output data on the one hand, and on the other hand, its output behavior characteristics were more complex, so it was more suitable to use the method proposed in this paper to carry out distributed photovoltaic output forecasting. Figure 4 This is the test set result diagram of the distributed photovoltaic power prediction model (attention-LSTM) for microgrids in heavily overloaded areas. According to the prediction, the photovoltaic load involved in the calculation reaches the maximum load at 19:15 in the evening, with a maximum value of about 98kW. The mean absolute error MAE between the predicted data and the actual data is 0.88973, and the root mean square error MSE is 0.1313.

[0091] Example 2

[0092] Based on the same inventive concept, the present invention also provides a distributed photovoltaic power prediction device for a microgrid in a heavily overloaded area, and the distributed photovoltaic power prediction device for a microgrid in a heavily overloaded area includes:

[0093] An acquisition module is used to obtain the modal key-value tag corresponding to the distributed photovoltaic output scenario of the microgrid in the heavily overloaded area;

[0094] A prediction module is used to use the modal key-value label as the input of a pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area, and obtain a photovoltaic power prediction result corresponding to the distributed photovoltaic output scenario output by the pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area.

[0095] Preferably, the training process of the pre-trained distributed photovoltaic power prediction model for the heavily overloaded regional microgrid includes:

[0096] The sliding window algorithm is used to obtain the historical data of distributed photovoltaic power in the microgrid in the heavily overloaded area;

[0097] Clustering the distributed photovoltaic power history data of the microgrid in the heavily overloaded area, and taking each type of distributed photovoltaic power history data as distributed photovoltaic power history data corresponding to a distributed photovoltaic output scenario;

[0098] Set the modal key-value labels of the distributed photovoltaic power history data corresponding to different distributed photovoltaic output scenarios, and use the distributed photovoltaic power history data corresponding to different distributed photovoltaic output scenarios to construct training data;

[0099] The training data is used to train the distributed photovoltaic power prediction model of the initial heavily overloaded regional microgrid to obtain the pre-trained distributed photovoltaic power prediction model of the heavily overloaded regional microgrid.

[0100] Furthermore, the distributed photovoltaic power prediction model of the microgrid in the initial heavily overloaded area includes a long short-term memory network, an attention mechanism network and a fully connected neural network connected in sequence.

[0101] Furthermore, the mathematical model of the long short-term memory network is as follows:

[0102] h t =O t *tanh(C t )

[0103] In the above formula, h t is the output feature of the long short-term memory network at time t, O t is the output gate output feature of the LSTM network at time t, C t is the output feature of the memory unit of the long short-term memory network at time t.

[0104] Furthermore, the output characteristics of the output gate of the long short-term memory network at time t are as follows:

[0105] O t =σ( <W o ,[h t-1 , x t ]>+b o )

[0106] The output characteristics of the memory unit of the long short-term memory network at time t are as follows:

[0107]

[0108] In the above formula, σ is the sigmoid function, W o is the output weight, h t-1 is the output feature of the long short-term memory network at time t-1, x t is the input sequence at time t, b o is the output bias, C t-1 is the output feature of the memory unit of the long short-term memory network at time t-1, f t is the output sequence of the forget gate at time t, is the candidate memory state, r t To preserve information weight.

[0109] Furthermore, the output sequence of the forget gate at time t is as follows:

[0110] f t =σ( <W f ,[h t-1 ,x t ]>+b f )

[0111] The candidate memory states are as follows:

[0112]

[0113] The retained information weights are as follows:

[0114] r t =σ( <W r ,[h t-1 ,x t ]>+b r )

[0115] In the above formula, W f is the forget weight matrix, b f is the forget bias, W c is the candidate memory weight, b c is the candidate bias, W r To preserve information weight, b r To retain the bias.

[0116] Furthermore, the mathematical model of the attention mechanism network is as follows:

[0117]

[0118] In the above formula, Attention is the output feature of the mathematical model of the attention mechanism network, α i is the normalized similarity value corresponding to the distributed photovoltaic output scenario i, Value i is the output feature of the long short-term memory network in distributed photovoltaic output scenario i, i∈[1,n], n is the number of distributed photovoltaic output scenarios.

[0119] Furthermore, the normalized similarity value corresponding to the distributed photovoltaic output scene i is as follows:

[0120]

[0121] In the above formula, e is a natural constant, SML i is the similarity between the target sequence and the modal key-value label corresponding to distributed photovoltaic output scenario i.

[0122] Furthermore, the similarity between the target sequence and the modal key-value label corresponding to the distributed photovoltaic output scenario i is as follows:

[0123]

[0124] In the above formula, Array is the target sequence, Key i is the modal key-value label corresponding to the distributed photovoltaic output scenario i.

[0125] Example 3

[0126] Based on the same inventive concept, the present invention also provides a computer device, which includes a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function, so as to implement the steps of a distributed photovoltaic power prediction method for a microgrid in a heavily overloaded area in the above embodiment.

[0127] Example 4

[0128] Based on the same inventive concept, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It can be understood that the computer-readable storage medium here can include both built-in storage media in a computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of a distributed photovoltaic power prediction method for a microgrid in a heavily overloaded area in the above embodiment.

[0129] Those skilled in the art will appreciate 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. Moreover, 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.

[0130] 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.

[0131] 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.

[0132] 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 in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A distributed photovoltaic power prediction method for a microgrid in a heavily overloaded area, characterized in that: The method comprises: Obtain the modal key-value tags corresponding to the distributed photovoltaic output scenarios of microgrids in heavily overloaded areas; The modal key-value label is used as the input of a pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area, and a photovoltaic power prediction result corresponding to the distributed photovoltaic output scenario output by the pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area is obtained.

2. The method according to claim 1, characterized in that The training process of the pre-trained distributed photovoltaic power prediction model for the heavily overloaded regional microgrid includes: The sliding window algorithm is used to obtain the historical data of distributed photovoltaic power in the microgrid in the heavily overloaded area; Clustering the distributed photovoltaic power history data of the microgrid in the heavily overloaded area, and taking each type of distributed photovoltaic power history data as distributed photovoltaic power history data corresponding to a distributed photovoltaic output scenario; Set the modal key-value labels of the distributed photovoltaic power history data corresponding to different distributed photovoltaic output scenarios, and use the distributed photovoltaic power history data corresponding to different distributed photovoltaic output scenarios to construct training data; The training data is used to train the distributed photovoltaic power prediction model of the initial heavily overloaded regional microgrid to obtain the pre-trained distributed photovoltaic power prediction model of the heavily overloaded regional microgrid.

3. The method according to claim 2, characterized in that The distributed photovoltaic power prediction model of the initial heavily overloaded regional microgrid includes a long short-term memory network, an attention mechanism network and a fully connected neural network connected in sequence.

4. The method according to claim 3, characterized in that The mathematical model of the long short-term memory network is as follows: h t =O t *tanh(C t ) In the above formula, h t is the output feature of the long short-term memory network at time t, O t is the output gate output feature of the LSTM network at time t, C t is the output feature of the memory unit of the long short-term memory network at time t.

5. The method according to claim 4, characterized in that The output gate output characteristics of the long short-term memory network at time t are as follows: The t =σ( <W o ,[h t-1 ,x t ]>+b o ) The output characteristics of the memory unit of the long short-term memory network at time t are as follows: In the above formula, σ is the sigmoid function, W o is the output weight, h t-1 is the output feature of the long short-term memory network at time t-1, x t is the input sequence at time t, b o is the output bias, C t-1 is the output feature of the memory unit of the long short-term memory network at time t-1, f t is the output sequence of the forget gate at time t, is the candidate memory state, r t To preserve information weight.

6. The method according to claim 5, characterized in that The output sequence of the forget gate at time t is as follows: f t =σ( <W f ,[h t-1 ,x t ]>+b f ) The candidate memory states are as follows: The retained information weights are as follows: r t =σ( <W r ,[h t-1 ,x t ]>+b r ) In the above formula, W f is the forget weight matrix, b f is the forget bias, W c is the candidate memory weight, b c is the candidate bias, W r To preserve information weight, b r To retain the bias.

7. The method according to claim 6, characterized in that The mathematical model of the attention mechanism network is as follows: In the above formula, Attention is the output feature of the mathematical model of the attention mechanism network, α i is the normalized similarity value corresponding to the distributed photovoltaic output scenario i, Value i is the output feature of the long short-term memory network in distributed photovoltaic output scenario i, i∈[1,n], n is the number of distributed photovoltaic output scenarios.

8. The method according to claim 7, characterized in that The normalized similarity value corresponding to the distributed photovoltaic output scenario i is as follows: In the above formula, e is a natural constant, SML i is the similarity between the target sequence and the modal key-value label corresponding to distributed photovoltaic output scenario i.

9. The method according to claim 8, characterized in that The similarity between the target sequence and the modal key-value label corresponding to the distributed photovoltaic output scenario i is as follows: In the above formula, Array is the target sequence, Key i is the modal key-value label corresponding to the distributed photovoltaic output scenario i.

10. A device for predicting distributed photovoltaic power in a microgrid in a heavily overloaded area based on any one of claims 1 to 9, characterized in that: The device comprises: An acquisition module is used to obtain the modal key value tag corresponding to the distributed photovoltaic output scenario of the microgrid in the heavily overloaded area; A prediction module is used to use the modal key-value label as the input of a pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area, and obtain a photovoltaic power prediction result corresponding to the distributed photovoltaic output scenario output by the pre-trained distributed photovoltaic power prediction model for a microgrid in a heavily overloaded area.

11. A computer device, characterized in that: include: one or more processors; The processor is configured to execute one or more programs; When the one or more programs are executed by the one or more processors, the distributed photovoltaic power prediction method for a microgrid in a heavily overloaded area as described in any one of claims 1 to 9 is implemented.

12. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, a distributed photovoltaic power prediction method for a microgrid in a heavily overloaded area as described in any one of claims 1 to 9 is implemented.

Images