Photovoltaic power prediction method and system based on artificial intelligence

Abstract

The invention relates to a photovoltaic power prediction method and system based on artificial intelligence, and the method comprises the steps: obtaining a historical weather cloud picture in a deployment and control region at a current moment, and constructing a similar historical weather cloud picture data set in the deployment and control region; constructing a current directed graph according to the similar historical weather cloud atlas data set; acquiring all weather cloud picture data in the monitoring area at a plurality of historical corresponding moments, and constructing a plurality of historical digraphs year by year according to the historical corresponding moment data of the plurality of windows before the current time period; and inputting the current directed graph into a weather cloud atlas development prediction network trained based on a historical directed graph for prediction, and obtaining the photovoltaic power generation capacity in the distribution and control area at the future moment. Compared with the prior art, the method has the advantages that the complex relation between the weather cloud atlas and the photovoltaic generating capacity can be deeply mined, and the correlation between the nodes can be accurately calculated, so that the slight difference of the change of the weather cloud atlas can be captured.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power prediction, and particularly to a photovoltaic power prediction method and system based on artificial intelligence. Background Art

[0002] With the transformation of the global energy structure and the rapid development of renewable energy, photovoltaic power generation has become one of the important sources of clean energy. However, the output power of a photovoltaic power generation system is affected by various factors, such as weather conditions, light intensity, temperature, etc., resulting in large fluctuations and uncertainties in the output power. Therefore, accurate prediction of photovoltaic power is of great significance for the stable operation of the power system and the efficient utilization of energy. Summary of the Invention

[0003] The purpose of the present invention is to overcome the defects of the above-mentioned existing technologies and provide a photovoltaic power prediction method and system based on artificial intelligence.

[0004] The purpose of the present invention can be achieved through the following technical solutions:

[0005] As the first aspect of the present invention, a photovoltaic power prediction method based on artificial intelligence is provided, and the steps include:

[0006] Obtain historical weather cloud maps within the controlled area at the current moment, and construct a dataset of similar historical weather cloud maps located within the controlled area;

[0007] Construct a current directed graph according to the dataset of similar historical weather cloud maps;

[0008] Obtain all weather cloud map data within the controlled area at multiple historical corresponding moments, and construct multiple historical directed graphs year by year according to the data at multiple historical corresponding moments in multiple windows before the current time period;

[0009] Input the current directed graph into a weather cloud map development prediction network trained based on the historical directed graphs for prediction, and obtain the photovoltaic power generation amount within the controlled area at a future moment.

[0010] As a preferred technical solution, the dataset of similar historical weather cloud maps includes:

[0011] The first dataset includes rainy season data and rainfall data located within the controlled area under similar historical weather cloud maps;

[0012] The second dataset includes average light data and photovoltaic power generation data located within the controlled area under similar historical weather cloud maps.

[0013] As a preferred technical solution, constructing the current directed graph according to the dataset of similar historical weather cloud maps is specifically as follows:

[0014] Construct a weather cloud map development node at the current moment based on multiple historical weather cloud maps, and each of the said development nodes represents a change vector of weather development respectively;

[0015] Take the photovoltaic power generation amount under the change vector corresponding to the development node as the data of this development node and perform vector embedding to obtain a node representation vector;

[0016] For the weather cloud map at the current moment, determine the future expected nodes according to the occurrence times of the development nodes in the weather cloud maps on future days corresponding to the historical data of the weather cloud map at the current moment in the similar historical weather cloud map dataset, and form a current directed graph.

[0017] As a preferred technical solution, the weather cloud map development prediction network includes multiple graph development prediction networks and multiple photovoltaic power generation amount data sets corresponding and associated with the graph development prediction networks, where:

[0018] For the said graph development prediction network, perform feature matching between the directed graph and the historical weather cloud maps in the controlled area at historical future moments after the current moment based on the node representation vector to obtain the weather cloud maps in the corresponding controlled area at the same moment;

[0019] The node representation vectors based on the historical weather cloud maps in the lower controlled area at historical future moments are stored in the photovoltaic power generation amount data set;

[0020] Extract the corresponding stored node representation vectors from the photovoltaic power generation amount data set corresponding and associated with the graph development prediction network for the weather cloud maps in the corresponding controlled area at the same moment predicted by the graph development prediction network to obtain photovoltaic power generation amount data.

[0021] As a preferred technical solution, the training process of the weather cloud map development prediction network is as follows:

[0022] Obtain all weather cloud map data in the controlled area at multiple historical corresponding moments, and after cleaning, obtain a first historical data set and a second historical data set respectively, where: the first historical data set is the future light length data under the historical similar weather cloud maps in the controlled area; the second historical data set is the average daily light duration data under the future light length data under the historical similar weather cloud maps in the controlled area;

[0023] Slice the first historical data set and the second historical data set respectively to obtain a set of first historical data subsets and a set of second historical data subsets corresponding to multiple historical corresponding moments;

[0024] Construct multiple historical directed graphs according to the first historical data subsets;

[0025] Using multiple second historical data subsets as labels, train a pre-constructed weather cloud map development prediction network with multiple historical directed graphs.

[0026] As a preferred technical solution, after the weather cloud map development prediction network predicts the photovoltaic power generation, use the second data set and multiple second historical data subsets as labels, and re-train the currently trained weather cloud map development prediction network with the current directed graph and multiple historical directed graphs; Based on the re-trained weather cloud map development prediction network, re-obtain the photovoltaic power generation in the controlled area at a future moment;

[0027] Among them, the second data set includes the average light intensity data and photovoltaic power generation data located in the controlled area under similar historical weather cloud maps; the second historical data set includes the average daily light duration data under the future light length data located in the controlled area under similar historical weather cloud maps.

[0028] As the second aspect of the present invention, provide a photovoltaic power prediction system based on artificial intelligence, the system includes:

[0029] A data acquisition module, which acquires the historical weather cloud map in the controlled area at the current moment and performs data cleaning to obtain a data set of similar historical weather cloud maps in the controlled area;

[0030] A data processing module, which constructs a current directed graph according to the data set of similar historical weather cloud maps in the controlled area;

[0031] A prediction module, input the current directed graph into the currently trained weather cloud map attention convolutional network for prediction, and obtain the photovoltaic power generation in the controlled area at a future moment; the weather cloud map attention convolutional network.

[0032] As a preferred technical solution, the data processing module constructs the current directed graph according to the data set of similar historical weather cloud maps, specifically as follows:

[0033] Construct weather cloud map development nodes at the current moment according to multiple historical weather cloud maps, and each of the development nodes represents a change vector of a weather development;

[0034] Use the photovoltaic power generation under the change vector corresponding to the development node as the data of the development node and perform vector embedding to obtain a node representation vector;

[0035] For the weather cloud map at the current moment, determine the future expected nodes according to the occurrence times of the development nodes in the weather cloud map on the future day corresponding to the historical data of the weather cloud map at the current moment in the data set of similar historical weather cloud maps, and form a current directed graph.

[0036] As a preferred technical solution, the weather cloud map development prediction network includes a plurality of map development prediction networks and a plurality of photovoltaic power generation data sets corresponding to and associated with the map development prediction networks, where:

[0037] The map development prediction network performs feature matching on the directed graph and the historical weather cloud maps within the deployment area at future historical moments after the current moment based on the node representation vectors, and obtains the weather cloud maps within the corresponding deployment area at the same moment;

[0038] The photovoltaic power generation data sets store the node representation vectors based on the historical weather cloud maps within the deployment area at future historical moments;

[0039] The weather cloud maps within the corresponding deployment area at the same moment predicted by the map development prediction network extract the corresponding stored node representation vectors from the photovoltaic power generation data set corresponding to and associated with the map development prediction network to obtain photovoltaic power generation data.

[0040] As a preferred technical solution, the prediction module is trained according to a plurality of historical directed graphs corresponding to the same historical moments in a plurality of windows before the current time period, specifically as follows:

[0041] Obtain all weather cloud map data within the deployment area at a plurality of historical corresponding moments, and after cleaning, obtain a first historical data set and a second historical data set respectively, where: the first historical data set is the future light length data under the historical similar weather cloud maps in the deployment area; the second historical data set is the average daily light duration data under the future light length data under the historical similar weather cloud maps in the deployment area;

[0042] Slice the first historical data set and the second historical data set respectively to obtain a set of first historical data subsets and a set of second historical data subsets corresponding to a plurality of historical corresponding moments;

[0043] Construct a plurality of historical directed graphs according to the first historical data subsets;

[0044] Use a plurality of second historical data subsets as labels, and use the plurality of historical directed graphs to train the pre-constructed weather cloud map development prediction network.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1) First, the present invention constructs a directed graph at the current moment. By capturing the changing trend of weather cloud images and the correlation relationship between nodes, the directed graph accurately reflects the potential impact of weather cloud images on future photovoltaic power generation. By constructing development nodes of multiple historical weather cloud images and performing vector embedding on the photovoltaic power generation as the data of the nodes, we obtain rich node representation vectors. At the same time, according to the occurrence times of weather cloud image nodes, the future expected nodes are determined, thus forming a complete current directed graph.

[0047] 2) The present invention uses the current directed graph and multiple historical directed graphs at the corresponding moments of multiple years in multiple windows before the current time period as input data and inputs them into a carefully trained weather cloud image development prediction network. The network consists of multiple graph development prediction networks and corresponding photovoltaic power generation data sets, and can deeply explore the complex relationship between weather cloud images and photovoltaic power generation. It realizes the accurate calculation of the correlation between nodes, thus capturing the subtle differences in the changes of weather cloud images.

[0048] 3) The present invention uses the second data set and multiple second historical data subsets as labels to preliminarily train the weather cloud image development prediction network. Then, according to the input data of the current directed graph and multiple historical directed graphs, the network is retrained and optimized, which not only improves the generalization ability of the network, but also ensures that the prediction results are closer to the actual situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a flowchart of a photovoltaic power prediction method based on artificial intelligence according to the present invention;

[0050] Figure 2 is a training flowchart of the weather cloud image development prediction network in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and gives the detailed implementation manner and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0052] Embodiment 1

[0053] The present invention provides as Figure 1 shown,

[0054] S1. Obtain historical weather cloud images in the controlled area at the current moment, and construct a data set of similar historical weather cloud images located in the controlled area;

[0055] S2. Construct a current directed graph according to the data set of similar historical weather cloud images;

[0056] S3. Obtain all weather cloud map data within the controlled area at multiple historical corresponding moments, and construct multiple historical directed graphs year by year based on the historical corresponding moment data of multiple windows before the current time period;

[0057] S4. Input the current directed graph into the weather cloud map development prediction network trained based on the historical directed graphs for prediction, and obtain the photovoltaic power generation amount within the controlled area at future moments.

[0058] In step S2, the obtained similar historical weather cloud map data set includes: a first data set, including rainy season data and rainfall data located within the controlled area under the similar historical weather cloud maps; a second data set, including average illumination data and photovoltaic power generation amount data located within the controlled area under the similar historical weather cloud maps.

[0059] In step S2, construct the current directed graph according to the similar historical weather cloud map data set as follows:

[0060] Construct weather cloud map development nodes at the current moment based on multiple historical weather cloud maps, and each of the development nodes represents a change vector of a weather development;

[0061] Take the photovoltaic power generation amount under the change vector corresponding to the development node as the data of this development node and perform vector embedding to obtain a node representation vector;

[0062] For the weather cloud map at the current moment, determine the future expected nodes according to the occurrence times of the development nodes in the weather cloud map on the future date corresponding to the historical data of the weather cloud map at the current moment in the similar historical weather cloud map data set, and form the current directed graph.

[0063] Among them, the weather cloud map development prediction network includes multiple graph development prediction networks and multiple photovoltaic power generation amount data sets corresponding and associated with the graph development prediction networks, where:

[0064] The graph development prediction network uses a convolutional neural network to perform feature matching between the directed graph and the historical weather cloud maps within the controlled area at historical future moments after the current moment based on the node representation vector, and obtains the weather cloud maps corresponding to the same moment within the controlled area;

[0065] The photovoltaic power generation amount data set stores the node representation vectors based on the historical weather cloud maps within the controlled area at historical future moments;

[0066] Extract the corresponding stored node representation vectors from the photovoltaic power generation amount data set corresponding and associated with the graph development prediction network according to the weather cloud maps corresponding to the same moment within the controlled area predicted by the graph development prediction network to obtain the photovoltaic power generation amount data.

[0067] In step S4, as Figure 2As shown in the figure, the training process of the weather cloud map development prediction network is as follows:

[0068] S41. Obtain all weather cloud map data within the controlled area at multiple historical corresponding moments. After cleaning, obtain the first historical data set and the second historical data set respectively, where: the first historical data set is the future light length data under the historical similar weather cloud maps in the controlled area; the second historical data set is the average daily light duration data under the future light length data in the historical similar weather cloud maps in the controlled area;

[0069] S42. Slice the first historical data set and the second historical data set respectively to obtain a set of first historical data subsets and a set of second historical data subsets corresponding to multiple historical corresponding moments;

[0070] S43. Construct multiple historical directed graphs according to the first historical data subsets;

[0071] S44. Use multiple second historical data subsets as labels, and use multiple historical directed graphs to train the pre-constructed weather cloud map development prediction network.

[0072] After the weather cloud map development prediction network predicts the photovoltaic power generation, use the second data set and multiple second historical data subsets as labels, and use the current directed graph and multiple historical directed graphs to retrain the currently trained weather cloud map development prediction network; based on the retrained weather cloud map development prediction network, re-obtain the photovoltaic power generation within the controlled area at future moments;

[0073] Embodiment 2

[0074] As another implementation manner of the present invention, the present invention also provides an artificial intelligence-based photovoltaic power prediction system, including:

[0075] A data acquisition module, which acquires the historical weather cloud maps within the controlled area at the current moment and performs data cleaning to obtain a first data set and a second data set, where:

[0076] The first data set includes the rainy season data and rainfall data within the controlled area under the similar historical weather cloud maps;

[0077] The second data set includes the average light data and photovoltaic power generation data within the controlled area under the similar historical weather cloud maps. Collect the historical weather cloud maps within the controlled area at the current moment from a wide range of data sources, and through a strict data cleaning process, eliminate redundant and incorrect information, and organize them into a first data set and a second data set. The first data set covers the future light length data under the similar weather cloud maps, as well as key information such as the rainy season and rainfall; the second data set contains the average daily light duration data under the future light length, as well as the average light and photovoltaic power generation data of the controlled area.

[0078] A data processing module that constructs a current directed graph based on the first data set.

[0079] Specifically, based on multiple historical weather cloud maps, multiple weather cloud map development nodes for the historical future days under the current weather cloud map are constructed. One node represents a vector change. The photovoltaic power generation amount under the vector change is used as the data of the node and vector embedding is performed to obtain the node representation vector.

[0080] According to the occurrence times of the weather cloud map nodes for the historical future days under the current weather cloud map, the nodes are used as future expected nodes among multiple nodes to form the current directed graph.

[0081] The directed graph at the current moment is constructed. By capturing the change trend of the weather cloud map and the correlation relationship between nodes, this directed graph accurately reflects the potential impact of the weather cloud map on future photovoltaic power generation. By constructing the development nodes of multiple historical weather cloud maps and performing vector embedding with the photovoltaic power generation amount as the data of the nodes, we obtain rich node representation vectors. At the same time, according to the occurrence times of the weather cloud map nodes, the future expected nodes are determined from the development nodes, thus forming a complete current directed graph.

[0082] A prediction module that inputs multiple historical directed graphs corresponding to the same historical moments in multiple windows before the current time period, together with the current directed graph, into the currently trained weather cloud map development prediction network for prediction to obtain the photovoltaic power generation amount in the controlled area at the future time.

[0083] Furthermore, the currently trained weather cloud map development prediction network includes:

[0084] S41. Obtain all weather cloud map data in the controlled area at multiple historical corresponding moments, and after cleaning, obtain the first historical data set and the second historical data set respectively, where:

[0085] The first historical data set is the future light length data under the historical similar weather cloud maps in the controlled area.

[0086] The second historical data set is the average daily light duration data under the future light length data.

[0087] S42. Slice the first historical data set and the second historical data set to obtain a set of first historical data subsets and a set of second historical data subsets corresponding to multiple historical corresponding moments.

[0088] S43. Construct multiple historical directed graphs according to the first historical data subsets.

[0089] S44, using the multiple second historical data subsets as labels, and using the multiple historical directed graphs to train the pre-constructed weather cloud map development prediction network.

[0090] Secondly, the weather cloud map development prediction network includes multiple map development prediction networks and multiple photovoltaic power generation data sets associated with them one by one, among which:

[0091] The multiple graph development prediction networks are historical weather cloud maps in the next control area at the historical future time after the current time, and the directed graphs corresponding to them are feature extracted to obtain the weather cloud maps in the control area corresponding to the same time;

[0092] Multiple photovoltaic power generation data sets are used to learn historical weather cloud maps in the control area at historical future moments, and multiple photovoltaic power generation data are expected.

[0093] By using the current directed graph and multiple historical directed graphs of corresponding moments in the yearly history of multiple windows before the current time period as input data, they are input into the carefully trained weather cloud map development prediction network. The network consists of multiple graph development prediction networks and corresponding photovoltaic power generation data sets, which can deeply explore the complex relationship between weather cloud maps and photovoltaic power generation, thereby capturing the subtle differences in weather cloud map changes.

[0094] In this embodiment, after the prediction module is generated, the second data set and multiple second historical data subsets are used as labels, and the current directed graph and multiple historical directed graphs are used to retrain the currently trained weather cloud map development prediction network. Then, the photovoltaic power generation in the control area at the future moment is re-obtained.

[0095] Through retraining, the system can further learn the relationship between the current directed graph and the historical directed graph, as well as the complex relationship between these directed graphs and photovoltaic power generation. This helps to capture more detailed information and improve the accuracy and reliability of the forecast. During the retraining process, the model parameters are adjusted and optimized according to the new data. This helps to eliminate the deviations that may exist during the initial training and make the model more in line with the actual photovoltaic power generation forecast needs.

[0096] 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.

[0097] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows 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 the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0098] These computer program instructions can 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, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0099] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0100] Specific embodiments of the present invention are used to elaborate on the principles and implementation manners of the present invention. The descriptions of the above embodiments are only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, based on the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

[0101] Embodiments of the present application also provide a specific implementation manner of an electronic device that can implement all the steps in the method in the above embodiments. The electronic device specifically includes the following:

[0102] A processor, a memory, a communications interface, and a bus;

[0103] Among them, the processor, the memory, and the communications interface communicate with each other through the bus;

[0104] The processor is used to call the computer program in the memory, and when the processor executes the computer program, all steps in the method in the above embodiments are implemented.

[0105] An embodiment of the present application further provides a computer-readable storage medium capable of implementing all steps in the method in the above embodiments. A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, all steps in the method in the above embodiments are implemented.

[0106] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the hardware + program type of embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, reference can be made to the partial description of the method embodiments. Although the method operation steps are provided as described in the embodiments or flowcharts in this specification, based on conventional or non-creative means, there may be more or fewer operation steps. The order of steps listed in the embodiments is only one way among many execution orders of steps and does not represent the only execution order. When actually executed in a device or terminal product, it can be executed in the order of the method shown in the embodiments or the drawings or executed in parallel (for example, in an environment of parallel processors or multi-threaded processing, or even in a distributed data processing environment). The term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, product or device including a series of elements not only includes those elements but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, product or device. Without further limitation, it does not exclude the existence of additional identical or equivalent elements in the process, method, product or device including the said elements. For the convenience of description, the above device is described by dividing it into various modules according to functions. Of course, when implementing the embodiments of this specification, the functions of each module can be realized in the same or multiple software and / or hardware, or the modules realizing the same function can be realized by a combination of multiple sub-modules or sub-units, etc. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in electrical, mechanical or other forms. The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general computer, a special computer, an embedded processor or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks

[0107] Those skilled in the art should understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of this specification can 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.) that contain computer-usable program code. Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. The key point of each embodiment is to illustrate the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. In the description of this specification, the description of reference terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the embodiments of this specification.

[0108] The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those of ordinary skill in the art can make many modifications and variations based on the concept of the present invention without creative work. Therefore, all technical solutions that can be obtained by those skilled in the art in the technical field of this application based on the concept of the present invention through logical analysis, reasoning, or limited experiments on the basis of the prior art should be within the protection scope determined by the claims.​

Claims

1. A photovoltaic power prediction method based on artificial intelligence, characterized in that the steps include: Obtain the historical weather cloud map in the control area at the current moment, and construct a similar historical weather cloud map dataset in the control area; Construct the current directed graph based on similar historical weather cloud map datasets; Acquire all weather cloud map data in the control area at multiple historical corresponding moments, and construct multiple historical directed graphs year by year based on the historical corresponding moment data of multiple windows before the current time period; The current directed graph is input into the weather cloud map development prediction network trained based on the historical directed graph to predict the photovoltaic power generation in the controlled area at the future moment.

2. The photovoltaic power prediction method based on artificial intelligence according to claim 1, characterized in that: The similar historical weather cloud map datasets include: The first data set includes rainy season data and rainfall data in the control area under the historical weather cloud map; The second data set includes average illumination data and photovoltaic power generation data in the controlled area under similar historical weather cloud maps.

3. The photovoltaic power prediction method based on artificial intelligence according to claim 2 is characterized in that: The current directed graph is constructed based on similar historical weather cloud map data sets, as follows: Constructing a weather cloud map development node at the current moment according to a plurality of historical weather cloud maps, each of the development nodes representing a change vector of weather development; The photovoltaic power generation under the change vector corresponding to the development node is used as the data of the development node and vector embedding is performed to obtain the node representation vector; For the current moment weather cloud map, the future expected nodes are determined according to the number of occurrences of development nodes in the weather cloud map of the future day corresponding to the historical data of the current moment weather cloud map in the similar historical weather cloud map data set, and the current directed graph is formed.

4. The photovoltaic power prediction method based on artificial intelligence according to claim 1, characterized in that: The weather cloud map development prediction network includes a plurality of map development prediction networks and a plurality of photovoltaic power generation data sets corresponding to the map development prediction networks, wherein: The graph development prediction network performs feature matching between the directed graph and the historical weather cloud map in the control area at the historical future time after the current time based on the node characterization vector, and obtains the weather cloud map in the control area at the same time; The photovoltaic power generation data set stores node representation vectors based on historical weather cloud maps in the next control area at historical future moments; The weather cloud map in the corresponding control area at the same time predicted by the graph development prediction network extracts the corresponding stored node representation vector from the photovoltaic power generation data set correspondingly associated with the graph development prediction network to obtain photovoltaic power generation data.

5. The photovoltaic power prediction method based on artificial intelligence according to claim 4 is characterized in that: The training process of the weather cloud map development prediction network is as follows: Acquire all weather cloud map data in the control area at multiple historical corresponding moments, and obtain the first historical data set and the second historical data set after cleaning, wherein: the first historical data set is the future sunshine length data under the historical similar weather cloud map of the control area; the second historical data set is the average sunshine duration data under the future sunshine length data under the historical similar weather cloud map of the control area; Slicing the first historical data set and the second historical data set respectively to obtain a first historical data subset set and a second historical data subset set corresponding to a plurality of historical corresponding moments; constructing a plurality of historical directed graphs according to the first historical data subset; The plurality of second historical data subsets are used as labels, and the pre-constructed weather cloud map development prediction network is trained using the plurality of historical directed graphs.

6. The photovoltaic power prediction method based on artificial intelligence according to claim 1, characterized in that: After the weather cloud map development prediction network predicts the photovoltaic power generation, the second data set and the plurality of second historical data subsets are used as labels, and the currently trained weather cloud map development prediction network is retrained using the current directed graph and the plurality of historical directed graphs; Based on the retrained weather cloud map development prediction network, the photovoltaic power generation in the control area at the future moment is retrieved; Among them, the second data set includes average illumination data and photovoltaic power generation data in the controlled area under similar historical weather cloud maps; the second historical data set includes average daylight duration data under future illumination length data in the controlled area under similar historical weather cloud maps.

7. A photovoltaic power prediction system based on artificial intelligence, characterized in that: The system comprises: The data acquisition module obtains the historical weather cloud map in the control area at the current moment and performs data cleaning to obtain a similar historical weather cloud map data set in the control area; The data processing module constructs the current directed graph based on similar historical weather cloud map data sets in the control area; Prediction module, the current directed graph is input into the currently trained weather cloud map attention convolutional network for prediction, and the photovoltaic power generation in the controlled area at the future moment is obtained; the weather cloud map attention convolutional network.

8. The photovoltaic power prediction system based on artificial intelligence according to claim 7, characterized in that: The data processing module constructs the current directed graph based on similar historical weather cloud map data sets, as follows: Constructing a weather cloud map development node at the current moment according to a plurality of historical weather cloud maps, each of the development nodes representing a change vector of weather development; The photovoltaic power generation under the change vector corresponding to the development node is used as the data of the development node and vector embedding is performed to obtain the node representation vector; For the current moment weather cloud map, the future expected nodes are determined according to the number of occurrences of development nodes in the weather cloud map of the future day corresponding to the historical data of the current moment weather cloud map in the similar historical weather cloud map data set, and the current directed graph is formed.

9. The photovoltaic power prediction system based on artificial intelligence according to claim 7, characterized in that: The weather cloud map development prediction network includes a plurality of map development prediction networks and a plurality of photovoltaic power generation data sets corresponding to the map development prediction networks, wherein: The graph development prediction network performs feature matching between the directed graph and the historical weather cloud map in the control area at the historical future time after the current time based on the node characterization vector, and obtains the weather cloud map in the control area at the same time; The photovoltaic power generation data set stores node representation vectors based on historical weather cloud maps in the next control area at historical future moments; The weather cloud map in the corresponding control area at the same time predicted by the graph development prediction network extracts the corresponding stored node representation vector from the photovoltaic power generation data set correspondingly associated with the graph development prediction network to obtain photovoltaic power generation data.

10. The photovoltaic power prediction system based on artificial intelligence according to claim 7, characterized in that: The prediction module is trained according to the corresponding multiple historical directed graphs of the corresponding historical moments of the multiple windows before the current time period, as follows: Acquire all weather cloud map data in the control area at multiple historical corresponding moments, and obtain the first historical data set and the second historical data set after cleaning, wherein: the first historical data set is the future sunshine length data under the historical similar weather cloud map of the control area; the second historical data set is the average sunshine duration data under the future sunshine length data under the historical similar weather cloud map of the control area; Slicing the first historical data set and the second historical data set respectively to obtain a first historical data subset set and a second historical data subset set corresponding to a plurality of historical corresponding moments; constructing a plurality of historical directed graphs according to the first historical data subset; The plurality of second historical data subsets are used as labels, and the pre-constructed weather cloud map development prediction network is trained using the plurality of historical directed graphs.

Images