Photovoltaic station tropical cyclone early warning method, device, equipment and medium

Abstract

The invention relates to the technical field of new energy, and discloses a photovoltaic station tropical cyclone early warning method, device, equipment and medium, the photovoltaic station tropical cyclone early warning method comprises the following steps: obtaining a fusion gray level cloud picture of satellite data; wherein the fused gray level cloud picture is obtained by fusing cloud picture data of at least two different channels; inputting the fused gray level cloud picture into a pre-trained Res-UNet model, and extracting the time sequence characteristics of the fused gray level cloud picture based on the Res-UNet model to obtain an extrapolated gray level cloud picture; performing semantic segmentation based on the tropical cyclone cloud block on the extrapolated gray level cloud image to obtain a tropical cyclone cloud block identification result; extracting moving features of the tropical cyclone from the extrapolated gray level cloud picture based on the tropical cyclone cloud block identification result; and carrying out tropical cyclone early warning on each photovoltaic station based on the movement characteristics. The tropical cyclone prediction based on satellite data for the photovoltaic stations can be realized.

Description

Technical Field

[0001] The present invention relates to the field of new energy technology, and in particular to a tropical cyclone early warning method, device, equipment and medium for photovoltaic stations. Background Art

[0002] When photovoltaic power stations are affected by tropical cyclones, photovoltaic modules may be blown off or damaged, and the support structure may be deformed or collapsed because it cannot withstand the typhoon brought by the tropical cyclone. This will not only cause direct economic losses and economic reconstruction, but also affect the normal power output of the modules, and pose a threat to the surrounding environment and personnel safety. At the same time, tropical cyclones are extreme weather phenomena that are disruptive, destructive, and mobile, causing interference and even damage to photovoltaic modules.

[0003] Therefore, it is necessary to issue tropical cyclone warnings for photovoltaic stations. Summary of the invention

[0004] In view of this, the present invention provides a tropical cyclone early warning method, device, equipment and medium for photovoltaic stations to solve the problem of lack of tropical cyclone early warning in photovoltaic stations in the related art.

[0005] A tropical cyclone early warning method for photovoltaic stations, comprising: obtaining a fused grayscale cloud image of satellite data; wherein the fused grayscale cloud image is obtained by fusing cloud image data of at least two different channels; inputting the fused grayscale cloud image into a pre-trained Res-UNet model, extracting the temporal features of the fused grayscale cloud image based on the Res-UNet model, and obtaining an extrapolated grayscale cloud image; performing semantic segmentation of the extrapolated grayscale cloud image based on tropical cyclone cloud blocks, and obtaining a tropical cyclone cloud block recognition result; extracting the movement features of the tropical cyclone from the extrapolated grayscale cloud image based on the tropical cyclone cloud block recognition result; and performing a tropical cyclone early warning for each photovoltaic station based on the movement features.

[0006] As an exemplary embodiment, the Res-UNet model uses multiple encoders and decoders of the U-Net model, each of the encoders and decoders is connected with multiple submodules, each submodule includes multiple residual connections, and each residual block includes a shortcut to achieve cross-layer connection. As an exemplary embodiment, the method for training the Res-UNet model includes: obtaining a historical fused grayscale cloud map of the satellite data; dividing the historical fused grayscale cloud map to obtain a first historical fused grayscale cloud map in a first time period and a second historical fused grayscale cloud map in a second time period; wherein the first time period is earlier than the second time period; using the first historical fused grayscale cloud map as an input feature and the second historical fused grayscale cloud map as a predicted label to train a pre-constructed Res-UNet model; during the model training process, continuously adjusting the model parameters of the constructed Res-UNet model so that the Res-UNet model can output the second historical fused grayscale cloud map based on the temporal change characteristics of the first historical fused grayscale cloud map.

[0007] As an exemplary embodiment, the extraction of the movement characteristics of the tropical cyclone in the extrapolated grayscale cloud map based on the tropical cyclone cloud block identification result includes: calculating the tropical cyclone wind speed of each of the extrapolated grayscale cloud maps based on the tropical cyclone cloud block identification result; calculating the movement vector of each tropical cyclone in the extrapolated grayscale cloud map at every two adjacent moments and the time difference of the grayscale cloud map based on the tropical cyclone cloud block identification result; calculating the speed and direction of the tropical cyclone based on the time difference and the movement vector; traversing each of the extrapolated grayscale cloud maps, synthesizing the speed, direction and wind speed of each tropical cyclone to obtain the movement trajectory of the tropical cyclone as the movement characteristic.

[0008] As an exemplary embodiment, the method of obtaining a fused grayscale cloud map of satellite data includes: obtaining first satellite data of a visible light channel and second satellite data of an infrared channel of a target satellite; determining first channel grayscale data and second channel grayscale data based on the first satellite data and the second satellite data; and fusing the first channel grayscale data and the second channel grayscale data according to a preset weight to obtain the fused grayscale cloud map.

[0009] As an exemplary embodiment, the extrapolated grayscale cloud image is subjected to semantic segmentation based on tropical cyclone cloud blocks to obtain a tropical cyclone cloud block recognition result, including: inputting the extrapolated grayscale cloud image into a pre-trained convolutional neural network, extracting the cloud block thickness and vortex center in the extrapolated grayscale cloud image based on semantic features represented by the convolutional neural network based on the grayscale cloud image region, and obtaining a cloud block having an actual thickness greater than a preset thickness and a vortex center as a tropical cyclone cloud block.

[0010] As an exemplary embodiment, the tropical cyclone warning for each photovoltaic station based on the movement characteristics includes: acquiring the location information of each photovoltaic station; performing spatiotemporal matching on the movement characteristics based on the location information to obtain the cyclone wind speed of each photovoltaic station; and obtaining warning information for each photovoltaic station based on the cyclone wind speed.

[0011] In a second aspect, the present invention provides a tropical cyclone early warning device for a photovoltaic station, the device comprising: an acquisition module, used to acquire a fused grayscale cloud image of satellite data; wherein the fused grayscale cloud image is obtained by fusing cloud image data of at least two different channels; a cloud image extrapolation module, used to input the fused grayscale cloud image into a pre-trained Res-UNet model, extract the temporal features of the fused grayscale cloud image based on the Res-UNet model, and obtain an extrapolated grayscale cloud image; a semantic segmentation module, used to perform semantic segmentation of the extrapolated grayscale cloud image based on tropical cyclone cloud blocks, and obtain a tropical cyclone cloud block recognition result; a movement feature extraction module, used to extract the movement features of the tropical cyclone in the extrapolated grayscale cloud image based on the tropical cyclone cloud block recognition result; and an early warning module, used to perform tropical cyclone early warnings for each photovoltaic station based on the movement features.

[0012] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the tropical cyclone early warning method for a photovoltaic station according to the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0013] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the tropical cyclone early warning method for a photovoltaic station according to the first aspect or any corresponding embodiment thereof.

[0014] The present invention provides a tropical cyclone early warning method for photovoltaic stations, comprising: obtaining a fused grayscale cloud image of satellite data; wherein the fused grayscale cloud image is obtained by fusing cloud image data of at least two different channels; inputting the fused grayscale cloud image into a pre-trained Res-UNet model, extracting the temporal features of the fused grayscale cloud image based on the Res-UNet model, and obtaining an extrapolated grayscale cloud image; performing semantic segmentation of tropical cyclone cloud blocks on the extrapolated grayscale cloud image, and obtaining a tropical cyclone cloud block recognition result; extracting the movement features of the tropical cyclone from the extrapolated grayscale cloud image based on the tropical cyclone cloud block recognition result; and performing tropical cyclone early warning on each photovoltaic station based on the movement features, thereby realizing tropical cyclone prediction based on satellite data for photovoltaic stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 is a schematic flow chart of a tropical cyclone early warning method for a photovoltaic station according to an embodiment of the present invention;

[0017] Figure 2 is a schematic flow chart of another tropical cyclone early warning method for a photovoltaic station according to an embodiment of the present invention;

[0018] Figure 3 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0020] When photovoltaic power stations are affected by tropical cyclones, photovoltaic modules may be blown off or damaged, and the support structure may be deformed or collapsed because it cannot withstand the typhoon brought by the tropical cyclone. This will not only cause direct economic losses and economic reconstruction, but also affect the normal power output of the modules, and pose a threat to the surrounding environment and personnel safety. At the same time, tropical cyclones are extreme weather phenomena that are disruptive, destructive, and mobile, causing interference and even damage to photovoltaic modules.

[0021] To avoid the photovoltaic field station in related technology

[0022] According to an embodiment of the present invention, an embodiment of a tropical cyclone warning method for a photovoltaic station is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0023] In this embodiment, a tropical cyclone early warning method for a photovoltaic station is provided. Figure 1 FIG. 4 is a flow chart of a tropical cyclone early warning method for a photovoltaic station according to an embodiment of the present invention. Figure 1As shown, the process includes the following steps:

[0024] Step S101, obtaining a fused grayscale cloud image of satellite data; wherein the fused grayscale cloud image is obtained by fusing cloud image data of at least two different channels.

[0025] Step S102: input the fused grayscale cloud image into a pre-trained Res-UNet model, extract the temporal features of the fused grayscale cloud image based on the Res-UNet model, and obtain an extrapolated grayscale cloud image.

[0026] The Res-UNet (Residual Network) generative deep learning model is a deep convolutional neural network based on residual connections, which can better extract low-frequency information and alleviate the problem of missing semantic information.

[0027] Specifically, the pre-trained Res-UNet model is used to extrapolate the grayscale cloud image based on the temporal features of the fused grayscale cloud image to obtain an extrapolated grayscale cloud image.

[0028] In one embodiment, a pre-trained Res-UNet model is used to perform meteorological extrapolation on the fused grayscale cloud map data based on the time series features of the time series data contained in the fused grayscale cloud map data of the first 4 hours and the time series data labels for the next 4 hours to obtain a cloud map every 10 minutes for the next 4 hours.

[0029] Step S103, performing semantic segmentation based on tropical cyclone cloud blocks on the extrapolated grayscale cloud image to obtain a tropical cyclone cloud block recognition result.

[0030] In this embodiment, the tropical cyclone contained in the grayscale cloud image is identified by identifying the semantic features in the extrapolated grayscale cloud image.

[0031] Since tropical cyclone clouds usually have vortices and are characterized by thick clouds, in this embodiment, the semantic features of each pixel in the extrapolated cloud image data can be first identified to obtain an initial category label representing its semantic category; further, target pixels with vortices and actual thickness greater than a preset thickness are screened out according to the initial category label of each pixel, and the target tropical cyclone area where the regional tropical cyclone composed of the target pixels is located is used as the tropical cyclone cloud block identification result.

[0032] In one embodiment, the extrapolated grayscale cloud image can be semantically segmented based on tropical cyclone clouds through tropical cyclone cloud recognition models such as machine learning and deep learning to obtain tropical cyclone cloud recognition results; wherein, the tropical cyclone cloud recognition model can be obtained through pre-annotated historical grayscale cloud images; specifically, historical grayscale cloud image data is obtained, and the historical grayscale cloud image data is pre-annotated based on the tropical cyclone area to obtain an annotation result, and the historical grayscale cloud image data and its corresponding annotation result are input as a training set into a pre-constructed machine learning or deep learning model for model training; during the model training process, the parameters of the model are continuously adjusted so that the pre-constructed machine learning or deep learning model can extract the semantic information represented in the historical grayscale cloud image, and output a tropical cyclone cloud recognition result based on the semantic information, the error with the annotation result satisfies a preset error.

[0033] Step S104: extracting the movement characteristics of the tropical cyclone from the extrapolated grayscale cloud image based on the tropical cyclone cloud block identification result.

[0034] In this embodiment, the movement characteristics of the tropical cyclone are extracted from the extrapolated grayscale cloud image based on the tropical cyclone cloud block identification result to perform tropical cyclone tracking.

[0035] Among them, illustratively, the extrapolated cloud map data is a cloud map representation every 10 minutes for the next 4 hours. Therefore, the movement characteristics of the tropical cyclone can be determined by comparing the position of the tropical cyclone in the cloud map before and after 10N minutes, where N is a positive integer not greater than 12; it should be understood that the smaller N is, the higher the accuracy of the extracted tropical cyclone movement characteristics.

[0036] In one embodiment, N is 1, that is, the movement characteristics of the tropical cyclone are determined by comparing the positions of the tropical cyclone in the cloud map 10 minutes before and after; specifically, the speed and direction of the tropical cyclone can be determined based on the movement vector of the pixel corresponding to the identified tropical cyclone vortex center in the cloud map 10 minutes before and after, and the wind speed characteristics are extracted based on the target tropical cyclone area, and the intensity level of the tropical cyclone is determined based on the wind speed characteristics; the intensity of the tropical cyclone, using speed, direction and intensity as the movement characteristics of the tropical cyclone.

[0037] Step S105: issuing a tropical cyclone warning to each photovoltaic station based on the movement characteristics.

[0038] In one embodiment, in order to implement tropical cyclone warnings for each photovoltaic station, time and space matching can be performed based on the movement characteristics of the tropical cyclone and the location of each photovoltaic station to obtain the predicted tropical cyclone speed of the photovoltaic station, and further divide the warning level according to the predicted tropical cyclone speed to send early warning warnings to the photovoltaic stations.

[0039] In one embodiment, the spatial matching of the photovoltaic stations may be accomplished by a proximity matching method.

[0040] In one embodiment, in order to match the 15-minute time resolution of the ultra-short-term prediction, the movement features are averaged forward and backward with a 15-minute resolution to achieve time matching.

[0041] In one embodiment, Table 1 is a schematic tropical cyclone grade forecast rule, see Table 1:

[0042] Table 1. Tropical cyclone grade forecast rules

[0043]

[0044] The method for early warning of tropical cyclones for photovoltaic stations of the present invention obtains a fused grayscale cloud image of satellite data; wherein the fused grayscale cloud image is obtained by fusing cloud image data of at least two different channels; the fused grayscale cloud image is input into a pre-trained Res-UNet model, and the temporal features of the fused grayscale cloud image are extracted based on the Res-UNet model to obtain an extrapolated grayscale cloud image; semantic segmentation of tropical cyclone cloud blocks is performed on the extrapolated grayscale cloud image to obtain a tropical cyclone cloud block recognition result; based on the tropical cyclone cloud block recognition result, the movement features of the tropical cyclone are extracted from the extrapolated grayscale cloud image; and tropical cyclone early warning is performed on each photovoltaic station based on the movement features, so that tropical cyclone prediction based on satellite data for photovoltaic stations can be realized.

[0045] As an exemplary embodiment, the Res-UNet model adopts multiple encoders and decoders of the U-Net model, each of the encoders and decoders is connected with multiple sub-modules, each sub-module includes multiple residual connections, and each residual block includes a shortcut to achieve cross-layer connection.

[0046] In this embodiment, the Res-UNet model is built based on U-Net, which adds residual connections on the basis of U-Net; specifically, each encoder and decoder is connected with multiple sub-modules, each sub-module includes multiple residual connections, and each residual block includes a shortcut to achieve cross-layer connection, which can better extract low-frequency information and alleviate the problem of missing semantic information.

[0047] As an exemplary embodiment, the method for training the Res-UNet model includes: obtaining a historical fused grayscale cloud map of the satellite data; dividing the historical fused grayscale cloud map to obtain a first historical fused grayscale cloud map within a first time period and a second historical fused grayscale cloud map within a second time period; wherein the first time period is earlier than the second time period; using the first historical fused grayscale cloud map as an input feature and the second historical fused grayscale cloud map as a prediction label to train a pre-constructed Res-UNet model; during the model training process, continuously adjusting the model parameters of the constructed Res-UNet model so that the Res-UNet model can output the second historical fused grayscale cloud map based on the temporal change characteristics of the first historical fused grayscale cloud map.

[0048] In this embodiment, the time scale of the historical fused grayscale cloud map is not less than 8 hours; the method for obtaining the historical fused grayscale cloud map can refer to the content recorded in step S101, which will not be repeated here.

[0049] Exemplarily, the technical solution of this embodiment is explained by taking the time scale of the historical fused grayscale cloud map as 8h as an example; after obtaining the historical fused grayscale cloud map with a time scale of 8h of the satellite data, the historical fused grayscale cloud map is divided to obtain a first historical fused grayscale cloud map within 0h-4h and a second historical fused grayscale cloud map within 4h-8h; the first historical fused grayscale cloud map within 0h-4h is used as an input feature, and 4h-8h is used as a prediction label to train a pre-constructed Res-UNet model; during the model training process, the model parameters of the constructed Res-UNet model are continuously adjusted so that the Res-UNet model can output the second historical fused grayscale cloud map based on the temporal change characteristics of the first historical fused grayscale cloud map.

[0050] As an exemplary embodiment, the extraction of the movement characteristics of the tropical cyclone in the extrapolated grayscale cloud map based on the tropical cyclone cloud block identification result includes: calculating the tropical cyclone wind speed of each of the extrapolated grayscale cloud maps based on the tropical cyclone cloud block identification result; calculating the movement vector of each tropical cyclone in the extrapolated grayscale cloud map at every two adjacent moments and the time difference of the grayscale cloud map based on the tropical cyclone cloud block identification result; calculating the speed and direction of the tropical cyclone based on the time difference and the movement vector; traversing each of the extrapolated grayscale cloud maps, synthesizing the speed, direction and wind speed of each tropical cyclone to obtain the movement trajectory of the tropical cyclone as the movement characteristic.

[0051] In this embodiment, the tropical cyclone wind speed of each of the extrapolated grayscale cloud images is calculated based on the tropical cyclone cloud block identification result; specifically, the wind speed characteristics of the target tropical cyclone area corresponding to the tropical cyclone cloud block identification result are extracted, and the tropical cyclone wind speed of each of the extrapolated grayscale cloud images is calculated based on the wind speed characteristics.

[0052] At the same time, based on the tropical cyclone cloud block identification results, the movement vector of each tropical cyclone in the extrapolated grayscale cloud map between every two adjacent moments is calculated; specifically, the movement vector can be determined by the unit pixel distance of the pixel point in the extrapolated grayscale cloud map between two adjacent moments in the horizontal and vertical directions and the actual distance represented by the unit pixel distance.

[0053] Furthermore, each of the extrapolated grayscale cloud images is traversed to obtain the speed, direction and intensity of the tropical cyclone in each extrapolated grayscale cloud image, and the speed, direction and intensity of each tropical cyclone are further synthesized to obtain the moving trajectory of the tropical cyclone as the moving feature.

[0054] In one embodiment, the tropical cyclone may be synthesized according to the time sequence of the extrapolated grayscale cloud images corresponding to the speed, direction, and intensity results of the tropical cyclone to obtain the moving track of the tropical cyclone.

[0055] As an exemplary embodiment, the method of obtaining a fused grayscale cloud map of satellite data includes: obtaining first satellite data of a visible light channel and second satellite data of an infrared channel of a target satellite; determining first channel grayscale data and second channel grayscale data based on the first satellite data and the second satellite data; and fusing the first channel grayscale data and the second channel grayscale data according to a preset weight to obtain the fused grayscale cloud map.

[0056] Specifically, the first satellite data and the second satellite data can be obtained by acquiring the visible light channel data and the infrared channel data of the sunflower satellite L1, and further processing the visible light data and the infrared channel data to obtain the first channel grayscale data and the second channel grayscale data, respectively; exemplarily, during the processing, atmospheric correction, standardization, and background color removal are performed in sequence to generate grayscale cloud map data corresponding to the visible light channel and grayscale cloud map data corresponding to the infrared channel; further, the grayscale cloud map data corresponding to the visible light channel and the grayscale cloud map data corresponding to the infrared channel are fused with equal weights to obtain a fused grayscale cloud map.

[0057] As an exemplary embodiment, the extrapolated grayscale cloud image is subjected to semantic segmentation based on tropical cyclone cloud blocks to obtain a tropical cyclone cloud block recognition result, including: inputting the extrapolated grayscale cloud image into a pre-trained convolutional neural network, extracting the cloud block thickness and vortex center in the extrapolated grayscale cloud image based on semantic features represented by the convolutional neural network for the grayscale cloud image region, and obtaining a cloud block having an actual thickness greater than a preset thickness and a vortex center as a tropical cyclone cloud block.

[0058] In this embodiment, the semantic segmentation algorithm based on deep learning uses convolutional neural network classification (fully convolutional network). Unlike ordinary networks, the classification layer of the convolutional neural network classification is a convolutional layer; specifically, the convolutional neural network assigns an initial category label to each pixel, and the encoder is expanded to an encoder-decoder architecture, in which the encoder uses a convolutional layer to extract the feature map of the extrapolated grayscale cloud map, and uses a pooling layer to reduce the size of the feature map to make it a lower dimensional representation; the decoder receives this representation and restores the spatial dimension by performing upsampling through transposed convolution, so that each transposed convolution can expand the size of the feature map.

[0059] Specifically, when the extrapolated grayscale cloud image is input into a pre-trained convolutional neural network, the convolution layer based on the convolutional neural network extracts the extrapolated grayscale cloud image to obtain a feature map of the extrapolated grayscale cloud image; further, the pooling layer performs a downsampling operation on the feature map to reduce the spatial dimension of the data while retaining important features, and further assigns an initial category label to each pixel through the fully connected layer as a semantic feature representing each pixel; further, based on the semantic features, the cloud block thickness and vortex center in the extrapolated grayscale cloud image are extracted to obtain a cloud block with an actual thickness greater than a preset thickness and a vortex center as a tropical cyclone cloud block.

[0060] This embodiment provides a tropical cyclone early warning device for a photovoltaic station, such as Figure 2 As shown, including:

[0061] The acquisition module 501 is used to acquire a fused grayscale cloud image of satellite data; wherein the fused grayscale cloud image is obtained by fusing cloud image data of at least two different channels;

[0062] A cloud map extrapolation module 502 is used to input the fused grayscale cloud map into a pre-trained Res-UNet model, extract the temporal features of the fused grayscale cloud map based on the Res-UNet model, and obtain an extrapolated grayscale cloud map;

[0063] The semantic segmentation module 503 is used to perform semantic segmentation of the extrapolated grayscale cloud image based on tropical cyclone cloud blocks to obtain tropical cyclone cloud block recognition results;

[0064] A movement feature extraction module 504, configured to extract movement features of the tropical cyclone in the extrapolated grayscale cloud image based on the tropical cyclone cloud block identification result;

[0065] The early warning module 505 is used to issue a tropical cyclone early warning to each photovoltaic station based on the movement characteristics.

[0066] It should be noted here that the examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the contents disclosed in the above embodiments.

[0067] It should be noted that the above modules as part of the device can be implemented by software or hardware, wherein the hardware environment includes a network environment.

[0068] An embodiment of the present invention also provides a computer device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus, the memory is used to store a computer program; the processor is used to execute a method in any of the above embodiments by running the computer program stored in the memory.

[0069] Figure 3 is a structural block diagram of an optional computer device according to an embodiment of the present application, such as Figure 3 As shown, it includes a processor 10, a communication interface 20, a memory 30 and a communication bus 40, wherein the processor 10, the communication interface 20 and the memory 30 communicate with each other through the communication bus 40, wherein:

[0070] A memory 30, for storing computer programs;

[0071] The processor 10 is used to implement the tropical cyclone early warning method for a photovoltaic station as described in any of the above embodiments when executing the computer program stored in the memory 30.

[0072] Optionally, in this embodiment, the communication bus may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0073] The communication interface is used for communication between the above-mentioned computer device and other devices.

[0074] The memory may include RAM, or may include non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0075] The above-mentioned processor can be a general-purpose processor, which can include but not be limited to: CPU (Central Processing Unit), NP (Network Processor), etc.; it can also be DSP (Digital Signal Processing), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0076] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments, and this embodiment will not be described in detail here.

[0077] It can be understood by those skilled in the art that Figure 3 The structure shown is for illustration only. The device for implementing any one of the methods in the above embodiments may be a terminal device, which may be a smart phone (such as an Android phone, an IOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (Mobile Internet Devices, MID), a PAD, or other terminal devices. Figure 3 It does not limit the structure of the above electronic device. For example, the terminal device may also include Figure 3 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 3 Different configurations shown.

[0078] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, which can include: a flash drive, ROM, RAM, a magnetic disk or an optical disk, etc.

[0079] As an exemplary embodiment, the present application also provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute any one of the method steps of the present embodiment when running.

[0080] Optionally, in this embodiment, the above-mentioned storage medium can be used to execute the program code of the method steps of the embodiment of the present application.

[0081] Optionally, in this embodiment, the storage medium may be located on at least one network device among a plurality of network devices in the network shown in the above embodiment.

[0082] Optionally, in this embodiment, the storage medium is configured to store data for executing the method in the above embodiment.

[0083] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments, which will not be described in detail in this embodiment.

[0084] Optionally, in this embodiment, the storage medium may include, but is not limited to, various media that can store program codes, such as a USB flash drive, a ROM, a RAM, a mobile hard disk, a magnetic disk, or an optical disk.

[0085] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0086] If the integrated units in the above embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling one or more computer devices (which may be personal computers, servers or network devices, etc.) to execute all or part of the steps of the method in the above embodiments.

[0087] In the several embodiments provided in the present application, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, 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 mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0088] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution provided in this embodiment.

[0089] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0090] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0091] The above are only preferred implementations of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A tropical cyclone early warning method for a photovoltaic station, characterized in that: The tropical cyclone early warning method for photovoltaic stations comprises: Acquire a fused grayscale cloud image of satellite data; wherein the fused grayscale cloud image is obtained by fusing cloud image data of at least two different channels; Inputting the fused grayscale cloud image into a pre-trained Res-UNet model, extracting the temporal features of the fused grayscale cloud image based on the Res-UNet model, and obtaining an extrapolated grayscale cloud image; Performing semantic segmentation based on tropical cyclone cloud blocks on the extrapolated grayscale cloud image to obtain tropical cyclone cloud block recognition results; Extracting the movement characteristics of the tropical cyclone in the extrapolated grayscale cloud image based on the tropical cyclone cloud block identification result; Based on the movement characteristics, tropical cyclone warnings are issued to each photovoltaic station.

2. The tropical cyclone early warning method for photovoltaic stations according to claim 1, characterized in that: The Res-UNet model adopts multiple encoders and decoders of the U-Net model, each of the encoders and decoders is connected to multiple sub-modules, each sub-module contains multiple residual connections, and each residual block includes a shortcut to achieve cross-layer connection.

3. The tropical cyclone early warning method for photovoltaic stations according to claim 2, characterized in that: The method of training the Res-UNet model includes: Obtaining a historical fused grayscale cloud image of the satellite data; Dividing the historical fused grayscale cloud map to obtain a first historical fused grayscale cloud map within a first time period and a second historical fused grayscale cloud map within a second time period; wherein the first time period is earlier than the second time period; The first historical fused grayscale cloud map is used as an input feature, and the second historical fused grayscale cloud map is used as a prediction label to train a pre-constructed Res-UNet model; during the model training process, the model parameters of the constructed Res-UNet model are continuously adjusted so that the Res-UNet model can output the second historical fused grayscale cloud map based on the temporal change characteristics of the first historical fused grayscale cloud map.

4. The tropical cyclone early warning method for photovoltaic stations according to claim 1, characterized in that: The extracting the movement characteristics of the tropical cyclone from the extrapolated grayscale cloud image based on the tropical cyclone cloud block identification result includes: Calculating the tropical cyclone wind speed of each of the extrapolated grayscale cloud images based on the tropical cyclone cloud block identification result; Calculate the moving vector of each tropical cyclone in the extrapolated grayscale cloud image and the time difference of the grayscale cloud image between two adjacent moments based on the tropical cyclone cloud block identification result; Calculate the speed and direction of the tropical cyclone based on the time difference and the movement vector; The extrapolated grayscale cloud images are traversed to synthesize the speed, direction and wind speed of each tropical cyclone to obtain the moving track of the tropical cyclone as the moving feature.

5. The tropical cyclone early warning method for photovoltaic stations according to claim 1, characterized in that: The step of obtaining a fused grayscale cloud image of satellite data includes: Acquire first satellite data of a visible light channel and second satellite data of an infrared channel of a target satellite; Determine first channel grayscale data and second channel grayscale data based on the first satellite data and the second satellite data; The first channel grayscale data and the second channel grayscale data are fused according to a preset weight to obtain the fused grayscale cloud map.

6. The tropical cyclone early warning method for photovoltaic stations according to claim 1, characterized in that: The performing semantic segmentation of the extrapolated grayscale cloud image based on tropical cyclone cloud blocks to obtain tropical cyclone cloud block recognition results includes: The extrapolated grayscale cloud image is input into a pre-trained convolutional neural network, and the cloud block thickness and vortex center in the extrapolated grayscale cloud image are extracted based on the semantic features represented by the convolutional neural network based on the grayscale cloud image area, so that the cloud blocks with actual thickness greater than the preset thickness and with vortex centers are obtained as tropical cyclone cloud blocks.

7. The tropical cyclone early warning method for photovoltaic stations according to claim 1, characterized in that: The tropical cyclone warning for each photovoltaic station based on the movement characteristics includes: Obtaining location information of each photovoltaic station; Performing spatiotemporal matching on the movement characteristics based on the position information to obtain the cyclone wind speed of each photovoltaic station; The early warning information of each photovoltaic station is obtained based on the cyclone wind speed.

8. A tropical cyclone early warning device for a photovoltaic station, characterized in that: The device comprises: An acquisition module, used to acquire a fused grayscale cloud image of satellite data; wherein the fused grayscale cloud image is obtained by fusing cloud image data of at least two different channels; A cloud map extrapolation module is used to input the fused grayscale cloud map into a pre-trained Res-UNet model, extract the temporal features of the fused grayscale cloud map based on the Res-UNet model, and obtain an extrapolated grayscale cloud map; A semantic segmentation module, used for performing semantic segmentation of the extrapolated grayscale cloud image based on tropical cyclone cloud blocks to obtain tropical cyclone cloud block recognition results; A movement feature extraction module, used for extracting the movement feature of the tropical cyclone in the extrapolated grayscale cloud image based on the tropical cyclone cloud block identification result; The early warning module is used to issue tropical cyclone early warning to each photovoltaic station based on the movement characteristics.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the tropical cyclone early warning method for a photovoltaic station according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the tropical cyclone early warning method for a photovoltaic station according to any one of claims 1 to 7.

Images