Strong convective weather forecasting and early warning method, system and equipment based on satellite data
By using multi-band satellite data fusion and generative deep learning models to identify the regions and intensities of severe convective weather, the problem of low forecast accuracy in existing technologies has been solved, enabling efficient early warning and safety assurance for photovoltaic power plants.
Patent Information
- Application Number
- CN202511713412.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-20
- Publication Date
- 2026-02-03
AI Technical Summary
Existing technologies suffer from low forecast accuracy in convective weather forecasting, particularly in their insufficient prediction of cloud formation and dissipation processes, leading to distorted forecast results.
Multi-band satellite data fusion processing is used to generate fused satellite cloud images. Generative deep learning prediction models are used to identify the regions and intensities of severe convective weather by comparing the characteristic parameters of cloud physical properties with preset thresholds. Combined with an early warning system, this provides early warnings for photovoltaic power plants.
This improves the accuracy and precision of severe convective weather forecasts, ensuring the safe operation of photovoltaic power plants and the stability of the power grid.
Smart Images
Figure CN121454652A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of weather forecasting, in particular to a strong convective weather forecasting and warning method, system and device based on satellite data. BACKGROUND
[0002] Strong convective weather such as thunderstorm, hail, etc. poses a serious threat to the safe operation of photovoltaic stations and the stability of the power grid. Therefore, it is crucial to accurately forecast and warn strong convective weather for photovoltaic stations.
[0003] In the prior art, some traditional nowcasting methods, such as Lagrangian persistence forecasting method based on cloud cluster movement vector extrapolation, can predict the movement of cloud clusters, but as the forecasting time increases, the prediction accuracy will decrease sharply, especially for the new generation and dissipation process of cloud clusters, which cannot be effectively predicted, resulting in problems such as cloud cluster accumulation and shape distortion in the prediction results. Some methods use traditional machine learning algorithms such as logistic regression and random forest to analyze satellite cloud images, but such methods have limited ability to capture complex, long-term nonlinear spatiotemporal evolution relationships in cloud image sequences, making it difficult to mine deep features, resulting in low prediction accuracy. SUMMARY
[0004] Therefore, the purpose of the present application is to provide a strong convective weather forecasting and warning method, system and device based on satellite data to overcome the problem of poor prediction of strong convective weather.
[0005] To achieve the above purpose, the present application adopts the following technical solutions: In a first aspect, the present application provides a strong convective weather forecasting and warning method based on satellite data, comprising: obtaining multi-band satellite data and performing fusion processing on the multi-band satellite data to generate a fused satellite cloud image; wherein the multi-band satellite data includes water vapor band data, visible light band data and infrared band data; inputting the fused satellite cloud image into a pre-trained generative deep learning prediction model to obtain a predicted fused satellite cloud image of a future time period; In the predicted fused satellite cloud image of the future time period, by comparing the feature parameters representing the physical characteristics of the cloud cluster with the preset recognition threshold, the area and intensity of the strong convective weather are identified; Based on the identified area and intensity of the strong convective weather, a strong convective weather warning is given for photovoltaic stations.
[0006] Further, in some embodiments of the present application, the multi-band satellite data is fused to generate a fused satellite cloud image, comprising: performing missing data filling on the multi-band satellite data; Based on the scaling processing, the visible light band data, the water vapor band data and the infrared band data in the multi-band satellite data filled with the missing data are scaled to a first preset range, and are saved as a visible light satellite cloud image, a water vapor satellite cloud image and an infrared satellite cloud image respectively; The visible light satellite cloud image, the water vapor satellite cloud image and the infrared satellite cloud image are fused, and a non-cloud area filtering processing is performed based on a threshold method and a time sequence moving comparison method, and are rescaled to the first preset range, to obtain the fused satellite cloud image.
[0007] Further, in some embodiments of the present application, the training process of the generative deep learning prediction model comprises: taking the WGAN network or the Res-UNet network as a basic network; taking the fused satellite cloud images arranged in time sequence within 24 hours before the target time as features, and taking the fused satellite cloud images arranged in time sequence within 24 hours after the target time as labels, to train the basic network to obtain the generative deep learning prediction model.
[0008] Further, in some embodiments of the present application, the feature parameter is a gray value of the fused satellite cloud image; In the fused satellite cloud image of the predicted future period, by comparing the feature parameter representing the physical characteristics of the cloud cluster with a preset identification threshold, the area and intensity of the severe convective weather are identified, comprising: scaling the gray value in the predicted future period of the fused satellite cloud image to a second preset range; identifying the area with the gray value greater than the preset identification threshold as the area of the severe convective weather, and determining the intensity of the severe convective weather based on the interval range where the gray value is located.
[0009] Further, in some embodiments of the present application, further comprising: counting the area of the severe convective weather and the power generation field station in the area within the prediction period; taking the time information of the prediction period as a time axis, to generate the moving track of the future severe convective weather.
[0010] Further, in some embodiments of the present application, the severe convective weather alarm for the photovoltaic power station based on the identified area and intensity of the severe convective weather comprises: determining the corresponding relationship between the gray value in the fused satellite cloud image and the preset warning standard; determining the warning information based on the gray value in the predicted future period of the fused satellite cloud image and the corresponding relationship, the warning information comprising a warning type, a warning level, an intensity type and a warning description.
[0011] Further, in some embodiments of the present application, the strong convective weather warning for the photovoltaic power station based on the identified area and intensity of the strong convective weather further comprises: spatially matching the identified area of the strong convective weather with the actual photovoltaic power station; based on the matching result, sending the early warning information to the corresponding photovoltaic power station.
[0012] Further, in some embodiments of the present application, it further comprises: using the newly acquired multi-band satellite data as a training sample to incrementally train the generative deep learning prediction model.
[0013] In a second aspect, the embodiments of the present application provide a strong convective weather forecasting and warning system based on satellite data, comprising: a fusion module configured to acquire multi-band satellite data and perform fusion processing on the multi-band satellite data to generate a fused satellite cloud image; wherein the multi-band satellite data comprises water vapor band data, visible light band data and infrared band data; a prediction module configured to input the fused satellite cloud image into a pre-trained generative deep learning prediction model to obtain a predicted fused satellite cloud image of a future time period; a forecasting module configured to identify an area and intensity of strong convective weather in the predicted fused satellite cloud image of the future time period by comparing a feature parameter representing physical characteristics of a cloud cluster with a preset identification threshold; a warning module configured to perform strong convective weather warning for a photovoltaic power station based on the identified area and intensity of the strong convective weather.
[0014] In a third aspect, the embodiments of the present application provide a strong convective weather forecasting and warning device based on satellite data, comprising a processor and a memory, wherein the processor is connected to the memory: wherein the processor is configured to call and execute a program stored in the memory; the memory is configured to store the program, and the program is at least used to execute the strong convective weather forecasting and warning method based on satellite data.
[0015] The application relates to the technical field of weather forecasting, in particular to a strong convective weather forecasting and early warning method, system and equipment based on satellite data. BRIEF DESCRIPTION OF DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.
[0017] Figure 1 FIG. 1 is a flowchart of the strong convective weather forecasting and early warning method based on satellite data provided by the embodiment of the present application; Figure 2 FIG. 2 is a structural diagram of the strong convective weather forecasting and early warning system based on satellite data provided by the embodiment of the present application; Figure 3 FIG. 3 is a structural diagram of the strong convective weather forecasting and early warning equipment based on satellite data provided by the embodiment of the present application. DETAILED DESCRIPTION
[0018] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described in detail below. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of the present application.
[0019] Figure 1 FIG. 1 is a flowchart of the strong convective weather forecasting and early warning method based on satellite data provided by the embodiment of the present application, please refer to Figure 1 The embodiment can include the following steps: S101, acquiring multi-band satellite data and performing fusion processing on the multi-band satellite data to generate a fused satellite cloud image.
[0020] The multi-band satellite data includes water vapor band data, visible light band data, and infrared band data.
[0021] Specifically, in the present application, the satellite is a stationary observation satellite, also known as a geostationary satellite, including Fengyun satellites, GOES series satellites, or Himawari satellites, etc., which are high temporal resolution observations that remain relatively stationary above a specific location on the Earth, capable of minute-level observations of the weather above the photovoltaic power station in the observation area.
[0022] In addition, in the present application, severe convective weather refers to convective wind (≥17.2 m / s) accompanied by thunderstorm phenomena, hail, short-term heavy precipitation (≥20 mm / h), and cloud layers with thick and white cloud features, which is one of the most devastating weather disasters, can damage photovoltaic component facilities, reduce photovoltaic power generation power, cause severe fluctuations in photovoltaic power station output, and even threaten the safety and stability of the power grid.
[0023] S102, input the fused satellite cloud image into the pre-trained generative deep learning prediction model to obtain a predicted fused satellite cloud image of a future period.
[0024] S103, in the predicted fused satellite cloud image of the future period, by comparing the feature parameters representing the physical characteristics of the cloud cluster with the preset identification threshold, the area and intensity of the severe convective weather are identified.
[0025] S104, based on the identified area and intensity of the severe convective weather, a severe convective weather alarm is performed for the photovoltaic power station.
[0026] The severe convective weather forecasting and warning method based on satellite data provided in the present application, by acquiring multi-band satellite data and performing fusion processing on the multi-band satellite data to generate a fused satellite cloud image; wherein the multi-band satellite data includes water vapor band data, visible light band data, and infrared band data; input the fused satellite cloud image into the pre-trained generative deep learning prediction model to obtain a predicted fused satellite cloud image of a future period; in the predicted fused satellite cloud image of the future period, by comparing the feature parameters representing the physical characteristics of the cloud cluster with the preset identification threshold, the area and intensity of the severe convective weather are identified; based on the identified area and intensity of the severe convective weather, a severe convective weather alarm is performed for the photovoltaic power station, which can greatly improve the effect of severe convective weather forecasting and ensure the safety of the photovoltaic power station.
[0027] Further, in some embodiments of the present application, the multi-band satellite data is fused to generate a fused satellite cloud image, including: filling missing data of the multi-band satellite data; based on scaling processing, scaling the visible light band data, the water vapor band data and the infrared band data in the multi-band satellite data after filling the missing data to be within a first preset range, and saving them as a visible light satellite cloud image, a water vapor satellite cloud image and an infrared satellite cloud image respectively; and fusing the visible light satellite cloud image, the water vapor satellite cloud image and the infrared satellite cloud image, and performing non-cloud area filtering processing based on threshold method and time sequence moving comparison method, and rescaling to be within the first preset range to obtain the fused satellite cloud image.
[0028] Specifically, the embodiments of the present application can use Himawari for prediction and early warning. In actual application, first, Himawari multi-band satellite data such as satellite L1 data is acquired through Himawari satellite visible light and infrared sensors, Himawari satellite visible light and infrared L1 data is downloaded, and stored in a cloud database. Then, the satellite L1 data and power generation data can be aligned according to time stamp to construct a time sequence data set, wherein the data acquisition frequency is 10 minutes / time.
[0029] Then, the data is preprocessed, including: using the mean value method of the front and rear time data to fill the missing data; and using the scaling processing method to scale the visible light band data, the water vapor band data and the infrared band data into a first preset range such as 0-255, and save them as visible light satellite cloud image, water vapor satellite cloud image and infrared satellite cloud image gray scale image respectively.
[0030] On this basis, the image fusion method of Gamma correction transition zone gray scale invariable principle is used to fuse the satellite cloud image, and the visible light satellite cloud image, the water vapor satellite cloud image and the infrared satellite cloud image are fused into a new satellite cloud image.
[0031] Then, the threshold method and the time sequence moving comparison method are used to remove the low image of the above fusion result, that is, to perform non-cloud area filtering processing, and the land, desert, snow mountain, river and lake are removed (the desert and snow mountain are set to 0 by using the time sequence moving comparison method to compare the cloud thickness and position of the cloud-like pixels moving with the front and rear time), and then rescaled to 0-1 to generate a satellite cloud image of 0-255 to obtain the fused satellite cloud image, so as to retain a reasonable and clean satellite cloud image.
[0032] Further, in some embodiments of the present application, the training process of the generative deep learning prediction model comprises: taking the WGAN network or the Res-UNet network as the basic network; taking the time-sequenced fusion satellite cloud image of 24 hours before the target time as the feature, and taking the time-sequenced fusion satellite cloud image of 24 hours after the target time as the label, training the basic network to obtain the generative deep learning prediction model.
[0033] Specifically, in the present application, based on the historical time fusion satellite cloud image, a generative deep learning algorithm (such as WGAN network, Res-UNet network, etc.) is used to generate a predicted fusion satellite cloud image of a future period.
[0034] In the training stage, the model can take the time-sequenced data (i.e. the time-sequenced fusion satellite cloud image) of 24 hours before the target time as the feature, and the time-sequenced data of 24 hours after the target time as the label, to train the generative deep learning prediction model. In actual application, the predicted power of the photovoltaic power station can be reported at the time of the fusion satellite cloud image prediction, so as to complete the prediction of the fusion satellite cloud image and obtain the fusion satellite cloud image prediction data, i.e. the predicted fusion satellite cloud image of the future period.
[0035] As mentioned above, in the present application, the basic network of the prediction model can be a WGAN network or a Res-UNet network.
[0036] Specifically, WGAN is an improved version of generative adversarial network (GAN) (Wasserstein Generative Adversarial Networks), which introduces the Wasserstein distance (also known as Earth Mover's Distance, EMD) as the loss function, so as to solve the problems of mode collapse and unstable training encountered in the training process of traditional GAN, thereby significantly improving the stability and generation quality of GAN. In the present application, WGAN outperforms Res-UNet network in the clarity of the satellite cloud image prediction result by relying on its unique loss function.
[0037] In the present application, the WGAN network is modified as follows compared with the GAN network: The last layer of the discriminator removes the sigmoid; The loss of the generator and the discriminator does not take the log; After updating the parameters of the discriminator each time, their absolute values are truncated to not more than a fixed constant c; No momentum-based optimization algorithm (including momentum and Adam) is used, and RMSProp and SGD are recommended.
[0038] On this basis, the current generated fusion satellite cloud picture is input into the pre-trained generative deep learning prediction model to obtain a predicted fusion satellite cloud picture of a future period, and then the predicted fusion satellite cloud picture of the future period is used for prediction and early warning of severe convective weather, including identifying the area and intensity of the severe convective weather in the predicted fusion satellite cloud picture of the future period by comparing the feature parameters representing the physical characteristics of the cloud cluster with a preset identification threshold. The identification process specifically includes first scaling the gray scale in the predicted fusion satellite cloud picture of the future period to be within a second preset range; then identifying the area with a gray scale value greater than the preset identification threshold as the area of the severe convective weather, and determining the intensity of the severe convective weather based on the interval range in which the gray scale value is located.
[0039] First of all, it needs to be pointed out that the occurrence of severe convective weather is often accompanied by cumulonimbus clouds, cloud layer clouds, and thick and white cloud clusters. In the present application, by incorporating the water vapor satellite cloud picture into the fusion satellite cloud picture, the recognition accuracy of cumulonimbus clouds and cloud layer clouds can be improved, thereby accurately completing the recognition of severe convective weather based on satellite data.
[0040] Specifically, first, the gray scale value in the predicted fusion satellite cloud picture is read, and weather division is performed based on the read gray scale value, such as scaling the gray scale value of the predicted fusion satellite cloud picture within the range of 0-1. In the present application, it is determined through statistics and existing knowledge that cumulonimbus clouds and cloud layer clouds appear above 0.655, so 0.655 is taken as the identification threshold, and the positions of the predicted fusion satellite cloud picture above 0.655 are identified as areas where severe convective weather occurs, and the intensity of the severe convective weather can be identified based on the specific gray scale value.
[0041] Further, in some embodiments of the present application, the area of the severe convective weather identified within the prediction period and the power generation field station within the area are also included, and the moving track of the future severe convective weather is generated with the time information of the prediction period as the time axis.
[0042] Specifically, according to the result of identifying the severe convective weather, the areas of the severe convective weather in each predicted fusion satellite cloud picture within the prediction period are counted to obtain the cloud cluster thickness, direction and position of the severe convective weather with the prediction period as the time axis, and finally the moving track of the future severe convective weather is formed based on the prediction time, cloud cluster thickness, cloud cluster direction and cloud cluster position, which facilitates the photovoltaic field station and power grid to take advance measures and coordinate strategy adjustment.
[0043] Further, in some embodiments of the present application, based on the identified area and intensity of severe convective weather, a severe convective weather warning is given for a photovoltaic power station, including: determining the correspondence between the gray scale in the fused satellite cloud image and the preset warning standard; determining the warning information based on the gray scale in the fused satellite cloud image of the predicted future period and the correspondence, wherein the warning information includes warning type, warning level, intensity type and warning description.
[0044] Specifically, the scaled gray scale value of the fused satellite cloud image can be divided into corresponding first, second, third and fourth warning levels according to industry standards, respectively corresponding to red, orange, yellow and blue warning of the severe convective weather standard warning, realizing four-level division of severe convective warning, so as to generate warning information (which can include warning type, warning level, intensity type and warning description, etc. as shown in the following table, of course, in some embodiments, the warning information can also include the cloud thickness range and cloud type in Table 1) after identification based on gray scale (the actual principle is to reflect the cloud thickness through the gray scale value, and to determine the prediction result based on the cloud thickness range, as shown in "cloud thickness range" in Table 1 below): Table 1 Warning division table
[0045] On this basis, the identified area of severe convective weather is spatio-temporally matched with the actual photovoltaic power station; based on the matching result, the warning information is sent to the corresponding photovoltaic power station and the warning is displayed.
[0046] Specifically, according to the above identification and prediction of the position and intensity of severe convective weather, spatio-temporal matching is performed with the photovoltaic power station. For example, spatial matching can be completed by using a nearby matching method, and time alignment is performed between the 15-minute resolution of the photovoltaic power station power and the prediction result of the severe convective weather, and then corresponding warning is performed.
[0047] In addition, in some embodiments of the present application, dynamic visualization of warning results and coordinated operation with the power grid can also be performed. For example, the warning information and the future movement trajectory of the severe convective weather are displayed in real time through a visualization platform; and the prediction result and the warning information are linked with the power grid dispatching system to report the power generation strategy (such as power generation plan) according to the warning information in real time.
[0048] In addition, in some embodiments of the present application, the newly acquired multi-band satellite data processed as training samples is periodically used to incrementally train the generative deep learning prediction model (the specific training process and principle can be understood by referring to the introduction of the above model training phase, which will not be described here) to optimize the model and ensure continuous improvement of the warning capability.
[0049] The method for forecasting and warning strong convective weather based on satellite data provided in the application can forecast high-definition, high-time resolution and high-quality satellite cloud images by introducing geostationary satellite data, fusing visible light satellite cloud images, water vapor satellite cloud images and infrared satellite cloud images, and using WGAN neural network training, can identify strong convective weather by using cloud cover as a threshold when cumulonimbus clouds and rain layer clouds cause strong convective weather, and can obtain fusion satellite cloud image feature data and fusion satellite cloud image prediction feature data by introducing geostationary satellite data, issue warning types, warning levels, strong convective weather intensity and strong convective weather area of strong convective weather through strong convective weather warning level information threshold falling area, and form future period strong convective weather track by statistically timing strong convective weather area, thereby improving the strong convective weather warning capability of photovoltaic power station.
[0050] Based on the same inventive concept, the application also provides a strong convective weather forecasting and warning system based on satellite data for implementing the above method embodiments, Figure 2 is a structural schematic diagram of the strong convective weather forecasting and warning system based on satellite data provided in the embodiments of the application, as Figure 2 shown, the system comprises: A fusion module 11 is configured to obtain multi-band satellite data and perform fusion processing on the multi-band satellite data to generate a fusion satellite cloud image, wherein the multi-band satellite data comprises water vapor band data, visible light band data and infrared band data. A prediction module 12 is configured to input the fusion satellite cloud image into a pre-trained generative deep learning prediction model to obtain a predicted fusion satellite cloud image of a future period. A forecasting module 13 is configured to identify the area and intensity of strong convective weather in the predicted fusion satellite cloud image of the future period by comparing a feature parameter representing cloud cluster physical characteristics with a preset identification threshold. A warning module 14 is configured to perform strong convective weather warning for photovoltaic power stations based on the identified area and intensity of strong convective weather.
[0051] As to the system in the above embodiments, the specific manner in which each module performs operations has been described in detail in the embodiments related to the method, and will not be described in detail here.
[0052] The application further provides a strong convective weather forecasting and warning device based on satellite data for implementing the above method embodiments. Figure 3 is a structural schematic diagram of the strong convective weather forecasting and warning device based on satellite data provided in the embodiments of the application, as Figure 3As shown, the satellite data-based severe convective weather forecasting and early warning device of the embodiment comprises a processor 21 and a memory 22, and the processor 21 is connected with the memory 22. The processor 21 is configured to call and execute a program stored in the memory 22; and the memory 22 is configured to store the program, and the program is used at least for executing the satellite data-based severe convective weather forecasting and early warning method in the above embodiment.
[0053] The specific implementation of the satellite data-based severe convective weather forecasting and early warning device provided by the embodiment of the present application can refer to the implementation of the satellite data-based severe convective weather forecasting and early warning method of any of the above embodiments, which will not be repeated here.
[0054] It can be understood that the same or similar parts in the above embodiments can be mutually referred to, and the content not described in detail in some embodiments can be referred to the same or similar content in other embodiments.
[0055] It should be noted that, in the description of the present application, the terms "first", "second", etc. are only used for the purpose of description, and cannot be understood as indicating or implying relative importance. In addition, in the description of the present application, unless otherwise specified, the meaning of "a plurality of" is at least two.
[0056] Any process or method descriptions in flow charts or otherwise described herein represent embodiments that can be understood as a sequence of steps of executable instructions for achieving a particular logic function or process, and the scope of the preferred embodiments of the present application includes additional implementations that can not be precisely shown or described herein, including implementations that can be performed in an order different from the order shown or discussed, including implementations that can be performed at substantially the same time or in reverse order, as will be understood by those skilled in the art of the embodiments to which the present application pertains.
[0057] It should be understood that each part of the present application can be realized by hardware, software, firmware or their combination. In the above implementation, a plurality of steps or methods can be realized by software or firmware stored in the memory and executed by a suitable instruction execution system. For example, if realized by hardware, and as in another implementation, it can be realized by any one or their combination of the following technologies known in the art: discrete logic circuit with logic gate circuit for implementing logic function on data signal, application specific integrated circuit with suitable combination logic gate circuit, programmable gate array (PGA), field programmable gate array (FPGA) and the like.
[0058] Those skilled in the art of the present technology can understand that all or part of the steps carried out by the above-mentioned embodiment method can be instructed by a program to complete the relevant hardware, and the program can be stored in a computer readable storage medium, and the program includes one or a combination of the steps of the method embodiment when executed.
[0059] In addition, each functional unit in each embodiment of the present application can be integrated in one processing module, or each unit can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware, or in the form of a software functional module. When the integrated module is realized in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.
[0060] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.
[0061] In the description of the present specification, the description referring to the terms "one embodiment", "some embodiments", "an example", "a specific example" or "some examples" etc. 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 present application. In the present specification, the illustrative description of the above terms does not necessarily mean the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.
[0062] Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.
Claims
1. A method for forecasting and issuing early warning of severe convective weather based on satellite data, characterized in that, include: Multi-band satellite data is acquired and fused to generate a fused satellite cloud image; wherein, the multi-band satellite data includes water vapor band data, visible light band data and infrared band data; The fused satellite cloud image is input into a pre-trained generative deep learning prediction model to obtain the predicted fused satellite cloud image for the future time period. In the fused satellite cloud images of the predicted future period, the regions and intensities of severe convective weather are identified by comparing the characteristic parameters representing the physical properties of cloud clusters with preset identification thresholds. Based on the identified areas and intensities of severe convective weather, severe convective weather alerts are issued for photovoltaic power plants.
2. The method for forecasting and warning severe convective weather based on satellite data according to claim 1, characterized in that, The multi-band satellite data is fused to generate a fused satellite cloud image, including: Fill in missing data in multi-band satellite data; Based on scaling processing, the visible light band data, water vapor band data and infrared band data in the multi-band satellite data after missing data filling are all scaled to the first preset range and saved as visible light satellite cloud image grayscale image, water vapor satellite cloud image grayscale image and infrared satellite cloud image grayscale image respectively. The grayscale images of visible light satellite cloud images, water vapor satellite cloud images, and infrared satellite cloud images are fused together, and non-cloud areas are filtered based on thresholding and time-series moving comparison methods. The images are then rescaled to a first preset range to obtain the fused satellite cloud image.
3. The method for forecasting and issuing early warning of severe convective weather based on satellite data according to claim 1, characterized in that, The training process of the generative deep learning prediction model includes: Based on WGAN or Res-UNet networks; The basic network is trained using fused satellite cloud images arranged chronologically 24 hours before the target time as features and fused satellite cloud images arranged chronologically 24 hours after the target time as labels to obtain a generative deep learning prediction model.
4. The method for forecasting and warning severe convective weather based on satellite data according to claim 1, characterized in that, The feature parameter is the grayscale value of the fused satellite cloud image; In the fused satellite cloud images of the predicted future time period, the regions and intensities of severe convective weather are identified by comparing characteristic parameters representing the physical properties of cloud clusters with preset identification thresholds, including: The grayscale values in the fused satellite cloud image for the predicted future time period are scaled to a second preset range; Areas with gray values greater than a preset recognition threshold are identified as areas of severe convective weather, and the intensity of severe convective weather is determined based on the range of gray values.
5. The method for forecasting and issuing early warning of severe convective weather based on satellite data according to claim 1, characterized in that, Also includes: Areas of severe convective weather identified within the statistical forecast period, and power plants within those areas; Using the forecast duration as the time axis, the movement trajectory of future severe convective weather is generated.
6. The method for forecasting and warning severe convective weather based on satellite data according to claim 4, characterized in that, The method of issuing severe convective weather alarms for photovoltaic power plants based on the identified areas and intensities of severe convective weather includes: Determine the correspondence between grayscale values in the fused satellite cloud images and preset early warning standards; Early warning information is determined based on the grayscale and correspondence in the fused satellite cloud imagery for predicted future time periods. The early warning information includes the early warning type, early warning level, intensity type, and early warning description.
7. The method for forecasting and warning severe convective weather based on satellite data according to claim 6, characterized in that, The method of issuing severe convective weather alarms for photovoltaic power plants based on the identified areas and intensities of severe convective weather also includes: Spatial matching of identified areas of severe convective weather with actual photovoltaic power stations; Based on the matching results, the warning information will be sent to the corresponding photovoltaic power station.
8. The method for forecasting and warning severe convective weather based on satellite data according to claim 6, characterized in that, Also includes: The newly acquired multi-band satellite data is used as training samples to incrementally train the generative deep learning prediction model.
9. A severe convective weather forecasting and early warning system based on satellite data, characterized in that, include: The fusion module is used to acquire multi-band satellite data and perform fusion processing on the multi-band satellite data to generate a fused satellite cloud image; wherein, the multi-band satellite data includes water vapor band data, visible light band data and infrared band data; The prediction module is used to input the fused satellite cloud image into a pre-trained generative deep learning prediction model to obtain the predicted fused satellite cloud image for the future time period. The forecast module is used to identify the region and intensity of severe convective weather by comparing characteristic parameters representing the physical properties of cloud clusters with preset identification thresholds in fused satellite cloud images for the predicted future period. The alarm module is used to issue severe convective weather warnings for photovoltaic power plants based on the identified area and intensity of severe convective weather.
10. A severe convective weather forecasting and early warning device based on satellite data, characterized in that, It includes a processor and a memory, wherein the processor is connected to the memory: The processor is used to call and execute the program stored in the memory; The memory is used to store the program, which is at least used to execute the severe convective weather forecasting and early warning method based on satellite data as described in any one of claims 1-8.