Meteorological prediction method and device, storage medium and electronic device

Abstract

The invention discloses a meteorological prediction method, and the method comprises the steps: obtaining meteorological collection of meteorological elements in a preset region, and obtaining the meteorological parameters of the meteorological elements in the preset region at the current moment; on the basis of the meteorological parameters of the meteorological elements in the preset area at the current moment, meteorological integration features of the meteorological elements in the preset area are obtained; processing the meteorological integration features to obtain meteorological region features, and performing region attention representation on the meteorological region features to obtain meteorological association features of meteorological elements of the preset region; and splicing the meteorological integration features and the meteorological association features, and predicting the meteorological parameters of the preset area at the target moment. The problem that the accuracy of multi-weather prediction is low in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and more specifically, to a weather forecasting method and device, a storage medium, and an electronic device. Background Art

[0002] With the rapid development of the new energy power industry, the importance of meteorological services has become increasingly apparent. Meteorology plays a vital role in the entire life cycle of new energy project planning and site selection, design selection, construction, operation and maintenance management. With the gradual access of large-scale wind power and photovoltaic power to the power grid, the unstable characteristics of new energy power represented by wind and solar energy have brought severe challenges to the balance and dispatch of the power grid. Therefore, the power grid will have more stringent power forecast assessment standards for each new energy power station, and accurate high-resolution grid-based and numerical meteorological forecasts can lay a good foundation for the accurate prediction of power and load of new energy power plants. In addition, new energy power stations are usually distributed over a wide area and have weak ability to resist meteorological disasters. Therefore, the demand for refined, industry-based, and indexed high-quality meteorological warnings is also stronger.

[0003] Traditional weather forecasting methods mainly rely on numerical weather forecast models and the experience of meteorologists, which often have limitations when dealing with complex and changeable weather systems. Among related technologies, the rapid development of artificial intelligence (AI) technology has brought new opportunities to the field of weather forecasting. AI technology, especially machine learning and deep learning, has demonstrated its powerful data processing and pattern recognition capabilities in many fields. Applying AI technology to weather forecasting can achieve in-depth mining and analysis of large amounts of meteorological data, thereby revealing the deep laws of weather changes and improving the accuracy and efficiency of high-resolution weather forecasts. However, high-resolution forecasts face problems such as large amounts of data and high computational complexity, resulting in low accuracy in weather forecasts.

[0004] Therefore, how to improve the accuracy of weather forecasts has become a pressing issue to be addressed in meteorological research. Summary of the invention

[0005] The embodiment of the present application provides a weather forecasting method to at least solve the problem of low accuracy of weather forecasting in the related art.

[0006] According to an embodiment of the present application, a weather forecasting method is provided, including: acquiring meteorological data for meteorological elements in a preset area to obtain meteorological parameters of the meteorological elements in the preset area at the current moment; obtaining meteorological integration features of the meteorological elements in the preset area based on the meteorological parameters of the meteorological elements in the preset area at the current moment; processing the meteorological integration features to obtain meteorological regional features, and performing regional attention characterization on the meteorological regional features to obtain meteorological correlation features of the meteorological elements in the preset area; splicing the meteorological integration features and the meteorological correlation features to predict the meteorological parameters of the preset area at the target moment.

[0007] In an exemplary embodiment, the meteorological integration characteristics of the meteorological elements in the preset area are obtained based on the meteorological parameters of the meteorological elements in the preset area at the current moment, including: dividing the meteorological parameters of the meteorological elements in the preset area at the current moment based on the attributes of the meteorological elements in the preset area to obtain multiple first meteorological parameters and multiple second meteorological parameters; splicing the multiple first meteorological parameters to obtain a first meteorological characteristic of a first dimension; splicing the multiple second meteorological parameters to obtain a second meteorological characteristic of a second dimension; and dimensionally mapping the first meteorological characteristics and the second meteorological characteristics to obtain the meteorological integration characteristics of the meteorological elements in the preset area.

[0008] In an exemplary embodiment, the meteorological integrated features are processed to obtain meteorological regional features, including: inputting the meteorological integrated features into a pre-trained meteorological detection model; using the meteorological detection model to perform two-dimensional spatial mapping on the meteorological integrated features to obtain meteorological integrated features after two-dimensional spatial mapping; dividing the meteorological integrated features after two-dimensional spatial mapping to obtain divided meteorological features; performing dimensional transformation on the divided meteorological features to obtain divided meteorological features after dimensional transformation; and splicing the divided meteorological features after dimensional transformation to obtain meteorological regional features.

[0009] In an exemplary embodiment, the meteorological area characteristics are characterized by regional attention to obtain meteorological correlation characteristics of the meteorological elements of the preset area, including: associating the meteorological area characteristics based on multiple model parameters set by the regional attention mechanism layer of a pre-trained meteorological detection model to determine the meteorological correlation characteristics of the meteorological elements of the preset area.

[0010] In an exemplary embodiment, the training steps of the meteorological detection model include: obtaining training samples; the training samples include sample meteorological parameters of meteorological elements in the sample area at sample times; inputting the training samples into the neural network model to be trained, and the neural network model to be trained performs training on the training samples until a preset condition is met, thereby obtaining a trained meteorological detection model.

[0011] In an exemplary embodiment, the training samples are input into a neural network model to be trained, and the neural network model to be trained performs training on the training samples until a preset condition is met, thereby obtaining a trained meteorological detection model, including: inputting the training samples into the neural network model to be trained, and obtaining a sample meteorological integrated feature of the meteorological elements of the sample area by the neural network model to be trained; performing two-dimensional spatial mapping on the sample meteorological integrated feature by the neural network model to be trained to obtain the sample meteorological integrated feature after two-dimensional spatial mapping; dividing the sample meteorological integrated feature after two-dimensional spatial mapping by the neural network model to be trained to obtain sample divided meteorological features, performing dimension conversion on the sample divided meteorological features to obtain the sample divided meteorological features after dimension conversion, and splicing the sample divided meteorological features after dimension conversion to obtain sample meteorological regional features; performing regional attention representation on the sample meteorological regional features by the neural network model to be trained according to the regional attention representation result, thereby stopping when a preset condition is met, thereby obtaining a trained meteorological detection model.

[0012] In an exemplary embodiment, the neural network model to be trained performs regional attention characterization on the sample meteorological regional features, and trains the neural network model to be trained according to the regional attention characterization result until a preset condition is met, thereby obtaining a trained meteorological detection model, including:

[0013] Based on the neural network model to be trained, the sample meteorological regional characteristics are characterized by regional attention, and sample meteorological related characteristics are output; based on the neural network model to be trained, the sample meteorological related characteristics and the sample meteorological integrated characteristics are spliced ​​to determine the sample meteorological forecast result; the loss between the sample meteorological actual result and the sample meteorological forecast result is determined, and the parameters of the neural network model to be trained are adjusted according to the loss until the preset conditions are met, so as to obtain a trained meteorological detection model.

[0014] According to another embodiment of the embodiment of the present application, a weather forecasting device is also provided, including: an acquisition module, used to acquire meteorological data of meteorological elements in a preset area, and obtain meteorological parameters of the meteorological elements in the preset area at the current moment; a first feature extraction module, used to obtain meteorological integration features of the meteorological elements in the preset area based on the meteorological parameters of the meteorological elements in the preset area at the current moment; a second feature extraction module, used to process the meteorological integration features to obtain meteorological regional features, and perform regional attention characterization on the meteorological regional features to obtain meteorological correlation features of the meteorological elements in the preset area; a prediction module, used to splice the meteorological integration features and the meteorological correlation features to predict the meteorological parameters of the preset area at the target moment.

[0015] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above method when running.

[0016] According to another aspect of the embodiments of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the method through the computer program.

[0017] According to another aspect of the embodiments of the present application, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0018] In an embodiment of the present application, by acquiring meteorological parameters of the meteorological elements in the preset area at the current moment through meteorological collection of meteorological elements in a preset area, the meteorological integration characteristics of the meteorological elements in the preset area are determined, and the meteorological integration characteristics are further processed to obtain meteorological regional characteristics, so that the meteorological regional characteristics can be characterized by regional attention to obtain meteorological correlation characteristics; and then meteorological forecasts can be made based on the meteorological integration characteristics and meteorological correlation characteristics. Since the meteorological correlation characteristics obtained by the regional attention characterization can establish the correlation between the characteristics from a regional perspective, and then by splicing the meteorological integration characteristics and the meteorological correlation characteristics, the meteorological parameters can be predicted, which can improve the accuracy of the meteorological forecast. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0020] Figure 1It is a hardware structure block diagram of a computer terminal of a weather forecasting method according to an embodiment of the present application;

[0021] Figure 2 is a flow chart of a weather forecasting method according to an embodiment of the present application;

[0022] Figure 3 is a schematic diagram of dividing windows according to an embodiment of the present application;

[0023] Figure 4 is another flow chart of the weather forecasting method according to an embodiment of the present application;

[0024] Figure 5 is a structural block diagram of a meteorological detection model according to an embodiment of the present application;

[0025] Figure 6 It is a structural block diagram of a weather forecasting device according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices; "plurality" means two or more.

[0028] The method embodiments provided in the embodiments of the present application can be executed in a computer terminal or similar computing device or cloud platform or independent physical server or software platform, wherein the above software platform is run through one or more servers. Taking running on a computer terminal as an example, Figure 1 is a hardware structure block diagram of a weather forecasting method according to an embodiment of the present application. Figure 1As shown, the computer terminal may include one or more ( Figure 1 Only one is shown in the figure) processor 102 and memory 104 for storing data. In an exemplary embodiment, the computer terminal may also include a transmission device 106 for communication functions and an input and output device 108, wherein the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. A person skilled in the art can understand that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Equivalent functions or comparisons shown Figure 1 A different configuration with more features is shown.

[0029] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as computer programs corresponding to the methods in the embodiments of the present application. The processor 102 executes various functional applications and data processing by running the computer programs stored in the memory 104, that is, the above-mentioned methods are implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0030] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0031] In this embodiment, a weather forecasting method is provided, which is applied to the above-mentioned computer device. Figure 2 : is a flow chart of a weather forecasting method according to an embodiment of the present application, the flow chart includes the following steps:

[0032] Step S202, acquiring meteorological parameters of the meteorological elements in the preset area at the current moment by performing meteorological collection on the meteorological elements in the preset area;

[0033] Among them, the preset area is an area within a certain boundary range. For example, meteorological elements of various urban areas in a province can be collected to obtain meteorological parameters of meteorological elements in different urban areas at the current moment; meteorological elements of different areas of a city can be collected to obtain meteorological parameters of meteorological elements in each area at the current moment. Meteorological elements can be air pressure, temperature, U wind, V wind, humidity, sea level pressure, gust wind speed, surface pressure, surface temperature, near-surface temperature, near-surface humidity, near-surface U wind, near-surface V wind, wind speed, wind direction, wind energy density, etc. Meteorological parameters are the measured values ​​corresponding to the meteorological elements respectively.

[0034] Step S204, obtaining meteorological integration characteristics of the meteorological elements in the preset area based on the meteorological parameters of the meteorological elements in the preset area at the current moment;

[0035] It can be understood that in order to learn the correlation between meteorological elements, the meteorological parameters of the meteorological elements at the current moment can be represented by meteorological elements, that is, the meteorological parameters of all meteorological elements can be integrated into a vector for subsequent regional attention representation.

[0036] In an exemplary embodiment, the meteorological integration characteristics of the meteorological elements in the preset area are obtained based on the meteorological parameters of the meteorological elements in the preset area at the current moment, including: dividing the meteorological parameters of the meteorological elements in the preset area at the current moment based on the attributes of the meteorological elements in the preset area to obtain multiple first meteorological parameters and multiple second meteorological parameters; splicing the multiple first meteorological parameters to obtain a first meteorological characteristic of a first dimension; splicing the multiple second meteorological parameters to obtain a second meteorological characteristic of a second dimension; and dimensionally mapping the first meteorological characteristics and the second meteorological characteristics to obtain the meteorological integration characteristics of the meteorological elements in the preset area.

[0037] It should be noted that the attributes of meteorological elements refer to the properties or characteristics of meteorological elements. For example, some meteorological elements belong to surface meteorology, some belong to high-altitude meteorological elements, and some belong to atmospheric meteorological elements. The vector dimensions of meteorological elements of the same type are the same, so the meteorological parameters of meteorological elements of the same dimension can be spliced ​​first to obtain the splicing results of meteorological parameters of meteorological elements of the same dimension, and then the splicing results are dimensionally mapped to obtain meteorological integration characteristics.

[0038] In some embodiments, the computer terminal can represent 16 meteorological elements, including air pressure, temperature, U wind, V wind, humidity, sea level pressure, gust wind speed, surface pressure, surface temperature, near surface temperature, near surface humidity, near surface U wind, near surface V wind, wind speed, wind direction, and wind energy density, as pres, t, u, v, q, prmsl, gust, sfcp, sfct, tn, qn, un, vn, wsp, wdir, and wpd, respectively. Where prmsl, gust, sfcp, sfct, tn, qn, un, vn∈R 1×lat×lon ,pres,t,u,v,q,wsp,wdir,wpd∈R layers×lat×lon .

[0039] First, the computer terminal can divide the meteorological parameters of the meteorological elements at the current moment according to the attributes of each meteorological element, wherein the attributes of prmsl, gust, sfcp, sfct, tn, qn, un, and vn are the same, and the attributes of pres, t, u, v, q, wsp, wdir, and wpd are the same. Furthermore, multiple first meteorological parameters and multiple second meteorological elements can be determined according to the meteorological elements with the same attributes. Among them, the first meteorological parameters of prmsl, gust, sfcp, sfct, tn, qn, un, and vn are spliced ​​in the first dimension to obtain the first meteorological feature. The second meteorological parameters of pres, t, u, v, q, wsp, wdir, and wpd are spliced ​​in the second dimension to obtain the second meteorological feature. After obtaining the first meteorological feature and the second meteorological feature, the first meteorological feature and the second meteorological feature can be dimensionally mapped to obtain a meteorological integration feature.

[0040] Step S206, processing the meteorological integration features to obtain meteorological regional features, and performing regional attention characterization on the meteorological regional features to obtain meteorological correlation features of the meteorological elements of the preset area;

[0041] Among them, the meteorological regional features are the regional features in the space extracted based on the meteorological integration features. Through the regional attention representation, the correlation between the regional features in the space can be learned to obtain the meteorological correlation features of the meteorological elements in the set area.

[0042] Step S208, combining the meteorological integration features and the meteorological correlation features to predict the meteorological parameters of the preset area at the target time.

[0043] In an exemplary embodiment, the meteorological parameters of the preset area at the target time are predicted based on the meteorological integration characteristics and the meteorological correlation characteristics, including: fusing the meteorological integration characteristics and the meteorological correlation characteristics to obtain target meteorological characteristics; and predicting the meteorological parameters of the preset area at the target time according to the target meteorological characteristics.

[0044] Specifically, the computer terminal can fuse the meteorological integration features and the meteorological correlation features to obtain target meteorological features, and predict the meteorological parameters of the preset area at the target time through the target meteorological features.

[0045] In an embodiment of the present application, by acquiring meteorological parameters of the meteorological elements in the preset area at the current moment through meteorological collection of meteorological elements in a preset area, the meteorological integration characteristics of the meteorological elements in the preset area are determined, and the meteorological integration characteristics are further processed to obtain meteorological regional characteristics, so that the meteorological regional characteristics can be characterized by regional attention to obtain meteorological correlation characteristics; and then meteorological forecasts can be made based on the meteorological integration characteristics and meteorological correlation characteristics. Since the meteorological correlation characteristics obtained by the regional attention characterization can establish the correlation between the characteristics from a regional perspective, and then by splicing the meteorological integration characteristics and the meteorological correlation characteristics, the meteorological parameters can be predicted, which can improve the accuracy of the meteorological forecast.

[0046] In an exemplary embodiment, the meteorological integrated features are processed to obtain meteorological regional features, including: inputting the meteorological integrated features into a pre-trained meteorological detection model; using the meteorological detection model to perform two-dimensional spatial mapping on the meteorological integrated features to obtain meteorological integrated features after two-dimensional spatial mapping; dividing the meteorological integrated features after two-dimensional spatial mapping to obtain divided meteorological features; performing dimensional transformation on the divided meteorological features to obtain divided meteorological features after dimensional transformation; and splicing the divided meteorological features after dimensional transformation to obtain meteorological regional features.

[0047] Specifically, the computer terminal can obtain the meteorological regional characteristics through the meteorological detection model. The meteorological integrated characteristics are mapped into a two-dimensional space through a single-layer feedforward neural network of the meteorological detection model. The formula is as follows:

[0048] I'=I*W1+b1;

[0049] Where I refers to the meteorological integration feature, W1 and b1 are training parameters, I' is the meteorological integration feature after two-dimensional space mapping, I'∈R^(H×W×1). Then I' is divided into windows with a window size of (a, a), where a satisfies that H and W are divisible. Figure 3As shown in FIG. 1 , a schematic diagram of the partition window is shown. After obtaining the partition meteorological features, the partition meteorological features can be dimensionally transformed, and the characteristic values ​​of the partition meteorological features can be converted into 1 dimension to obtain the partition meteorological features after dimension transformation. Finally, the partition meteorological features after dimension transformation are obtained and spliced ​​to obtain Windows∈R^(HW / aa×aa), that is, the meteorological regional features.

[0050] In the above embodiment, the meteorological integrated features are processed through the meteorological detection model to obtain meteorological regional features, and then the correlation between regional features can be learned based on the meteorological regional features to obtain regional correlation features, and the accuracy of meteorological forecasts can be improved through regional correlation features.

[0051] In an exemplary embodiment, the meteorological area characteristics are characterized by regional attention to obtain meteorological correlation characteristics of the meteorological elements of the preset area, including: associating the meteorological area characteristics based on multiple model parameters set by the regional attention mechanism layer of a pre-trained meteorological detection model to determine the meteorological correlation characteristics of the meteorological elements of the preset area.

[0052] Specifically, the computer terminal can learn the correlation between regional features through the meteorological detection model. Specifically, the correlation between regional features in the two-dimensional space can be learned through the regional attention mechanism layer of the meteorological detection model. The formula is as follows:

[0053] AttentionRegion=Softmax((Windows W2+b2)×(Windows

[0054] W3+b3)T / aa)(Windows W4+b4)

[0055] Among them, AttentionRegion is the obtained regional attention representation, W2, b2, W3, b3, W4, b4 are model parameters. Convert the dimension to AttentionRegionMap∈R H×W×1 , AttentionRegionMap is the obtained regional attention representation map.

[0056] In an exemplary embodiment, the training steps of the meteorological detection model include: obtaining training samples; the training samples include sample meteorological parameters of meteorological elements in the sample area at sample times; inputting the training samples into the neural network model to be trained, and the neural network model to be trained performs training on the training samples until a preset condition is met, thereby obtaining a trained meteorological detection model.

[0057] It is understandable that the meteorological detection model is pre-trained. The training samples are sample meteorological parameters of meteorological elements in the sample area at the sample time. The trained meteorological detection model is obtained by inputting the training samples into the trained neural network model until the preset conditions are met, such as reaching the set threshold of the number of iterations.

[0058] In an exemplary embodiment, the training samples are input into a neural network model to be trained, and the neural network model to be trained performs training on the training samples until a preset condition is met, thereby obtaining a trained meteorological detection model, including: inputting the training samples into the neural network model to be trained, and obtaining a sample meteorological integrated feature of the meteorological elements of the sample area by the neural network model to be trained; performing two-dimensional spatial mapping on the sample meteorological integrated feature by the neural network model to be trained to obtain the sample meteorological integrated feature after two-dimensional spatial mapping; dividing the sample meteorological integrated feature after two-dimensional spatial mapping by the neural network model to be trained to obtain sample divided meteorological features, performing dimension conversion on the sample divided meteorological features to obtain the sample divided meteorological features after dimension conversion, and splicing the sample divided meteorological features after dimension conversion to obtain sample meteorological regional features; performing regional attention representation on the sample meteorological regional features by the neural network model to be trained according to the regional attention representation result, thereby stopping when a preset condition is met, thereby obtaining a trained meteorological detection model.

[0059] It should be noted that in the process of training the neural network model, the sample meteorological integration characteristics of the meteorological elements in the sample area are first obtained, and then the sample meteorological integration characteristics are mapped in two-dimensional space to obtain the sample meteorological integration characteristics after two-dimensional space mapping. Based on the sample meteorological integration characteristics, division and dimension conversion processing are performed to obtain the sample meteorological regional characteristics.

[0060] In an exemplary embodiment, the neural network model to be trained performs regional attention characterization on the sample meteorological regional characteristics, and trains the neural network model to be trained according to the regional attention characterization results until a preset condition is met, thereby obtaining a trained meteorological detection model, including: performing regional attention characterization on the sample meteorological regional characteristics based on the neural network model to be trained, and outputting sample meteorological correlation characteristics; splicing the sample meteorological correlation characteristics and the sample meteorological integration characteristics based on the neural network model to be trained, and determining a sample meteorological prediction result; determining the loss between the sample meteorological actual result and the sample meteorological prediction result, and adjusting the parameters of the neural network model to be trained according to the loss, thereby stopping until a preset condition is met, thereby obtaining a trained meteorological detection model.

[0061] It can be understood that when the regional attention characterization of the sample meteorological regional features is performed based on the neural network model to be trained, it can be implemented based on the attention mechanism layer of the neural network model. Through the regional attention characterization, the sample meteorological related features are output. On this basis, the sample meteorological related features and the sample meteorological integrated features are spliced ​​by the neural network model to be trained to determine the sample meteorological forecast results. The real sample meteorological results can be a predetermined label, that is, the sample meteorological parameters of the meteorological elements in the sample area at the target sample time. Through the loss between the label and the sample meteorological forecast result, the parameters of the neural network model to be trained are adjusted according to the loss, such as w1, b1, w2, b2, w3, b3, w4, b4, etc., until the preset conditions are met and the meteorological detection model that has been trained is obtained.

[0062] It is understandable that in order to better understand the process of the above-mentioned weather forecasting method, the above-mentioned weather forecasting method is described below in combination with an optional embodiment, but it is not used to limit the technical solution of the embodiment of the present application.

[0063] With the rapid development of the new energy power industry, the importance of meteorological services has become increasingly apparent. Meteorology is a key technology for the new energy industry to transform from "relying on the weather" to "knowing the weather" in an intelligent way, and plays a vital role in the entire life cycle of new energy project planning, site selection, design selection, construction, operation and maintenance management. With the gradual access of large-scale wind power and photovoltaic power to the power grid, the unstable characteristics of new energy power represented by wind power and solar energy have brought severe challenges to the balance and dispatch of the power grid. Therefore, the power grid will have more stringent power forecast assessment standards for each new energy power station, and accurate high-resolution grid-based and numerical meteorological forecasts can lay a good foundation for the accurate prediction of power power and load of new energy power stations. In addition, new energy power stations are usually distributed over a wide area and have weak ability to resist meteorological disasters. Therefore, there is a stronger demand for refined, industry-based, and indexed high-quality meteorological warnings. In summary, high-resolution, high-quality meteorological services directly affect the economic benefits of power generation companies, are related to the safe production and stable operation of the power system, and play an increasingly important role in the production and operation of new energy power.

[0064] At present, the accurate forecasting methods of new energy meteorological services require large computing resources and long computing time, making it difficult to promote and deploy in new energy power plants. In addition, due to the large uncertainty of the estimation of atmospheric initial conditions, traditional numerical models have different degrees of forecast errors, which cannot meet the needs of high-resolution meteorological forecasts at the kilometer and hour levels in key new energy areas such as deserts, mountains, and coastal areas with insufficient observations and complex dynamic mechanisms in the industry. However, the emergence of data-driven artificial intelligence big model algorithms provides a new way to solve this challenge. In recent years, the rapid development of artificial intelligence big models has brought revolutionary progress to human society. Subsequently, big models have continued to emerge, and have achieved outstanding performance in fields including natural language processing, visual tasks, and multimodal applications. Since meteorological data also has the characteristics of large data volume, and the existing numerical forecasting methods require massive computing resources, researchers have begun to pay attention to the construction of global meteorological big models. Although the stability of the performance of meteorological big models still needs further verification and exploration in business and scientific research, their forecast accuracy and timeliness of some meteorological elements are comparable to or even surpass those of traditional numerical weather forecast models. Artificial intelligence technology represented by large model technology will inevitably have a transformative impact on future weather forecasts and related tasks, and is expected to achieve higher-resolution weather forecasting capabilities, greatly improve forecasting efficiency, promote the smooth grid connection of weather-dependent new energy industries represented by wind power and photovoltaics, and advance new energy construction.

[0065] Aiming at the demand of new energy meteorological services, this application proposes a meteorological forecasting method, referring to Figure 4 As shown, it is an overall flow chart of the meteorological forecasting method, which mainly includes meteorological parameter input at the current moment, meteorological element representation, meteorological element association learning, meteorological element forecast and meteorological output at the target moment. Through the meteorological forecasting method of this application, the forecast of 2km spatial resolution of meteorological elements such as temperature, wind speed and humidity can be achieved.

[0066] In the meteorological input part at the current moment, this application uses the normalized values ​​of 16 meteorological elements at time T0, including air pressure, temperature, U wind, V wind, humidity, sea level pressure, gust wind speed, surface pressure, surface temperature, near surface temperature, near surface humidity, near surface U wind, near surface V wind, wind speed, wind direction, and wind energy density, as input. The meteorological parameter output at the target moment is the forecast value of these 16 meteorological elements at time T1.

[0067] In the meteorological element representation part, the first meteorological feature can be obtained: WeatherFeatureDim1 = Concat(prmsl,gust,sfcp,sfct,tn,qn,un,vn)∈R^(8×lat×lon), and the second meteorological feature can be obtained: WeatherFeatureDim2 = Concat(pres,t,u,v,q,wsp,wdir,wpd)∈R^(layers×8×lat×lon); by splicing WeatherFeatureDim1 and WeatherFeatureDim2, and then performing dimension mapping, the meteorological element feature is obtained: WeatherFeature = Concat(WeatherFeatureDim1,WeatherFeatureDim2)∈R^((layers+1)×8×lat×lon)->WeatherFeature∈R^(8(layers+1)×lat×lon). Then, with WeatherFeature as input, CNN and MLP are used to obtain the representation WeatherFeatureMap.

[0068] The meteorological element association learning and weather forecasting are realized through the meteorological detection model. Figure 5 As shown in the figure, it is a structural block diagram of the meteorological detection model. In this application, Swin Transformer is used as the basic architecture, and the regional attention mechanism is embedded in the Swin Transformer architecture. The meteorological detection model can specifically include multiple modules such as patch partition, linear embedding, Swin Transformer block, and regional attention mechanism.

[0069] To summarize, the optional embodiments of the present application obtain meteorological parameters of the meteorological elements in the preset area at the current moment by acquiring meteorological collection of meteorological elements in the preset area, determine the meteorological integration characteristics of the meteorological elements in the preset area, and further process the meteorological integration characteristics to obtain meteorological regional characteristics, so that the meteorological regional characteristics can be characterized by regional attention to obtain meteorological correlation characteristics; and then meteorological forecasts can be made based on the meteorological integration characteristics and meteorological correlation characteristics. Since the meteorological correlation characteristics obtained by the regional attention characterization can establish the correlation between the characteristics from the regional perspective, and then by splicing the meteorological integration characteristics and the meteorological correlation characteristics, the meteorological parameters can be predicted, which can improve the accuracy of the meteorological forecast.

[0070] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0071] The present application also provides a structural block diagram of a weather forecasting device. Figure 6 is a structural block diagram of a weather forecasting device according to an embodiment of the present application; Figure 6 As shown, including:

[0072] An acquisition module 62 is used to acquire meteorological parameters of meteorological elements in a preset area at a current moment by performing meteorological collection on meteorological elements in the preset area;

[0073] A first feature extraction module 64 is used to obtain the meteorological integration features of the meteorological elements in the preset area based on the meteorological parameters of the meteorological elements in the preset area at the current moment;

[0074] The second feature extraction module 66 is used to process the meteorological integration feature to obtain a meteorological regional feature, and perform regional attention characterization on the meteorological regional feature to obtain a meteorological correlation feature of the meteorological element of the preset area;

[0075] The prediction module 68 is used to combine the meteorological integration features and the meteorological correlation features to predict the meteorological parameters of the preset area at the target time.

[0076] Through the above-mentioned device, meteorological collection is carried out on the meteorological elements in the preset area, and the meteorological parameters of the meteorological elements in the preset area at the current moment are obtained, the meteorological integration characteristics of the meteorological elements in the preset area are determined, and the meteorological integration characteristics are further processed to obtain meteorological regional characteristics, so that the meteorological regional characteristics can be characterized by regional attention to obtain meteorological correlation characteristics; then, meteorological prediction can be carried out based on the meteorological integration characteristics and meteorological correlation characteristics, because the meteorological correlation characteristics obtained by the regional attention characterization can establish the correlation between the characteristics from the regional perspective, and then by splicing the meteorological integration characteristics and the meteorological correlation characteristics, the meteorological parameters are predicted, which can improve the accuracy of the meteorological prediction.

[0077] In an exemplary embodiment, the first feature extraction module 64 is also used to divide the meteorological parameters of the meteorological elements at the current moment based on the attributes of the meteorological elements in the preset area to obtain multiple first meteorological parameters and multiple second meteorological parameters; splice the multiple first meteorological parameters to obtain the first meteorological feature of the first dimension; splice the multiple second meteorological parameters to obtain the second meteorological feature of the second dimension; and perform dimension mapping on the first meteorological feature and the second meteorological feature to obtain the meteorological integration feature of the meteorological elements in the preset area.

[0078] In an exemplary embodiment, the second feature extraction module 66 is also used to input the meteorological integrated features into a pre-trained meteorological detection model; the meteorological detection model performs two-dimensional spatial mapping on the meteorological integrated features to obtain the meteorological integrated features after two-dimensional spatial mapping; divides the meteorological integrated features after two-dimensional spatial mapping to obtain divided meteorological features; performs dimensional transformation on the divided meteorological features to obtain divided meteorological features after dimensional transformation; and splices the divided meteorological features after dimensional transformation to obtain meteorological regional features.

[0079] In an exemplary embodiment, the second feature extraction module 66 is also used to associate the meteorological regional features based on multiple model parameters set by the regional attention mechanism layer of the pre-trained meteorological detection model to determine the meteorological correlation characteristics of the meteorological elements of the preset area.

[0080] In an exemplary embodiment, the above-mentioned device also includes a training module for obtaining training samples; the training samples include sample meteorological parameters of meteorological elements in the sample area at sample times; the training samples are input into the neural network model to be trained, and the neural network model to be trained is trained for the training samples until a preset condition is met, thereby obtaining a trained meteorological detection model.

[0081] In an exemplary embodiment, the training module is also used to input the training samples into the neural network model to be trained, and obtain the sample meteorological integrated features of the meteorological elements of the sample area by the neural network model to be trained; the neural network model to be trained performs two-dimensional spatial mapping on the sample meteorological integrated features to obtain the sample meteorological integrated features after two-dimensional spatial mapping; the neural network model to be trained divides the sample meteorological integrated features after two-dimensional spatial mapping to obtain sample divided meteorological features, performs dimension conversion on the sample divided meteorological features to obtain the sample divided meteorological features after dimension conversion, and splices the sample divided meteorological features after dimension conversion to obtain sample meteorological regional features; the neural network model to be trained performs regional attention representation on the sample meteorological regional features, and trains the neural network model to be trained according to the regional attention representation results until the preset conditions are met, thereby obtaining a trained meteorological detection model.

[0082] In an exemplary embodiment, the training module is also used to perform regional attention characterization on the sample meteorological regional features based on the neural network model to be trained, and output sample meteorological related features; splice the sample meteorological related features and the sample meteorological integrated features based on the neural network model to be trained to determine the sample meteorological forecast results; determine the loss between the sample meteorological actual results and the sample meteorological forecast results, adjust the parameters of the neural network model to be trained according to the loss, and stop when the preset conditions are met to obtain a trained meteorological detection model.

[0083] An embodiment of the present application further provides a storage medium, which includes a stored program, wherein the program executes any of the above methods when it is run.

[0084] Optionally, in this embodiment, the storage medium may be configured to store program codes for executing the following steps:

[0085] S1, obtaining meteorological parameters of meteorological elements in a preset area at the current moment by collecting meteorological elements in the preset area;

[0086] S2, obtaining meteorological integration characteristics of the meteorological elements in the preset area based on the meteorological parameters of the meteorological elements in the preset area at the current moment;

[0087] S3, processing the meteorological integration features to obtain meteorological regional features, and performing regional attention characterization on the meteorological regional features to obtain meteorological correlation features of meteorological elements in the preset area;

[0088] S4, combining the meteorological integration features and the meteorological correlation features to predict the meteorological parameters of the preset area at the target time.

[0089] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0090] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0091] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0092] S1, obtaining meteorological parameters of meteorological elements in a preset area at the current moment by collecting meteorological elements in the preset area;

[0093] S2, obtaining meteorological integration characteristics of the meteorological elements in the preset area based on the meteorological parameters of the meteorological elements in the preset area at the current moment;

[0094] S3, processing the meteorological integration features to obtain meteorological regional features, and performing regional attention characterization on the meteorological regional features to obtain meteorological correlation features of meteorological elements in the preset area;

[0095] S4, combining the meteorological integration features and the meteorological correlation features to predict the meteorological parameters of the preset area at the target time.

[0096] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store program codes.

[0097] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps in any one of the above method embodiments are implemented.

[0098] An embodiment of the present application further provides another computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0099] An embodiment of the present application also provides a computer program, which includes computer instructions stored in a computer-readable storage medium; a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps in any one of the above method embodiments.

[0100] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0101] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0102] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A weather forecasting method, characterized in that: include: Acquire meteorological parameters of meteorological elements in a preset area at the current moment by collecting meteorological elements in the preset area; Based on the meteorological parameters of the meteorological elements in the preset area at the current moment, obtaining meteorological integration characteristics of the meteorological elements in the preset area; Processing the meteorological integration features to obtain meteorological regional features, and performing regional attention characterization on the meteorological regional features to obtain meteorological correlation features of meteorological elements in the preset area; The meteorological integration features and the meteorological correlation features are spliced ​​together to predict the meteorological parameters of the preset area at the target time.

2. The method according to claim 1, characterized in that: The obtaining of the meteorological integration characteristics of the meteorological elements in the preset area based on the meteorological parameters of the meteorological elements in the preset area at the current moment includes: Based on the attributes of the meteorological elements in the preset area, the meteorological parameters of the meteorological elements at the current moment are divided to obtain a plurality of first meteorological parameters and a plurality of second meteorological parameters; splicing the multiple first meteorological parameters to obtain a first meteorological feature of a first dimension; splicing the plurality of second meteorological parameters to obtain a second meteorological feature of a second dimension; The first meteorological feature and the second meteorological feature are dimensionally mapped to obtain meteorological integration features of the meteorological elements of the preset area.

3. The method according to claim 1, characterized in that The processing of the meteorological integration features to obtain meteorological regional features includes: Inputting the meteorological integration features into a pre-trained meteorological detection model; The meteorological detection model performs two-dimensional spatial mapping on the meteorological integration features to obtain the meteorological integration features after two-dimensional spatial mapping; Dividing the meteorological integration features after the two-dimensional space mapping to obtain divided meteorological features; Performing dimension conversion on the divided meteorological features to obtain the divided meteorological features after dimension conversion; The meteorological characteristics of the divisions after the dimension conversion are spliced ​​to obtain meteorological regional characteristics.

4. The method according to claim 1, characterized in that The performing regional attention characterization on the meteorological regional features to obtain meteorological correlation features of meteorological elements in the preset area includes: Based on multiple model parameters set by the regional attention mechanism layer of the pre-trained meteorological detection model, the meteorological regional characteristics are associated to determine the meteorological correlation characteristics of the meteorological elements in the preset area.

5. The method according to claim 3, characterized in that: The training steps of the meteorological detection model include: Acquire training samples; the training samples include sample meteorological parameters of meteorological elements in a sample area at sample times; The training samples are input into the neural network model to be trained, and the neural network model to be trained is trained for the training samples until a preset condition is met, thereby obtaining a trained meteorological detection model.

6. The method according to claim 5, characterized in that The training samples are input into the neural network model to be trained, and the neural network model to be trained performs training on the training samples until a preset condition is met, thereby obtaining a trained meteorological detection model, including: Inputting the training samples into the neural network model to be trained, and obtaining the sample meteorological integration features of the meteorological elements of the sample area from the neural network model to be trained; The neural network model to be trained performs two-dimensional spatial mapping on the sample meteorological integrated features to obtain the sample meteorological integrated features after two-dimensional spatial mapping; The neural network model to be trained divides the sample meteorological integration features after the two-dimensional space mapping to obtain sample divided meteorological features, performs dimension conversion on the sample divided meteorological features to obtain the sample divided meteorological features after dimension conversion, and splices the sample divided meteorological features after dimension conversion to obtain sample meteorological regional features; The neural network model to be trained performs regional attention characterization on the sample meteorological area characteristics, and the neural network model to be trained is trained according to the regional attention characterization results until a preset condition is met, thereby obtaining a trained meteorological detection model.

7. The method according to claim 6, characterized in that The neural network model to be trained performs regional attention characterization on the sample meteorological regional features, and trains the neural network model to be trained according to the regional attention characterization result until a preset condition is met, thereby obtaining a trained meteorological detection model, including: Based on the neural network model to be trained, the sample meteorological regional features are characterized by regional attention, and the sample meteorological correlation features are output; Based on the neural network model to be trained, the sample meteorological correlation features and the sample meteorological integration features are spliced ​​to determine the sample meteorological prediction result; The loss between the sample meteorological actual result and the sample meteorological forecast result is determined, and the parameters of the neural network model to be trained are adjusted according to the loss until the preset conditions are met, thereby obtaining a trained meteorological detection model.

8. A weather forecasting device, characterized in that: include: An acquisition module is used to acquire meteorological parameters of meteorological elements in a preset area at a current moment by performing meteorological collection on meteorological elements in the preset area; A first feature extraction module, configured to obtain meteorological integration features of meteorological elements in the preset area based on meteorological parameters of the meteorological elements in the preset area at a current moment; A second feature extraction module is used to process the meteorological integration features to obtain meteorological regional features, and perform regional attention characterization on the meteorological regional features to obtain meteorological correlation features of meteorological elements in the preset area; The prediction module is used to combine the meteorological integration features and the meteorological correlation features to predict the meteorological parameters of the preset area at the target time.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method described in any one of claims 1 to 7 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 7 through the computer program.

Images