Artificial intelligence weather forecasting method and device guided by physical information of wind farm

Abstract

This disclosure relates to a method and apparatus for AI-guided weather forecasting using physical information of a wind farm. It includes: acquiring first historical data of multiple key meteorological elements in a target wind farm, wherein the influence of each key meteorological element on the weather in the target wind farm is greater than a preset threshold; using a weather prediction model for the target wind farm to predict the weather based on the first historical data of the multiple key meteorological elements, obtaining predicted weather data for the target wind farm at future times; since the weather prediction model is pre-trained using multiple meteorological parameters in a coarse-resolution mode, and then fine-tuned using historical data of multiple key meteorological elements that have a significant impact on the weather, a micro-scale weather prediction model is obtained to adapt to the corresponding wind farm for weather forecasting, improving the resolution and efficiency of weather forecasting while reducing computational resource consumption, thereby lowering the cost of weather forecasting.

Description

Technical Field

[0001] This disclosure relates to the field of wind farm information prediction technology, and in particular to a method and apparatus for guiding artificial intelligence weather forecasting based on the physical information of a wind farm. Background Technology

[0002] As a representative of new energy sources, weather-based power generation methods are characterized by intermittency and volatility, which significantly impacts the continuous and stable operation of the power system. More accurate weather forecasts can predict short-term weather conditions for wind farms, thereby making wind power output more predictable and ensuring the stable operation of the power system.

[0003] In existing technologies, weather forecasting for wind farms generally employs methods based on solving numerical partial differential equations or mathematical statistical modeling. However, existing weather forecasting methods suffer from low resolution, low efficiency, and high cost. Therefore, there is an urgent need to provide a more efficient and accurate weather forecasting method. Summary of the Invention

[0004] To address the aforementioned technical issues, this disclosure provides a method and apparatus for AI-guided weather forecasting using physical information from wind farms.

[0005] Firstly, this disclosure provides a method for guiding artificial intelligence weather forecasting using physical information from wind farms, including:

[0006] First historical data of multiple key meteorological elements in the target wind farm are obtained, wherein the influence of the multiple key meteorological elements on the weather in the target wind farm is greater than a preset threshold.

[0007] Using the weather prediction model of the target wind farm, weather prediction is performed on the first historical data of the various key meteorological elements to obtain the predicted weather data of the target wind farm at future times;

[0008] The weather prediction model is pre-trained using historical data of various meteorological parameters in different wind farms, and then fine-tuned using second historical data of various key meteorological parameters from the target wind farm.

[0009] Secondly, this disclosure provides a physical information-guided artificial intelligence weather forecasting device for wind farms, comprising:

[0010] The first acquisition module is used to acquire the first historical data of multiple key meteorological elements in the target wind farm, wherein the influence of the multiple key meteorological elements on the weather in the target wind farm is greater than a preset threshold.

[0011] The weather forecasting module is used to use the weather forecasting model of the target wind farm to perform weather forecasting on the first historical data of the various key meteorological elements, and obtain the predicted weather data of the target wind farm at future times.

[0012] The weather prediction model is pre-trained using historical data of various meteorological parameters in different wind farms, and then fine-tuned using second historical data of various key meteorological parameters from the target wind farm.

[0013] Thirdly, embodiments of this disclosure also provide an electronic device, including:

[0014] One or more processors;

[0015] Storage device for storing one or more programs.

[0016] When one or more programs are executed by one or more processors, the one or more processors implement the methods provided in the first aspect.

[0017] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method provided in the first aspect.

[0018] The technical solution provided in this disclosure has the following advantages compared with the prior art:

[0019] This disclosure discloses a method and apparatus for guiding artificial intelligence weather forecasting based on the physical information of a wind farm. The method acquires first historical data of multiple key meteorological elements in a target wind farm, wherein the influence of these elements on the weather in the target wind farm is greater than a preset threshold. A weather prediction model for the target wind farm is used to predict the weather based on the first historical data of these key meteorological elements, yielding predicted weather data for the target wind farm at future times. The weather prediction model is pre-trained using historical data of multiple meteorological parameters from different wind farms, and then fine-tuned using second historical data of multiple key meteorological parameters from the target wind farm. This approach, through pre-training with a coarse-resolution model and fine-tuning with historical data of multiple key meteorological elements that significantly influence weather, obtains a micro-scale weather prediction model adapted to the specific wind farm, improving the resolution and efficiency of weather forecasting while reducing computational resource consumption and lowering forecasting costs. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0021] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A flowchart illustrating a wind farm physical information-guided artificial intelligence weather forecasting method provided in this embodiment of the disclosure;

[0023] Figure 2 A flowchart illustrating another method for guiding artificial intelligence weather forecasting using physical information of a wind farm, as provided in this embodiment of the disclosure;

[0024] Figure 3 A schematic diagram of the structure of a wind farm physical information-guided artificial intelligence weather forecasting device provided in this embodiment of the disclosure;

[0025] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0026] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0027] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0028] The related technologies employ methods based on solving numerical partial differential equations, specifically: solving fluid dynamics and thermodynamic equations (such as the Navier-Stokes equations and continuity equations) under given initial and boundary conditions to obtain the future atmospheric evolution. These methods can characterize weather change mechanisms at the physical level and have strong theoretical support. However, the drawbacks of this method are that, in practical situations, it requires high resolution, and when solving for fine-resolution weather processes, it consumes significant computational resources, resulting in high costs and low computational efficiency.

[0029] Related technologies based on mathematical statistical modeling methods specifically employ regression models, time series models (such as ARIMA), probability distribution models (such as normal distribution and stable distribution), and stochastic simulation methods (such as Monte Carlo, Markov chains, and kernel density estimation). However, the drawback of this approach is its over-reliance on historical data for fitting, its insufficient ability to capture nonlinear relationships under complex meteorological conditions, and its difficulty in reflecting the true differences and local disturbance characteristics at local locations.

[0030] Among related technologies, artificial intelligence-based methods are also used to predict weather at wind farms. Specifically, this involves analyzing historical weather data using large-scale meteorological models to make kilometer-level weather forecasts. However, the drawback of this method is that, relying on historical weather data to predict future weather, it is essentially similar to methods based on mathematical statistical modeling, considering only a single parameter and thus limiting prediction accuracy.

[0031] To solve the above problems, the following will combine... Figures 1-2 This disclosure describes a method for guiding artificial intelligence weather forecasting using the physical information of a wind farm, as provided in an embodiment. In this embodiment, the method can be executed by an electronic device or a server. The electronic device may include devices with communication capabilities such as tablets, desktop computers, and laptops, or devices simulated by virtual machines or simulators. The server may be a cloud server or a server cluster.

[0032] Figure 1 A flowchart illustrating a wind farm physical information-guided artificial intelligence weather forecasting method provided in an embodiment of this disclosure is shown.

[0033] like Figure 1 As shown, the physical information of this wind farm guiding an artificial intelligence weather forecasting method may include the following steps.

[0034] S110. Obtain the first historical data of multiple key meteorological elements in the target wind farm, wherein the influence of multiple key meteorological elements on the weather in the target wind farm is greater than the preset threshold.

[0035] When weather forecasting is required for wind farms, in order to improve the accuracy of weather forecasts, electronic devices acquire historical data from the wind farms that have a significant impact on weather, and use this data as the raw data to drive weather forecasting.

[0036] Among them, the first historical data can be meteorological element data from several days, hours, or minutes prior to the current time.

[0037] Among them, since the influence of key meteorological elements on weather is greater than the preset threshold, key meteorological elements are recorded as meteorological parameters that are highly correlated with weather. The preset threshold can be a pre-set empirical value.

[0038] Optionally, multiple key meteorological elements, including: surface velocity, compressive velocity, compressive temperature, and surface temperature difference.

[0039] Specifically, the methods for determining key meteorological elements include, but are not limited to, the following: determining the impact of each meteorological parameter on the weather in the target wind farm; and acquiring multiple meteorological parameters with an impact greater than a preset threshold as multiple key meteorological elements.

[0040] The method for determining the impact of each meteorological parameter on the weather in the target wind farm includes: obtaining second historical data for each meteorological parameter from a preset three-dimensional region of the target wind farm; for each meteorological parameter, using any one of the second historical data points as the grid center and a preset number of second historical data points as surrounding grid points; calculating the data difference between the second historical data at the grid center and the second historical data at the surrounding grid points; and conducting a correlation analysis between the data difference and the weather in the target wind farm to determine the impact.

[0041] Specifically, the electronic device uses physical equations (such as the Navier-Stokes equations) as its theoretical basis, selects second historical data of various meteorological parameters of a preset three-dimensional region of the target wind farm, constructs a grid using multiple second historical data for each meteorological parameter, and analyzes the correlation between the data difference of each meteorological parameter and the weather by calculating the data difference between the second historical data at the center of the grid and the second historical data of the surrounding grid points, thereby obtaining the influence of each meteorological parameter on the weather. Then, the influence of each meteorological parameter on the weather is compared with a preset threshold, and meteorological parameters with an influence greater than the preset threshold are obtained as key meteorological elements.

[0042] S120. Using the weather prediction model of the target wind farm, weather prediction is performed on the first historical data of multiple key meteorological elements to obtain the predicted weather data of the target wind farm at future times. The weather prediction model is pre-trained using historical data of multiple meteorological parameters in different wind farms, and then fine-tuned using the second historical data of multiple key meteorological parameters among the multiple meteorological parameters of the target wind farm.

[0043] In this embodiment, before predicting the weather at the target wind farm, the electronic device can also acquire a weather prediction model for the target wind farm based on coarse-resolution pre-training and fine-resolution fine-tuning. Specifically, the electronic device can use historical data of various meteorological parameters from different wind farms to acquire a pre-trained model, thereby capturing the nonlinear interactions and dynamic evolution patterns among various complex meteorological elements under coarse-resolution (kilometer-level resolution) mode. Then, the pre-trained model is transferred to the target wind farm, and fine-tuned using the fine-resolution (hundred-meter-level resolution) meteorological data of the target wind farm itself to obtain the weather prediction model.

[0044] Among them, the weather prediction model is a lightweight model with low complexity, which is easy to transfer and can be adapted to various different scenarios for weather prediction.

[0045] Forecasted weather data refers to weather information used for power generation at power plants. Optionally, forecasted weather data may include one or more combinations of wind speed, wind direction, and sunlight intensity.

[0046] Furthermore, after acquiring first-historical data of various key meteorological elements, the electronic equipment directly processes the first-historical data using a weather prediction model to obtain the predicted weather data for the target wind farm at future times.

[0047] This disclosure discloses a method for guiding artificial intelligence weather forecasting based on the physical information of a wind farm. The method acquires first historical data of multiple key meteorological elements in a target wind farm, wherein the influence of these elements on the weather at the target wind farm exceeds a preset threshold. A weather prediction model for the target wind farm is used to predict the weather based on the first historical data of these key meteorological elements, yielding predicted weather data for the target wind farm at future times. The weather prediction model is pre-trained using historical data of multiple meteorological parameters from different wind farms, and then fine-tuned using second historical data of multiple key meteorological parameters from the target wind farm. This approach, through pre-training with a coarse-resolution model and fine-tuning with historical data of multiple key meteorological elements that significantly influence weather, obtains a micro-scale weather prediction model adapted to the specific wind farm, improving the resolution and efficiency of weather forecasting while reducing computational resource consumption and lowering forecasting costs.

[0048] In another embodiment of this disclosure, the specific implementation method of S120 will be explained in detail.

[0049] Figure 2 A flowchart illustrating another wind farm physical information-guided artificial intelligence weather forecasting method provided in this disclosure embodiment is shown.

[0050] like Figure 2 As shown, the physical information of this wind farm guiding an artificial intelligence weather forecasting method may include the following steps.

[0051] S210. Obtain the first historical data of multiple key meteorological elements in the target wind farm, wherein the influence of multiple key meteorological elements on the weather in the target wind farm is greater than the preset threshold.

[0052] S210 is similar to S110, and will not be described in detail here.

[0053] S220. Based on the coding network in the weather prediction model, the first historical data of multiple key meteorological elements are coded respectively to obtain the coding features corresponding to the multiple key meteorological elements.

[0054] The weather prediction model is pre-trained using historical data of various meteorological parameters from different wind farms, and then fine-tuned using second historical data of several key meteorological parameters from the target wind farm.

[0055] Optionally, various key meteorological elements include one-dimensional and two-dimensional meteorological elements. For example, surface temperature difference is a one-dimensional meteorological element, while surface velocity, compressive surface velocity, and compressive surface temperature are all two-dimensional meteorological elements.

[0056] In some embodiments, multiple key meteorological elements include one-dimensional meteorological elements; then, the specific implementation method of S220 includes: performing linear processing on the first historical data of the one-dimensional meteorological elements based on the linear layer in the coding network to obtain the mapping features of the one-dimensional meteorological elements; performing activation processing on the mapping features of the one-dimensional meteorological elements based on the first activation layer in the coding network to obtain the activation features of the one-dimensional meteorological elements; and performing normalization processing on the activation features of the one-dimensional meteorological elements based on the first normalization layer in the coding network to obtain the coding features of the one-dimensional meteorological elements.

[0057] In other embodiments, multiple key meteorological elements include two-dimensional meteorological elements; then, the specific implementation method of S220 includes: performing convolution processing on the first historical data of the two-dimensional meteorological elements based on the convolution layer in the coding network to obtain the convolution features of the two-dimensional meteorological elements; performing activation processing on the convolution features of the two-dimensional meteorological elements based on the second activation layer in the coding network to obtain the activation features of the two-dimensional meteorological elements; and performing encoding features on the activation features of the two-dimensional meteorological elements based on the second normalization layer in the coding network.

[0058] It should be noted that, regardless of whether it is a one-dimensional or two-dimensional meteorological element, after being encoded by the coding network, each meteorological element is presented in the form of a two-dimensional vector. Specifically, one dimension of the two-dimensional vector is used to identify the meteorological element, clarifying which category of meteorological element the two-dimensional vector corresponds to, and the other dimension is used to identify the mapping result of the meteorological element in high-dimensional space.

[0059] In this way, the information carrying capacity and expression accuracy of features are significantly improved through high-dimensional space mapping, providing support for the accurate interaction and complementarity of information between elements in the subsequent association and fusion process.

[0060] S230. Based on the fusion network in the weather prediction model, the coding features corresponding to multiple key meteorological elements are fused to obtain the fused features of multiple key meteorological elements.

[0061] The specific implementation method of S230 includes, but is not limited to, the following method: a multi-head attention network based on a fusion network is used to perform weighted summation on the coding features corresponding to multiple key meteorological elements to obtain the fusion features of multiple key meteorological elements.

[0062] Specifically, electronic devices utilize multi-head attention mechanisms to construct weight matrices among high-dimensional features of different meteorological elements, thereby achieving deep fusion and information complementarity of multiple features and effectively uncovering potential correlations among meteorological elements.

[0063] Therefore, by utilizing a fusion network based on a multi-head attention mechanism, key information between different meteorological elements can be efficiently captured and utilized, providing universal meteorological correlation characteristics for subsequent targeted output of microscale weather.

[0064] S240. Based on the prediction network in the weather prediction model, weather prediction is performed on the fusion characteristics of multiple key meteorological elements to obtain the predicted weather data for the wind farm.

[0065] The specific implementation method of S240 includes, but is not limited to, the following methods: based on the first linear layer in the prediction network, the fusion features of multiple key meteorological elements are linearly processed to obtain the first linear features of the fusion features; based on the activation layer in the prediction network, the first linear features of the fusion features are activated to obtain the activation features of the fusion features; based on the second linear layer in the prediction network, the activation features of the fusion features are linearly processed to obtain the predicted weather data of the wind farm.

[0066] Specifically, the electronic device employs a two-layer linear structure and uses activation layers to construct a lightweight network while suppressing overfitting. Optionally, the activation layers can utilize the multi-layer nonlinear mapping capability of a multi-layer perceptron (MLP) network to accurately transform the fused features into point-based weather data, thereby enhancing the fitting ability of the complex nonlinear relationship between grid meteorological elements and fine-resolution point-based weather data.

[0067] In this way, by encoding and fusing the first historical data of multiple meteorological elements through weather prediction models, not only is the problem of fusion difficulties caused by differences in the original feature dimensions and numerical magnitudes of different meteorological elements solved, but more importantly, a general multivariate high-dimensional output modeling paradigm is provided. Since this paradigm does not rely on the exclusive features of a specific region, but is based on the standardized processing logic of general meteorological elements, it can be easily transferred to other different regions, greatly improving the model's versatility and promotion value.

[0068] This disclosure also provides a wind farm physical information-guided artificial intelligence weather forecasting device for implementing the above-described wind farm physical information-guided artificial intelligence weather forecasting method. The following is in conjunction with... Figure 3 The following explanation is provided. In this embodiment, the physical information of the wind farm guiding the artificial intelligence weather forecasting device can be an electronic device or a server. The electronic device can include devices with communication functions such as tablets, desktop computers, and laptops, or devices simulated by virtual machines or simulators. The server can be a cloud server or a server cluster.

[0069] Figure 3 A schematic diagram of the structure of a wind farm physical information-guided artificial intelligence weather forecasting device provided in an embodiment of this disclosure is shown.

[0070] like Figure 3 As shown, the physical information of the wind farm guiding the artificial intelligence weather forecasting device 300 may include:

[0071] The first acquisition module 310 is used to acquire the first historical data of multiple key meteorological elements in the target wind farm, wherein the influence of the multiple key meteorological elements on the weather in the target wind farm is greater than a preset threshold.

[0072] The weather forecasting module 320 is used to use the weather forecasting model of the target wind farm to perform weather forecasting on the first historical data of the multiple key meteorological elements, and obtain the predicted weather data of the target wind farm at future times.

[0073] The weather prediction model is pre-trained using historical data of various meteorological parameters in different wind farms, and then fine-tuned using second historical data of various key meteorological parameters from the target wind farm.

[0074] This disclosure discloses a wind farm physical information-guided artificial intelligence weather forecasting device. It acquires first historical data of multiple key meteorological elements in a target wind farm, wherein the influence of these elements on the weather in the target wind farm exceeds a preset threshold. Using a weather prediction model for the target wind farm, it performs weather prediction on the first historical data of these key meteorological elements to obtain predicted weather data for the target wind farm at future times. The weather prediction model is pre-trained using historical data of multiple meteorological parameters from different wind farms, and then fine-tuned using second historical data of multiple key meteorological parameters from the target wind farm. Thus, through pre-training with a coarse-resolution model and fine-tuning with historical data of multiple key meteorological elements that significantly influence weather, a micro-scale weather prediction model is obtained to adapt to the corresponding wind farm for weather forecasting. This improves the resolution and efficiency of weather forecasting while reducing computational resource consumption, thereby lowering the cost of weather forecasting.

[0075] In some embodiments of this disclosure, the various key meteorological elements include:

[0076] Surface velocity, compressive surface velocity, compressive surface temperature, and surface temperature difference.

[0077] In some embodiments of this disclosure, the weather forecasting module 320 includes:

[0078] The coding unit is used to encode the first historical data of the multiple key meteorological elements based on the coding network in the weather prediction model, so as to obtain the coding features corresponding to the multiple key meteorological elements respectively.

[0079] The fusion unit is used to fuse the coding features corresponding to the various key meteorological elements based on the fusion network in the weather prediction model, so as to obtain the fusion features of the various key meteorological elements.

[0080] The prediction unit is used to perform weather prediction based on the prediction network in the weather prediction model, using the fusion characteristics of the multiple key meteorological elements, to obtain the predicted weather data for the wind farm.

[0081] In some embodiments of this disclosure, the multiple key meteorological elements include one-dimensional meteorological elements; the encoding unit is specifically used for:

[0082] Based on the linear layer in the coding network, the first historical data of the one-dimensional meteorological element is linearly processed to obtain the mapping characteristics of the one-dimensional meteorological element.

[0083] Based on the first activation layer in the coding network, the mapping features of the one-dimensional meteorological element are activated to obtain the activation features of the one-dimensional meteorological element.

[0084] Based on the first normalization layer in the coding network, the activation features of the one-dimensional meteorological element are normalized to obtain the coding features of the one-dimensional meteorological element.

[0085] In some embodiments of this disclosure, the various key meteorological elements include two-dimensional meteorological elements; the encoding unit is specifically used for:

[0086] Based on the convolutional layers in the coding network, the first historical data of the two-dimensional meteorological elements are convolved to obtain the convolutional features of the two-dimensional meteorological elements.

[0087] Based on the second activation layer in the coding network, the convolutional features of the two-dimensional meteorological elements are activated to obtain the activation features of the two-dimensional meteorological elements.

[0088] Based on the second normalization layer in the coding network, the activation features of the two-dimensional meteorological elements are used to obtain the coding features of the two-dimensional meteorological elements.

[0089] In some embodiments of this disclosure, the fusion unit is specifically used for:

[0090] Based on the multi-head attention network of the fusion network, the coding features corresponding to the various key meteorological elements are weighted and summed to obtain the fusion features of the various key meteorological elements.

[0091] In some embodiments of this disclosure, the prediction unit is specifically used for:

[0092] Based on the first linear layer in the prediction network, the fusion features of the multiple key meteorological elements are linearly processed to obtain the first linear features of the fusion features;

[0093] Based on the activation layer in the prediction network, the first linear feature of the fused feature is activated to obtain the activated feature of the fused feature;

[0094] Based on the second linear layer in the prediction network, the activation features of the fused features are linearly processed to obtain the predicted weather data for the wind farm.

[0095] In some embodiments of this disclosure, the device further includes:

[0096] The impact determination module is used to determine the impact of each meteorological parameter on the weather in the target wind farm.

[0097] The second acquisition module is used to acquire a variety of meteorological parameters whose influence is greater than the preset threshold, as the various key meteorological elements.

[0098] In some embodiments of this disclosure, the influence determination module is specifically used for:

[0099] Second historical data for each meteorological parameter are obtained from a preset three-dimensional region of the target wind farm;

[0100] For each meteorological parameter, any second historical data point is used as the grid center, and a preset number of second historical data points are used as the surrounding grid points.

[0101] Calculate the data difference between the second historical data of the grid center and the second historical data of the surrounding grid points;

[0102] The degree of influence is determined by performing a correlation analysis between the data difference and the weather in the target wind farm.

[0103] It should be noted that, Figure 3 The physical information of the wind farm shown guides the AI ​​weather forecasting device 300 to perform [its tasks]. Figures 1-2 The various steps in the method embodiment shown are implemented. Figures 1-2 The processes and effects in the method embodiments shown are not described in detail here.

[0104] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure is shown.

[0105] like Figure 4 As shown, the electronic device may include a processor 401 and a memory 402 storing computer program instructions.

[0106] Specifically, the processor 401 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0107] Memory 402 may include a large-capacity storage for information or instructions. For example, and not limitingly, memory 402 may include a hard disk drive (HDD), a floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 402 may include removable or non-removable (or fixed) media. Where appropriate, memory 402 may be internal or external to the integrated gateway device. In a particular embodiment, memory 402 is a non-volatile solid-state memory. In a particular embodiment, memory 402 includes read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (Electrically Programmable ROM, EPROM), an electrically erasable programmable PROM (EEPROM), an electrically alterable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0108] The processor 401 reads and executes computer program instructions stored in the memory 402 to perform the steps of the wind farm physical information-guided artificial intelligence weather forecasting method provided in this embodiment of the disclosure.

[0109] In one example, the electronic device may also include a transceiver 403 and a bus 404. Wherein, as... Figure 4 As shown, the processor 401, memory 402 and transceiver 403 are connected via bus 404 and communicate with each other.

[0110] Bus 404 includes hardware, software, or both. For example, and not limitingly, a bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industrial Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 404 may include one or more buses. Although specific buses are described and illustrated in the embodiments of this application, this application considers any suitable bus or interconnection.

[0111] The following are embodiments of a computer-readable storage medium provided in this disclosure. This computer-readable storage medium and the wind farm physical information-guided artificial intelligence weather forecasting method in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the computer-readable storage medium, please refer to the embodiments of the wind farm physical information-guided artificial intelligence weather forecasting method described above.

[0112] This embodiment provides a storage medium containing computer-executable instructions. When executed by a computer processor, these instructions are used to perform a wind farm physical information-guided artificial intelligence weather forecasting method, including:

[0113] First historical data of multiple key meteorological elements in the target wind farm are obtained, wherein the influence of the multiple key meteorological elements on the weather in the target wind farm is greater than a preset threshold.

[0114] Using the weather prediction model of the target wind farm, weather prediction is performed on the first historical data of the various key meteorological elements to obtain the predicted weather data of the target wind farm at future times;

[0115] The weather prediction model is pre-trained using historical data of various meteorological parameters in different wind farms, and then fine-tuned using second historical data of various key meteorological parameters from the target wind farm.

[0116] Of course, the computer-executable instructions provided in the embodiments of this disclosure are not limited to the above-described method operations, but can also perform related operations in the wind farm physical information-guided artificial intelligence weather forecasting method provided in any embodiment of this disclosure.

[0117] Based on the above description of the implementation methods, those skilled in the art can clearly understand that this disclosure can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer cloud platform (which can be a personal computer, server, or network cloud platform, etc.) to execute the wind farm physical information-guided artificial intelligence weather forecasting method provided in the various embodiments of this disclosure.

[0118] Note that the above description is merely a preferred embodiment and the technical principles employed in this disclosure. Those skilled in the art will understand that this disclosure is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this disclosure. Therefore, although this disclosure has been described in detail through the above embodiments, it is not limited to the above embodiments. Many other equivalent embodiments may be included without departing from the concept of this disclosure, and the scope of this disclosure is determined by the scope of the appended claims.

Claims

1. A method for AI-guided weather forecasting using physical information from wind farms, characterized in that, include: First historical data of multiple key meteorological elements in the target wind farm are obtained, wherein the influence of the multiple key meteorological elements on the weather in the target wind farm is greater than a preset threshold. Using the weather prediction model of the target wind farm, weather prediction is performed on the first historical data of the various key meteorological elements to obtain the predicted weather data of the target wind farm at future times; The weather prediction model is pre-trained using historical data of various meteorological parameters in different wind farms, and then fine-tuned using second historical data of various key meteorological parameters from the target wind farm. Also includes: Determine the degree of influence of each meteorological parameter on the weather in the target wind farm; Multiple meteorological parameters with an impact greater than the preset threshold are acquired as the multiple key meteorological elements; Determining the impact of each meteorological parameter on the weather in the target wind farm includes: Second historical data for each meteorological parameter are obtained from a preset three-dimensional region of the target wind farm; For each meteorological parameter, any second historical data point is used as the grid center, and a preset number of second historical data points are used as the surrounding grid points. Calculate the data difference between the second historical data of the grid center and the second historical data of the surrounding grid points; The degree of influence is determined by performing a correlation analysis between the data difference and the weather in the target wind farm.

2. The method according to claim 1, characterized in that, The key meteorological elements include: Surface velocity, compressive surface velocity, compressive surface temperature, and surface temperature difference.

3. The method according to claim 1, characterized in that, The method of using the weather prediction model of the target wind farm to perform weather prediction on the first historical data of the various key meteorological elements, and obtaining the predicted weather data of the target wind farm at future times, includes: Based on the coding network in the weather prediction model, the first historical data of the various key meteorological elements are coded respectively to obtain the coding features corresponding to the various key meteorological elements. Based on the fusion network in the weather prediction model, the coding features corresponding to the various key meteorological elements are fused to obtain the fusion features of the various key meteorological elements. Based on the prediction network in the weather prediction model, weather prediction is performed on the fusion characteristics of the multiple key meteorological elements to obtain the predicted weather data for the wind farm.

4. The method according to claim 3, characterized in that, The multiple key meteorological elements include one-dimensional meteorological elements; the first historical data of the multiple key meteorological elements are encoded based on the coding network in the weather prediction model to obtain the coding features corresponding to the multiple key meteorological elements, including: Based on the linear layer in the coding network, the first historical data of the one-dimensional meteorological element is linearly processed to obtain the mapping characteristics of the one-dimensional meteorological element. Based on the first activation layer in the coding network, the mapping features of the one-dimensional meteorological element are activated to obtain the activation features of the one-dimensional meteorological element. Based on the first normalization layer in the coding network, the activation features of the one-dimensional meteorological element are normalized to obtain the coding features of the one-dimensional meteorological element.

5. The method according to claim 3, characterized in that, The multiple key meteorological elements include two-dimensional meteorological elements; the first historical data of the multiple key meteorological elements are encoded based on the coding network in the weather prediction model to obtain the coding features corresponding to the multiple key meteorological elements, including: Based on the convolutional layers in the coding network, the first historical data of the two-dimensional meteorological elements are convolved to obtain the convolutional features of the two-dimensional meteorological elements. Based on the second activation layer in the coding network, the convolutional features of the two-dimensional meteorological elements are activated to obtain the activation features of the two-dimensional meteorological elements. Based on the second normalization layer in the coding network, the activation features of the two-dimensional meteorological elements are used to obtain the coding features of the two-dimensional meteorological elements.

6. The method according to claim 3, characterized in that, The fusion network based on the weather prediction model fuses the coding features corresponding to the various key meteorological elements to obtain fused features of the various key meteorological elements, including: Based on the multi-head attention network of the fusion network, the coding features corresponding to the various key meteorological elements are weighted and summed to obtain the fusion features of the various key meteorological elements.

7. The method according to claim 3, characterized in that, The prediction network based on the weather prediction model performs weather prediction on the fusion features of the multiple key meteorological elements to obtain predicted weather data for the wind farm, including: Based on the first linear layer in the prediction network, the fusion features of the multiple key meteorological elements are linearly processed to obtain the first linear features of the fusion features; Based on the activation layer in the prediction network, the first linear feature of the fused feature is activated to obtain the activated feature of the fused feature; Based on the second linear layer in the prediction network, the activation features of the fused features are linearly processed to obtain the predicted weather data for the wind farm.

8. A physical information-guided artificial intelligence weather forecasting device for wind farms, characterized in that, include: The first acquisition module is used to acquire the first historical data of multiple key meteorological elements in the target wind farm, wherein the influence of the multiple key meteorological elements on the weather in the target wind farm is greater than a preset threshold. The weather forecasting module is used to use the weather forecasting model of the target wind farm to perform weather forecasting on the first historical data of the various key meteorological elements, and obtain the predicted weather data of the target wind farm at future times. The weather prediction model is pre-trained using historical data of various meteorological parameters in different wind farms, and then fine-tuned using second historical data of various key meteorological parameters from the target wind farm. Also includes: The impact determination module is used to determine the impact of each meteorological parameter on the weather in the target wind farm. The second acquisition module is used to acquire multiple meteorological parameters with an impact greater than the preset threshold, as the multiple key meteorological elements; The influence determination module is specifically used for: Second historical data for each meteorological parameter are obtained from a preset three-dimensional region of the target wind farm; For each meteorological parameter, any second historical data point is used as the grid center, and a preset number of second historical data points are used as the surrounding grid points. Calculate the data difference between the second historical data of the grid center and the second historical data of the surrounding grid points; The degree of influence is determined by performing a correlation analysis between the data difference and the weather in the target wind farm.

Images