Wind resource prediction method, device and equipment for medium and micro-scale terrains and medium

Abstract

The invention provides a wind resource prediction method, device and equipment for medium and micro-scale terrains, and a medium, and the method comprises the steps: extracting the surface roughness data of a target terrain, carrying out the fluid dynamics numerical simulation processing of the target terrain after grid division, and determining the three-dimensional flow field data; performing data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data and the surface roughness data, and determining a fixed-dimension terrain elevation matrix and a fixed-dimension surface roughness matrix; carrying out obstruction coefficient calculation on the fixed-dimension terrain elevation matrix in the incoming flow direction, and determining wind field obstruction characteristics of the target terrain; and inputting the fixed-dimension terrain elevation matrix, the wind field obstruction features and the fixed-dimension surface roughness matrix into a wind resource prediction model to carry out convolution down-sampling, nonlinear transformation, a self-attention mechanism, convolution up-sampling and feature fusion, and predicting a wind resource result map of the target terrain. And a wind resource result of any terrain can be quickly and accurately predicted.

Description

Technical Field

[0001] The present application relates to the technical field of wind resource prediction, and in particular to a method, device, equipment and medium for predicting wind resources for meso- and micro-scale terrain. Background Art

[0002] At present, wind resource prediction mainly relies on numerical weather prediction models, statistical analysis methods and traditional machine learning algorithms. Numerical weather prediction models simulate and predict future wind speed and direction by solving atmospheric physics equations, combining terrain data and real-time weather reports. However, existing wind resource prediction technologies have the following major defects in data collection, model performance and production deployment: (1) Difficulty in data collection: The collection of wind resource data mainly relies on a limited number of measurement points that are discretely distributed in time and space. Although these measurement points can obtain wind resource information within a certain range, from the perspective of the overall wind field, they are only local samples and cannot comprehensively and continuously cover the entire wind field space. (2) Insufficient model performance: Traditional methods such as numerical simulation methods and machine learning methods have problems such as low accuracy and slow computing speed. For example, numerical simulation requires a lot of computing resources and has a slow prediction speed; while machine learning methods lack accuracy, especially under complex terrain and irregular wind direction conditions, the prediction effect is limited. (3) Low accuracy of wind resource prediction. Therefore, how to improve the accuracy of wind resource prediction has become a technical issue that cannot be underestimated. Summary of the invention

[0003] In view of this, the purpose of this application is to provide a method, device, equipment and medium for wind resource prediction of meso-micro scale terrain. Through the wind resource prediction model, the wind resource results of any meso-micro scale terrain can be quickly and accurately predicted, while enhancing the adaptability and versatility of the model and reducing dependence on large amounts of historical data and high-performance hardware.

[0004] The present application provides a method for predicting wind resources for meso-microscale terrains, the method comprising: Extracting surface roughness data of the target terrain, and performing fluid dynamics numerical simulation processing on the target terrain after meshing to determine three-dimensional flow field data of the target terrain; Performing data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data, and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix; Calculating the barrier coefficient of the fixed-dimensional terrain elevation matrix in the incoming flow direction to determine the wind field barrier characteristics of the target terrain; The fixed-dimensional terrain elevation matrix, the wind field barrier features and the fixed-dimensional surface roughness matrix are input into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain; wherein the wind resource prediction model is obtained by training the initial network model with the target height fixed-dimensional flow field matrix as the training target, and the initial network model is obtained by adding multiple Transformer layers to the UNet architecture convolutional neural network model.

[0005] In a possible implementation manner, the data preprocessing of the grid coordinates of the target terrain, the three-dimensional flow field data, and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix includes: Linearly interpolating the underlying surface grid coordinates in the grid coordinates of the target terrain into a two-dimensional terrain matrix of fixed dimension, normalizing the original terrain height of each terrain data point based on the lowest height in the two-dimensional terrain matrix, and determining the fixed-dimensional terrain elevation matrix; Based on the linear interpolation, the surface roughness data of each terrain data point is converted, outliers are removed, and maximum and minimum values ​​are normalized to determine the fixed-dimensional surface roughness matrix.

[0006] In a possible implementation manner, calculating the barrier coefficient of the fixed-dimensional terrain elevation matrix in the incoming flow direction to determine the wind field barrier characteristics of the target terrain includes: For each terrain data point, starting from the terrain data point, judging point by point along the upwind straight line, to determine whether the terrain elevation data of the upwind terrain data point on the upwind straight line is greater than the terrain elevation data of the terrain data point; If so, determining the blocking coefficient of the upwind terrain data point to the terrain data point based on the height difference between the terrain elevation data of the upwind terrain data point and the terrain elevation data of the terrain data point and the straight-line distance between the upwind terrain data point and the terrain data point; Accumulating the barrier coefficient of each terrain data point to determine the comprehensive barrier coefficient of each terrain data point; The maximum and minimum values ​​of the plurality of comprehensive blocking coefficients are normalized to determine the wind field blocking characteristics of the target terrain.

[0007] In a possible implementation manner, the blocking coefficient of the upwind terrain data point to the terrain data point is determined by the following steps:

[0008]

[0009] in, For terrain data points The barrier factor, is the upwind terrain data point on the upwind straight line, The value range is , z is the terrain elevation data, t is the height difference, L is the terrain length of the fixed-dimension terrain elevation data, is a symbolic function.

[0010] In a possible implementation manner, the fixed-dimensional terrain elevation matrix, the wind field barrier feature, and the fixed-dimensional surface roughness matrix are input into a wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling, and feature fusion to predict the wind resource result map of the target terrain, including: Based on the multiple coding layers of the wind resource prediction model, convolution downsampling feature extraction is performed on the fixed-dimensional terrain elevation matrix, the wind field barrier feature and the fixed-dimensional surface roughness matrix in turn to determine coding feature maps of different scales; Based on multiple Transformer layers, a nonlinear transformation and a self-attention mechanism are performed on the deep coding feature map output by the deepest coding layer to extract implicit features of the deep coding feature map containing local information and global information; Based on the multiple decoding layers of the wind resource prediction model, the implicit features and the multiple coded feature maps are subjected to convolution up-sampling processing, and in the decoding process, the decoded feature map output by the decoding layer and the corresponding coded feature map are subjected to feature fusion by means of skip connection; The convolution layer based on the wind resource prediction model performs convolution processing on the decoding feature map output by the last decoding layer to generate the wind resource result map; wherein the wind resource result map displays the wind speed modulus at each terrain data point.

[0011] In a possible implementation manner, the wind resource prediction model is determined by the following steps: Inputting a sample fixed-dimensional terrain elevation matrix, a sample wind field barrier feature, and a sample fixed-dimensional surface roughness matrix of a plurality of sample terrains into an initial network model to determine a predicted wind resource result map for each of the sample terrains; Based on the linear interpolation algorithm, the sample three-dimensional flow field data of each sample terrain is converted, outliers are removed, and maximum and minimum values ​​are processed, and the wind speed component of each sample terrain data point at the height of the ground target is extracted from the processed sample three-dimensional flow field data; The actual wind speed modulus of each sample terrain at the target height is calculated based on the wind speed component of each sample terrain data point; The initial network model is iteratively trained based on the mean square error and the mean absolute value error between the actual wind speed modulus of each sample terrain and the wind speed modulus in the corresponding predicted wind resource result map to determine the wind resource prediction model.

[0012] In a possible implementation manner, the initial network model is iteratively trained based on the mean square error and the mean absolute value error between the actual wind speed modulus of each sample terrain and the wind speed modulus in the corresponding predicted wind resource result map to determine the wind resource prediction model, including: Detecting whether the mean square error and the mean absolute error are less than or equal to a preset threshold; If any one of the above is no, then the hyperparameters and network parameters of the initial network model are adjusted, and the iterative training of the initial network model after the parameter adjustment is continued; If both are true, the initial network model is used as the wind resource prediction model.

[0013] The embodiment of the present application also provides a wind resource prediction device for meso-micro scale terrain, the wind resource prediction device comprising: An acquisition module is used to extract surface roughness data of the target terrain, and to perform fluid dynamics numerical simulation processing on the target terrain after grid division to determine the three-dimensional flow field data of the target terrain; A preprocessing module, used for performing data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix; A wind field barrier feature construction module is used to calculate the barrier coefficient of the fixed-dimensional terrain elevation matrix in the incoming flow direction to determine the wind field barrier feature of the target terrain; The wind resource prediction module is used to input the fixed-dimensional terrain elevation matrix, the wind field barrier features and the fixed-dimensional surface roughness matrix into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain; wherein, the wind resource prediction model is obtained by training the initial network model with the target height fixed-dimensional flow field matrix as the training target, and the initial network model is a UNet architecture convolutional neural network model with multiple Transformer layers added.

[0014] An embodiment of the present application also provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory through the bus, and when the machine-readable instructions are executed by the processor, the steps of the wind resource prediction method for meso- and micro-scale terrain are performed as described above.

[0015] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the above-mentioned method for predicting wind resources for meso- and micro-scale terrains are executed.

[0016] The embodiments of the present application provide a method, device, equipment and medium for predicting wind resources for meso-microscale terrain. The wind resource prediction method includes: extracting surface roughness data of the target terrain, and performing fluid dynamics numerical simulation processing on the target terrain after gridding to determine the three-dimensional flow field data of the target terrain; performing data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix; calculating the barrier coefficient in the incoming flow direction of the fixed-dimensional terrain elevation matrix to determine the wind field barrier characteristics of the target terrain; inputting the fixed-dimensional terrain elevation matrix, the wind field barrier characteristics and the fixed-dimensional surface roughness matrix into a wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain; wherein the wind resource prediction model is obtained by training an initial network model with a fixed-dimensional flow field matrix at a target height as a training target, and the initial network model is obtained by adding multiple Transformer layers to a UNet architecture convolutional neural network model. The beneficial effects of the present application are as follows: the wind resource prediction model can quickly and accurately predict the wind resource results of arbitrary terrain at meso- and micro-scales, while enhancing the adaptability and versatility of the model and reducing dependence on large amounts of historical data and high-performance hardware.

[0017] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 A flow chart of a method for predicting wind resources for meso-microscale terrain provided in an embodiment of the present application; Figure 2 A schematic diagram of the UI interface display of the wind resource result map provided in the embodiment of the present application; Figure 3 One of the structural schematic diagrams of a wind resource prediction device for meso-micro scale terrain provided in an embodiment of the present application; Figure 4 A second structural schematic diagram of a wind resource prediction device for meso-micro scale terrain provided in an embodiment of the present application; Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme 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 a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application usually described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, each other embodiment obtained by those skilled in the art without making creative work belongs to the scope of protection of the present application.

[0021] First, the application scenarios to which the present application is applicable are introduced. The present application can be applied to the field of wind resource prediction technology.

[0022] Research has found that current wind resource forecasting mainly relies on numerical weather forecasting models, statistical analysis methods, and traditional machine learning algorithms. Numerical weather forecasting models simulate and predict future wind speed and direction by solving atmospheric physics equations, combining terrain data and real-time weather reports. However, existing wind resource forecasting technologies have the following major defects in data collection, model performance, and production deployment: (1) Difficulty in data collection: The collection of wind resource data mainly relies on a limited number of measurement points that are discretely distributed in time and space. Although these measurement points can obtain wind resource information within a certain range, from the perspective of the overall wind field, they are only local samples and cannot fully and continuously cover the entire wind field space. (2) Insufficient model performance: Traditional methods such as numerical simulation methods and machine learning methods have problems such as low accuracy and slow computing speed. For example, numerical simulation requires a lot of computing resources and has a slow prediction speed; while machine learning methods lack accuracy, especially under complex terrain and irregular wind direction conditions, the prediction effect is limited. (3) Low accuracy of wind resource prediction. Therefore, how to improve the accuracy of wind resource prediction has become a technical issue that cannot be underestimated.

[0023] Based on this, an embodiment of the present application provides a method for predicting wind resources for meso- and micro-scale terrain. The wind resource prediction model can quickly and accurately predict the wind resource results for any meso- and micro-scale terrain, while enhancing the adaptability and versatility of the model and reducing dependence on large amounts of historical data and high-performance hardware.

[0024] See also Figure 1 , Figure 1 This is a flow chart of a method for predicting wind resources for meso-microscale terrain provided in an embodiment of the present application. Figure 1 As shown in, the wind resource prediction method provided by the embodiment of the present application includes: S101: extracting surface roughness data of a target terrain, and performing fluid dynamics numerical simulation processing on the target terrain after meshing, to determine three-dimensional flow field data of the target terrain.

[0025] In this step, the surface roughness data of the target terrain obtained from the terrain database is first extracted, and then the target terrain is meshed, and the target terrain after meshing is subjected to fluid dynamics numerical simulation to determine the three-dimensional flow field data of the target terrain.

[0026] Here, a rectangular area of ​​L×L size is cut out from the terrain database according to the given latitude and longitude coordinates, and then the area is meshed discretized at a specified resolution. Then, according to the set target flow and boundary conditions, the target terrain is subjected to fluid dynamics numerical simulation to determine the three-dimensional flow field data of the target terrain. In addition, the surface roughness data of the target terrain in the selected area is extracted and saved.

[0027] In a specific embodiment, according to the predetermined longitude and latitude coordinates, a rectangular area of ​​the target point with a size of L×L is cut from the terrain database, and the cut rectangular area is gridded at a resolution of 25m to obtain the target terrain. Grid data. For the target calculation area, set the boundary inflow conditions. Furthermore, set the inflow direction to easterly wind, and the inflow wind speed profile to the wind speed profile with a thermal stability of 5. Set the bottom surface of the calculation domain to a no-slip boundary, and the rest to free boundary conditions. According to the initial boundary conditions, the RANS model is used to perform numerical simulation on the target calculation domain. After the flow stabilizes, select the three-dimensional flow field results of the final fixed time step as the three-dimensional flow field data corresponding to the target area, and extract and output the roughness data of the target area.

[0028] S102: Preprocessing the grid coordinates of the target terrain, the three-dimensional flow field data, and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix.

[0029] In this step, data preprocessing is performed on the grid coordinates, three-dimensional flow field data and surface roughness data of the target terrain to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix.

[0030] Here, data preprocessing includes terrain data format conversion, data cleaning, and data normalization.

[0031] In a possible implementation manner, the data preprocessing of the grid coordinates of the target terrain, the three-dimensional flow field data, and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix includes: A: Linearly interpolate the underlying surface grid coordinates in the grid coordinates of the target terrain to a two-dimensional terrain matrix of fixed dimension, normalize the original terrain height of each terrain data point based on the lowest height in the two-dimensional terrain matrix, and determine the fixed-dimensional terrain elevation matrix.

[0032] Here, the underlying surface grid coordinates in the grid coordinates of the target terrain are linearly interpolated to a two-dimensional terrain matrix of fixed dimension, and the original terrain height of each terrain data point is normalized according to the lowest height in the two-dimensional terrain matrix to determine the fixed-dimensional terrain elevation matrix.

[0033] Among them, from the three-dimensional flow field data obtained by CFD calculation, the grid coordinate information of the underlying surface is first extracted as the initial terrain data. Then, the height of the original terrain data is normalized with the lowest point in the area as the benchmark to obtain unified relative height information. Subsequently, the processed terrain elevation data is converted into a matrix with fixed dimensions through linear interpolation to ensure that it can adapt to the input requirements of the model.

[0034] B: Based on the linear interpolation, the surface roughness data of each terrain data point is converted, outliers are removed, and maximum and minimum values ​​are normalized to determine the fixed-dimensional surface roughness matrix.

[0035] Here, the surface roughness data of each terrain data point is transformed, outliers are removed, and maximum and minimum values ​​are normalized according to linear interpolation to determine a fixed-dimensional surface roughness matrix.

[0036] Among them, the CFD calculation examples that have not converged are eliminated through outlier identification to ensure data quality. The maximum and minimum values ​​of terrain elevation data and surface roughness data are counted respectively, and the minimum value of the current terrain data is subtracted from each terrain data point to unify the relative height benchmark. The maximum and minimum values ​​of all data types (terrain elevation data and surface roughness data) are normalized, as shown in the following formula:

[0037] in, is the normalized value, is the original data value, is the minimum value of the data type. The maximum value for this data type.

[0038] Here, according to different wind direction conditions, the terrain data is relatively rotated with the wind direction as the negative direction of the X-axis as the reference, so that the rotated terrain uniformly takes the negative direction of the X-axis as the incoming flow direction, which is convenient for the subsequent use of a unified input form for the network model.

[0039] S103: Calculating the barrier coefficient of the fixed-dimensional terrain elevation matrix in the incoming flow direction to determine the wind field barrier characteristics of the target terrain.

[0040] In this step, the barrier coefficient in the incoming flow direction is calculated for the terrain elevation data in the fixed-dimensional terrain elevation matrix to determine the wind field barrier characteristics of the target terrain, that is, to numerically and figuratively express the interactive relationship between the terrain and the incoming flow.

[0041] In a possible implementation manner, calculating the barrier coefficient of the fixed-dimensional terrain elevation matrix in the incoming flow direction to determine the wind field barrier characteristics of the target terrain includes: a: For each terrain data point, take the terrain data point as the starting point and judge point by point along the upwind straight line to determine whether the terrain elevation data of the upwind terrain data point on the upwind straight line is greater than the terrain elevation data of the terrain data point.

[0042] Here, for each terrain data point, a point-by-point judgment is made along the upwind straight line starting from the terrain data point to determine whether the terrain elevation data of the upwind terrain data point on the upwind straight line is greater than the terrain elevation data of the terrain data point.

[0043] b: If so, the blocking coefficient of the upwind terrain data point to the terrain data point is determined based on the height difference between the terrain elevation data of the upwind terrain data point and the terrain elevation data of the terrain data point and the straight-line distance between the upwind terrain data point and the terrain data point.

[0044] Here, if yes, the blocking coefficient of the upwind terrain data point to the terrain data point is determined according to the height difference between the terrain elevation data of the upwind terrain data point and the terrain elevation data of the terrain data point and the straight-line distance between the upwind terrain data point and the terrain data point.

[0045] In a possible implementation manner, the blocking coefficient of the upwind terrain data point to the terrain data point is determined by the following steps:

[0046]

[0047] in, For terrain data points The barrier factor, is the upwind terrain data point on the upwind straight line, The value range is , z is the terrain elevation data, t is the height difference, L is the terrain length of the fixed-dimension terrain elevation data, is a symbolic function.

[0048] Here, L can be set to 256. The role is to judge when Greater than Calculate the barrier coefficient when .

[0049] The wind field barrier characteristics can be broken down into two parts: height difference and upwind straight-line distance. , height difference For example, the set of points on the upwind straight line is , and The closer the distance is, the greater the barrier coefficient value is, indicating that the straight line point against the wind For terrain data points The greater the flow barrier is, the greater the The gradual increase of the terrain data points As the distance increases, the barrier of the upwind straight line point to the terrain data point decreases. The height difference is a linear coefficient in the calculation formula of the incoming flow barrier coefficient, which means that the higher the height of the upwind straight line point is than the current data point, the stronger the incoming flow barrier is.

[0050] c: Accumulate the barrier coefficient of each terrain data point to determine the comprehensive barrier coefficient of each terrain data point.

[0051] Here, the barrier coefficient of each terrain data point is accumulated to determine the comprehensive barrier coefficient of each terrain data point.

[0052] c: performing maximum and minimum normalization processing on the multiple comprehensive barrier coefficients to determine the wind field barrier characteristics of the target terrain.

[0053] Here, the maximum and minimum values ​​of multiple comprehensive barrier coefficients are normalized to determine the wind field barrier characteristics of the target terrain.

[0054] S104: Input the fixed-dimensional terrain elevation matrix, the wind field barrier features and the fixed-dimensional surface roughness matrix into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain.

[0055] In this step, the fixed-dimensional terrain elevation matrix, wind field barrier features and fixed-dimensional surface roughness matrix are input into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain.

[0056] Among them, the wind resource result map of the target terrain shows the wind speed modulus of each terrain data point.

[0057] Among them, the wind resource prediction model is obtained by training the initial network model with a target height fixed-dimensional flow field matrix as the training target, and the initial network model is obtained by adding multiple Transformer layers to the UNet architecture convolutional neural network model.

[0058] Among them, the target height fixed-dimensional flow field matrix is ​​to parse and load the original three-dimensional flow field data file (.vtk format), from which the flow field information of the 100-meter altitude layer is extracted. The nearest neighbor interpolation algorithm is used to accurately obtain the wind speed component data (u, v, w) at this altitude layer from the three-dimensional flow field. Subsequently, based on the wind speed component data, the wind speed modulus U is calculated by vector synthesis. Finally, the wind speed modulus data in the original coordinate system is resampled into a fixed-dimensional matrix using linear interpolation to ensure its compatibility with the model input format.

[0059] Among them, in this application, in addition to using the UNet architecture convolutional neural network model to construct the initial network model, it can also be constructed according to other variant architectures such as TransUNet, and this part is not specifically limited.

[0060] In a possible implementation manner, the fixed-dimensional terrain elevation matrix, the wind field barrier feature, and the fixed-dimensional surface roughness matrix are input into a wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling, and feature fusion to predict the wind resource result map of the target terrain, including: (1): Based on the multiple coding layers of the wind resource prediction model, the fixed-dimensional terrain elevation matrix, the wind field barrier features and the fixed-dimensional surface roughness matrix are sequentially convolved and down-sampled to extract features, and coding feature maps of different scales are determined.

[0061] Here, the encoding layer uses convolution kernels to extract features from wind field barrier features, fixed-dimensional terrain elevation matrix, and fixed-dimensional surface roughness matrix. During the encoding process, the size of the feature map gradually decreases, while the number of channels gradually increases, and the downsampling operation is performed after each encoding layer to halve the spatial size of the feature map and increase the number of channels, thereby obtaining encoded feature maps of different scales.

[0062] (2): Based on multiple Transformer layers, the deep coding feature map output by the deepest coding layer is subjected to nonlinear transformation and self-attention mechanism processing to extract implicit features of the deep coding feature map containing local information and global information.

[0063] Here, the Transformer layer is different from the classic UNet. The model proposed in this scheme introduces multiple Transformer layers in the deep encoding stage, so that the input image can be converted and serialized, and then the implicit features containing local and global information in the deep encoding feature map are extracted through multiple Transformer layers. These features are then reshaped to match the size of the deepest encoding layer.

[0064] (3): Based on the multiple decoding layers of the wind resource prediction model, the implicit features and the multiple encoded feature maps are subjected to convolution up-sampling processing, and during the decoding process, the decoded feature map output by the decoding layer and the corresponding encoded feature map are subjected to feature fusion by means of jump connection.

[0065] Here, convolution up-sampling is performed on implicit features and multiple encoded feature maps according to multiple decoding layers of the wind resource prediction model, and feature fusion is performed on the decoded feature map output by the decoding layer and the corresponding encoded feature map through jump connections during the decoding process.

[0066] In the decoding stage, the model gradually restores the spatial resolution of the feature map through the transposed convolution layer while reducing the number of channels. This process reconstructs the spatial details of the image and fuses feature maps of different scales to enhance the model's ability to recognize complex structures. The model uses a full-scale jump connection method, which allows the feature maps of the encoding layer and the decoding layer to be combined during the decoding process to achieve full-scale feature fusion.

[0067] (4): The convolution layer of the wind resource prediction model performs convolution processing on the decoding feature map output by the last decoding layer to generate the wind resource result map; wherein the wind resource result map displays the wind speed modulus at each terrain data point.

[0068] Here, the convolution layer performs convolution processing on the decoded feature map output by the last decoding layer to generate a wind resource result map.

[0069] Among them, the role and functions of the wind resource prediction model are as follows: interactive interface: provides a simple and efficient UI, and you can get the wind resource prediction results by clicking; data output: supports the export of prediction results, and provides multiple formats such as npy and vtk to facilitate subsequent analysis and application; real-time prediction: combines the latest meteorological and terrain data, dynamically updates the prediction results, and ensures the timeliness of the output.

[0070] For further information, see Figure 2 , Figure 2 This is a schematic diagram of the UI interface display of the wind resource result map provided in the embodiment of the present application. Figure 2 As shown in the output interface of wind resource forecast, the wind speed modulus of the target terrain at any terrain data point is displayed. Figure 2 The color streamlines shown in the figure represent different colors corresponding to different wind speed moduli, so that users can easily distinguish the distribution of wind resource concentration under the target terrain.

[0071] In a specific embodiment, according to the wind field data acquisition module, a rectangular area of ​​L×L size is cut out from the terrain database according to the given longitude and latitude coordinates, and then the area is grid discretized at a specified resolution. Then, according to the set target incoming flow and boundary conditions, the target terrain is numerically simulated by CFD to obtain the three-dimensional flow field of the target terrain, and the elevation data and roughness data of the selected area are extracted and saved. The wind field data acquisition module is connected to the external data source through an interface, and the collected surface roughness data and three-dimensional flow field data are transmitted to the wind field data processing module for processing to obtain a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix. The fixed-dimensional terrain elevation matrix is ​​sent to the wind field barrier feature construction module to calculate the barrier coefficient of the terrain elevation data in the fixed-dimensional flow field matrix in the incoming flow direction, and determine the wind field barrier characteristics of the target terrain. The wind field barrier characteristics, the fixed-dimensional terrain elevation matrix, and the fixed-dimensional surface roughness matrix are input into the wind resource prediction model to predict the wind resource result map of the target terrain.

[0072] In a possible implementation manner, the wind resource prediction model is determined by the following steps: I: Inputting the sample fixed-dimensional terrain elevation matrix, sample wind field barrier characteristics and sample fixed-dimensional surface roughness matrix of multiple sample terrains into the initial network model to determine the predicted wind resource result map of each sample terrain.

[0073] Here, the sample wind field barrier characteristics of multiple sample terrains, the sample fixed-dimensional terrain elevation matrix and the sample fixed-dimensional surface roughness matrix are input into the initial network model to determine the predicted wind resource result map of each sample terrain.

[0074] Among them, the processing process of the predicted wind resource result map of the sample terrain is consistent with that of the wind resource result map of the target terrain, which will not be repeated in this part.

[0075] II: Based on the linear interpolation algorithm, the sample three-dimensional flow field data of each sample terrain is converted, outliers are removed, and maximum and minimum values ​​are processed. The wind speed component of each sample terrain data point at the height of the ground target is extracted from the processed sample three-dimensional flow field data.

[0076] Here, data processing is performed on the sample three-dimensional flow field data of each sample terrain, and the wind speed component of each sample terrain data point at a height above the ground target is extracted from the sample three-dimensional flow field data after data processing.

[0077] III: Calculate the actual wind speed modulus of each sample terrain based on the wind speed component of each sample terrain data point.

[0078] Here, the actual wind speed modulus of each sample terrain is calculated according to the wind speed component of each sample terrain data point.

[0079] IV: The initial network model is iteratively trained based on the mean square error and the mean absolute value error between the actual wind speed modulus of each sample terrain and the wind speed modulus in the corresponding predicted wind resource result map to determine the wind resource prediction model.

[0080] Here, the initial network model is iteratively trained according to the mean square error and mean absolute value error between the actual wind speed modulus of each sample terrain and the wind speed modulus in the corresponding predicted wind resource result map to determine the wind resource prediction model.

[0081] In a possible implementation manner, the initial network model is iteratively trained based on the mean square error and the mean absolute value error between the actual wind speed modulus of each sample terrain and the wind speed modulus in the corresponding predicted wind resource result map to determine the wind resource prediction model, including: i: Detect whether the mean square error and the mean absolute error are less than or equal to a preset threshold.

[0082] Here, it is detected whether the mean square error and the mean absolute value error are less than or equal to a preset threshold.

[0083] ii: If any one of the above is no, then the hyperparameters and network parameters of the initial network model are adjusted, and the iterative training of the initial network model after the parameter adjustment is continued.

[0084] Here, if any one of them is no, the hyperparameters and network parameters of the initial network model are adjusted, and the iterative training of the initial network model after parameter adjustment is continued.

[0085] iii: If both are true, the initial network model is used as the wind resource prediction model.

[0086] In a specific embodiment, terrain data and CFD wind resource simulation data are obtained, and data format conversion and normalization algorithms are configured for the obtained terrain data and CFD wind resource simulation data to ensure data quality and build a data storage and management system that supports large-scale data processing. A wind field barrier feature extraction algorithm is configured to expand the data feature dimension. Terrain height, wind field barrier and roughness features are integrated. Based on the UNet network, a new initial network model is designed, and the Transformer layer is embedded between deep encoders to improve the model feature learning ability. Prepare sufficient historical data for model training and verification, and configure high-performance computing devices such as GPUs to accelerate training. Use techniques such as mean square error, mean absolute error and hyperparameter tuning to optimize the model generalization performance and prediction accuracy, establish a regular model update mechanism, and adapt to data and environmental changes to obtain the final wind resource prediction model.

[0087] The wind resource prediction method for meso- and micro-scale terrain implemented in this application has the following specific advantages: using advanced CFD technology, selecting nationally representative terrain data, obtaining high-precision simulation results, and significantly improving data quality. Innovative integration of UNet and Transformer architectures to achieve second-level response and meet real-time prediction needs. Maintain high accuracy in complex terrain and variable wind directions. The wind resource prediction model has powerful feature extraction capabilities, excellent generalization performance, adaptability to different regions and wind resource conditions, high robustness, ensuring that the prediction results are stable and reliable, and the predicted wind resource result map can be intuitively displayed on the UI interface.

[0088] The embodiment of the present application provides a method for predicting wind resources for meso-microscale terrain, the method comprising: extracting surface roughness data of the target terrain, and performing fluid dynamics numerical simulation processing on the target terrain after gridding to determine the three-dimensional flow field data of the target terrain; performing data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix; calculating the barrier coefficient in the incoming flow direction of the fixed-dimensional terrain elevation matrix to determine the wind field barrier characteristics of the target terrain; inputting the fixed-dimensional terrain elevation matrix, the wind field barrier characteristics and the fixed-dimensional surface roughness matrix into a wind resource prediction model to perform convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain; wherein the wind resource prediction model is obtained by training an initial network model with a fixed-dimensional flow field matrix of target height as a training target, and the initial network model is obtained by adding multiple Transformer layers to a UNet architecture convolutional neural network model. The wind resource prediction model can quickly and accurately predict the wind resource results of any meso- and micro-scale terrain, while enhancing the adaptability and versatility of the model and reducing dependence on large amounts of historical data and high-performance hardware.

[0089] See also Figure 3 , Figure 4 , Figure 3 One of the structural schematic diagrams of a wind resource prediction device for meso-micro scale terrain provided in an embodiment of the present application; Figure 4 This is a second structural diagram of a wind resource prediction device for meso-micro scale terrain provided in an embodiment of the present application. Figure 3 As shown in , the wind resource prediction device 300 for meso-micro scale terrain includes: The acquisition module 310 is used to extract surface roughness data of the target terrain, and perform fluid dynamics numerical simulation processing on the target terrain after grid division to determine the three-dimensional flow field data of the target terrain; A preprocessing module 320 is used to perform data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix; A wind field barrier feature construction module 330 is used to calculate the barrier coefficient in the incoming flow direction of the fixed-dimensional terrain elevation matrix to determine the wind field barrier feature of the target terrain; The wind resource prediction module 340 is used to input the fixed-dimensional terrain elevation matrix, the wind field barrier features and the fixed-dimensional surface roughness matrix into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain; wherein the wind resource prediction model is obtained by training the initial network model with the target height fixed-dimensional flow field matrix as the training target, and the initial network model is a UNet architecture convolutional neural network model with multiple Transformer layers added.

[0090] Further, when the preprocessing module 320 is used to perform data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data, and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix, the preprocessing module 320 is specifically used to: Linearly interpolating the underlying surface grid coordinates in the grid coordinates of the target terrain into a two-dimensional terrain matrix of fixed dimension, normalizing the original terrain height of each terrain data point based on the lowest height in the two-dimensional terrain matrix, and determining the fixed-dimensional terrain elevation matrix; Based on the linear interpolation, the surface roughness data of each terrain data point is converted, outliers are removed, and maximum and minimum values ​​are normalized to determine the fixed-dimensional surface roughness matrix.

[0091] Further, when the wind field barrier feature construction module 330 is used to calculate the barrier coefficient in the incoming flow direction of the fixed-dimensional terrain elevation matrix to determine the wind field barrier feature of the target terrain, the wind field barrier feature construction module 330 is specifically used to: For each terrain data point, starting from the terrain data point, judging point by point along the upwind straight line, to determine whether the terrain elevation data of the upwind terrain data point on the upwind straight line is greater than the terrain elevation data of the terrain data point; If so, determining the blocking coefficient of the upwind terrain data point to the terrain data point based on the height difference between the terrain elevation data of the upwind terrain data point and the terrain elevation data of the terrain data point and the straight-line distance between the upwind terrain data point and the terrain data point; Accumulating the barrier coefficient of each terrain data point to determine the comprehensive barrier coefficient of each terrain data point; The maximum and minimum values ​​of the plurality of comprehensive blocking coefficients are normalized to determine the wind field blocking characteristics of the target terrain.

[0092] Furthermore, the wind field blocking feature construction module 330 determines the blocking coefficient of the upwind terrain data point to the terrain data point through the following steps:

[0093]

[0094] in, For terrain data points The barrier factor, is the upwind terrain data point on the upwind straight line, The value range is , z is the terrain elevation data, t is the height difference, L is the terrain length of the fixed-dimension terrain elevation data, is a symbolic function.

[0095] Further, when the wind resource prediction module 340 is used to input the fixed-dimensional terrain elevation matrix, the wind field barrier feature, and the fixed-dimensional surface roughness matrix into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling, and feature fusion to predict the wind resource result map of the target terrain, the wind resource prediction module 340 is specifically used to: Based on the multiple coding layers of the wind resource prediction model, convolution downsampling feature extraction is performed on the fixed-dimensional terrain elevation matrix, the wind field barrier feature and the fixed-dimensional surface roughness matrix in turn to determine coding feature maps of different scales; Based on multiple Transformer layers, a nonlinear transformation and a self-attention mechanism are performed on the deep coding feature map output by the deepest coding layer to extract implicit features of the deep coding feature map containing local information and global information; Based on the multiple decoding layers of the wind resource prediction model, the implicit features and the multiple coded feature maps are subjected to convolution up-sampling processing, and in the decoding process, the decoded feature map output by the decoding layer and the corresponding coded feature map are subjected to feature fusion by means of skip connection; The convolution layer based on the wind resource prediction model performs convolution processing on the decoding feature map output by the last decoding layer to generate the wind resource result map; wherein the wind resource result map displays the wind speed modulus at each terrain data point.

[0096] Further, such as Figure 4 As shown, the wind resource prediction device 300 for meso-micro scale terrain further includes a model training module 350. The model training module 350 determines the wind resource prediction model through the following steps: Inputting a sample fixed-dimensional terrain elevation matrix, a sample wind field barrier feature, and a sample fixed-dimensional surface roughness matrix of a plurality of sample terrains into an initial network model to determine a predicted wind resource result map for each of the sample terrains; Based on the linear interpolation algorithm, the sample three-dimensional flow field data of each sample terrain is converted, outliers are removed, and maximum and minimum values ​​are processed, and the wind speed component of each sample terrain data point at the height of the ground target is extracted from the processed sample three-dimensional flow field data; The actual wind speed modulus of each sample terrain at the target height is calculated based on the wind speed component of each sample terrain data point; The initial network model is iteratively trained based on the mean square error and the mean absolute value error between the actual wind speed modulus of each sample terrain and the wind speed modulus in the corresponding predicted wind resource result map to determine the wind resource prediction model.

[0097] Further, when the model training module 350 is used to iteratively train the initial network model based on the mean square error and the mean absolute error between the actual wind speed modulus based on each sample terrain and the wind speed modulus in the corresponding predicted wind resource result map to determine the wind resource prediction model, the model training module 350 is specifically used to: Detecting whether the mean square error and the mean absolute error are less than or equal to a preset threshold; If any one of the above is no, then the hyperparameters and network parameters of the initial network model are adjusted, and the iterative training of the initial network model after the parameter adjustment is continued; If both are true, the initial network model is used as the wind resource prediction model.

[0098] The embodiment of the present application provides a wind resource prediction device for meso-micro scale terrain, and the wind resource prediction device includes: an acquisition module, which is used to extract surface roughness data of the target terrain, and perform fluid dynamics numerical simulation processing on the target terrain after grid division to determine the three-dimensional flow field data of the target terrain; a preprocessing module, which is used to perform data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix; a wind field barrier feature construction module, which is used to perform a barrier coefficient on the fixed-dimensional terrain elevation matrix in the incoming flow direction. Calculate and determine the wind field barrier characteristics of the target terrain; the wind resource prediction module is used to input the fixed-dimensional terrain elevation matrix, the wind field barrier characteristics and the fixed-dimensional surface roughness matrix into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain; wherein the wind resource prediction model is obtained by training the initial network model with the target height fixed-dimensional flow field matrix as the training target, and the initial network model is obtained by adding multiple Transformer layers to the UNet architecture convolutional neural network model. The wind resource prediction model can quickly and accurately predict the wind resource results of any terrain at the micro-scale, while enhancing the adaptability and versatility of the model and reducing the dependence on large amounts of historical data and high-performance hardware.

[0099] See also Figure 5 , Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 5 As shown in , the electronic device 500 includes a processor 510 , a memory 520 and a bus 530 .

[0100] The memory 520 stores machine-readable instructions executable by the processor 510. When the electronic device 500 is running, the processor 510 communicates with the memory 520 via the bus 530. When the machine-readable instructions are executed by the processor 510, the above-mentioned Figure 1 The steps of the method for predicting wind resources for meso- and micro-scale terrain in the method embodiment shown in the figure, and the specific implementation methods can be found in the method embodiment, which will not be repeated here.

[0101] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program can execute the above-mentioned Figure 1 The steps of the method for predicting wind resources for meso- and micro-scale terrain in the method embodiment shown in the figure, and the specific implementation methods can be found in the method embodiment, which will not be repeated here.

[0102] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0103] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0104] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0105] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0106] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application can essentially or partly contribute to the prior art or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0107] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for predicting wind resources in meso- and micro-scale terrain, characterized in that: The wind resource prediction method comprises: Extracting surface roughness data of the target terrain, and performing fluid dynamics numerical simulation processing on the target terrain after meshing to determine three-dimensional flow field data of the target terrain; Performing data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data, and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix; Calculating the barrier coefficient of the fixed-dimensional terrain elevation matrix in the incoming flow direction to determine the wind field barrier characteristics of the target terrain; The fixed-dimensional terrain elevation matrix, the wind field barrier features and the fixed-dimensional surface roughness matrix are input into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain; wherein the wind resource prediction model is obtained by training the initial network model with the target height fixed-dimensional flow field matrix as the training target, and the initial network model is obtained by adding multiple Transformer layers to the UNet architecture convolutional neural network model.

2. The wind resource prediction method according to claim 1, characterized in that: The data preprocessing of the grid coordinates of the target terrain, the three-dimensional flow field data and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix includes: Linearly interpolating the underlying surface grid coordinates in the grid coordinates of the target terrain into a two-dimensional terrain matrix of fixed dimension, normalizing the original terrain height of each terrain data point based on the lowest height in the two-dimensional terrain matrix, and determining the fixed-dimensional terrain elevation matrix; Based on the linear interpolation, the surface roughness data of each terrain data point is converted, outliers are removed, and maximum and minimum values ​​are normalized to determine the fixed-dimensional surface roughness matrix.

3. The wind resource prediction method according to claim 1, characterized in that: The calculating of the barrier coefficient of the fixed-dimensional terrain elevation matrix in the incoming flow direction to determine the wind field barrier characteristics of the target terrain includes: For each terrain data point, starting from the terrain data point, judging point by point along the upwind straight line, to determine whether the terrain elevation data of the upwind terrain data point on the upwind straight line is greater than the terrain elevation data of the terrain data point; If so, determining the blocking coefficient of the upwind terrain data point to the terrain data point based on the height difference between the terrain elevation data of the upwind terrain data point and the terrain elevation data of the terrain data point and the straight-line distance between the upwind terrain data point and the terrain data point; Accumulating the barrier coefficient of each terrain data point to determine the comprehensive barrier coefficient of each terrain data point; The maximum and minimum values ​​of the plurality of comprehensive blocking coefficients are normalized to determine the wind field blocking characteristics of the target terrain.

4. The wind resource prediction method according to claim 3, characterized in that: Determine the blocking coefficient of the upwind terrain data point to the terrain data point through the following steps: in, For terrain data points The barrier factor, is the upwind terrain data point on the upwind straight line, The value range is , z is the terrain elevation data, t is the height difference, L is the terrain length of the fixed-dimension terrain elevation data, is a symbolic function.

5. The wind resource prediction method according to claim 1, characterized in that: The fixed-dimensional terrain elevation matrix, the wind field barrier features, and the fixed-dimensional surface roughness matrix are input into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling, and feature fusion to predict the wind resource result map of the target terrain, including: Based on the multiple coding layers of the wind resource prediction model, convolution downsampling feature extraction is performed on the fixed-dimensional terrain elevation matrix, the wind field barrier feature and the fixed-dimensional surface roughness matrix in turn to determine coding feature maps of different scales; Based on multiple Transformer layers, a nonlinear transformation and a self-attention mechanism are performed on the deep coding feature map output by the deepest coding layer to extract implicit features of the deep coding feature map containing local information and global information; Based on the multiple decoding layers of the wind resource prediction model, the implicit features and the multiple coded feature maps are subjected to convolution up-sampling processing, and in the decoding process, the decoded feature map output by the decoding layer and the corresponding coded feature map are subjected to feature fusion by means of skip connection; The convolution layer based on the wind resource prediction model performs convolution processing on the decoding feature map output by the last decoding layer to generate the wind resource result map; wherein the wind resource result map displays the wind speed modulus at each terrain data point.

6. The wind resource prediction method according to claim 1, characterized in that: The wind resource prediction model is determined by the following steps: Inputting a sample fixed-dimensional terrain elevation matrix, a sample wind field barrier feature, and a sample fixed-dimensional surface roughness matrix of a plurality of sample terrains into an initial network model to determine a predicted wind resource result map for each of the sample terrains; Based on the linear interpolation algorithm, the sample three-dimensional flow field data of each sample terrain is converted, outliers are removed, and maximum and minimum values ​​are processed, and the wind speed component of each sample terrain data point at the height of the ground target is extracted from the processed sample three-dimensional flow field data; The actual wind speed modulus of each sample terrain at the target height is calculated based on the wind speed component of each sample terrain data point; The initial network model is iteratively trained based on the mean square error and the mean absolute value error between the actual wind speed modulus of each sample terrain and the wind speed modulus in the corresponding predicted wind resource result map to determine the wind resource prediction model.

7. The wind resource prediction method according to claim 6, characterized in that: The initial network model is iteratively trained based on the mean square error and the mean absolute value error between the actual wind speed modulus of each sample terrain and the wind speed modulus in the corresponding predicted wind resource result map to determine the wind resource prediction model, including: Detecting whether the mean square error and the mean absolute error are less than or equal to a preset threshold; If any one of the above is no, then the hyperparameters and network parameters of the initial network model are adjusted, and the iterative training of the initial network model after the parameter adjustment is continued; If both are true, the initial network model is used as the wind resource prediction model.

8. A wind resource prediction device for meso-micro scale terrain, characterized in that: The wind resource prediction device comprises: An acquisition module is used to extract surface roughness data of the target terrain, and to perform fluid dynamics numerical simulation processing on the target terrain after grid division to determine the three-dimensional flow field data of the target terrain; A preprocessing module, used for performing data preprocessing on the grid coordinates of the target terrain, the three-dimensional flow field data and the surface roughness data to determine a fixed-dimensional terrain elevation matrix and a fixed-dimensional surface roughness matrix; A wind field barrier feature construction module is used to calculate the barrier coefficient of the fixed-dimensional terrain elevation matrix in the incoming flow direction to determine the wind field barrier feature of the target terrain; The wind resource prediction module is used to input the fixed-dimensional terrain elevation matrix, the wind field barrier features and the fixed-dimensional surface roughness matrix into the wind resource prediction model for convolution downsampling, nonlinear transformation, self-attention mechanism, convolution upsampling and feature fusion to predict the wind resource result map of the target terrain; wherein, the wind resource prediction model is obtained by training the initial network model with the target height fixed-dimensional flow field matrix as the training target, and the initial network model is a UNet architecture convolutional neural network model with multiple Transformer layers added.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the method for predicting wind resources for meso-micro-scale terrain as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for predicting wind resources for meso-micro-scale terrains as described in any one of claims 1 to 7 are executed.

Images