High-frequency short-term wind speed prediction method for complex terrain

Abstract

The invention discloses a high-frequency short-term wind speed prediction method for a complex terrain, and the method comprises the steps: building a high-frequency short-term wind speed prediction model which comprises a CNN layer and an LSTM layer based on the collection of numerical weather prediction data of different isobaric surfaces; the CNN layer is used for capturing common spatial correlation of wind speed, wind direction and related data and generating a feature map; the two LSTM layers extract time sequence dynamic features from the wind speed feature map and the wind direction feature map respectively; and after the full connection layer integrates the features, the output layer gives a wind speed prediction result. And training the model by using a training set, evaluating and adjusting parameters through a verification set, and optimizing by using a joint loss function. The high-frequency short-term wind speed prediction model shows more excellent performance in high-frequency short-term wind speed prediction, the space-time limitation of a physical model is solved, the dependence on historical data is reduced, more reliable and accurate short-term wind speed prediction in complex terrains is realized, and the high-frequency short-term wind speed prediction model has important significance in wind energy management, power grid stability and energy production optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of wind speed prediction, and in particular to a method for short-term wind speed prediction in complex terrain based on spatiotemporal multi-task joint optimization. Background Art

[0002] Wind power is a crucial component of the global energy transition. However, its inherent volatility and intermittency pose significant challenges to its integration into power systems. Since wind power is proportional to the cube of wind speed, even small fluctuations in wind speed can result in significant changes in wind power. Therefore, accurate wind speed forecasting is crucial for effective wind energy management, ensuring grid stability, optimizing energy production, and reducing operating costs. Traditional wind speed forecasting methods are primarily categorized into physical models and data-driven approaches. Physical models, particularly those based on numerical weather prediction (NWP), use real-time meteorological and geographic data to mathematically simulate the atmosphere and predict wind speed. NWP models perform well in medium- and long-term wind speed forecasting. However, the extensive and complex computational requirements of these models result in low temporal and spatial resolution, making them less suitable for short-term and high-frequency forecasting. Data-driven models, including statistical and machine learning methods, rely on historical wind speed information to generate correlation-based forecasts. Traditional statistical models, such as the autoregressive integrated moving average (ARIMA) model and its extensions, have been widely used, but due to their inherent linear assumptions, they often struggle to capture the nonlinear dynamics of wind speed. Summary of the Invention

[0003] The present invention provides a method for short-term wind speed prediction in complex terrain based on spatiotemporal multi-task joint optimization, which can solve the above problems.

[0004] In order to solve the above problems, the technical solutions adopted by the present invention are as follows: A high-frequency short-term wind speed prediction method for complex terrain includes the following steps: (1) Collect numerical weather forecast data of different isobaric surfaces, perform the first preprocessing to form model forecast input data, merge the model forecast input data with the data of the automatic observation station equipment, perform the second preprocessing, and then divide the data into training set and validation set; (2) Construct a high-frequency short-term wind speed prediction model, which includes an input layer, a shared layer, an independent layer, a fully connected layer, and an output layer; the shared layer is a CNN layer, which is used to capture the common spatial correlation between wind speed, wind direction, and numerical weather forecast data of different isobaric surfaces, and obtain wind speed-related spatial feature maps and wind direction-related spatial feature maps; there are two independent layers, both of which are LSTM layers, which are used to capture the temporal dynamic characteristics of wind speed from the wind speed-related spatial feature maps, and to capture the temporal dynamic characteristics of wind direction from the wind direction-related spatial feature maps; the fully connected layer contains two hidden layers, which are used to obtain the integrated wind speed characteristics based on the temporal dynamic characteristics of wind speed, and to obtain the integrated wind direction characteristics based on the temporal dynamic characteristics of wind direction; the output layer is used to obtain the wind speed prediction result based on the integrated wind speed characteristics and wind direction characteristics; (3) Based on the multi-forecast task joint optimization method, the high-frequency short-term wind speed prediction model is trained using the training set. During the training process, the high-frequency short-term wind speed prediction model is evaluated using the validation set and the model parameters are adjusted. The loss function used in the training process is a weighted loss function of the combined wind speed loss function and the wind direction loss function. (4) Use the trained high-frequency short-term wind speed prediction model to predict the future wind speed of the target site.

[0005] As a further description of the above technical solution, in step (1), the numerical weather forecast data include the average wind speed, zonal wind component, meridional wind component, and average wind speed at the isobaric surfaces of 850 hPa, 700 hPa, 600 hPa, 500 hPa, 400 hPa, 300 hPa, 250 hPa, 200 hPa, and 100 hPa, respectively; and the average wind speed at a height of 10 meters at a monitoring station; each numerical weather forecast data is integrated into a unified 15-minute interval data set at different time resolutions.

[0006] As a further description of the above technical solution, in step (1), the method for collecting numerical weather forecast data of different isobaric surfaces includes: inputting the coordinates of the wind speed observation equipment, finding the coordinates near 3 km of the numerical model weather forecast as the grid center, and finding the grid within the coordinates and the nearby adjacent grids, and then extracting the numerical weather forecast data of different isobaric surfaces.

[0007] As a further description of the above technical solution, in step (1), the first preprocessing includes missing and outlier processing, as well as normalization of variables; the second preprocessing is to eliminate model weather forecast data with a value less than a threshold of 0.2 through the Pearson correlation coefficient.

[0008] As a further description of the above technical solution, step (4) also includes applying inverse normalization processing to the wind speed prediction result.

[0009] As a further description of the above technical solution, in step (2), the high-frequency short-term wind speed prediction model framework is determined based on grid search.

[0010] As a further description of the above technical solution, the CNN layer consists of two hidden layers. Each hidden layer of the CNN layer uses 16 channels and the convolution kernel size is 3×3; each hidden layer of the fully connected layer has 100 neurons.

[0011] As a further description of the above technical solution, during the training process of the high-frequency short-term wind speed prediction model, the Dropout rate is set to 0.25; the weights of the wind speed loss function and the wind direction loss function are 0.7 and 0.3, respectively. During the training process of the high-frequency short-term wind speed prediction model, a polynomial attenuation strategy is adopted based on the Adam optimizer to adjust the learning rate, with an initial learning rate of 0.001 and a weight attenuation parameter of 0.1; the high-frequency short-term wind speed prediction model is used for 6-hour prediction tasks, 12-hour prediction tasks, and 24-hour prediction tasks. The batch size of the 6-hour prediction task is 12, the 12-hour prediction task is 32, and the 24-hour prediction task is 64; the entire training of the high-frequency short-term wind speed prediction model is set to 500 cycles, and an early stopping mechanism is introduced to avoid model overfitting.

[0012] As a further description of the above technical solution, in step (3), the server configuration used for training the high-frequency short-term wind speed prediction model is: Intel(R) Xeon(R) Gold 5218 CPU @ 2.30GHz CPU, Quadro RTX 6000GPU, Ubuntu 16.04.6 LTS operating system.

[0013] As a further description of the above technical solution, in step (3), the model evaluation indicators used include mean absolute error and root mean square error.

[0014] Compared with the prior art, the present invention has the following beneficial effects: (1) This paper proposes an innovative model framework for high-frequency, short-term wind speed forecasting in complex terrain, effectively addressing the limitations of traditional wind speed forecasting methods in short-term and high-frequency forecasting. By combining CNN layers and LSTM layers, the model can simultaneously capture the spatial correlation and temporal dynamic characteristics between wind speed, wind direction, and numerical weather forecast data at different isobaric surfaces, thereby achieving more accurate wind speed forecasting.

[0015] (2) The high-frequency short-term wind speed prediction model WindNet in the present invention is trained by combining the weighted loss function of the wind speed loss function and the wind direction loss function, and the model parameters are further optimized. This enables WindNet to achieve more stable and accurate short-term wind speed prediction in complex terrain, providing strong support for power grid stability, energy production and reduction of operating costs, and has significant practical value and application prospects.

[0016] (3) The WindNet model of the present invention demonstrates excellent performance under conditions of complex terrain and extreme wind speed variations. Compared with benchmark models such as the traditional NWP model, LSTM model, and CNN-LSTM model, WindNet demonstrates higher accuracy in 6-hour, 12-hour, and 24-hour forecasts. This not only overcomes the shortcomings of physical models in terms of temporal and spatial resolution, but also reduces dependence on historical data, providing more reliable forecast results for wind energy management.

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more clearly understood, embodiments of the present invention are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0019] Figure 1 is a flow chart of a method for high-frequency short-term wind speed prediction in complex terrain according to an embodiment of the present invention; Figure 2 It is a framework diagram of the high-frequency short-term wind speed prediction model described in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0021] Please refer to Figure 1 and Figure 2 , an embodiment of the present invention provides a method for high-frequency short-term wind speed prediction in complex terrain, comprising the following steps: (1) Input the coordinates of the wind speed observation equipment, find the coordinates of the numerical model weather forecast near 3 km as the grid center, find the grid within the coordinates and the adjacent grids nearby, extract and collect the numerical weather forecast data of different isobaric surfaces, and then perform the first preprocessing to form the model forecast input data; merge the model forecast input data with the data of the observation automatic station equipment, and then perform the second preprocessing, and divide it into training set, validation set and test set.

[0022] In this embodiment, numerical weather prediction data were collected from three automatic weather monitoring stations located near the wind farm between January 1 and July 31, 2024. Stations S1 and S2 are located in the lower reaches of the Jinsha River in Sichuan Province, with Station S1 located in a mountainous area and Station S2 located in a canyon terrain. These stations are located in the dry and hot Jinsha River Valley, an area with frequent localized convective weather and high temperatures. Station S3 is located in plain terrain near Panzhihua City, Sichuan Province, with a subtropical monsoon climate. All automatic weather monitoring stations are managed by the China National Atmospheric Monitoring Center. Raw wind speed (m / s) and wind direction (degrees) data were recorded at 15-minute intervals. The numerical weather prediction (NWP) data used in this invention were generated by the Weather Research and Forecasting (WRF) model (version 3.9.1) with a horizontal resolution of 3 km × 3 km and a temporal resolution of 1 hour. NWP forecasts are provided every 24 hours (i.e., NWP predicts the weather conditions for the next 24 hours). The forecast content includes the average wind speed (m / s), the zonal wind component U and the meridional wind component V at different isobaric surfaces (850hPa, 700hPa, 600hPa, 500hPa, 400hPa, 300hPa, 250hPa, 200hPa, 100hPa), as well as the average wind speed (m / s) at a height of 10 meters (i.e. the height of the monitoring station).

[0023] In this embodiment, the method for performing a first preprocessing on the collected numerical weather forecast data includes: 1) Missing and outlier processing; 2) In addition, due to the differences in magnitude and units between different meteorological variables, the present invention also uses the maximum and minimum normalization method to normalize the variables during preprocessing, thereby ensuring that all training features have the same weight and accelerating model training. The specific formula is as follows: , Among them, x min and x max Respectively represent the minimum and maximum values ​​of each variable in the sample data. x represents a sample in the original data, x n represents the normalized sample. After normalization, the values ​​of each meteorological variable are all within the consistent range of [0, 1].

[0024] The data after normalization processing is the model forecast input data (it should be noted that since normalization processing is performed here, after the wind speed forecast results are obtained using the high-frequency short-term wind speed forecast model, inverse normalization processing needs to be applied to the wind speed forecast results to ensure that the final wind speed forecast results have physical meaning).

[0025] In this embodiment, in order to improve the temporal resolution of the data, the present invention synthesizes the model forecast input data and the data of the automatic observation station equipment at a temporal resolution of 15 minutes.

[0026] The synthesized data were preprocessed for the second time, that is, the model weather forecast data with a value less than the threshold of 0.2 were eliminated through the Pearson correlation coefficient. The final numerical weather forecast dataset contained 20,352 records.

[0027] After the second preprocessing, the resulting numerical weather forecast dataset was used to generate samples using a sliding window approach, sliding forward one step (15 minutes) at a time to divide the dataset into training, validation, and test sets. Specifically, the first five months of data (January 1 to May 31) were used for training, the sixth month (June 1 to June 30) for validation, and the final month (July 1 to July 31) for testing.

[0028] (2) Construct a high-frequency short-term wind speed prediction model, which includes input layer, shared layer, independent layer, fully connected layer and output layer.

[0029] In this embodiment, CNN-LSTM is used as the basic architecture to build an integrated model called "WindNet" for short-term wind speed prediction in complex terrain. The basic structure of the WindNet model is as follows: Figure 1 As shown in Figure 2 , the present invention uses NWP forecast data for surrounding areas on different isobaric surfaces prior to the target timestamp as input. For example, if the current timestamp is 13:00 and the present invention needs to predict wind speed for the next six hours (i.e., from 13:00 to 19:00, with the target timestamp being 19:00), the input data is NWP forecast data at 15-minute intervals from 13:00 to 19:00 (i.e., the input length is 6 × 4 = 24). These inputs enable the present invention to fully utilize existing forecast results and provide future temporal trend information that cannot be obtained using only historical time series wind speed data.

[0030] Unlike a simple CNN-LSTM framework, this paper designs a multi-task learning (MTL) model to jointly optimize short-term wind speed and direction forecasts, taking into account the similar local spatial correlations between wind speed and wind direction. Specifically, the convolutional layers are designed as shared layers to capture the common spatial correlations between wind speed, wind direction, and NWP forecast data on different isobaric surfaces, resulting in wind speed-related spatial feature maps and wind direction-related spatial feature maps. This shared CNN layer efficiently extracts spatial features related to both data types from the surrounding area and captures the complex nonlinear relationship between wind speed at station height and different isobaric surfaces.

[0031] Afterwards, LSTM is constructed as an independent layer. These two independent LSTM layers can capture the temporal dynamic characteristics of wind speed and wind direction respectively, thereby effectively modeling the temporal dependency of each feature.

[0032] In this embodiment, the optimal model structure of WindNet was determined through grid search. The CNN architecture used in this invention consists of two hidden layers, without a pooling layer. Each hidden layer uses 16 channels and a convolution kernel size of 3×3. The fully connected layer includes two hidden layers, one for deriving integrated wind speed features based on the temporal dynamic features of wind speed, and the other for deriving integrated wind direction features based on the temporal dynamic features of wind direction. Each hidden layer of the fully connected layer has 100 neurons.

[0033] The output layer is used to obtain wind speed prediction results based on the integrated wind speed characteristics and wind direction characteristics.

[0034] (3) Based on the multi-forecast task joint optimization method, the high-frequency short-term wind speed prediction model is trained using the training set. During the training process, the high-frequency short-term wind speed prediction model is evaluated using the validation set and the model parameters are adjusted. The loss function used in the training process is a weighted loss function of the joint wind speed loss function and the wind direction loss function.

[0035] To further integrate the temporal information of wind speed and direction, this paper performs multi-task joint optimization, combining the information of the two loss functions of wind speed and wind direction to further improve wind speed prediction capabilities. The final output of the model is multiple continuous prediction values ​​up to the target timestamp.

[0036] Specifically, the present invention takes wind speed prediction as the main task and wind direction prediction as the auxiliary task, that is, the weighted loss function F=0.7L+0.3L1, where L is the wind speed loss function and L1 is the wind direction loss function.

[0037] The wind speed loss function L is: , Where i represents the i-th training sample, n is the number of training samples, f(x) represents the wind speed or wind direction prediction result, and y represents the true value of wind speed or wind direction.

[0038] During the training of the high-frequency short-term wind speed prediction model, the dropout rate was set to 0.25. A polynomial decay (Poly) strategy was used to adjust the learning rate based on the Adam optimizer. The initial learning rate was 0.001, and the weight decay parameter was 0.1. The high-frequency short-term wind speed prediction model was used for 6-hour, 12-hour, and 24-hour prediction tasks. The batch size varied depending on the task: 12 for the 6-hour prediction task, 32 for the 12-hour prediction task, and 64 for the 24-hour prediction task. The entire training of the high-frequency short-term wind speed prediction model was set to 500 epochs, and an early stopping mechanism was introduced to prevent model overfitting.

[0039] In this embodiment, the evaluation indicators used for the prediction task (regression) include mean absolute error (MAE) and root mean square error (RMSE), where yes The actual wind speed at the moment, yes Wind speed forecast at the moment: Mean Absolute Error (MAE): , Root Mean Square Error (RMSE): , (4) Use the trained high-frequency short-term wind speed prediction model to predict the future wind speed of the target site, and apply inverse normalization processing to the wind speed prediction results.

[0040] To demonstrate the effectiveness of our WindNet model, we conducted a multi-model cross-comparison. For comparative analysis, we compared our WindNet model with NWP, LSTM, and CNN-LSTM models using the same input data. The optimal architectures for these models were also determined via grid search. All experiments were conducted using the Keras deep learning framework. The server configuration for model training was as follows: an Intel(R) Xeon(R) Gold 5218 CPU @ 2.30GHz, a Quadro RTX 6000 GPU, and an Ubuntu 16.04.6 LTS operating system. We compared the performance of NWP, LSTM, CNN-LSTM, and WindNet on wind speed forecasts for the next 6, 12, and 24 hours (with a temporal resolution of 15 minutes). Table 1 shows that model performance varied across different sites, with site S3 performing significantly better, owing to its location in a plain area. All models showed a gradual decline in prediction accuracy for longer prediction steps.

[0041] Table 1 Multi-model cross-comparison results

[0042] Among all models, the LSTM model performed the worst, with a MAE of 1.39 and an RMSE of 1.69 for the 6-hour forecast, a MAE of 1.45 and an RMSE of 1.79 for the 12-hour forecast, and a MAE of 1.60 and an RMSE of 2.03 for the 24-hour forecast. In contrast, the simple CNN-LSTM model performed better for short-term wind speed forecasting, reducing the MAE by 13.67%, 8.28%, and 8.13% for the 6-hour, 12-hour, and 24-hour forecasts, respectively, and reducing the RMSE by 8.88%, 8.38%, and 12.32%. This improvement is primarily attributed to the model's ability to process sequential data and further extract high-dimensional features from the input data. Among all models, "WindNet" performed the best, significantly outperforming the other baseline models, with MAE reductions of 20.83%, 15.79%, and 10.20%, and RMSE reductions of 20.13%, 14.02%, and 9.55%, respectively. In particular, the model's ability to capture sudden fluctuations and extremes in wind speed highlights its exceptional predictive power. This improvement is primarily due to the model's joint optimization strategy, which effectively exploits the local spatial correlation between wind speed and direction, thereby improving forecast accuracy. These results highlight the robustness of our proposed model, WindNet, particularly under diverse and complex terrain conditions, and its ability to provide accurate high-temporal and high-resolution forecasts, making it a valuable tool for precise short-term wind speed forecasting.

[0043] In summary, the WindNet model of an embodiment of the present invention is capable of achieving 6-hour, 12-hour, and 24-hour wind speed forecasts for target sites. WindNet utilizes numerical weather prediction (NWP) data from different isobaric surfaces in the nearby area as input features, combining the advantages of NWP and deep learning. This integration enables WindNet to incorporate real-time meteorological information. The multi-task learning framework introduced in this invention captures the complex interdependencies between wind speed and wind direction, jointly optimizing wind speed and wind direction forecasts, enhancing the model's generalization capabilities and reducing its reliance on training samples. Furthermore, the present invention performs meticulous variable selection for each site, retaining only the most relevant predictors from multiple isobaric surfaces, ensuring that the model focuses on physically meaningful inputs while eliminating noise and redundancy.

[0044] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A high-frequency short-term wind speed prediction method for complex terrain, characterized by: The following steps are involved: (1) Collect numerical weather forecast data of different isobaric surfaces, perform the first preprocessing to form model forecast input data, merge the model forecast input data with the data of the automatic observation station equipment, perform the second preprocessing, and then divide the data into training set and validation set; (2) Construct a high-frequency short-term wind speed prediction model, which includes an input layer, a shared layer, an independent layer, a fully connected layer, and an output layer; the shared layer is a CNN layer, which is used to capture the common spatial correlation between wind speed, wind direction, and numerical weather forecast data of different isobaric surfaces, and obtain wind speed-related spatial feature maps and wind direction-related spatial feature maps; There are two independent layers, both of which are LSTM layers, which are used to capture the temporal dynamic characteristics of wind speed from the wind speed-related spatial feature map, and the temporal dynamic characteristics of wind direction from the wind direction-related spatial feature map; The fully connected layer contains two hidden layers, which are used to obtain the integrated wind speed features based on the temporal dynamic features of wind speed and the integrated wind direction features based on the temporal dynamic features of wind direction; The output layer is used to obtain wind speed prediction results based on the integrated wind speed characteristics and wind direction characteristics; (3) Based on the multi-forecast task joint optimization method, the high-frequency short-term wind speed prediction model is trained using the training set. During the training process, the high-frequency short-term wind speed prediction model is evaluated using the validation set and the model parameters are adjusted. The loss function used in the training process is a weighted loss function of the combined wind speed loss function and the wind direction loss function. (4) Use the trained high-frequency short-term wind speed prediction model to predict the future wind speed of the target site.

2. The high-frequency short-term wind speed prediction method for complex terrain according to claim 1, characterized in that: In step (1), the numerical weather forecast data include the average wind speed, zonal wind component, meridional wind component, and average wind speed at the isobaric surfaces of 850 hPa, 700 hPa, 600 hPa, 500 hPa, 400 hPa, 300 hPa, 250 hPa, 200 hPa, and 100 hPa, respectively; and the average wind speed at a 10-meter monitoring station height. Each numerical weather forecast data set is integrated into a unified 15-minute interval data set at different time resolutions.

3. The high-frequency short-term wind speed prediction method for complex terrain according to claim 1, characterized in that: In step (1), the method for collecting numerical weather forecast data of different isobaric surfaces includes: inputting the coordinates of the wind speed observation equipment, finding the coordinates near 3 km of the numerical model weather forecast as the grid center, and finding the grid within the coordinates and the nearby adjacent grids, and then extracting the numerical weather forecast data of different isobaric surfaces.

4. The high-frequency short-term wind speed prediction method for complex terrain according to claim 1, characterized in that: In step (1), the first preprocessing includes missing and outlier processing, as well as normalization of variables; the second preprocessing is to eliminate model weather forecast data with a value less than the threshold of 0.2 through the Pearson correlation coefficient.

5. The high-frequency short-term wind speed prediction method for complex terrain according to claim 4, characterized in that: Step (4) also includes applying inverse normalization to the wind speed prediction results.

6. The high-frequency short-term wind speed prediction method for complex terrain according to claim 1, characterized in that: In step (2), the high-frequency short-term wind speed prediction model framework is determined based on grid search.

7. The high-frequency short-term wind speed prediction method for complex terrain according to claim 6, characterized in that: The CNN layer consists of two hidden layers. Each hidden layer of the CNN layer uses 16 channels and the convolution kernel size is 3×3; each hidden layer of the fully connected layer has 100 neurons.

8. The high-frequency short-term wind speed prediction method for complex terrain according to claim 7, characterized in that: During the training of the high-frequency short-term wind speed prediction model, the dropout rate was set to 0.25; the weights of the wind speed loss function and wind direction loss function were 0.7 and 0.3, respectively. During the training of the high-frequency short-term wind speed prediction model, a polynomial decay strategy was used to adjust the learning rate based on the Adam optimizer, with an initial learning rate of 0.001 and a weight decay parameter of 0.

1. The high-frequency short-term wind speed prediction model was used for 6-hour prediction tasks, 12-hour prediction tasks, and 24-hour prediction tasks. The batch size was 12 for the 6-hour prediction task, 32 for the 12-hour prediction task, and 64 for the 24-hour prediction task. The entire training of the high-frequency short-term wind speed prediction model was set to 500 epochs, and an early stopping mechanism was introduced to prevent model overfitting.

9. The high-frequency short-term wind speed prediction method for complex terrain according to claim 1, characterized in that: In step (3), the server configuration used for training the high-frequency short-term wind speed prediction model is: Intel(R) Xeon(R) Gold 5218 CPU @2.30GHz CPU, Quadro RTX 6000 GPU, Ubuntu 16.04.6 LTS operating system.

10. The high-frequency short-term wind speed prediction method for complex terrain according to claim 1, characterized in that: In step (3), the model evaluation indicators used include mean absolute error and root mean square error.

Images