Wind farm wake and power prediction method based on generative adversarial network model
By adopting wake and power prediction methods based on the generation of adversarial network models in wind farms, the problems of low accuracy and high numerical simulation calculation cost in the prior art are solved, and efficient and accurate wake prediction is achieved.
Patent Information
- Application Number
- CN202311734604.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-15
- Publication Date
- 2025-03-11
- Estimated Expiration
- 2043-12-15
AI Technical Summary
In the prior art, the analytical wake model is convenient to predict but has low accuracy, while the numerical simulation method has high accuracy but high calculation cost, making it difficult to obtain both.
The wind farm wake and power prediction method based on the generative adversarial network model is adopted. By constructing a single fan wake prediction model and a multi-fan wake overlay prediction model, the Transformer model and a convolutional neural network are used for training and verification.
It achieves the wake prediction accuracy close to numerical simulation while ensuring efficient calculations, can be quickly deployed in different wind farms, and further improves prediction performance through actual measured data.
Smart Images
Figure CN117744709B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of wind power generation, and in particular to a method for predicting wind farm wake and power based on a generative adversarial network model. Background Art
[0002] As one of the main renewable energy sources in the global energy market, wind energy has grown rapidly in recent years. In wind farms, the wake effect is the key factor causing power loss of downstream wind turbines, and accurate prediction of wind farm wake can significantly improve the power generation efficiency of large-scale wind farms. Therefore, wake prediction technology is of great significance for improving the economic benefits of wind farms. Wind turbine wake refers to the feather-like area in the downstream area of the unit, which is characterized by reduced wind speed and increased turbulence intensity. Under conditions such as the wind direction is completely aligned with the wind turbine formation, the wind speed is lower than the rated wind speed, or the layout is close, the wake loss may be as high as 70%.
[0003] The two mainstream methods for predicting wind turbine wakes are analytical wake models and numerical simulations. The analytical wake model is easy to use and deploy, but it cannot accurately capture the complex turbulence changes in the wake. High-fidelity numerical simulations can significantly improve the accuracy of wake predictions, but the computational cost is extremely high when solving large-scale wind farm environmental problems. Summary of the invention
[0004] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is: to provide a wind farm wake and power prediction method based on a generative adversarial network model, which solves the problem that the analytical wake model method in the prior art is convenient for prediction but has low accuracy, while the numerical simulation method has high accuracy but high computational cost, and it is difficult to have both.
[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:
[0006] A method for predicting wind farm wake and power based on a generative adversarial network model comprises the following steps:
[0007] 1) A single wind turbine wake prediction model is constructed based on the transformer model and conditional generative adversarial network. The wake flow field dataset is generated through empirical analytical wake model and numerical simulation to train and verify the model.
[0008] 2) A multi-wind turbine wake superposition prediction model is constructed based on a convolutional neural network. The wake flow field dataset is generated through empirical analytical wake model and numerical simulation to train and verify the model.
[0009] 3) Predicting the wake velocity field, turbulence field and total power generation of the wind farm wake based on the single wind turbine wake prediction model and the multi-wind turbine wake superposition prediction model.
[0010] As an optimization, step 1) includes the following steps:
[0011] 101) The one-dimensional input vector composed of the wind turbine yaw angle, thrust coefficient and other relevant unit parameter specifications, the three-dimensional input array composed of the three-dimensional flow field velocity and turbulence information of the inflow profile, and the characteristic map data composed of random noise generated by Gaussian distribution are used as input parameters of the entire model;
[0012] 102) embedding the vision-based Vision Transformer into the generator network in the conditional generative adversarial network, wherein a specific moving window attention mechanism module is inserted into the Vision Transformer to perform partitioned attention mechanism processing on the feature map data, and the size of the feature map doubles each time it passes through the module; each moving window attention mechanism module is followed by a block fusion layer to fuse the feature map data between different channels;
[0013] 103) The feature map data after the combination module of multiple moving window attention mechanism modules and block fusion layers is further subjected to pixel fusion and ordinary convolutional layers to generate the final predicted tail flow field information;
[0014] 104) The predicted wake field information is further input into the discriminator network for true or false judgment. The discriminator network is composed of a stack of basic modules of the residual network, and finally outputs the true or false probability judgment of the near, middle and far wakes;
[0015] 105) combining the generator network and the discriminator network to form a complete conditional generative adversarial network, thereby constructing a single wind turbine wake prediction model;
[0016] 106) Pre-training stage: using the training data set generated by the empirical analytical wake model method to conduct preliminary training on the single wind turbine wake prediction model;
[0017] 107) Fine-tuning stage: The wake flow field data set generated by the numerical simulation method is used to further train and verify the single wind turbine wake prediction model;
[0018] 108) After the pre-training stage and the fine-tuning stage, the final single wind turbine wake prediction model is obtained.
[0019] As an optimization, in step 101),
[0020] The one-dimensional input vector is passed through a parameter mapping network to obtain a three-dimensional tensor with the same shape as the three-dimensional input array, and then added to the three-dimensional input array and input into the generator network.
[0021] As an optimization, the added array is split into blocks of a specific size before entering the generator model.
[0022] As an optimization, step 2) includes the following steps:
[0023] 201) using a stacked residual network basic module to construct a multi-fan wake superposition prediction model, the input of which is the prediction result of the single fan wake prediction model in step 1);
[0024] 202) Pre-training stage: The wake flow field data set generated by the empirical analytical wake model is used to conduct preliminary training on the multi-wind turbine wake superposition prediction model;
[0025] 203) Fine-tuning stage: The wake flow field data set generated by the numerical simulation method is used to further train and verify the multi-wind turbine wake superposition prediction model;
[0026] 204) After the pre-training stage and the fine-tuning stage, the final multi-wind turbine wake superposition prediction model is obtained.
[0027] As an optimization, the wake flow field data set is a wake wind speed and turbulence data set generated under different incoming flows and yaw angles.
[0028] As an optimization, step 3) includes the following steps:
[0029] 301) Setting a standard reference wind direction, rotating the wind turbine coordinates in the wind farm so that the wind turbines are arranged in order from upstream to downstream of the wind direction;
[0030] 302) According to the rotated coordinate system, the single wake field is calculated starting from the most upstream wind turbine in sequence, that is, the incoming flow is input into the single wind turbine wake prediction model to obtain the single wake field of the most upstream wind turbine, and then starting from the second wind turbine, the single wake field results calculated by each wind turbine according to the single wind turbine wake prediction model and the single wake field result of the previous wind turbine are input into the multi-wind turbine wake superposition prediction model to obtain the multi-wake field prediction results, and the calculations are accumulated to the most downstream wind turbine in sequence to obtain the final wake field results;
[0031] 303) according to the wake field result, the average inflow wind speed of each wind turbine is calculated by obtaining the wind speed information at the coordinate point at the cross section of each wind turbine impeller;
[0032] 304) According to the inflow wind speed of each wind turbine, the wake velocity field and turbulence field of the wind farm wake and the total power generation are predicted according to the preset relationship curve.
[0033] Compared with the prior art, this application has the following beneficial effects:
[0034] The present invention aims to solve the problem of rapid prediction of flow field and power in wind farms under different incoming flow conditions. Based on the generative adversarial network and Transformer model in deep learning technology, a wake calculation method that combines the advantages of analytical wake model and numerical simulation method is proposed. By adopting deep learning model technology, the wake prediction accuracy close to that of numerical simulation can be achieved while ensuring efficient calculation. Relying on data-driven model training methods, it can be quickly deployed in various types of wind farms; secondly, the prediction performance can be further improved through the measured data of the wind farm, avoiding the traditional parameter identification steps, and the application is more efficient. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 A single wind turbine wake prediction model constructed based on a transformer model and a conditional generative adversarial network in the present invention;
[0036] Figure 2 It is a flowchart of the wake field prediction of multiple wind turbine wake superposition in the present invention;
[0037] Figure 3 The single wind turbine wake prediction in the present invention is 1-predicted wind speed v / hub height wind speed v in the y-axis direction under the set working conditions h wake data;
[0038] Figure 4 The single wind turbine wake prediction in the present invention is the turbulence change rate in the y-axis direction under the set working conditions ΔI / hub height wind speed v h wake data;
[0039] Figure 5 The single wind turbine wake prediction in the present invention is 1-predicted wind speed v / hub height wind speed v in the z-axis direction under the set working conditions h wake data;
[0040] Figure 6 is the turbulence change rate ΔI / hub height wind speed v in the z-axis direction under the set working conditions predicted by the single wind turbine wake in the present invention h wake data;
[0041] Figure 7 The wake cloud diagrams of different cross sections for the single fan wake prediction under set working conditions in the present invention;
[0042] Figure 8 A schematic diagram of the layout of wind turbines for predicting the wake field of multiple wind turbine wake superposition in the present invention;
[0043] Fig. 9 For Figure 8 The Y-axis wake prediction data of the wind farm with wind turbine layout;
[0044] Fig.10For Figure 8 The z-axis wake prediction data of the wind farm with wind turbine layout;
[0045] Fig.11 For Figure 8 Cross-sectional wake cloud diagram of a wind farm with wind turbine layout. DETAILED DESCRIPTION
[0046] The present invention will be further described in detail below in conjunction with the accompanying drawings.
[0047] For specific implementation: see Figure 1-11 ,
[0048] In order to achieve the dual advantages of high efficiency and accuracy, the present invention proposes a wake prediction framework based on machine learning technology. The machine learning-based model will be trained and tested using a large number of wake data sets generated by CFD simulations. Once well-trained, the machine learning-based wake prediction model can effectively characterize the wake distribution of a wind farm, with a calculation speed comparable to that of a traditional analytical wake model, while maintaining good accuracy equal to that of a high-fidelity numerical simulation. The framework is expected to be applicable to the modeling of single or multiple wakes under various operating conditions, and further deployed in application problems of wind farms, such as power prediction, layout optimization, or intelligent control.
[0049] Specifically, the present invention proposes a method for predicting the wind farm wake and power based on a generative adversarial network model, comprising the following steps:
[0050] 1) A single wind turbine wake prediction model is constructed based on the transformer model and conditional generative adversarial network, and the model is trained and verified by generating a training data set through analytical wake model and numerical simulation. Figure 1 shown.
[0051] Specifically, step 1) includes the following steps:
[0052] 101) The one-dimensional input vector composed of the wind turbine yaw angle, thrust coefficient and other relevant unit parameter specifications, the three-dimensional input array composed of the three-dimensional flow field velocity and turbulence information of the inflow profile, and the feature map data composed of the random noise generated by the Gaussian distribution are used as the input parameters of the entire model; the one-dimensional input vector is passed through the parameter mapping network to obtain a three-dimensional tensor with the same shape as the three-dimensional input array, and then added to the three-dimensional input array and input into the generator network. The array obtained by addition is divided into blocks of a specific size before entering the generator model to reduce the number of tokens in subsequent processing to improve training efficiency.
[0053] 102) embedding the vision-based Vision Transformer into the generator network in the conditional generative adversarial network, wherein a specific moving window attention mechanism module is inserted into the Vision Transformer to perform partitioned attention mechanism processing on the feature map data, and the size of the feature map doubles each time it passes through the module; each moving window attention mechanism module is followed by a block fusion layer to fuse the feature map data between different channels;
[0054] 103) The feature map data after the combination module of multiple moving window attention mechanism modules and block fusion layers is further subjected to pixel fusion and ordinary convolutional layers to generate the final predicted tail flow field information;
[0055] 104) The predicted wake field information and the real wake data are further input into the discriminator network for true or false judgment. The discriminator network is composed of a stack of basic modules of a residual network, and finally outputs the true or false probability judgment of the near, middle and far wakes through a residual connection module and a multi-layer perceptron. The real wake data can be obtained by data collection, analysis of wake models or numerical simulation calculations.
[0056] 105) combining the generator network and the discriminator network to form a complete conditional generative adversarial network, thereby constructing a single wind turbine wake prediction model;
[0057] 106) Pre-training stage: using the training data set generated by the empirical analytical wake model method to conduct preliminary training on the single wind turbine wake prediction model;
[0058] 107) Fine-tuning stage: The wake flow field data set generated by the numerical simulation method is used to further train and verify the single wind turbine wake prediction model;
[0059] 108) After the pre-training stage and the fine-tuning stage, the final single wind turbine wake prediction model is obtained.
[0060] 2) A multi-wind turbine wake superposition prediction model is constructed based on a convolutional neural network. The model is trained and verified by generating a training data set through analytical wake model and numerical simulation. Figure 2 As shown,
[0061] Specifically, step 2) includes the following steps:
[0062] 201) using a stacked residual network basic module to construct a multi-fan wake superposition prediction model, the input of which is the prediction result of the single fan wake prediction model in step 1);
[0063] 202) Pre-training stage: The wake flow field data set generated by the empirical analytical wake model is used to conduct preliminary training on the multi-wind turbine wake superposition prediction model;
[0064] 203) Fine-tuning stage: The wake flow field data set generated by the numerical simulation method is used to further train and verify the multi-wind turbine wake superposition prediction model;
[0065] 204) After the pre-training stage and the fine-tuning stage, the final multi-wind turbine wake superposition prediction model is obtained.
[0066] The training data set is a wake wind speed and turbulence data set generated under different incoming flows and yaw angles.
[0067] 3) Predicting the wake field and power of the wind farm wake based on the single wind turbine wake prediction model and the multi-wind turbine wake superposition prediction model.
[0068] Specifically, step 3) includes the following steps:
[0069] 301) Set a standard reference wind direction, and rotate the wind turbine coordinates in the wind farm so that the wind turbines are arranged in sequence from upstream to downstream of the wind direction.
[0070] 302) According to the rotated coordinate system, the single wake field is calculated from the most upstream wind turbine in sequence, that is, the incoming flow is input into the single wind turbine wake prediction model to obtain the single wake field of the most upstream wind turbine, and then starting from the second wind turbine, the single wake field results calculated by each wind turbine according to the single wind turbine wake prediction model and the single wake field results of the previous wind turbine are input into the multi-wind turbine wake superposition prediction model to obtain the multi-wake field prediction results, and then accumulated to the most downstream wind turbine in sequence, so as to obtain the final wake field result. That is, the single wake field is calculated from the most upstream wind turbine in sequence, that is, the incoming flow is input into the single wake prediction model to obtain the wake field considering only the most upstream wind turbine, and then the wake fields of the second, third, fourth, etc. are calculated, wherein starting from the second wind turbine, after each calculation of the single wake field, the current result and the wake field obtained by the previous calculation need to be input into the multi-wake prediction model to obtain the multi-wake prediction results, and so on, and the final wake field result can be obtained by repeatedly accumulating to the last wind turbine.
[0071] 303) according to the wake field result, the average inflow wind speed of each wind turbine is calculated by obtaining the wind speed information at the coordinate point at the cross section of each wind turbine impeller;
[0072] 304) According to the inflow wind speed of each wind turbine, the power of each wind turbine is predicted according to the corresponding power curve to obtain the power generation power of the entire wind farm.
[0073] The present invention,
[0074] 1) For single wind turbine wake prediction: the inflow profile with different wind shear is divided into six layers of input feature maps, including u-direction wind speed, v-direction wind speed, w-direction wind speed, u-direction turbulence, v-direction turbulence, and w-direction turbulence. The wind speeds at the hub height are 8m / s, 12m / s, and 15m / s respectively. At the same time, the wind turbine yaw parameters, impeller diameter, hub height and other unit parameters are combined and input into Figure 1 The single wake prediction model shown can be obtained Figure 3-7 Wind speed turbulence profiles and flow field information at different downstream distances in each plane;
[0075] Specifically, the cross sections at the downstream distances of 4D, 6D, 8D, 10D, 12D, and 15D were respectively cut. Figure 3 where are the wind speed at hub height v h =8m / s, turbulence I u = 8%, yaw angle θ = 20°, and hub height wind speed v h =8m / s, turbulence I u =8%, yaw angle θ = 30°, wake data in the y-axis direction, the horizontal axis represents 1-predicted wind speed v / hub height wind speed v h , the ordinate represents the ratio of the distance from the x-axis along the y-axis to the hub diameter, the red line and the green line represent the prediction results of the multi-turbine wake superposition prediction model after the pre-training stage and the fine-tuning stage, respectively, × represents the data of the analytical wake model proposed by Ishihara, and ○ represents the data of the Reynolds stress turbulence and brake disc coupling model. Figure 4 is the turbulence change rate ΔI / hub height wind speed v in the wake data in the y-axis direction under the working condition parameters h of changes.
[0076] Figure 5 Wind speed at hub height v h =12m / s, turbulence I u = 8%, yaw angle θ = 20°, and hub height wind speed v h =12m / s, turbulence I u Wake data in the z-axis direction under two working conditions: =8% and yaw angle θ=30°. Figure 6 are respectively the wind speed at hub height v h =8m / s, turbulence I u = 8%, yaw angle θ = 20°, and hub height wind speed v h =8m / s, turbulence I u =8%, yaw angle θ = 30°, the turbulence change rate in the wake data in the z-axis direction ΔI / hub height wind speed v h of changes.
[0077] Figure 7 are respectively the wind speed at hub height v h =15m / s, turbulence I u = 8%, yaw angle θ = 20°, and hub height wind speed v h =15m / s, turbulence I u Cloud map data of wind speed turbulence profiles at different downstream distances under two operating parameters: θ = 8% and yaw angle θ = 30°.
[0078] 2) For the prediction of multiple wind turbine wakes: According to the prediction method of single wind turbine wake, the wake field information under single wind turbine working condition can be predicted in advance; the wake field results of each wind turbine can be further predicted according to Figure 2 The prediction process of the multi-wake prediction model is shown in the figure. After combining the input, the wake field information of multiple wake superposition can be obtained. The layout of the wind turbine is as follows: Figure 8 As shown, the prediction results are Figure 9-11 shown.
[0079] Specifically, use Figure 8 When the wind turbine layout is shown, there are two operating parameters, one of which is the hub height wind speed v h =8m / s, turbulence I u = 8%, yaw angle θ = 0°, the other is hub height wind speed v h =8m / s, turbulence I a =8%, yaw angle θ = 10°. The prediction results of the multi-wake prediction model of the present invention are as follows: Figure 9-11 As shown in the figure, the upper part is the first operating condition parameters, and the lower part is the second operating condition parameters. The cross sections at 9D, 11D, 13D, 15D, 17D and 20D away from the wind farm are respectively intercepted along the x-axis direction. The wake wind speed in the y-axis direction is as follows Fig. 9 As shown, the horizontal axis represents 1-predicted wind speed v / hub height wind speed v h , the ordinate represents the ratio of the distance from the x-axis along the y-axis to the hub diameter, the red line and the green line represent the prediction results of the multi-turbine wake superposition prediction model after the pre-training stage and the fine-tuning stage, × represents the data of the analytical wake model proposed by Ishihara, and ○ represents the data of the Reynolds stress turbulence and brake disc coupling model. The wake wind speed in the z-axis direction is as follows Fig.10 As shown, the ordinate represents the ratio of height to fan hub diameter, and the others are Figure 5 The wake cloud diagram of its cross section is as follows: Fig.11 As shown by Figure 7 The wake distribution data at different cross-sectional locations can be seen more intuitively.
[0080] The present invention aims to solve the problem of rapid prediction of flow field and power in wind farms under different incoming flow conditions. Based on the generative adversarial network and Transformer model in deep learning technology, a wake calculation method that combines the advantages of analytical wake model and numerical simulation method is proposed. By adopting deep learning model technology, the wake prediction accuracy close to that of numerical simulation can be achieved while ensuring efficient calculation. Relying on data-driven model training methods, it can be quickly deployed in various types of wind farms; secondly, the prediction performance can be further improved through the measured data of the wind farm, avoiding the traditional parameter identification steps, and the application is more efficient.
[0081] Although the embodiments of the present invention have been shown and described, it is apparent to those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and basis of the present invention. The scope of the present invention is defined by the appended claims and their equivalents. Therefore, the embodiments of the present invention are merely illustrative examples of the present invention. The embodiments of the present invention do not constitute limitations on the present invention from any point of view.
Claims
1. A method for predicting wind farm wake and power based on a generative adversarial network model, characterized in that: The steps include: 1) A single wind turbine wake prediction model is constructed based on the transformer model and conditional generative adversarial network. The wake flow field dataset is generated through empirical analytical wake model and numerical simulation to train and verify the model. 2) A multi-wind turbine wake superposition prediction model is constructed based on a convolutional neural network. The wake flow field dataset is generated through empirical analytical wake models and numerical simulations to train and verify the model. 3) Predicting the wake velocity field, turbulence field and total power generation of the wind farm wake according to the single wind turbine wake prediction model and the multi-wind turbine wake superposition prediction model; wherein, Step 3) includes the following steps: 301) Set the standard reference wind direction and rotate the wind turbine coordinates in the wind farm so that the wind turbines are arranged in order from upstream to downstream of the wind direction; 302) According to the rotated coordinate system, the single wake field is calculated starting from the most upstream wind turbine in sequence, that is, the incoming flow is input into the single wind turbine wake prediction model to obtain the single wake field of the most upstream wind turbine, and then starting from the second wind turbine, the single wake field results calculated by each wind turbine according to the single wind turbine wake prediction model and the single wake field result of the previous wind turbine are input into the multi-wind turbine wake superposition prediction model to obtain the multi-wake field prediction results, and the calculation is accumulated to the most downstream wind turbine in sequence, so as to obtain the final wake field result; 303) According to the wake field result, the average inflow wind speed of each wind turbine is calculated by obtaining wind speed information at the coordinate points at the cross section of each wind turbine impeller; 304) According to the inflow wind speed of each wind turbine, the wake velocity field, turbulence field and total power generation of the wind farm wake are predicted according to the preset relationship curve.
2. The method for predicting wind farm wake and power based on a generative adversarial network model according to claim 1, characterized in that: Step 1) includes the following steps: 101) The one-dimensional input vector consisting of the wind turbine yaw angle, thrust coefficient and other relevant unit parameter specifications, the three-dimensional input array consisting of the three-dimensional flow field velocity and turbulence information of the inflow profile, and the characteristic map data consisting of random noise generated by Gaussian distribution are used as input parameters of the entire model; 102) The vision-based Vision Transformer is embedded into the generator network in the conditional generative adversarial network, wherein a specific moving window attention mechanism module is inserted into the Vision Transformer to perform partitioned attention mechanism processing on the feature map data, and the size of the feature map doubles each time it passes through the module; each moving window attention mechanism module is followed by a block fusion layer to fuse the feature map data between different channels; 103) The feature map data after the combination module of multiple moving window attention mechanism modules and block fusion layers is further processed by pixel fusion and ordinary convolution layer to generate the final predicted tail flow field information; 104) The predicted wake field information is further input into the discriminator network for true or false judgment. The discriminator network is composed of the basic modules of the residual network stacked together, and finally outputs the true or false probability judgment of the near, middle and far wakes; 105) Combining the generator network and the discriminator network to form a complete conditional generative adversarial network, thereby constructing a single wind turbine wake prediction model; 106) Pre-training stage: The single wind turbine wake prediction model is initially trained using a training data set generated by the empirical analytical wake model method; 107) Fine-tuning stage: The wake flow field data set generated by numerical simulation method is used to further train and verify the single wind turbine wake prediction model; 108) After the pre-training stage and the fine-tuning stage, the final single wind turbine wake prediction model is obtained.
3. The method for predicting wind farm wake and power based on a generative adversarial network model according to claim 2, characterized in that: In step 101), The one-dimensional input vector is passed through a parameter mapping network to obtain a three-dimensional tensor with the same shape as the three-dimensional input array, and then added to the three-dimensional input array and input into the generator network.
4. The method for predicting wind farm wake and power based on a generative adversarial network model according to claim 3 is characterized in that: The added array is split into blocks of a specific size before entering the generator model.
5. The method for predicting wind farm wake and power based on a generative adversarial network model according to claim 1, characterized in that: Step 2) includes the following steps: 201) A stacked residual network basic module is used to construct a multi-wind turbine wake superposition prediction model, and the input is the prediction result of the single wind turbine wake prediction model in step 1); 202) Pre-training stage: The wake flow field dataset generated by the empirical analytical wake model is used to conduct preliminary training on the multi-wind turbine wake superposition prediction model; 203) Fine-tuning stage: The wake flow field data set generated by numerical simulation method is used to further train and verify the multi-wind turbine wake superposition prediction model; 204) After the pre-training stage and the fine-tuning stage, the final multi-wind turbine wake superposition prediction model is obtained.
6. The method for predicting wind farm wake and power based on a generative adversarial network model according to claim 1, characterized in that: The wake flow field data set is a wake wind speed and turbulence data set generated under different incoming flows and yaw angles.