A wind direction prediction and yaw control method and system for a wind turbine

Abstract

This invention discloses a wind direction prediction and yaw control method for wind turbines. A wind direction detector acquires the wind direction of each wind turbine in the wind farm in real time, predicts the wind direction based on the intrinsic mode components of each target, and superimposes and reconstructs the prediction results to output the wind direction prediction result. It calculates whether the angle α between the average wind direction and the nacelle axis during the yaw threshold time period is greater than the minimum value of the yaw deviation threshold. The yaw motor execution module adjusts the yaw angle according to the yaw deviation threshold under different yaw control modes. This wind direction prediction and yaw control method guides the yaw system's actions through accurate wind direction prediction, improving the yaw system's wind alignment accuracy and reducing yaw waiting time, thus achieving rapid wind alignment. Furthermore, by reclassifying the yaw control parameters, it avoids the problem of severe wear on the yaw system's mechanical components caused by frequent start-stop operations in low wind speed areas.

Description

Technical Field

[0001] This invention relates to the field of power generation, and in particular to a method and system for wind direction prediction and yaw control of wind turbine generators. Background Technology

[0002] The rapid development of the global economy and society has exacerbated the contradiction between the supply of primary energy sources such as oil and environmental protection, accelerating the outbreak of the traditional energy crisis. Wind energy, as a clean and environmentally friendly renewable energy source, is increasingly favored by countries worldwide. my country's carbon peaking and carbon neutrality strategy has further promoted the development of wind power generation. Wind power generation mainly utilizes wind to drive a turbine to rotate, which in turn drives a generator to rotate, cutting magnetic field lines and converting wind energy into mechanical work. This mechanical work then drives a rotor to rotate, ultimately generating electricity. Currently, wind farms are widespread throughout China. However, with increasing age, many wind farms face problems such as reduced power characteristic coefficients and wind power conversion efficiency. With the advent of grid parity, increasing wind turbine power generation has become a focus for many owners.

[0003] The yaw system plays an indispensable role in wind turbine generators. On one hand, it effectively adjusts the nacelle's alignment with the wind direction; on the other hand, it reduces fatigue load on the turbine, increases its lifespan, and prevents stress imbalances between the turbine and blades. The main factor affecting the yaw control system is wind direction. However, influenced by geographical factors, especially in mountainous areas, wind speed and direction are unstable, changing frequently and significantly. To rationally utilize wind resources, reduce the adverse effects of wind direction and speed characteristics, ensure safe grid connection of wind power, reduce the number of yawing maneuvers and yawing errors, and improve wind alignment accuracy, research on yaw control systems and strategies is particularly important.

[0004] With the development of technology, research on the optimization control of wind turbine units is increasing to fully utilize wind resources and increase power generation. However, compared with generator systems and pitch systems, research on yaw control systems is relatively limited. To reduce the impact of yaw error on wind power performance, most wind turbine units currently adopt active yaw control based on wind direction measurement. However, the contradiction between rapid changes in wind direction and slow yaw follow-up actions limits the performance of this control method. Finally, with the exploitation of low-speed wind resources, due to frequent changes in low-speed wind direction, directly transplanting traditional yaw strategies would require the nacelle to continuously restart wind-following operations to effectively align with the wind, thus affecting the lifespan of the yaw bearing. Summary of the Invention

[0005] The technical problem this invention aims to solve is to provide a wind direction prediction and yaw control method for wind turbines. This method guides the yaw system's actions through accurate wind direction prediction, thereby improving the yaw system's wind alignment accuracy and reducing yaw waiting time, achieving rapid wind alignment. By reclassifying the yaw control parameters and increasing the yaw deviation threshold and yaw time threshold in low wind speed ranges, the problem of severe wear on the yaw system's mechanical components caused by frequent start-stop operations in low wind speed areas is avoided.

[0006] To address the aforementioned technical problems, this invention provides a wind turbine wind direction prediction and yaw control method. The method includes a wind direction detector, a data acquisition module, a data prediction module, a wind turbine yaw module, and a yaw motor execution module. The method comprises the following steps: the wind direction detector acquires the wind direction of each wind turbine in the wind farm in real time; the data acquisition module acquires the wind direction data from the wind direction detector in real time and sends it to the data prediction module; the data prediction module decomposes the wind direction data into target intrinsic mode components (IMF) data of different frequencies according to a data decomposition strategy, predicts each target IMF data, and superimposes and reconstructs the prediction results to output a wind direction prediction result; the wind turbine yaw module calculates whether the angle α between the average wind direction and the nacelle axis during the yaw threshold time period is greater than the minimum value of the yaw deviation threshold based on the wind direction prediction result; if not, the yaw system does not operate; if so, different yaw deviation thresholds are applied under different yaw control modes according to the wind speed range; and the yaw motor execution module adjusts the yaw angle according to the yaw deviation thresholds under different yaw control modes.

[0007] Preferably, a set of wind direction data x(t) over a period of time is acquired. Gaussian white noise of the same length, following a normal distribution, is added to the wind direction data x(t), and the Gaussian white noise is normalized. Then, empirical mode decomposition (EMD) is used to decompose the data to obtain intrinsic mode components (IMFs). This process is repeated continuously, adding different normally distributed Gaussian white noise each time. The average of all the corresponding IMFs obtained each time is calculated to obtain a target IMF. Multiple target IMFs are obtained from the original wind direction data sequence x(t), forming the signal intrinsic mode combination of the wind direction data sequence x(t).

[0008]

[0009] The first few groups of target intrinsic mode components (IMFs) with higher energy and correlation coefficients are classified as high-frequency IMFs, while the last few groups of target intrinsic mode components (IMFs) with lower energy and correlation coefficients are classified as low-frequency IMFs.

[0010] Where x(t) is the original wind direction data sequence, c i(t) represents the i-th intrinsic mode component (IMF), r n The remaining component of Res.

[0011] Preferably, wind direction prediction based on the sequence data of the intrinsic mode components (IMFs) of each target includes: using a neural network strategy to predict the high-frequency component IMFs to obtain high-frequency component prediction values;

[0012] The least squares support vector machine model is used to predict the low-frequency component IMF and Res residual component to obtain the predicted values ​​of the low-frequency component IMF and Res residual component.

[0013] Preferably, the neural network is a backpropagation neural network (BPNN). The backpropagation neural network consists of an input layer, hidden layers, and an output layer. The number of nodes in the hidden layer is determined by the number of nodes in the input layer and the number of nodes in the output layer. If the number of nodes in the output layer is m and the number of nodes in the input layer is n, then the number of nodes in the hidden layer is: Where 'a' is an integer between 0 and 10, the number of hidden layers is one or more, and the transfer formula from the input layer to the hidden layer is: The transfer formula from the hidden layer to the output layer is as follows: The activation function of the hidden layer is: w ij w represents the connection weights between neurons in the input layer and the hidden layer. jk a represents the connection weights between neurons in the hidden layer and the output layer. j b is the threshold for the hidden layer. k Let Q be the output layer threshold. k Let x be the predicted value of the high-frequency component. i Let i be the high-frequency component IMF, i = 1, 2, ..., n.

[0014] Preferably, the calculation formula for the least squares support vector machine model is:

[0015]

[0016] Where: α i Let K(x) represent the Lagrange multipliers of the corresponding components. i x j ) represents the symmetric matrix of the kernel function of the least squares support vector machine model, b represents the bias, f(x) is the predicted value of the low-frequency component IMF and the residual component, x is the low-frequency component IMF and the Res residual component, and N is the number of data in the low-frequency component IMF and the Res residual component.

[0017] Wherein: the specific expression of the kernel function is:

[0018]

[0019] σ is the kernel parameter.

[0020] Preferably, the wind direction detector is used to acquire the wind direction at each wind turbine in the wind farm in real time, and is installed at a suitable position on the wind turbine. The wind direction detector is communicatively connected to the data acquisition module. The data acquisition module is used to acquire wind direction data from the wind direction detector and store the wind direction data. The data prediction module is communicatively connected to the data acquisition module. The wind turbine yaw module is communicatively connected to the data prediction module. The wind turbine yaw module is communicatively connected to the yaw motor execution module.

[0021] Preferably, the wind speed range is divided into: cut-in wind speed, peak wind speed, rated wind speed, and cut-out wind speed;

[0022] The phrase "execute yaw deviation thresholds under different yaw control modes according to the range of wind speed" specifically includes: first round of yaw deviation threshold selection: determining whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the cut-in wind speed and less than the peak wind speed;

[0023] If so, the wind turbine yaw module controls the execution of the yaw deviation threshold in the low wind speed control mode.

[0024] If not, proceed to the second round of yaw deviation threshold selection;

[0025] The second round of yaw deviation threshold selection involves determining whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the peak wind speed but less than the rated wind speed. If the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the peak wind speed but less than the rated wind speed, then the wind turbine yaw module controls the yaw deviation threshold in the wind speed control mode.

[0026] If not, proceed to the third round of yaw deviation threshold selection;

[0027] The third round of yaw deviation threshold selection: It determines whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the rated wind speed and less than the cut-out wind speed. If the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the rated wind speed and less than the cut-out wind speed, then the wind turbine yaw module controls the execution of the yaw deviation threshold in the high wind speed control mode.

[0028] If not, proceed to the fourth round of yaw deviation threshold selection;

[0029] Fourth round of yaw deviation threshold selection: Determine whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the cut-out wind speed.

[0030] If so, then enter shutdown mode.

[0031] Preferably, the yaw deviation threshold includes a yaw error threshold and a yaw error duration threshold.

[0032] To address the aforementioned technical problems, this invention also discloses a control system for the wind direction prediction and yaw control method applied to the aforementioned wind turbine generator. The control system includes a wind direction detector, a data acquisition module, a data prediction module, a wind turbine yaw module, and a yaw motor execution module. The wind direction detector acquires the wind direction of each wind turbine generator in the wind farm in real time. The data acquisition module acquires the wind direction data from the wind direction detector in real time and sends it to the data prediction module. The data prediction module decomposes the wind direction data according to a data decomposition strategy to obtain sequence data of target intrinsic mode components (IMFs) at different frequencies. It then predicts the wind direction for each IMF, superimposes and reconstructs the prediction results, and outputs the wind direction prediction result. The wind turbine yaw module uses the wind direction prediction result to calculate whether the angle α between the average wind direction and the nacelle axis during the yaw threshold time period is greater than the minimum value of the yaw deviation threshold. If not, the yaw system does not operate; if so, it executes the yaw deviation threshold under different yaw control modes according to the wind speed range. The yaw motor execution module adjusts the yaw angle according to the yaw deviation threshold under different yaw control modes.

[0033] The wind direction prediction and yaw control method for wind turbines of this invention involves a wind direction detector acquiring the wind direction of each wind turbine in the wind farm in real time. A data acquisition module collects the wind direction data from the wind direction detector and sends it to the data prediction module. The data prediction module decomposes the wind direction data into sequence data of target intrinsic mode components (IMFs) at different frequencies according to a data decomposition strategy. Wind direction prediction is performed on the data of each IMF, and the prediction results are superimposed and reconstructed to output the wind direction prediction result. The wind turbine yaw module uses the wind direction prediction result to calculate whether the angle α between the average wind direction and the nacelle axis during the yaw threshold time period is greater than the yaw threshold. The minimum yaw deviation threshold is set. If not, the yaw system does not operate. If it is, the yaw deviation threshold under different yaw control modes is executed according to the wind speed range. The yaw motor execution module adjusts the yaw angle according to the yaw deviation threshold under different yaw control modes. The wind direction prediction and yaw control method of this wind turbine guides the operation of the yaw system through accurate wind direction prediction, so as to improve the wind alignment accuracy of the yaw system and reduce the yaw waiting time, and achieve rapid wind alignment. By re-dividing the yaw control parameters and increasing the yaw deviation threshold and yaw time threshold in the low wind speed range, the problem of severe wear of the mechanical components of the yaw system caused by frequent start-stop of the yaw system in the low wind speed area is avoided. Attached Figure Description

[0034] Figure 1This is an overall flowchart of the wind direction prediction and yaw control method for wind turbines of the present invention;

[0035] Figure 2 This is a system connection diagram of the wind direction prediction and yaw control method for wind turbines of the present invention;

[0036] Figure 3 This is a flowchart of the data decomposition strategy for the wind direction prediction and yaw control method of the wind turbine of the present invention;

[0037] Figure 4 This is a flowchart of the wind direction prediction strategy of the wind direction prediction and yaw control method for wind turbines of the present invention. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the overall implementation scheme, embodiments, and accompanying drawings. It should be understood that the overall implementation scheme and specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0039] The implementation of the present invention will be described in detail below with reference to the overall implementation scheme.

[0040] Implementation Case 1

[0041] Please see Figure 1 and Figure 2 , Figure 1 This is an overall flowchart of the wind direction prediction and yaw control method for wind turbines of the present invention. Figure 2 This is a system connection diagram of the wind direction prediction and yaw control method for wind turbines of the present invention;

[0042] This embodiment discloses a wind turbine wind direction prediction and yaw control method, providing a wind direction detector 10, a data acquisition module 20, a data prediction module 30, a wind turbine yaw module 40, and a yaw motor execution module 50. The method includes the following steps: the wind direction detector 10 acquires the wind direction of each wind turbine in the wind farm in real time; the data acquisition module 20 acquires the wind direction data of the wind direction detector 10 in real time and sends it to the data prediction module 30; the data prediction module 30 decomposes the wind direction data according to the data decomposition strategy to obtain the sequence data of the target intrinsic mode components (IMFs) of different frequencies, and divides the sequence data of each target IMF and the residual component (Res) into high-frequency component IMF, low-frequency component IMF, and residual component Res by the energy-correlation coefficient method. The wind direction prediction result is output by superimposing and reconstructing the prediction results of each component sequence data. The wind turbine yaw module 40 calculates whether the angle α between the average wind direction and the nacelle axis during the yaw threshold time period is greater than the minimum value of the yaw deviation threshold based on the prediction results of the wind direction prediction module 30. If not, the yaw system does not operate. If so, the yaw deviation threshold under different yaw control modes is executed according to the wind speed range. The yaw motor execution module 50 adjusts the yaw angle according to the yaw deviation threshold under different yaw control modes.

[0043] Example 2

[0044] Please see Figure 3 , Figure 3 This is a flowchart of the data decomposition strategy for the wind direction prediction and yaw control method of the wind turbine of the present invention;

[0045] This embodiment is based on Embodiment 1. In this embodiment, the data decomposition strategy is as follows: Obtain a set of wind direction data x(t) over a period of time; add Gaussian white noise of the same length that follows a normal distribution to the wind direction data x(t), and normalize the Gaussian white noise; then decompose using the empirical mode decomposition method to obtain intrinsic mode components (IMFs); repeat the above steps continuously, adding different normally distributed Gaussian white noise each time; average all the corresponding IMFs obtained each time to obtain a target intrinsic mode component; obtain multiple target intrinsic mode components (IMFs) through the original wind direction data sequence x(t), forming a signal intrinsic mode combination of the wind direction data sequence x(t):

[0046]

[0047] The first few groups of target intrinsic mode components (IMFs) with higher energy and correlation coefficients are classified as high-frequency IMFs, while the last few groups of target intrinsic mode components (IMFs) with lower energy and correlation coefficients are classified as low-frequency IMFs.

[0048] Where x(t) is the original wind direction data sequence, c i (t) represents the i-th intrinsic mode component (IMF), r n The remaining component of Res.

[0049] The number of decompositions required during the decomposition process In the formula, N is the number of decompositions, e is the standard deviation of the added white noise, and ε is the allowable decomposition error, which is taken as 2%. The deviation value of Gaussian noise is set to 0.2.

[0050] Example 3

[0051] Please see Figure 4 , Figure 4 This is a flowchart of the wind direction prediction strategy of the wind direction prediction and yaw control method for wind turbines of the present invention.

[0052] This embodiment is based on Embodiment 2. In this embodiment, the wind direction prediction of the sequence data of each target intrinsic mode component (IMF) includes: using a neural network strategy to predict the high-frequency component IMF to obtain the high-frequency component prediction value T1.

[0053] The low-frequency component IMF and the residual component Res are predicted using a least-squares support vector machine model to obtain the predicted values ​​of low-frequency component T2 and residual component T3. The predicted values ​​T1, T2, and T3 are then superimposed to reconstruct the final predicted value.

[0054] The neural network used is a backpropagation neural network (BPNN). The backpropagation neural network consists of an input layer, hidden layers, and an output layer. The number of nodes in the hidden layer is determined by the number of nodes in the input layer and the number of nodes in the output layer. The number of nodes in the output layer is m, and the number of nodes in the input layer is n. Therefore, the number of nodes in the hidden layer is: Where 'a' is an integer between 0 and 10, the number of hidden layers is one or more, and the transfer formula from the input layer to the hidden layer is: The transfer formula from the hidden layer to the output layer is as follows: The activation function of the hidden layer is: w ij w represents the connection weights between neurons in the input layer and the hidden layer. jk a represents the connection weights between neurons in the hidden layer and the output layer. j b is the threshold for the hidden layer. k Let Q be the output layer threshold. k Let x be the predicted value of the high-frequency component. i Let i be the high-frequency component IMF, i = 1, 2, ..., n.

[0055] The calculation formula for the least squares support vector machine model is as follows:

[0056]

[0057] Where: α i Let K(x) represent the Lagrange multipliers of the corresponding components. i x j ) represents the symmetric matrix of the kernel function of the least squares support vector machine model, b represents the bias, f(x) is the predicted value of the low-frequency component IMF and the residual component, x is the low-frequency component IMF and the Res residual component, and N is the number of data in the low-frequency component IMF and the Res residual component.

[0058] Wherein: the specific expression of the kernel function is:

[0059]

[0060] σ is the kernel parameter, and the value range of the kernel parameter is set to 0.01 to 100.

[0061] Example 4

[0062] This embodiment is based on Embodiment 1. In this embodiment, the range of wind speed is divided into: cut-in wind speed, peak wind speed, rated wind speed and cut-out wind speed.

[0063] The phrase "execute yaw deviation thresholds under different yaw control modes according to the range of wind speed" specifically includes: first round of yaw deviation threshold selection: determining whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the cut-in wind speed and less than the peak wind speed;

[0064] If so, the wind turbine yaw module controls the execution of the yaw deviation threshold in the low wind speed control mode.

[0065] If not, proceed to the second round of yaw deviation threshold selection;

[0066] The second round of yaw deviation threshold selection involves determining whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the peak wind speed but less than the rated wind speed. If the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the peak wind speed but less than the rated wind speed, then the wind turbine yaw module controls the yaw deviation threshold in the wind speed control mode.

[0067] If not, proceed to the third round of yaw deviation threshold selection;

[0068] The third round of yaw deviation threshold selection: It determines whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the rated wind speed and less than the cut-out wind speed. If the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the rated wind speed and less than the cut-out wind speed, then the wind turbine yaw module controls the execution of the yaw deviation threshold in the high wind speed control mode.

[0069] If not, proceed to the fourth round of yaw deviation threshold selection;

[0070] Fourth round of yaw deviation threshold selection: Determine whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the cut-out wind speed.

[0071] If so, then enter shutdown mode.

[0072] In this embodiment, the yaw deviation threshold includes a yaw error threshold and a yaw error duration threshold.

[0073] Example 5

[0074] This embodiment discloses a control system for wind direction prediction and yaw control of wind turbine generators. The control system includes a wind direction detector 10, a data acquisition module 20, a data prediction module 30, a wind turbine yaw module 40, and a yaw motor execution module 50. The wind direction detector acquires the wind direction of each wind turbine generator in the wind farm in real time. The data acquisition module 20 acquires the wind direction data from the wind direction detector 10 in real time and sends it to the data prediction module 30. The data prediction module 30 decomposes the wind direction data according to a data decomposition strategy to obtain target intrinsic mode components (IMFs) of different frequencies. The sequence data is used to predict the wind direction of each target intrinsic mode component (IMF) data. The prediction results are superimposed and reconstructed to output the wind direction prediction result. The wind turbine yaw module calculates whether the angle α between the average wind direction and the nacelle axis during the yaw threshold time period is greater than the minimum value of the yaw deviation threshold based on the wind direction prediction result. If not, the yaw system does not operate. If so, the yaw deviation threshold under different yaw control modes is executed according to the wind speed range. The yaw motor execution module adjusts the yaw angle according to the yaw deviation threshold under different yaw control modes.

[0075] The wind direction prediction and yaw control method of this wind turbine guides the yaw system's actions by accurately predicting the wind direction, thereby improving the yaw system's wind alignment accuracy and reducing yaw waiting time, achieving rapid wind alignment. By reclassifying the yaw control parameters and increasing the yaw deviation threshold and yaw time threshold in the low wind speed range, the problem of severe wear on the yaw system's mechanical components caused by frequent start-stop operations in low wind speed areas is avoided.

[0076] It should be understood that the above are merely preferred embodiments of the present invention and should not be construed as limiting the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A wind direction prediction and yaw control method for a wind turbine, characterized in that, The method provides a wind direction detector, a data acquisition module, a data prediction module, a wind turbine yaw module, and a yaw motor execution module. The method includes the following steps: the wind direction detector acquires the wind direction of each wind turbine in the wind farm in real time; the data acquisition module acquires the wind direction data from the wind direction detector in real time and sends it to the data prediction module; the data prediction module decomposes the wind direction data into target intrinsic mode components (IMF) sequence data of different frequencies according to a data decomposition strategy, predicts each target IMF sequence data, and superimposes and reconstructs the prediction results to output the wind direction prediction result; the wind turbine yaw module calculates whether the angle α between the average wind direction and the nacelle axis during the yaw threshold time period is greater than the minimum value of the yaw deviation threshold based on the wind direction prediction result; if not, the yaw system does not operate; if so, it executes the yaw deviation threshold under different yaw control modes according to the wind speed range; the yaw motor execution module adjusts the yaw angle according to the yaw deviation threshold under different yaw control modes. The data decomposition strategy is as follows: Obtain a set of wind direction data x(t) over a period of time; add Gaussian white noise of the same length that follows a normal distribution to the wind direction data x(t), and normalize the Gaussian white noise; then decompose using the empirical mode decomposition method to obtain intrinsic mode components (IMFs); repeat the above steps continuously, adding different normally distributed Gaussian white noise each time; average all the corresponding IMFs obtained each time to obtain a target intrinsic mode component; obtain multiple target intrinsic mode components (IMFs) from the original wind direction data sequence x(t), forming a signal intrinsic mode combination of the wind direction data sequence x(t). ， The first few groups of target intrinsic mode components (IMFs) with higher energy and correlation coefficients are classified as high-frequency IMFs, while the last few groups of target intrinsic mode components (IMFs) with lower energy and correlation coefficients are classified as low-frequency IMFs. in, This is the original wind direction data sequence. For the first Each intrinsic mode component (IMF) The remaining component of Res.

2. The wind direction prediction and yaw control method of a wind turbine generator unit according to claim 1, characterized by, Wind direction prediction based on the sequence data of the intrinsic mode components (IMFs) of each target includes: using a neural network strategy to predict the high-frequency component IMFs to obtain the predicted values ​​of the high-frequency components; The least squares support vector machine model is used to predict the low-frequency component IMF and Res residual component to obtain the predicted values ​​of the low-frequency component IMF and Res residual component.

3. The wind direction prediction and yaw control method for wind turbine units according to claim 2, characterized in that, The neural network strategy includes employing a backpropagation neural network, which consists of an input layer, hidden layers, and an output layer. The number of nodes in the hidden layer is determined by the number of nodes in the input layer and the number of nodes in the output layer. The number of nodes in the output layer is... The number of input layer nodes is The number of hidden layer nodes is obtained: ;in, The integer is between 0 and 10, the number of hidden layers is one or more, and the transfer formula from the input layer to the hidden layer is: The transfer formula from the hidden layer to the output layer is as follows: ,j=1,2,…,l,k=1,2,…,m, for: , These are the connection weights between neurons in the input layer and the hidden layer. These are the connection weights between neurons in the hidden layer and the output layer. This is the threshold for the hidden layer. Let the output layer threshold be denoted as... The predicted value of the high-frequency component. Let i be the high-frequency component IMF, i = 1, 2, ..., n.

4. The wind direction prediction and yaw control method for wind turbine units according to claim 2, characterized in that, The calculation formula for the least squares support vector machine model is as follows: ， in: Denotes the Lagrange multipliers of the corresponding components. This represents the symmetric matrix of the kernel function in the least squares support vector machine model. Indicates deviation, These are the predicted values ​​for the low-frequency component IMF and the residual component Res. The low-frequency component IMF and the remaining Res component are defined as follows, where N is the number of data points in the low-frequency component IMF and the remaining Res component. Wherein: the specific expression of the kernel function is: ， is a nuclear parameter.

5. The wind direction prediction and yaw control method for wind turbine units according to claim 1, characterized in that, The wind direction detector is used to acquire the wind direction at each wind turbine in the wind farm in real time. It is installed at a suitable position on the wind turbine. The wind direction detector is communicatively connected to the data acquisition module. The data acquisition module is used to acquire wind direction data from the wind direction detector and store the wind direction data. The data prediction module is communicatively connected to the data acquisition module. The wind turbine yaw module is communicatively connected to the data prediction module. The wind turbine yaw module is communicatively connected to the yaw motor execution module.

6. The wind direction prediction and yaw control method for wind turbine units according to claim 1, characterized in that, The wind speed range is divided into: cut-in wind speed, peak wind speed, rated wind speed, and cut-out wind speed; The "execution of yaw deviation thresholds under different yaw control modes according to the range of wind speed" specifically includes: the first round of yaw deviation threshold selection: determining whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the cut-in wind speed and less than the peak wind speed; If so, the wind turbine yaw module controls the execution of the yaw deviation threshold in the low wind speed control mode. If not, proceed to the second round of yaw deviation threshold selection; The second round of yaw deviation threshold selection: It is determined whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the peak wind speed but less than the rated wind speed; if the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the peak wind speed but less than the rated wind speed, then the wind turbine yaw module controls the yaw deviation threshold in the wind speed control mode. If not, proceed to the third round of yaw deviation threshold selection; The third round of yaw deviation threshold selection: It is determined whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the rated wind speed and less than the cut-out wind speed; if the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the rated wind speed and less than the cut-out wind speed, then the wind turbine yaw module controls the execution of the yaw deviation threshold in the high wind speed control mode. If not, proceed to the fourth round of yaw deviation threshold selection; Fourth round of yaw deviation threshold selection: Determine whether the wind speed corresponding to the wind direction predicted by the data prediction module is greater than the cut-out wind speed. If so, then enter shutdown mode.

7. The wind direction prediction and yaw control method of a wind turbine generator unit according to claim 6, characterized by, The yaw deviation threshold includes a yaw error threshold and a yaw error duration threshold.

8. A control system applied to the wind direction prediction and yaw control method of claim 1, characterized by, The control system includes a wind direction detector, a data acquisition module, a data prediction module, a wind turbine yaw module, and a yaw motor execution module. The wind direction detector acquires the wind direction of each wind turbine in the wind farm in real time. The data acquisition module acquires the wind direction data from the wind direction detector in real time and sends it to the data prediction module. The data prediction module decomposes the wind direction data into sequence data of target intrinsic mode components (IMFs) at different frequencies according to a data decomposition strategy. It then predicts the wind direction for each IMF, superimposes and reconstructs the prediction results, and outputs the wind direction prediction result. The wind turbine yaw module uses the wind direction prediction result to calculate whether the angle α between the average wind direction and the nacelle axis during the yaw threshold time period is greater than the minimum value of the yaw deviation threshold. If not, the yaw system does not operate. If so, it executes the yaw deviation threshold under different yaw control modes according to the wind speed range. The yaw motor execution module adjusts the yaw angle according to the yaw deviation threshold under different yaw control modes. The data decomposition strategy is as follows: Obtain a set of wind direction data x(t) over a period of time; add Gaussian white noise of the same length that follows a normal distribution to the wind direction data x(t), and normalize the Gaussian white noise; then decompose using the empirical mode decomposition method to obtain intrinsic mode components (IMFs); repeat the above steps continuously, adding different normally distributed Gaussian white noise each time; average all the corresponding IMFs obtained each time to obtain a target intrinsic mode component; obtain multiple target intrinsic mode components (IMFs) from the original wind direction data sequence x(t), forming a signal intrinsic mode combination of the wind direction data sequence x(t). ， The first few groups of target intrinsic mode components (IMFs) with higher energy and correlation coefficients are classified as high-frequency IMFs, while the last few groups of target intrinsic mode components (IMFs) with lower energy and correlation coefficients are classified as low-frequency IMFs. in, This is the original wind direction data sequence. For the first Each intrinsic mode component (IMF) The remaining component of Res.

Images