A wind turbine control method, device, equipment and storage medium

Abstract

This invention discloses a wind turbine control method, device, equipment, and storage medium, applied in the field of wind power technology. It uses an unscented Kalman filter to determine the aerodynamic thrust of the wind turbine based on wind speed, generator speed, pitch angle, and blade root bending moment. Based on the aerodynamic thrust, it determines the critical state variables of the wind turbine and judges whether these critical state variables exceed control thresholds. When none of the critical state variables exceed the control thresholds, the pitch angle limit is determined as the initial default limit value. When any critical state variable exceeds the control threshold, machine learning is used to determine the pitch angle limit value. Based on the pitch angle limit value and the pitch controller output value, a pitch angle control command is generated, and the wind turbine pitch is controlled based on the pitch angle control command. This invention, based on an unscented Kalman filter and machine learning, determines the pitch angle limit value according to critical state variables, improving the safety and economy of wind turbine operation.

Description

Technical Field

[0001] This invention relates to the field of wind power technology, and in particular to a wind turbine control method, a wind turbine control device, an electronic device, and a computer-readable storage medium. Background Technology

[0002] Wind turbines are trending towards larger sizes with longer blades and taller towers, and users are increasingly demanding higher standards for operational safety and economy, posing challenges to wind turbine control. Specifically, in pitch control, conventional methods rely on power lookup tables to set pitch angle limits, thereby controlling the aerodynamic thrust of the turbine under high wind conditions and reducing load. However, this method is indirect, lacking fine-grained adjustment and exhibiting response lag. Therefore, improving pitch control methods to address these issues is beneficial for enhancing the safety and economy of wind turbines. Summary of the Invention

[0003] The purpose of this invention is to provide a wind turbine control method, device, equipment, and storage medium, which are applied in the field of wind power technology. The method determines the aerodynamic thrust of the wind turbine in real time through an unscented Kalman filter, determines key state variables based on the aerodynamic thrust, determines pitch angle limit values ​​based on the key state variables, and performs pitch control of the wind turbine based on the pitch angle limit values, thereby improving the safety and economy of wind turbine operation.

[0004] To solve the above-mentioned technical problems, the present invention provides a wind turbine control method, comprising:

[0005] The aerodynamic thrust of the wind turbine is determined by an unscented Kalman filter based on wind speed, generator speed, blade pitch angle and blade root bending moment.

[0006] Based on the aerodynamic thrust, the key state variables of the wind turbine are determined, and it is determined whether the key state variables exceed the control threshold.

[0007] When none of the key state variables exceed the control threshold, the pitch angle limit value is determined as the initial default limit value;

[0008] When a critical state quantity exceeds the control threshold, the pitch angle limit value is determined based on the critical state quantity.

[0009] Based on the pitch angle limit value and the pitch controller output value, a pitch angle control command is generated, and the wind turbine pitch is controlled based on the pitch angle control command.

[0010] Optionally, the aerodynamic thrust of the wind turbine can be determined using an unscented Kalman filter based on wind speed, generator speed, blade pitch angle, and blade root bending moment, including:

[0011] The state equation of the unscented Kalman filter is constructed; the inputs of the state equation are the wind speed, the generator speed and the blade pitch angle, and the state variable of the state equation is the aerodynamic thrust.

[0012] Generate state variable sigma points, update the state variable sigma points based on the state equation, and determine the prior state mean and prior state covariance based on the updated state variable sigma points.

[0013] Construct the observation equations for the unscented Kalman filter; the observations in the observation equations are the leaf root bending moments.

[0014] Update the observation sigma points based on the observation equation; determine the predicted mean and predicted covariance of the observations based on the updated observation sigma points;

[0015] The state-observation covariance is determined based on the prior state mean and the observation prediction mean; the Kalman gain is determined based on the state-observation covariance and the observation prediction covariance.

[0016] Based on the Kalman gain, the posterior state mean and posterior state covariance are determined, and the posterior state mean is determined as the output aerodynamic thrust.

[0017] Optionally, determining the pitch angle limit value based on the key state variables includes:

[0018] The proxy model is trained using simulation data or experimental data based on the key state variables and the pitch angle limit values;

[0019] The key state variables are input into the trained surrogate model to obtain the pitch angle limit value output by the model.

[0020] Optionally, a pitch angle control command is generated based on the pitch angle limit value and the pitch controller output value, including:

[0021] When the output value of the pitch controller is less than the pitch angle limit value, the pitch angle control command is generated based on the pitch angle limit value so that the wind turbine controls the pitch angle to the pitch angle limit value when executing the pitch angle control command.

[0022] When the pitch controller output value is not less than the pitch angle limit value, the pitch angle control command is generated based on the pitch controller output value, so that when the wind turbine executes the pitch angle control command, the pitch angle is controlled to be the pitch controller output value.

[0023] Optionally, the key state variables of the wind turbine are determined based on the aerodynamic thrust, including:

[0024] Determine the dynamic model for each of the key state variables, and input the aerodynamic thrust into the dynamic model to obtain the key state variables output by the model.

[0025] Optionally, the control threshold is determined based on safety indicators and economic indicators; the economic indicator is power generation and the safety indicator is safe load.

[0026] Optionally, the key state quantities include nacelle acceleration, hub load, yaw load, tower load, and power generation.

[0027] To solve the above-mentioned technical problems, the present invention provides a wind turbine control device, comprising:

[0028] The first module is used to determine the aerodynamic thrust of the wind turbine based on wind speed, generator speed, blade pitch angle and blade root bending moment using an unscented Kalman filter.

[0029] The second module is used to determine the key state variables of the wind turbine based on the aerodynamic thrust, and to determine whether the key state variables exceed the control threshold.

[0030] The third module is used to determine the pitch angle limit value as the initial default limit value when none of the key state quantities exceed the control threshold.

[0031] The fourth module is used to determine the pitch angle limit value based on the critical state quantity when the critical state quantity exceeds the control threshold.

[0032] The fifth module is used to generate a pitch angle control command based on the pitch angle limit value and the output value of the pitch controller, and to control the pitch of the wind turbine based on the pitch angle control command.

[0033] To solve the above-mentioned technical problems, the present invention provides an electronic device, comprising:

[0034] Memory, used to store computer programs;

[0035] A processor is used to implement the wind turbine control method described above when executing the computer program.

[0036] To address the aforementioned technical problems, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the aforementioned wind turbine control method.

[0037] As can be seen, this invention uses an unscented Kalman filter to determine the aerodynamic thrust of the wind turbine based on wind speed, generator speed, pitch angle, and blade root bending moment; it then determines the critical state variables of the wind turbine based on the aerodynamic thrust and judges whether these critical state variables exceed control thresholds; when none of the critical state variables exceed the control thresholds, the pitch angle limit value is determined as the initial default limit value; when any critical state variable exceeds the control threshold, the pitch angle limit value is determined based on the critical state variable; a pitch angle control command is generated based on the pitch angle limit value and the pitch controller output value, and the wind turbine pitch is controlled based on the pitch angle control command. This invention, based on an unscented Kalman filter and machine learning, determines the pitch angle limit value according to critical state variables, improving the safety and economy of wind turbine operation. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0039] Figure 1 A flowchart of a wind turbine control method provided in an embodiment of the present invention;

[0040] Figure 2 This is a wind turbine pitch control logic diagram provided in an embodiment of the present invention;

[0041] Figure 3 This is a structural block diagram of a wind turbine control device provided in an embodiment of the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Wind turbines are trending towards larger sizes with longer blades and taller towers, and users are increasingly demanding higher standards for operational safety and economy, posing challenges to wind turbine control. Specifically, in pitch control, conventional methods rely on power lookup tables to set pitch angle limits, thereby controlling the aerodynamic thrust of the turbine under high wind conditions and reducing load. However, this method is indirect, lacking fine-grained adjustment and exhibiting response lag. Therefore, improving pitch control methods to address these issues is beneficial for enhancing the safety and economy of wind turbines.

[0044] Wind turbine generators are nonlinear systems. Traditional Kalman filters are only suitable for linear systems and not for state estimation of nonlinear systems. Extended Kalman filters use linearization functions to approximate nonlinear systems, which inevitably introduces errors. In contrast, unscented Kalman filters directly approximate the probability distribution after nonlinear transformation, resulting in more accurate state estimation for nonlinear systems.

[0045] Therefore, this invention determines the aerodynamic thrust of the wind turbine in real time based on an unscented Kalman filter, determines the key state variables of the turbine based on the aerodynamic thrust, and determines the pitch angle limit value based on the key state variables. This method is more direct than conventional methods, and can achieve rapid and precise adjustment, thereby improving the safety and economy of wind turbine operation.

[0046] The following combination Figure 1 , Figure 1 A flowchart of a wind turbine control method provided in an embodiment of the present invention, the method may include:

[0047] S101: The aerodynamic thrust of the wind turbine is determined by an unscented Kalman filter based on wind speed, generator speed, blade pitch angle and blade root bending moment.

[0048] This embodiment can determine the aerodynamic thrust of a wind turbine using an unscented Kalman filter based on wind speed, generator speed, pitch angle, and blade root moment. This embodiment does not limit the specific methods for obtaining wind speed, generator speed, pitch angle, and blade root moment; generally, these parameters can be measured in real time using sensors. Specifically, wind speed can be measured using anemometers, lidar, etc.; generator speed can be measured using photoelectric sensors, encoders, etc.; pitch angle can be measured using encoders, etc.; and blade root moment can be measured using strain gauge sensors, fiber optic grating sensors, cantilever beam sensors, etc.

[0049] This embodiment does not limit the specific method of determining the aerodynamic thrust of a wind turbine using an unscented Kalman filter based on wind speed, generator speed, blade pitch angle, and blade root bending moment. The unscented Kalman filter is a nonlinear estimation algorithm that approximates the state distribution by weighting sampling points (sigma points) and uses the Kalman filter framework to achieve state estimation.

[0050] First, the state equations of the unscented Kalman filter are constructed. The inputs to the state equations are wind speed, generator speed, and blade pitch angle, and the state variable is aerodynamic thrust, as shown in the following equation:

[0051] ;

[0052] ;

[0053] ;

[0054] ;

[0055] In the formula, x is the state variable, which is the aerodynamic thrust D in this state equation; u is the input variable, including wind speed v, wind turbine speed ω and blade pitch angle j; w is the process noise, which usually satisfies the Gaussian distribution N(0,Q), and Q is the process noise covariance matrix; f is the nonlinear function reflecting the state transition, with subscript k representing the time step and superscript T representing the matrix transpose.

[0056] Subsequently, state variables sigma points are generated, and the state variables sigma points are updated based on the state equations. The prior state mean and prior state covariance are then determined based on the updated state variables sigma points.

[0057] The sigma point that generates the state variables:

[0058] ;

[0059] ,i takes 1~n;

[0060] i can be n+1 to 2n;

[0061] In the formula, Let sigma be a state variable. There are 2n+1 such sigma points for an n-dimensional state variable. The superscript (i) represents the index of the sigma point. Let P be the mean of the state variables, P be the covariance of the state variables, and λ be the scaling parameter; operators Let represent the i-th column of the matrix obtained by performing Cholesky decomposition on matrix E. Cholesky decomposition is a matrix decomposition method that decomposes a symmetric positive definite matrix into the product of a lower triangular matrix and its transpose.

[0062] Update the sigma point of the state variables using the state equation:

[0063] ;

[0064] In the formula, The updated state variable sigma point, with the subscript k|k-1 representing the k-time step data predicted at time step k-1.

[0065] Calculate the prior state mean and covariance:

[0066] ;

[0067] ;

[0068] In the formula, The mean of the prior states, For the prior state covariance, W m and W c These are the weighting coefficients.

[0069] The formula for calculating the weighting coefficient is as follows:

[0070] ;

[0071] ;

[0072] i can be 1 to 2n;

[0073] ;

[0074] In the formula, λ is the scaling parameter, representing the degree of diffusion of the sigma points. For a Gaussian distribution, α is the first candidate parameter, used to control the distribution state of the sampling points, and β is the second candidate parameter, a non-negative weighting coefficient that can be used to combine the momentum of higher-order terms in the equation. The third parameter is used to ensure that the matrix (n+λ)P is a positive semi-definite matrix. For a Gaussian distribution, α=0.001 and β=2 are generally chosen. =3-n.

[0075] Furthermore, the observation sigma point is updated using the observation equation. The observation equation for the unscented Kalman filter is constructed; the observation quantity in the observation equation is the leaf root bending moment. The observation equation for the unscented Kalman filter is shown in the following equation:

[0076] ;

[0077] ;

[0078] ;

[0079] In the formula, z represents the observation, which is the leaf root bending moment M in this observation equation, and the subscript k represents the time step; v is the measurement noise, which usually follows a Gaussian distribution N(0,R), R is the measurement noise covariance matrix, and h is the nonlinear observation function.

[0080] The observation sigma points are updated based on the observation equation, which can be obtained from the observation equation:

[0081] ;

[0082] In the formula, The updated observation sigma points.

[0083] Determine the predicted mean and predicted covariance of the observations based on the updated observation sigma points:

[0084] ;

[0085] ;

[0086] In the formula, The predicted mean of the observed values, To predict the covariance of the observations.

[0087] The state-observation covariance is determined based on the prior state mean and the observation prediction mean; the Kalman gain is determined based on the state-observation covariance and the observation prediction covariance.

[0088] Calculate the state-observation covariance:

[0089] ;

[0090] In the formula, This is the state-observation covariance.

[0091] Calculate the Kalman gain:

[0092] ;

[0093] In the formula, K k For Kalman gain.

[0094] Finally, based on the Kalman gain, the posterior state mean and posterior state covariance are determined, and the posterior state mean is determined as the output aerodynamic thrust.

[0095] Update post-verification status:

[0096] ;

[0097] ;

[0098] In the formula, Let z be the posterior state mean. 0k P is the actual measurement value at time step k. k Let be the posterior state covariance.

[0099] This embodiment can be taken The aerodynamic thrust output by the unscented Kalman filter at time step k is given, and... and P k Using the k+1 time step as the initial value, the above iterative process is repeated to achieve real-time calculation of aerodynamic thrust.

[0100] S102: Determine the key state variables of the wind turbine based on aerodynamic thrust, and determine whether the key state variables exceed the control threshold.

[0101] This embodiment can determine the key state variables of a wind turbine based on aerodynamic thrust. This embodiment does not limit the number or type of key state variables, and the specific settings can be based on actual applications. For example, the key state variables of a wind turbine cover various state variables that affect the operational safety and economy of the wind turbine, including but not limited to nacelle acceleration, hub load, yaw load, tower load, and power generation.

[0102] This embodiment does not limit the specific method of determining the key state variables of the wind turbine based on aerodynamic thrust. Generally, the dynamic model of each key state variable can be determined by deriving from the dynamic model, and the key state variables output by the model are obtained by inputting the aerodynamic thrust into the dynamic model.

[0103] For example, the formula for calculating the fore-and-aft acceleration of the cabin is as follows:

[0104] ;

[0105] In the formula, a nac Let D be the fore-and-aft acceleration of the nacelle, and k be the aerodynamic thrust. tow For the equivalent stiffness at the top of the tower, ξ tow For the equivalent damping at the top of the tower, x nac For the forward and backward displacement of the cabin, v nac Forward and backward speed of the cabin, m nac For cabin mass.

[0106] Furthermore, this embodiment can determine whether a critical state quantity exceeds a control threshold. Each critical state quantity corresponds to a control threshold, and the pitch angle limit value is determined by the exceedance of the critical state quantity and its corresponding control threshold.

[0107] This embodiment does not limit the method of determining the control threshold. It can generally be determined based on safety indicators and economic indicators. The economic indicator can be the power generation capacity, and the safety indicator can be the safe load.

[0108] S103: When none of the critical state variables exceed the control threshold, the pitch angle limit value is set as the initial default limit value.

[0109] S104: When a critical state variable exceeds the control threshold, determine the pitch angle limit value based on the critical state variable.

[0110] In this embodiment, an initial default limit value (e.g., 0 degrees) can be set. When the key state variables do not exceed the control threshold, the pitch angle limit value is determined as the initial default limit value.

[0111] In this embodiment, when a critical state variable exceeds the control threshold, a pitch angle limit value can be determined based on the critical state variable. This embodiment does not limit the specific method of determining the pitch angle limit value based on the critical state variable; it can be set according to the actual application. Generally, it can be determined through a surrogate model. The surrogate model is a mathematical model that reflects the correspondence between the critical state variable and the pitch angle limit value. The types of models include, but are not limited to, Kriging models, neural network models, etc., and the model parameters are obtained through machine learning methods.

[0112] Specifically, a surrogate model is trained using machine learning based on simulation or experimental data of key state variables and pitch angle limits. The surrogate model can include, but is not limited to, Kriging models and neural network models. The key state variables are input into the trained surrogate model to obtain the output pitch angle limits. For example, the basic mathematical expression of a Kriging model is:

[0113] ;

[0114] In the formula, X is the input critical state variable, and H... T Let t(X) be the regression function, θ be the regression coefficient, and t(X) be a random function reflecting the uncertainty of the model, satisfying a normal distribution N(0, σ). 2 ), where σ is the standard deviation.

[0115] In other words, the key state variables of the wind turbine calculated in real time are compared with the corresponding control thresholds. If none of the key state variables exceed the control thresholds, the pitch angle limit is set to the initial default limit; if any state variable exceeds the control threshold, the pitch angle limit is adjusted in real time according to the correspondence between the key state variables and the pitch angle limit.

[0116] S105: Generates pitch angle control commands based on pitch angle limit values ​​and pitch controller output values, and controls the pitch of the wind turbine based on the pitch angle control commands.

[0117] This embodiment can generate pitch angle control commands based on the pitch angle limit value and the output value of the pitch controller, and control the pitch of the wind turbine based on the pitch angle control commands. The pitch controller in this embodiment can be a PI (Proportion Integration) controller.

[0118] Specifically, when the pitch controller output value is less than the pitch angle limit value, a pitch angle control command can be generated based on the pitch angle limit value so that the wind turbine can control the pitch angle to the pitch angle limit value when executing the pitch angle control command.

[0119] When the pitch controller output value is not less than the pitch angle limit value, a pitch angle control command can be generated based on the pitch controller output value so that the wind turbine can control the pitch angle to the pitch controller output value when executing the pitch angle control command.

[0120] In this embodiment, the pitch control logic of the wind turbine can be as follows: Figure 2 As shown, the aerodynamic thrust is determined by an unscented Kalman filter based on the real-time collected wind speed, generator speed, pitch angle, and blade root bending moment; the key state variables are calculated in real time based on the aerodynamic thrust, and the pitch angle limit value is set by the relationship between the key state variables and the control threshold; the pitch angle control command is generated based on the pitch angle limit value and the pitch controller output value to perform pitch control.

[0121] Based on the above embodiments, the present invention determines the pitch angle limit value of the wind turbine in real time based on the unscented Kalman filter and machine learning, and performs pitch control based on the pitch angle limit value, thereby improving the safety and economy of wind turbine operation.

[0122] The following combination Figure 3 , Figure 3 This is a structural block diagram of a wind turbine control device provided in an embodiment of the present invention. The device may include:

[0123] The first module 100 is used to determine the aerodynamic thrust of the wind turbine based on wind speed, generator speed, blade pitch angle and blade root bending moment using an unscented Kalman filter.

[0124] The second module 200 is used to determine the key state variables of the wind turbine based on aerodynamic thrust and to determine whether the key state variables exceed the control threshold.

[0125] The third module 300 is used to determine the pitch angle limit value as the initial default limit value when none of the key state variables exceed the control threshold.

[0126] The fourth module 400 is used to determine the pitch angle limit value based on the critical state quantity when a critical state quantity exceeds the control threshold.

[0127] The fifth module 500 is used to generate pitch angle control commands based on the pitch angle limit value and the pitch controller output value, and to control the pitch of the wind turbine based on the pitch angle control commands.

[0128] Based on the above embodiments, the present invention determines the pitch angle limit value of the wind turbine in real time based on the unscented Kalman filter and machine learning, and performs pitch control based on the pitch angle limit value, thereby improving the safety and economy of wind turbine operation.

[0129] Based on the above embodiments, the first module 100 may include:

[0130] The first unit is used to construct the state equations of the unscented Kalman filter; the inputs to the state equations are wind speed, generator speed and blade pitch angle, and the state variable of the state equations is aerodynamic thrust.

[0131] The second unit is used to generate the state variable sigma point, update the state variable sigma point based on the state equation, and determine the prior state mean and prior state covariance based on the updated state variable sigma point.

[0132] The third unit is used to construct the observation equations for the unscented Kalman filter; the observations in the observation equations are the leaf root bending moments.

[0133] The fourth unit is used to update the observation sigma points based on the observation equation; and to determine the predicted mean and predicted covariance of the observations based on the updated observation sigma points.

[0134] The fifth unit is used to determine the state-observation covariance based on the prior state mean and the observation prediction mean; and to determine the Kalman gain based on the state-observation covariance and the observation prediction covariance.

[0135] The sixth unit is used to determine the posterior state mean and posterior state covariance based on the Kalman gain, and to determine the posterior state mean as the output aerodynamic thrust.

[0136] Based on the above embodiments, the fourth module 400 may include:

[0137] Unit 7 is used to train surrogate models using machine learning based on simulation or experimental data of key state variables and pitch angle limits.

[0138] The eighth unit is used to input key state variables into the trained surrogate model to obtain the pitch angle limit value output by the model.

[0139] Based on the above embodiments, the fifth module 500 may include:

[0140] The ninth unit is used to generate a pitch angle control command based on the pitch angle limit value when the output value of the pitch controller is less than the pitch angle limit value, so that the wind turbine can control the pitch angle to the pitch angle limit value when executing the pitch angle control command.

[0141] The tenth unit is used to generate the pitch angle control command based on the pitch controller output value when the pitch controller output value is not less than the pitch angle limit value, so that the wind turbine can control the pitch angle to the pitch controller output value when executing the pitch angle control command.

[0142] Based on the above embodiments, the second module 200 may include:

[0143] Unit 11 is used to determine the dynamic model for each key state variable. Aerodynamic thrust is input into the dynamic model to obtain the key state variables output by the model.

[0144] Based on the above embodiments, the control threshold is determined based on safety indicators and economic indicators; the economic indicator is power generation and the safety indicator is safe load.

[0145] Based on the above embodiments, key state quantities include nacelle acceleration, hub load, yaw load, tower load, and power generation.

[0146] Based on the above embodiments, the present invention also provides an electronic device, which may include a memory and a processor. The memory stores a computer program, and when the processor calls the computer program in the memory, it can implement the steps provided in the above embodiments. Of course, the device may also include various necessary network interfaces, a power supply, and other components.

[0147] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by an execution terminal or processor, can implement the method provided in the embodiments of the present invention; the storage medium may include various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0148] In this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, without necessarily requiring or implying any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A wind turbine control method, characterized in that, include: The aerodynamic thrust of the wind turbine is determined by an unscented Kalman filter based on wind speed, generator speed, blade pitch angle and blade root bending moment. Based on the aerodynamic thrust, the key state variables of the wind turbine are determined, and it is determined whether the key state variables exceed the control threshold. When none of the key state variables exceed the control threshold, the pitch angle limit value is determined as the initial default limit value; When a critical state quantity exceeds the control threshold, the pitch angle limit value is determined based on the critical state quantity. Based on the pitch angle limit value and the pitch controller output value, a pitch angle control command is generated, and the wind turbine pitch is controlled based on the pitch angle control command.

2. The wind turbine control method according to claim 1, characterized in that, The aerodynamic thrust of a wind turbine is determined using an unscented Kalman filter based on wind speed, generator speed, blade pitch angle, and blade root bending moment, including: The state equation of the unscented Kalman filter is constructed; the inputs of the state equation are the wind speed, the generator speed and the blade pitch angle, and the state variable of the state equation is the aerodynamic thrust. Generate state variable sigma points, update the state variable sigma points based on the state equation, and determine the prior state mean and prior state covariance based on the updated state variable sigma points. Construct the observation equations for the unscented Kalman filter; the observations in the observation equations are the leaf root bending moments. Update the observation sigma points based on the observation equation; determine the predicted mean and predicted covariance of the observations based on the updated observation sigma points; The state-observation covariance is determined based on the prior state mean and the observation prediction mean; the Kalman gain is determined based on the state-observation covariance and the observation prediction covariance. Based on the Kalman gain, the posterior state mean and posterior state covariance are determined, and the posterior state mean is determined as the output aerodynamic thrust.

3. The wind turbine control method according to claim 1, characterized in that, Determining the pitch angle limit value based on the key state variables includes: The proxy model is trained using simulation data or experimental data based on the key state variables and the pitch angle limit values; The key state variables are input into the trained surrogate model to obtain the pitch angle limit value output by the model.

4. The wind turbine control method according to claim 1, characterized in that, Based on the pitch angle limit value and the pitch controller output value, a pitch angle control command is generated, including: When the output value of the pitch controller is less than the pitch angle limit value, the pitch angle control command is generated based on the pitch angle limit value so that the wind turbine controls the pitch angle to the pitch angle limit value when executing the pitch angle control command. When the pitch controller output value is not less than the pitch angle limit value, the pitch angle control command is generated based on the pitch controller output value, so that when the wind turbine executes the pitch angle control command, the pitch angle is controlled to be the pitch controller output value.

5. The wind turbine control method according to claim 1, characterized in that, Based on the aerodynamic thrust, the key state variables of the wind turbine are determined, including: Determine the dynamic model for each of the key state variables, and input the aerodynamic thrust into the dynamic model to obtain the key state variables output by the model.

6. The wind turbine control method according to claim 1, characterized in that, The control threshold is determined based on safety and economic indicators; the economic indicator is power generation and the safety indicator is safe load.

7. The wind turbine control method according to claim 1, characterized in that, The key state parameters include nacelle acceleration, hub load, yaw load, tower load, and power generation.

8. A wind turbine control device, characterized in that, include: The first module is used to determine the aerodynamic thrust of the wind turbine based on wind speed, generator speed, blade pitch angle and blade root bending moment using an unscented Kalman filter. The second module is used to determine the key state variables of the wind turbine based on the aerodynamic thrust, and to determine whether the key state variables exceed the control threshold. The third module is used to determine the pitch angle limit value as the initial default limit value when none of the key state quantities exceed the control threshold. The fourth module is used to determine the pitch angle limit value based on the critical state quantity when the critical state quantity exceeds the control threshold. The fifth module is used to generate a pitch angle control command based on the pitch angle limit value and the output value of the pitch controller, and to control the pitch of the wind turbine based on the pitch angle control command.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the wind turbine control method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, implement the wind turbine control method as described in any one of claims 1 to 7.

Images