Wind power plant fatigue suppression active power control method based on data mechanism hybrid modeling

Abstract

The invention provides a wind power plant fatigue suppression active control method based on data mechanism hybrid modeling, and belongs to the technical field of wind power plants, and the method comprises the steps: building a continuous state space model, carrying out the discretization of the continuous state space model through a sampling period, and obtaining a discrete state space model, calculating a threshold value of a switching control strategy of a wind speed region based on the discrete state space model, and further obtaining a prediction model of main shaft torque and a prediction model of tower thrust, and recording the prediction models as wind turbine generator linear prediction models; an equivalent fatigue load linearization model is obtained through main shaft torque time sequence data, and after the equivalent fatigue load linearization model and a wind turbine generator linear prediction model are combined, a data mechanism dual-drive fatigue load model is obtained; and performing active optimization control solution on the data mechanism dual-drive fatigue load model to obtain an active power distribution optimal solution meeting the safety and stability constraints of the wind power plant.

Description

Technical Field

[0001] The invention belongs to the technical field of wind farms, and in particular relates to a wind farm fatigue suppression active power control method based on data mechanism hybrid modeling. Background Art

[0002] As the penetration rate of wind power continues to increase, the weak support and low inertia characteristics of wind power systems pose a severe challenge to the safety and stability of the power grid. Wind power access to the power grid requires inertia and primary frequency response capabilities, but in the process of wind farms participating in frequency regulation, the frequent fluctuations in output active power will increase the fatigue load of mechanical components such as the shaft system and tower of wind turbines, thereby leading to increased maintenance costs and reduced operating efficiency. At present, wind farms mainly respond to system frequency changes in a centralized manner, that is, the wind farm control system generates active power deviations according to the frequency response strategy and distributes them to the units in the station.

[0003] In the existing active power allocation method, the allocation strategy mainly depends on the available power ratio of each wind turbine, while ignoring the specific wind speed conditions and control inertia of the wind turbine location. This approach not only fails to effectively utilize the optimal allocation of resources in the wind farm, but may also further increase the wear of the turbines, thereby affecting the long-term operating efficiency and economy of the entire wind farm.

[0004] To this end, the present invention proposes a wind farm fatigue suppression active power control method based on data mechanism hybrid modeling. Summary of the invention

[0005] The present invention provides a wind farm fatigue suppression active power control method based on data mechanism hybrid modeling, so as to at least solve the problem that the existing active power allocation method cannot effectively utilize the optimal configuration of resources in the wind farm, aggravates the wear degree of the unit, and further affects the long-term operation efficiency and economy of the entire wind farm.

[0006] The embodiment of the present application provides a wind farm fatigue suppression active power control method based on data mechanism hybrid modeling, the method comprising: Step S1: establishing a continuous state space model, discretizing the continuous state space model using a sampling period to obtain a discrete state space model, and calculating the threshold of the switching control strategy of the wind speed region of the control system of the wind turbine generator system based on the discrete state space model; Step S2: based on the threshold of the switching control strategy of the wind speed area of ​​the control system of the wind turbine generator set, a prediction model of the main shaft torque and a prediction model of the tower thrust are obtained, which are recorded as the linear prediction model of the wind turbine generator set; Step S3: Obtain an equivalent fatigue load linearization model from the spindle torque time series data. After combining the equivalent fatigue load linearization model with the wind turbine linear prediction model, a fatigue load model driven by both data and mechanism is obtained. Step S4: Solve the active power optimization control for the fatigue load model driven by both data and mechanism to obtain the optimal solution of active power distribution that satisfies the safety and stability constraints of the wind farm.

[0007] Further, in Step S1, a continuous state space model is established, specifically including: Step S11: Establish a non - linear mechanism model for the wind turbine. Step S12: Based on the non - linear mechanism model of the wind turbine, construct a continuous state space model.

[0008] Further, in Step S11, establishing a non - linear mechanism model for the wind turbine includes: Step S111: Establish a derivative equation for the generator speed increment, and its expression is:

[0009] In the formula, represents the derivative of the generator speed increment, represents the gearbox transmission ratio, represents the equivalent mass combining the blade mass and the engine mass, represents the generator torque increment, represents the aerodynamic torque increment, represents at the generator torque measured at time represents at the aerodynamic torque measured at time Step S112: Establish a derivative equation for the filtered generator speed increment, and its expression is:

[0010] In the formula, represents the derivative of the filtered generator speed increment, represents the filter time constant, represents the filtered generator speed increment, represents the generator speed increment; Step S113: Define the increment to obtain:

[0011] In the formula, represents the pitch angle, represents the filtered generator speed, represents the rated value of the generator speed, represents the proportional gain of the fan PI controller, represents the integral gain of the fan PI controller, represents the Laplace operator; Step S114: Based on the derivative equation of the filtered generator speed increment established in step S112 and the increment obtained in step S113 , obtain the derivative equation of the increment :

[0012] (4) In the formula, represents the derivative of the increment ; Step S115: Perform approximate calculation according to the Taylor series at the operating point to obtain the aerodynamic torque increment and the generator torque increment; The expression of the aerodynamic torque increment is:

[0013] In the formula, represents the increment of the active power reference value of the th fan; The expression of the generator torque increment is: (6) In the formula, represents the partial derivative of the generator torque at the point with respect to the generator speed , and its expression is: (7) (8) In the formula, , represents the wind speed value at the blade at the moment , represents the blade, represents as a function of , represents the wind energy utilization coefficient; represents the partial derivative of the generator torque at the point with respect to the increment , and its expression is: (9) In the formula, represents the air density, Denote the blade; Step S116: Calculate and ; The expression of

[0014]

[0015] Denote the tip speed ratio, Denote the pitch angle, Denote the difference of adjacent units, Denote the difference of adjacent units, Denote the wind energy utilization coefficient, Denote the wind energy utilization coefficient at (n,m) in the look-up table, Denote the wind energy utilization coefficient at (n,m + 1) in the look-up table; The expression of (11) In the formula, Denote the wind energy utilization coefficient at (n + 1,m) in the look-up table.

[0016] Furthermore, in step S12, a continuous state space model is constructed based on the non-linear mechanism model of the wind turbine, including: Step S121: Construct a continuous state space model under high wind speed conditions, and its expression is: (12)

[0017]

[0018]

[0019] ; Step S122: Based on the continuous state space model under high wind speed conditions, ignore the terms related to the pitch angle to obtain a continuous state space model under low wind speed conditions, and its expression is:

[0020]

[0021]

[0022] .

[0023] Further, in step S1, the continuous state - space model is discretized using a sampling period to obtain a discrete state - space model, including: Using the sampling period Discretize the continuous state - space model under low - wind - speed conditions and the continuous state - space model under high - wind - speed conditions respectively; Discretize the continuous state - space model under low - wind - speed conditions to obtain a discrete state - space model under low - wind - speed conditions, and its expression is:

[0024]

[0025]

[0026] (13) In the formula, represents the step index, is the state variable at time is the state variable at time , , are all discrete state - space matrices after sampling; Discretize the continuous state - space model under high - wind - speed conditions to obtain a discrete state - space model under high - wind - speed conditions, and its expression is:

[0027]

[0028]

[0029] (14).

[0030] Further, in step S1, based on the discrete state - space model, calculate the threshold of the switching control strategy for the wind - speed region of the control system of the wind turbine, specifically including: From the low - speed shaft motion equation, we get:

[0031] In the formula, is the derivative of the rotor speed, and further we can get:

[0032]

[0033]

[0034] Therefore,

[0035] when the current operating point is in the high wind speed region, i.e., , will decrease as the increment of the active power reference value of the th wind turbine increases. If , the control system of the wind turbine will transition to the low wind speed region, and the threshold of the switching control strategy for the region where this operating point is located is , and the calculation process is as follows: According to equation (13), we get:

[0036] where and are the matrix elements of and respectively. Since , we get:

[0037] According to , it is deduced that , and its expression is:

[0038] When the current working point is in the low wind speed region, i.e., , will increase as decreases; if , the control system of the wind turbine will transition to the high wind speed, and the threshold of the switching control strategy for the region where this operating point is located is , and the calculation process is as follows: According to equation (14),

[0039] In the formula, and are the matrix elements of and respectively. According to calculated, we get , and its expression is: .

[0040] Further, in step S2, based on the threshold values of the switching control strategy for the wind speed regions of the wind turbine control system, a prediction model for the main shaft torque and a prediction model for the tower thrust are obtained, including: Calculate the main shaft torque under the following four cases respectively For the increment of the active power reference value Partial derivative : If , the wind turbine control system will stay at high wind speed. According to equations (5) and (6), equation (17) is transformed into:

[0041]

[0042]

[0043] Based on equations (13) and (23), we get:

[0044] Therefore,

[0045] Among them, , and The subscripts of represent the case index; If , the wind turbine control system will stay at low wind speed. According to equations (5) and (6), equation (17) is transformed into:

[0046] Among them,

[0047]

[0048] According to equations (14) and (26), we get:

[0049] Therefore,

[0050] Among them ; If , the wind turbine control system will transition from high wind speed to low wind speed. In this case, Is divided into two parts: and , Part operates at high wind speed, while Part operates at low wind speeds; Therefore, for approximate calculation, we get: (29) Therefore,

[0051]

[0052] ; If , the wind turbine control system will transition from low wind speeds to high wind speeds, divided into two parts: and , Part operates at low wind speeds, Part operates at high wind speeds; Therefore, for approximate calculation, we get: (31) Therefore,

[0053]

[0054] .

[0055] Furthermore, in step S2, based on the threshold of the switching control strategy for the wind speed region of the wind turbine control system, a prediction model for the main shaft torque and a prediction model for the tower thrust are obtained, and it also includes: Calculate the tower thrust respectively in the following four cases for the partial derivative of the active power reference value increment : : If , the wind turbine control system remains at high wind speeds, and calculate the increment of the tower thrust , and its expression is:

[0056] ; Based on Equation (24) and Equation (44), we get:

[0057] Therefore,

[0058] where, , and The subscript of represents the case index; If , the wind turbine control system stays at low wind speed and calculates the increment of tower thrust , and its expression is:

[0059] ; Based on Equation (14) and Equation (36), we get:

[0060] Therefore,

[0061] where, ; If , the wind turbine control system transitions from high wind speed to low wind speed and approximately calculates , and we get: (39) Therefore,

[0062] In the formula, ; If , the wind turbine control system transitions from low wind speed to high wind speed and approximately calculates , and we get: (41) Therefore,

[0063] In the formula, ; According to in Equation (33) and Equation (36) and are respectively and functions of, that is:

[0064]

[0065] function is non-linear and is described by a look-up table obtained based on the blade geometry. The inputs of this look-up table are and ; According to and , determine the operating point through a look-up table The power coefficient at is, that is , where and are the corresponding row index and column index respectively; Calculate and :

[0066]

[0067]

[0068]

[0069] The prediction model of the main shaft torque is:

[0070] The prediction model of the tower thrust is:

[0071] where is the active power reference value of the wt-th wind turbine, is the current active power output of the wind turbine generator set, , , , are all obtained from the formulas derived above, is the current tower thrust, is the current main shaft torque, is the increment of the active power reference value of the wt-th wind turbine.

[0072] Furthermore, in step S3, through the time series data of the main shaft torque, an equivalent fatigue load linearization model is obtained. After combining the equivalent fatigue load linearization model with the linear prediction model of the wind turbine generator set, a fatigue load model driven by data mechanism is obtained, specifically including: Generate time series data of the main shaft torque through the prediction model of the main shaft torque. Using a sliding window with a window length of 10, slice the time series data of the main shaft torque to obtain a time series data set of the main shaft torque and , is the data of the first 9 seconds before the time window as historical input data, is the data of the 10th second of the time window as prediction data. Through the rainflow counting method, calculate the load cycle amplitude and mean value of the main shaft torque data of each sliding window to obtain the equivalent fatigue load:

[0073] In the formula, are the load amplitudes at all levels, is the number of cycles of this level of load, m is the Wohler exponent of the S-N curve, and N depends on the design life of the wind turbine, is the equivalent fatigue load of this segment of the load signal, and the equivalent fatigue load time series is obtained from this ; Construct the observation function , and the observation function includes a cubic polynomial basis function, random Fourier basis functions with fundamental frequencies of 0.1 and 0.01, and a delay basis function with a 10-step delay. The following matrix is constructed from this:

[0074]

[0075]

[0076] Construct the state equation through the Koopman operator, and its expression is:

[0077] Calculate the Koopman matrices A and B using the extended dynamic mode decomposition (EDMD) method, i.e.:

[0078] Vectorize the optimization problem to obtain:

[0079] Among them, is the Kronecker product, and by solving the pseudo-inverse, we obtain:

[0080] Construct the output equation:

[0081]

[0082]

[0083]

[0084] Among them, is the error term. Calculate the Koopman matrices C and D using the extended dynamic mode decomposition (EDMD) method, i.e.:

[0085] By solving the pseudo-inverse, we obtain:

[0086] An equivalent fatigue load linearization model based on the data-driven method of Koopman operator modeling is thus obtained, and its expression is: After combining the equivalent fatigue load linearization model with the wind turbine linear prediction model, a fatigue load model driven by both data and mechanism is obtained, and the linearity of the model is maintained. The inputs of the fatigue load model driven by both data and mechanism are: wind speed, generator speed, output power, pitch angle, generator torque, active power reference, main shaft torque, and tower thrust, and the output is the equivalent fatigue load.

[0087] Furthermore, in step S4, active power optimization control is solved for the fatigue load model driven by both data and mechanism to obtain the optimal solution of active power distribution that satisfies the safety and stability constraints of the wind farm, specifically including: Construct the active power reduction optimization objective as:

[0088] where is the equivalent fatigue load of each wind turbine calculated according to the fatigue load model driven by both data and mechanism, and is obtained according to from the look-up table constructed in the following text. The constraint conditions of the optimization problem are:

[0089] where is the active power reference value, and are the maximum available power and minimum available power of a single unit respectively, is the power grid dispatching instruction, is the change amount of active power within 1 time step, is the maximum power ramp constraint of a single unit; Taking the fatigue load model driven by both data and mechanism as the prediction model, the distributed model predictive control algorithm is used to optimize and solve the fatigue load model driven by both data and mechanism, and the solution algorithm is the trust region constraint algorithm; The central controller calculates the active power reference of each wind turbine in the wind farm as the initial point according to the traditional proportional distribution algorithm. At the same time, the distributed controller of each wind turbine generates a wind turbine fatigue load look-up table according to the prediction model at a scale of 10 W and sends it to the central controller. The central controller solves the optimization problem through the trust region constraint algorithm and sends the control strategy to each wind turbine controller; The iterative steps of the trust region constraint algorithm specifically include: Step S41: Set the initial iteration point according to the average distribution strategy , and set the initial trust region radius , define the convergence threshold , gradient norm Stop iteration when Step S42: At the current iteration point , construct a sub-problem that includes the objective function and constraints. The objective function uses a quadratic approximation model, and its expression is:

[0090] In the formula, is the Hessian matrix, is the trial step, and at this time, the constraint should be satisfied:

[0091] Linearize the constraint conditions through the first-order Taylor expansion, and its expression is:

[0092] ; Step S43: Construct the Lagrangian function:

[0093] Among them, and are Lagrange multipliers, and . By solving the partial derivatives of the Lagrangian function and combining the KKT conditions, obtain the solution of the sub-problem. The KKT conditions are:

[0094] Step S44: Calculate the actual descent and the predicted descent , and calculate the ratio of the actual descent to the predicted descent. Its expression is:

[0095] The ratio of the actual descent to the predicted descent is used to measure the matching degree between the predicted descent of the prediction model and the actual descent; Step S45: Update the trust region radius: If , it is considered that the quality of the prediction model is good, and at this time, increase the trust region radius: ; If , it is considered that the quality of the prediction model is poor, and at this time, decrease the trust region radius: ; Otherwise, the trust region radius remains unchanged; Step S46: Update the iteration point: If , accept the update, p, otherwise reject the update. When the iteration end condition is satisfied, stop the iteration; After the iteration is completed, an optimal solution for active power distribution that satisfies the safety and stability constraints of the wind farm is obtained.

[0096] It can be seen from the above technical solutions that the present invention has the following advantages: In the active power control method for wind farm fatigue suppression based on data-mechanism hybrid modeling provided by this application, on the premise of ensuring the safe and stable operation of the wind farm, the active power is accurately distributed according to the working states of each wind turbine and the specific wind speed and wind conditions, and the distributed model predictive control algorithm is used to optimize and solve the fatigue load model driven by data and mechanism to meet the requirements of real-time operation, realizing the effective reduction of the whole-field fatigue load.

[0097] The present invention introduces a linearized prediction model of the main shaft torque based on physical mechanism and a fatigue load calculation method based on data-driven, transforming the non-convex optimization problem into a convex optimization problem that is easy to solve, thereby significantly reducing the fatigue damage risk of wind turbines while ensuring the efficient operation of the wind farm.

[0098] The convex optimization problem is accelerated by generating a look-up table form through a distributed controller, and the whole-field fatigue load suppression of the wind farm is realized under the premise of satisfying the safety and stability constraints of the wind farm. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] In order to more clearly illustrate the technical solutions of this application, the drawings required for description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of this application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0100] Figure 1 FIG. is a flowchart of an active power control method for wind farm fatigue suppression based on data-mechanism hybrid modeling.

[0101] Figure 2 FIG. is a prediction effect diagram of the main shaft torque of an active power control method for wind farm fatigue suppression based on data-mechanism hybrid modeling.

[0102] Figure 3 FIG. is a prediction effect diagram of the equivalent fatigue load linearization model of an active power control method for wind farm fatigue suppression based on data-mechanism hybrid modeling. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0103] To make the application purpose, features, and advantages of this application more obvious and understandable, specific embodiments and accompanying drawings will be used below to clearly and completely describe the technical solutions protected by this application. Obviously, the embodiments described below are only a part of the embodiments of this application, rather than all embodiments. Based on the embodiments in this patent, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this patent.

[0104] The technical solutions proposed in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0105] Figure 1 It is a flowchart of a wind farm fatigue suppression active power control method based on data-mechanism hybrid modeling provided for the embodiments of this application. As Figure 1 shown, a wind farm fatigue suppression active power control method based on data-mechanism hybrid modeling provided for the embodiments of this application specifically includes the following steps: Step S1: Based on preliminary research, it is found that the main shaft torque and tower thrust are the fatigue loads that cause faults in wind turbines. Therefore, a continuous state-space model is established, and the continuous state-space model is discretized using a sampling period to obtain a discrete state-space model. Based on the discrete state-space model, the thresholds of the switching control strategy for the wind speed region of the control system of the wind turbine are calculated, that is, the sensitivity of the main shaft torque and the sensitivity of the tower thrust; through the incremental state-space model and discretization processing, the sensitivity of the main shaft torque and the sensitivity of the tower thrust are calculated in this invention, and the thresholds of the switching control strategy for the wind speed region are obtained, so as to accurately calculate the changes in the main shaft torque and tower thrust of the wind turbine under different wind speed conditions, and be able to more accurately predict and control the fatigue loads of the wind turbine during the process of participating in power grid frequency regulation, thereby reducing maintenance costs and extending the service life of the wind turbine.

[0106] Step S2: Based on the thresholds of the switching control strategy for the wind speed region of the control system of the wind turbine, a prediction model for the main shaft torque and a prediction model for the tower thrust are obtained, denoted as the wind turbine linear prediction model; Step S3: Through the main shaft torque time series data, an equivalent fatigue load linearization model is obtained. After combining the equivalent fatigue load linearization model with the wind turbine linear prediction model, a data-mechanism dual-driven fatigue load model is obtained; Step S4: Perform active power optimization control solution on the data-mechanism dual-driven fatigue load model to obtain the optimal solution of the active power distribution that satisfies the safety and stability constraints of the wind farm.

[0107] Set the time of the operating point as , and the wind speed is a variable that can be both measured and estimated. In this invention, for An estimation was carried out. At the wind speed value at the blade is and it is assumed to be constant within a short control period. At the measured power output of the generator is , at the measured generator speed is , at the measured filtered generator speed is , at the measured pitch angle is . Based on the wind speed value of at , the measured power output of the generator of at , the measured generator speed of at , the measured filtered generator speed of at , and the measured pitch angle of at , the aerodynamic torque at can be obtained as and the generator torque at as .

[0108] Based on the linearization of the simplified non - linear wind turbine model at the operating point, the incremental state - space model under high wind speed conditions is derived through the following steps. Ignoring the terms related to the pitch angle, the incremental state - space model applicable to low wind speed conditions is derived based on the incremental state - space model under high wind speed conditions. The symbol Δ represents the increment of the variable.

[0109] In step S1, a continuous state - space model is established, specifically including: Step S11: Establish a non - linear mechanism model for the wind turbine; Step S12: Construct a continuous state - space model based on the non - linear mechanism model of the wind turbine.

[0110] In step S11, a non - linear mechanism model for the wind turbine is established, including: Step S111: Establish a derivative equation for the increment of the generator speed, and its expression is:

[0111] where, represents the derivative of the increment of the generator speed, represents the gearbox transmission ratio, represents the equivalent mass combining the blade mass and the engine mass, represents the generator torque increment, represents the aerodynamic torque increment, represents at the generator torque measured at the moment, represents at the aerodynamic torque measured at the moment; Step S112: Establish a derivative equation for the filtered generator speed increment, and its expression is:

[0112] In the formula, represents the derivative of the filtered generator speed increment, represents the filter time constant, represents the filtered generator speed increment, represents the generator speed increment; Step S113: Define the increment to obtain:

[0113] In the formula, represents the pitch angle, represents the filtered generator speed, represents the rated value of the generator speed, represents the proportional gain of the wind turbine PI controller, represents the integral gain of the wind turbine PI controller, represents the Laplace operator; Step S114: Based on the derivative equation of the filtered generator speed increment established in Step S112 and the increment obtained in Step S113, obtain the derivative equation of the increment :

[0114] (4) In the formula, represents the derivative of the increment ; The variables considered in the incremental state space model include: wind speed , power output , generator speed , filtered speed and pitch angle .

[0115] Step S115: The calculation of the wind turbine is non - linear. According to the Taylor series, approximate calculation is carried out at the operating point to obtain the aerodynamic torque increment and the generator torque increment; The expression for the pneumatic torque increment is:

[0116] In the formula, represents the increment of the reference value of the active power of the th fan; The expression for the generator torque increment is: (6) In the formula, represents the partial derivative of the generator torque at the point with respect to the generator speed , and its expression is: (7) (8) In the formula, , represents the wind speed value at the blade at the moment, represents the blade, represents as a function of , represents the wind energy utilization coefficient; represents the partial derivative of the generator torque at the point with respect to the increment , and its expression is: (9) In the formula, represents the air density, represents the blade; Step S116: Calculate and ; The expression for

[0117]

[0118] represents the tip speed ratio, represents the pitch angle, represents the difference in between adjacent units, represents the difference in between adjacent units, represents the wind energy utilization coefficient, represents the wind energy utilization coefficient at (n,m) in the look-up table Denote the wind energy utilization coefficient at (n, m+1) in the lookup table; The expression of is: (11) In the formula, Denote the wind energy utilization coefficient at (n+1, m) in the lookup table; is a non-linear function, Obtained through the lookup table, which is generated based on the simulation results of the NREL5MW wind turbine high-fidelity model. The input parameters are and , the difference of between adjacent units is , the difference of between adjacent units is , according to and , the power coefficient at the operating point can be obtained by querying the table, that is , where and are the corresponding row index and column index respectively.

[0119] In step S12, a continuous state space model is constructed based on the non-linear mechanism model of the wind turbine, including: Step S121: Construct a continuous state space model under high wind speed conditions based on formulas (1)-(11), and its expression is: (12)

[0120]

[0121] ; Step S122: Based on the continuous state space model under high wind speed conditions, ignore the terms related to the pitch angle to obtain the continuous state space model under low wind speed conditions, and its expression is:

[0122]

[0123]

[0124] .

[0125] In step S1, the continuous state space model is discretized using the sampling period to obtain a discrete state space model, including: Adopt a sampling period Discretize the continuous state-space model under low wind speed conditions and the continuous state-space model under high wind speed conditions respectively; Discretize the continuous state-space model under low wind speed conditions to obtain the discrete state-space model under low wind speed conditions, and its expression is:

[0126]

[0127]

[0128] (13) In the formula, represents the step index, is the state variable at time is the state variable at time , , are all discrete state-space matrices after sampling; Discretize the continuous state-space model under high wind speed conditions to obtain the discrete state-space model under high wind speed conditions, and its expression is:

[0129]

[0130]

[0131] (14).

[0132] In step S1, based on the discrete state-space model, calculate the threshold of the switching control strategy for the wind speed region of the control system of the wind turbine, specifically including: calculate the threshold of the switching control strategy for the control system of the wind turbine to transition to the low wind speed region and the threshold of the switching control strategy for the control system of the wind turbine to transition to the high wind speed region ; From the low-speed shaft motion equation, we get:

[0133] In the formula, is the derivative of the rotor speed, and further we can get:

[0134]

[0135]

[0136] Therefore,

[0137] During the sampling period, the wind speed condition of the wind turbine may switch between high wind speed and low wind speed. When the current operating point is in the high wind speed region, that is , will decrease with the increase of the active power reference value increment of the th wind turbine. If , the control system of the wind turbine will transition to the low wind speed region, and the threshold of the switching control strategy for the region where this operating point is located is , and the calculation process is as follows: According to equation (13), we get:

[0138] where and are the matrix elements of and respectively. Since , we get:

[0139] According to , it is deduced that , and its expression is:

[0140] It should be noted that the control system has two control strategies, one is for high wind speed and the other is for low wind speed. When the current operating point is in the high wind speed region, that is , will decrease with the increase of the active power reference value increment of the th wind turbine. If , the control system of the wind turbine will execute according to the control strategy under low wind speed conditions.

[0141] It should be further noted that this operating point refers to the operating point where the two control strategies of the wind turbine control system switch, that is, the operating point where the high wind speed operation strategy and the low wind speed operation strategy switch. And for the wind turbine control system, when other state variables are determined, the switching of the control strategy is only related to , and the threshold of the switching control strategy is .

[0142] When the current working point is in the low wind speed region, that is , will increase as decreases; if , the wind turbine control system will transition to high wind speed, and the threshold of the switching control strategy for the region where this operating point is located is , and the calculation process is as follows: According to Equation (14),

[0143] In the formula and are respectively and matrix elements of, and are calculated according to to obtain , and its expression is: .

[0144] In step S2, based on the threshold of the switching control strategy for the wind speed region of the wind turbine control system, a prediction model of the main shaft torque and a prediction model of the tower thrust are obtained, including: Calculate the main shaft torque respectively in the following four cases for the partial derivative of the active power reference value increment : Case 1: High wind speed High wind speed.

[0145] If , the wind turbine control system will stay at high wind speed. According to Equations (5) and (6), Equation (17) is transformed into:

[0146]

[0147]

[0148] Based on Equations (13) and (23), we get:

[0149] Therefore,

[0150] where , and subscripts of represent the case index; Case 2: Low wind speed Low wind speed.

[0151] If , the wind turbine control system will remain at low wind speed. According to Equation (5) and Equation (6), Equation (17) is transformed into:

[0152] Wherein,

[0153]

[0154] According to Equation (14) and Equation (26), we get:

[0155] Therefore,

[0156] Where ; Case 3: High wind speed Low wind speed.

[0157] If , the wind turbine control system will transition from high wind speed to low wind speed. In this case, is divided into two parts: and , The part operates at high wind speed, while The part operates at low wind speed; Therefore, for the approximate calculation of , we get: (29) Therefore,

[0158]

[0159] ; Case 4: Low wind speed High wind speed.

[0160] If , the wind turbine control system will transition from low wind speed to high wind speed, is divided into two parts: and , The part operates at low wind speed, The part operates at high wind speed; Therefore, for the approximate calculation of , we get: (31) Therefore,

[0161]

[0162] 。

[0163] For generality, the subscripts of and are omitted, and based on the measured values, and are calculated and sent to the wind farm controller to formulate an optimal scheduling algorithm.

[0164] In step S2, based on the thresholds of the switching control strategy for the wind speed regions of the control system of the wind turbine generator, a prediction model of the main shaft torque and a prediction model of the tower thrust are obtained, and it further includes: Calculating the tower thrust under the following four cases respectively for the partial derivative of the active power reference value increment : Case 1: High wind speed High wind speed.

[0165] If , the wind turbine generator control system remains at high wind speed, and the increment of the tower thrust is calculated, and its expression is:

[0166] ; Based on Equation (24) and Equation (44), it is obtained that:

[0167] Therefore,

[0168] where , and subscripts represent the case index; Case 2: Low wind speed Low wind speed.

[0169] If , the wind turbine generator control system stays at low wind speed, and the increment of the tower thrust is calculated, and its expression is:

[0170] ; Based on Equation (14) and Equation (36), it is obtained that:

[0171] Therefore,

[0172] wherein, ; Case 3: High wind speed Low wind speed.

[0173] If , the wind turbine control system transitions from high wind speed to low wind speed. For approximate calculation, we get: (39) Therefore,

[0174] In the formula, ; Case 4: Low wind speed High wind speed.

[0175] If , the wind turbine control system transitions from low wind speed to high wind speed. For approximate calculation, we get: (41) Therefore,

[0176] In the formula, ; According to and in formula (33) and formula (36) respectively, and are functions of

[0177]

[0178] Function is non - linear and is described by a look - up table obtained according to the blade geometry. The input of this look - up table is and ; According to and , the power coefficient at the operating point is determined through the look - up table, that is , wherein, and are the corresponding row index and column index respectively; Calculation and :

[0179]

[0180]

[0181]

[0182] For generality, the subscripts of and are omitted. According to the measured values, and are calculated and sent to the wind farm controller to formulate an optimal scheduling algorithm.

[0183] Figure 2 is the prediction effect diagram of the main shaft torque.

[0184] The prediction model of the main shaft torque is:

[0185] The prediction model of the tower thrust is:

[0186] Among them, is the active power reference value of the wt-th wind turbine, is the current active power output of the wind turbine, , , , are all obtained from the formulas derived above, is the current tower thrust, is the current main shaft torque, is the increment of the active power reference value of the wt-th wind turbine.

[0187] It should be noted that , , , are all obtained from the derived formulas , , , , , , , obtained.

[0188] In step S3, an equivalent fatigue load linearization model is obtained from the main shaft torque time series data. After combining the equivalent fatigue load linearization model with the wind turbine linear prediction model, a fatigue load model driven by data mechanism is obtained, specifically including: Slice the main shaft torque time series data with a sliding window of window length 10 to obtain a time series data set of the main shaft torque And , Take the data of the first 9 seconds before the time window as historical input data, Take the data of the 10th second of the time window as prediction data. Through the rain flow counting method, calculate the load cycle amplitude and mean value of the main shaft torque data of each sliding window to obtain the equivalent fatigue load:

[0189] In the formula, Is the load amplitude of each level, Is the number of cycles of this level of load, m is the Wohler index of the S-N curve, and N depends on the design life of the wind turbine, Is the equivalent fatigue load of this section of the load signal, and thus an equivalent fatigue load time series is obtained ; Construct an observation function , The observation function Includes a cubic polynomial basis function, random Fourier basis functions with fundamental frequencies of 0.1 and 0.01, and a delay basis function with a 10-step delay. Thus, the following matrix is constructed:

[0190]

[0191]

[0192] Construct a state equation through the Koopman operator, and its expression is:

[0193] Use the extended dynamic mode decomposition EDMD method to calculate the Koopman matrices A and B, that is:

[0194] Vectorize the optimization problem to obtain:

[0195] Among them, Is the Kronecker product, and by solving with the pseudoinverse, we get:

[0196] Construct the output equation:

[0197]

[0198]

[0199]

[0200] where, is the error term. Calculate the Koopman matrices C and D by using the method of extended dynamic mode decomposition (EDMD), that is:

[0201] Solve through the pseudo-inverse to obtain:

[0202] Thus, an equivalent fatigue load linearization model based on the data-driven method of Koopman operator modeling is obtained. Figure 3 is the prediction effect diagram of the equivalent fatigue load linearization model, and its expression is:

[0203]

[0204] is the load data in the first 9 seconds before the next time window. is the load data in the first 9 seconds before the current time window. is the predicted load data at the 10th second of the current time window. is the observation function.

[0205] After combining the equivalent fatigue load linearization model with the wind turbine linear prediction model, a fatigue load model driven by both data and mechanism is obtained, and the model is kept linear. The inputs of the fatigue load model driven by both data and mechanism are: wind speed, generator speed, output power, pitch angle, generator torque, active power reference, main shaft torque, and tower thrust, and the output is the equivalent fatigue load.

[0206] The wind turbine linear prediction model here includes the prediction model of the main shaft torque and the prediction model of the tower thrust.

[0207] In step S4, perform active optimization control solution on the fatigue load model driven by both data and mechanism to obtain the optimal solution of the active power distribution that meets the safety and stability constraints of the wind farm, specifically including: Construct the active load shedding optimization objective as:

[0208] where, The equivalent fatigue loads of each wind turbine calculated according to the fatigue load model driven by data mechanism and data are obtained according to a look-up table. The constraint conditions of the optimization problem are:

[0209] where is the reference value of active power, and are the maximum available power and the minimum available power of a single unit respectively, is the grid dispatching instruction, is the change in active power within one time step, is the maximum power ramp constraint of a single unit; Taking the fatigue load model driven by data mechanism and data as the prediction model, the distributed model predictive control algorithm is used to optimize and solve the fatigue load model driven by data mechanism and data, and the solution algorithm is the trust region constraint algorithm; The central controller calculates the active power reference of each wind turbine in the wind farm according to the traditional proportional distribution algorithm as the initial point. At the same time, the distributed controller of each wind turbine generates a wind turbine fatigue load look-up table according to the prediction model at a scale of 10 W and sends it to the central controller. The central controller solves the optimization problem through the trust region constraint algorithm and issues the control strategy to each wind turbine controller; A dual-loop communication method is adopted to improve the reliability of data transmission.

[0210] The iterative steps of the trust region constraint algorithm specifically include: Step S41: Set the initial iteration point according to the average distribution strategy, and set the initial trust region radius , define the convergence threshold , and stop the iteration when the gradient norm ; Step S42: At the current iteration point , construct a sub-problem including the objective function and constraint conditions. The objective function uses a quadratic approximation model, and its expression is:

[0211] where is the Hessian matrix, is the trial step, and the following constraints should be satisfied at this time:

[0212] Linearize the constraint conditions through the first-order Taylor expansion, and its expression is:

[0213] ; Step S43: Construct the Lagrangian function:

[0214] where and are Lagrange multipliers, and , and the solution of the sub-problem is obtained by solving the partial derivatives of the Lagrangian function and combining the KKT conditions. The KKT conditions are:

[0215] Step S44: Calculate the actual descent and the predicted descent , and calculate the ratio of the actual descent to the predicted descent. Its expression is:

[0216] The ratio of the actual descent to the predicted descent is used to measure the matching degree between the predicted descent of the prediction model and the actual descent; Step S45: Update the trust region radius: If , it is considered that the quality of the prediction model is good, and the trust region radius is increased at this time: ; If , it is considered that the quality of the prediction model is poor, and the trust region radius is decreased at this time: ; Otherwise, the trust region radius remains unchanged; Step S46: Update the iteration point: If , the update is accepted, p, otherwise the update is rejected. When the iteration end condition is satisfied, the iteration stops; After the iteration is completed, the optimal solution of the active power distribution that satisfies the safety and stability constraints of the wind farm is obtained and sent to each wind turbine for execution.

[0217] By comparing with the traditional proportional distribution algorithm, the present invention realizes a 35% reduction in the main shaft fatigue load of the whole field in a 10×5MW wind farm on the premise of satisfying the operation constraints of the wind farm.

[0218] After the present invention adopts the look-up table calculation acceleration, the solution speed can achieve real-time solution (with 1s as the control time step and the solution time for 10 wind turbines being 0.54s).

[0219] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

[0220] Without departing from the principles and spirit of the present invention, these changes, modifications, substitutions, and variations to the embodiments still fall within the protection scope of the present invention.

Claims

1. A wind farm fatigue suppression active power control method based on data mechanism hybrid modeling, characterized in that: The method comprises: Step S1: establishing a continuous state space model, discretizing the continuous state space model using a sampling period to obtain a discrete state space model, and calculating the threshold of the switching control strategy of the wind speed region of the control system of the wind turbine generator system based on the discrete state space model; Step S2: based on the threshold of the switching control strategy of the wind speed area of ​​the control system of the wind turbine generator set, a prediction model of the main shaft torque and a prediction model of the tower thrust are obtained, which are recorded as the linear prediction model of the wind turbine generator set; Step S3: obtaining an equivalent fatigue load linearization model through the main shaft torque time series data, and combining the equivalent fatigue load linearization model with the wind turbine linear prediction model to obtain a fatigue load model driven by both data and mechanism; Step S4: performing active power optimization control solution on the fatigue load model driven by both data and mechanism to obtain an optimal solution for active power distribution that meets the safety and stability constraints of the wind farm.

2. The wind farm fatigue suppression active power control method based on data mechanism hybrid modeling according to claim 1 is characterized in that: In step S1, a continuous state space model is established, which specifically includes: Step S11: establishing a nonlinear mechanism model for a wind turbine generator set; Step S12: constructing a continuous state space model based on the nonlinear mechanism model of the wind turbine generator system.

3. The wind farm fatigue suppression active power control method based on data mechanism hybrid modeling according to claim 2 is characterized in that: In step S11, a nonlinear mechanism model of the wind turbine is established, including: Step S111: Establish a derivative equation of the generator speed increment, which is expressed as follows: In the formula, represents the derivative of the generator speed increment, Indicates the gearbox transmission ratio, represents the equivalent mass of the blade mass and the engine mass combined, represents the generator torque increment, represents the pneumatic torque increment, Indicated in The generator torque measured at the moment, Indicated in The pneumatic torque measured at the moment; Step S112: Establish a derivative equation of the generator speed increment after filtering, the expression of which is: In the formula, represents the derivative of the filtered generator speed increment, represents the filter time constant, represents the generator speed increment after filtering, Indicates the generator speed increment; Step S113: Define increment get: In the formula, represents the pitch angle, represents the generator speed after filtering, Indicates the rated speed of the generator. represents the proportional gain of the fan PI controller, represents the integral gain of the fan PI controller, represents the Laplace operator; Step S114: Based on the derivative equation of the generator speed increment after filtering established in step S112 and the increment obtained in step S113 , get the increment The derivative equation of is: （4） In the formula, Indicates increment The derivative of Step S115: performing approximate calculation at the working point according to the Taylor series to obtain the pneumatic torque increment and the generator torque increment; The expression of aerodynamic torque increment is: In the formula, Indicates The increase in the active power reference value of the typhoon generator; The expression of the generator torque increment is: （6） In the formula, Indicated in Generator torque at point For generator speed The partial derivative of is expressed as: （7） （8） In the formula, , Indicated in The wind speed value at the blade at the moment, Indicates leaves, express about The function of represents the wind energy utilization coefficient; Indicated in Generator torque at point For increment The partial derivative of is expressed as: （9） In the formula, represents the air density, Indicates leaves; Step S116: Calculation and ; The expression is: represents the tip speed ratio, represents the pitch angle, Represents the adjacent unit The difference, Represents the adjacent unit The difference, represents the wind energy utilization coefficient, represents the wind energy utilization coefficient at (n,m) in the lookup table, represents the wind energy utilization coefficient at (n,m+1) in the lookup table; The expression is: （11） In the formula, Represents the wind energy utilization coefficient at (n+1,m) in the lookup table.

4. The wind farm fatigue suppression active power control method based on data mechanism hybrid modeling according to claim 3 is characterized in that: In step S12, a continuous state space model is constructed based on the nonlinear mechanism model of the wind turbine generator set, including: Step S121: construct a continuous state space model under high wind speed conditions, the expression of which is: （12） ； Step S122: Based on the continuous state space model under high wind speed, the terms related to the pitch angle are ignored to obtain the continuous state space model under low wind speed, which is expressed as: 。 5. The wind farm fatigue suppression active power control method based on data mechanism hybrid modeling according to claim 4 is characterized in that: In step S1, the continuous state space model is discretized using a sampling period to obtain a discrete state space model, including: Sampling period Discretization is performed on the continuous state space model under low wind speed and the continuous state space model under high wind speed respectively; The continuous state space model under low wind speed is discretized to obtain the discrete state space model under low wind speed, which is expressed as: （13） In the formula, represents the step index, for The state variables at time, for The state variables at time, , , They are all discrete state space matrices after sampling; The continuous state space model under high wind speed is discretized to obtain the discrete state space model under high wind speed, which is expressed as: （14）。 6. The wind farm fatigue suppression active power control method based on data mechanism hybrid modeling according to claim 5 is characterized in that: In step S1, based on the discrete state space model, the threshold of the switching control strategy of the wind speed area of ​​the control system of the wind turbine generator system is calculated, which specifically includes: From the low speed shaft motion equation: In the formula, is the derivative of the rotor speed, and further we can get: therefore, When the current operating point is in the high wind speed area, that is, , Will follow the Active power reference value increment of typhoon generator decreases with the increase of , the control system of the wind turbine will transition to the low wind speed area. The threshold of the switching control strategy in this operating point area is , the calculation process is as follows: According to formula (13), we get: in and They are and The matrix elements of ,get: according to , it can be deduced that , whose expression is: When the current working point is in the low wind speed area, that is, hour, Will follow decreases and increases; if , the wind turbine control system will transition to high wind speed, and the threshold of the switching control strategy in the area where this operating point is located is , the calculation process is as follows: According to formula (14), In the formula and They are and The matrix elements of Calculated , whose expression is: 。 7. The wind farm fatigue suppression active power control method based on data mechanism hybrid modeling according to claim 6 is characterized in that: In step S2, based on the threshold of the switching control strategy of the wind speed area of ​​the control system of the wind turbine generator system, a prediction model of the main shaft torque and a prediction model of the tower thrust are obtained, including: Calculate the spindle torque in the following four cases Increment of active power reference value The partial derivative of : if , the wind turbine control system will stay at high wind speed. According to equations (5) and (6), equation (17) is transformed into: Based on equation (13) and equation (23), we get: therefore, in, , and The subscripts of indicate case indices; if , the wind turbine control system will stay at low wind speed. According to equations (5) and (6), equation (17) is transformed into: in, According to formula (14) and formula (26), we can get: therefore, in ; if , the wind turbine control system will transition from high wind speed to low wind speed. In this case, It is divided into two parts: and , Some operate in high wind speeds, while Some operate in low wind speeds; Therefore, Approximate calculations yield: (29) therefore, ； if , the wind turbine control system will transition from low wind speed to high wind speed, It is divided into two parts: and , Some operate in low wind speeds. Some operate in high wind speeds; Therefore, Approximate calculations yield: (31) therefore, 。 8. The wind farm fatigue suppression active power control method based on data mechanism hybrid modeling according to claim 7 is characterized in that: In step S2, based on the threshold of the switching control strategy of the wind speed area of ​​the control system of the wind turbine generator system, a prediction model of the main shaft torque and a prediction model of the tower thrust are obtained, which also includes: Calculate the tower thrust in the following four cases Increment of active power reference value The partial derivative of : if , the wind turbine control system maintains high wind speed and calculates the tower thrust The increment of is expressed as: ； Based on equation (24) and equation (44), we get: therefore, in, , and The subscripts of indicate case indices; if , the wind turbine control system stays at low wind speed and calculates the tower thrust The increment of is expressed as: ； Based on equation (14) and equation (36), we get: therefore, in, ; if , the wind turbine control system transitions from high wind speed to low wind speed. Approximate calculations yield: (39) therefore, In the formula, ; if , the wind turbine control system transitions from low wind speed to high wind speed. Approximate calculations yield: (41) therefore, In the formula, ; According to formula (33) and formula (36), and They are and The function is: function is nonlinear and is described by a lookup table derived from the blade geometry whose input is and ; according to and , determine the operating point by looking up the table The power coefficient at ,in, and are the corresponding row and column indices respectively; calculate and : The prediction model of spindle torque is: The prediction model of tower thrust is: in, is the active power reference value of the wtth wind turbine, is the current active power output of the wind turbine, , , , All of them are derived from the above derived formulas. is the current tower thrust, is the current spindle torque, is the increment of the active power reference value of the wtth wind turbine.

9. The wind farm fatigue suppression active power control method based on data mechanism hybrid modeling according to claim 8 is characterized in that: In step S3, an equivalent fatigue load linearization model is obtained through the main shaft torque time series data. After combining the equivalent fatigue load linearization model with the wind turbine linear prediction model, a fatigue load model driven by both data and mechanism is obtained, which specifically includes: The main shaft torque prediction model is used to generate the main shaft torque time series data. The main shaft torque and tower thrust time series data are sliced ​​with a sliding window of 10 window length to obtain the main shaft torque and tower thrust time series data sets. and , The data of the first 9 seconds of the time window is used as historical input data. The 10th second data of the time window is used as the prediction data. The load cycle amplitude and mean of the main shaft torque data of each sliding window are calculated by the rain flow counting method to obtain the equivalent fatigue load: In the formula, is the load amplitude at each level, is the number of cycles of this level of load, m is the Wohler index of the SN curve, and N depends on the design life of the wind turbine. is the equivalent fatigue load of this load signal, and the equivalent fatigue load time series is obtained ; Constructing the observation function , the observation function Including the third-order polynomial basis function, the random Fourier basis function with base frequencies of 0.1 and 0.01, and the delayed basis function with 10 steps of delay, the following matrix is ​​constructed: The state equation is constructed by Koopman operator, and its expression is: The Koopman matrices A and B are calculated using the extended dynamic mode decomposition (EDMD) method, namely: Vectorizing the optimization problem: in, is the Kronecker product, and by solving the pseudo-inverse, we get: Construct the output equation: in, is the error term, and the Koopman matrix C and D are calculated using the extended dynamic mode decomposition EDMD method, namely: By solving the pseudo-inverse, we get: Thus, the equivalent fatigue load linearization model based on the Koopman operator modeling method of data-driven method is obtained, and its expression is: The load data for the first 9 seconds of the next time window. The load data for the first 9 seconds of the current time window. Forecast load data for the 10th second of the current time window, is the observation function; After combining the equivalent fatigue load linearization model with the wind turbine linear prediction model, a data-mechanism dual-driven fatigue load model is obtained, and the model linearity is maintained. The input of the data-mechanism dual-driven fatigue load model is: wind speed, generator speed, output power, pitch angle, generator torque, active power reference, main shaft torque and tower thrust, and the output is the equivalent fatigue load.

10. The wind farm fatigue suppression active power control method based on data mechanism hybrid modeling according to claim 9, characterized in that: In step S4, active power optimization control is performed on the fatigue load model driven by both data and mechanism to obtain the optimal solution for active power distribution that meets the safety and stability constraints of the wind farm, specifically including: The optimization objective of active load shedding is constructed as follows: in, The equivalent fatigue load of each fan is calculated based on the fatigue load model driven by the data mechanism. The optimization problem constraints are obtained by looking up the table: in, is the active power reference value, and are the maximum available power and the minimum available power of a single machine, respectively. is the grid dispatch instruction, is the change in active power within 1 time step, is the maximum power climbing constraint of a single machine; The fatigue load model driven by both data and mechanism is used as a prediction model, and a distributed model predictive control algorithm is used to optimize and solve the fatigue load model driven by both data and mechanism. The solution algorithm is a trust region constraint algorithm. The central controller calculates the active power reference of each wind turbine in the wind farm as the initial point according to the traditional proportional allocation algorithm. At the same time, the distributed controller of each wind turbine generates a wind turbine fatigue load lookup table based on the prediction model at a scale of 10W and sends it to the central controller. The central controller solves the optimization problem through the trust domain constraint algorithm and sends the control strategy to each wind turbine controller. The iterative steps of the trust region constraint algorithm specifically include: Step S41: Setting the initial iteration point according to the average distribution strategy , and set the initial trust domain radius , define the convergence threshold , the gradient norm Stop iteration when Step S42: At the current iteration point , construct a sub-problem including the objective function and constraints. The objective function uses a quadratic approximation model, and its expression is: In the formula, is the Hessian matrix, This is a trial step, and the constraints should be satisfied at this time: The constraints are linearized by Taylor's first-order expansion, and the expression is: ； Step S43: construct the Lagrangian function: in, and is a Lagrange multiplier, and , by solving the partial derivatives of the Lagrangian function and combining the KKT conditions, the solution of the subproblem is obtained. The KKT conditions are: Step S44: Calculate the actual descent amount and predicted decline , and calculate the ratio of the actual decline to the predicted decline, the expression is: The ratio of the actual decline to the predicted decline is used to measure the degree of match between the predicted decline of the forecast model and the actual decline; Step S45: Update the trust domain radius: if , then the prediction model is considered to be of good quality, and the radius of the trust region is increased: ; if , then the prediction model is considered to be of poor quality, and the radius of the trust region is reduced: ; Otherwise, the radius of the trust domain remains unchanged; Step S46: Update iteration point: if , then accept the update, p, otherwise refuse to update, and stop iterating when the iteration end condition is met; After completing the iteration, the optimal solution for active power allocation that meets the safety and stability constraints of the wind farm is obtained.

Images