Wind power plant active wake flow control method and device based on machine-field cooperation

Abstract

The invention relates to a wind power plant active wake flow control method and device based on machine-field cooperation. The method is characterized in that a field control layer and a machine control layer are cooperatively operated; calculating the real-time wind speed of each fan in the wind power plant based on a Gaussian rotation-energy conservation wake flow superposition model describing the asymmetric characteristics of the wake flow and the multi-wake flow interactive superposition of the whole wind power plant; in the field control layer control, the optimal yaw angle under different wind condition combinations is obtained by taking the maximization of the total output power of the wind power plant as a target; in the control of the machine control layer, yaw angle tracking errors and the fatigue load of a yaw bearing are considered, model prediction control is applied within a continuous time range, and the yaw angle of each fan is dynamically adjusted according to the optimal yaw angle provided by the field control layer. Compared with the prior art, the method has the advantages that the generating capacity is remarkably improved and the fatigue load is reduced in a randomly fluctuating wind field environment, and the method has remarkable applicability in a large offshore wind plant.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of offshore wind farm active wake control, and particularly to a machine-farm coordinated yaw control method considering dynamic wind conditions. BACKGROUND

[0002] With the continuous expansion of offshore wind farm scale, the wake effect between wind turbines is increasingly significant. In large-scale wind farms, the wake generated by upstream wind turbines can cause up to 20% power loss. The wake effect is manifested in that after the wind turbine extracts energy from the airflow, the downstream wind speed decreases and is accompanied by an increase in turbulence intensity. Therefore, active wake control is crucial to improve the power generation efficiency and operational reliability of wind farms. The common wake control techniques currently include yaw control, variable pitch control and torque control, among which yaw control has become the main control method because it can effectively guide the deflection of the wake and reduce the power loss downstream.

[0003] The research on yaw control mainly focuses on wake modeling and control strategy design. As a key link between wind farm dynamics and control models, the wake model directly affects the effectiveness of the yaw strategy. Computational fluid dynamics models (CFD) such as large eddy simulation can accurately depict the wake situation and verify the asymmetric characteristics of the wake under yaw conditions, including wake deflection, expansion and other phenomena. However, although the CFD model has high accuracy, its computational cost is huge and it is difficult to apply to real-time control of large-scale wind farms. Therefore, most studies use engineering wake models to improve computational efficiency, including the classic Jensen, Gaussian, Frandsen and Larsen models. These models are computationally efficient, but they rely on static assumptions and wake symmetry simplifications.

[0004] To improve the adaptability and accuracy of engineering wake models, various improvement strategies have been proposed by scholars. For example, introducing yaw wake correction, considering dynamic turbulence intensity, proposing an aggregated reduced-order model, and simplifying system modeling by scaling wind turbine characteristics. Although the above methods improve the modeling accuracy, they still face the problem of limited adaptability under complex inflow conditions. In addition, data-driven methods have also received attention in recent years, such as machine learning models trained based on historical data, but their computational complexity is usually 1-2 orders of magnitude higher than traditional engineering models. Therefore, balancing computational efficiency and model accuracy in wake modeling is still a core challenge faced by current research.

[0005] On the basis of the above model, many studies have designed corresponding control algorithms, mainly focusing on single-time optimization strategies under static assumptions. Existing methods mostly fail to fully consider the dynamic characteristics of offshore wind energy resources, especially the control performance under conditions of severe fluctuations in wind speed and direction. For example, the measured data of the Rudong wind farm in Jiangsu Province, China, show that the fluctuation amplitude of wind direction within each minute can exceed 30°, and the wind speed change often reaches 2 m / s. At the same time, wind direction mutations frequently occur within consecutive time periods. These changes in wind farm environment may be influenced by factors such as changes in the boundary layer height and meteorological disturbances, further exacerbating the uncertainty of the wake. Therefore, the real-time adaptability of active wake control strategies needs to be strengthened.

[0006] Chinese Patent Application CN202411986934.7 discloses a multi-machine cooperative yaw control method considering dynamic wind direction, which builds a single-machine yaw optimization model in the existing patent scheme, synchronously improves the total power generation and reduces the fatigue load, and uses a serial optimization method for multi-round iterative optimization of each wind turbine. Due to the complexity of the objective function solved in the control system, the solving time is relatively long, and in the actual operation process, yaw lag and frequent yawing may occur, which has a great impact on power generation efficiency and mechanical bearing of the unit, and is not suitable for field engineering application. SUMMARY

[0007] The purpose of the present application is to solve the problems of the prior art, such as not fully considering the dynamic characteristics of offshore wind energy resources and poor real-time adaptability of active wake control strategies, and to provide an offshore wind farm active wake control method and device based on machine-farm cooperation.

[0008] The purpose of the present application can be achieved by the following technical solutions:

[0009] As a first aspect of the present application, a wind farm active wake control method based on machine-farm cooperation is provided, characterized in that the method is cooperatively operated by a field control layer and a machine control layer.

[0010] Based on the Gaussian vorticity-energy conservation wake superposition model depicting the asymmetric characteristics of the wake and the interactive superposition of the full-field multi-wake of the wind farm, the real-time wind speed of each wind turbine in the wind farm is calculated.

[0011] In the field control layer control, the optimal yaw angle under different wind condition combinations is obtained to maximize the total output power of the wind farm.

[0012] In the machine control layer control, the yaw angle tracking error and the fatigue load of the yaw bearing are considered, and model predictive control is applied within a continuous time range to dynamically adjust the yaw angle of each wind turbine according to the optimal yaw angle obtained by the field control layer.

[0013] As a preferred technical solution, the field control layer adopts a hybrid optimization method of gradient descent and local traversal. First, a feasible solution is obtained using the gradient descent algorithm. Then, a performance-based local traversal search is performed around this feasible solution. The specific steps are as follows:

[0014] The wind direction is divided into multiple discrete intervals, and the wind speed is divided into multiple intervals;

[0015] For each wind speed-wind direction combination, the wake region is iteratively updated using wind farm simulation tools, and the actual wind speed of the wind turbine is calculated based on the Gaussian-curl energy conservation wake model.

[0016] Calculate the total field output power based on the objective function and constraints of the field control layer;

[0017] Set a maximum number of iterations, calculate the wind turbine's gradient in each iteration, and update the yaw angle;

[0018] If the difference between the yaw angles of the two consecutive iterations is less than the convergence threshold, then output the current yaw angle and exit the iteration.

[0019] Based on the yaw angle obtained through iteration, a local traversal optimization is performed within a set local search range, and the optimal yaw angle corresponding to the maximum power across the entire field is selected.

[0020] The optimal yaw angle for each wind speed-direction combination is obtained by iterative analysis, and then used for real-time yaw angle control in the engine control layer.

[0021] As a preferred technical solution, the objective function of the field control layer aims to maximize the total output power P of the wind farm:

[0022]

[0023] In the formula, v i (α i ,θ) represents the yaw angle α of wind turbine i. i The actual wind speed at the wind direction angle θ; α i P is the yaw angle of wind turbine i; N is the total number of wind turbines in the wind farm; i The output power of fan i:

[0024]

[0025] In the formula, ρ is the air density; A is the swept area of ​​the wind turbine; C P v is the power coefficient; i f represents the actual wind speed of fan i; p This is the power attenuation coefficient.

[0026] As a preferred technical solution, the actual wind speed of the fan is calculated by using a Gaussian-curl energy conservation wake model, a curl term based on the principle of angular momentum conservation is introduced, the total energy loss of the fan i is the superposition result of the influence of each upstream fan, the actual wind speed v i The calculation is as follows:

[0027]

[0028] Wherein, Γ is the fan set affected only by a single wake; The wind speed received by the fan i is:

[0029]

[0030] Wherein, v0 is the far-field incoming wind speed, D is the fan impeller diameter, k a is the wake depth coefficient, k b is the wake diffusion coefficient, α j is the yaw angle of the fan j, δ is the wake diffusion width, y i is the radial distance of the fan i, x i is the axial distance of the fan i:

[0031] x i =Lcos(|θ i -θ+0.3C T α j |)

[0032] Wherein, L is the distance between the fan i and j; θ is the wind direction angle; θ i is the angle between the line connecting the fan i and j and the y-axis; C T is the thrust coefficient.

[0033] As a preferred technical solution, the model predictive control strategy is adopted in the machine control layer, the future state is predicted within a limited time domain, and the rolling optimization is performed to determine the optimal yaw control sequence, and the specific control process is as follows:

[0034] The disturbance estimation fed back by the disturbance observer is used to correct the control input to complete the disturbance compensation, and then the system state at the next moment is predicted based on the wind turbine state update and the system output state space equation according to the corrected control input;

[0035] The minimum objective function of the rolling solution optimization problem is solved within the optimization time window, and the optimal control sequence is continuously updated;

[0036] The controller only executes the first control action in the current optimization sequence, and sends the control action to the yaw actuator; at the same time, the real-time response data of the actuator is collected, which is used for the next state estimation and disturbance observer update;

[0037] The disturbance observer estimates the external disturbance based on the deviation between the predicted yaw angle and the actual measured value, and feeds the disturbance estimation back to the prediction model for disturbance compensation.

[0038] As a preferred technical solution, the control target function of the machine control layer comprehensively considers the yaw angle tracking error and the fatigue load of the yaw bearing, and the reference trajectory is given by the optimal yaw angle generated by the field control layer:

[0039]

[0040] Wherein, λ is the weight coefficient of multi-objective control; e i,t+k and L i,t+k is the yaw angle tracking error and the fatigue load of the yaw bearing of the machine group i at t+k moment;

[0041]

[0042] Wherein, α i,t+k , α i,t+k is the yaw angle of the machine group i at t+k moment and the yaw angle reference input; K yaw is the fatigue load stiffness coefficient; β is the gain coefficient for considering the influence of wind speed v on load estimation.

[0043] As a preferred technical solution, the state space model of the wind turbine i state update is represented as:

[0044] X i,t+k+1 =A s X i,t+k +B s u i,t+k +ζ i,t+k ,k=0,…,T-1

[0045] Wherein, A S is the state transition matrix, B S is the control input matrix, and ζ i,t+k is the external disturbance term;

[0046] The system output equation is represented as:

[0047] Y i,t+k =C s X i,t+k +n i,t+k ,k=0,…,T-1

[0048] Wherein, Y i,t+k is the output vector, including the predicted yaw angle C S is the output matrix, and n i,t+k is the measurement noise.

[0049] As a preferred technical solution, the constraint condition of the machine-control layer includes that the yaw angle and yaw angle velocity are subjected to mechanical safety constraints, a yaw start threshold constraint is introduced, a fatigue load constraint caused by yaw is introduced, and an energy consumption constraint in the yaw process is introduced.

[0050] As a second aspect of the present application, a wind farm active wake control device based on machine-field cooperation is provided, which comprises a memory, a processor, and a program stored in the memory, and the processor implements the wind farm active wake control method based on machine-field cooperation as described above when executing the program.

[0051] As a third aspect of the present application, a storage medium is provided, which stores a program, and the program implements the wind farm active wake control method based on machine-field cooperation as described above when executed.

[0052] Compared with the prior art, the present application has the following beneficial effects:

[0053] 1) The present application proposes a machine-field cooperative active wake control method. In the field control layer, based on the full-field wake interaction model, the ideal solution set of full-field power optimization is solved offline from the perspective of full-field wake interaction, and in the machine control layer, the multi-objective optimization of power and load is realized in the form of tracking the optimal solution. Compared with the prior art scheme of establishing a yaw optimization model for a single wind turbine and performing multi-round iterative optimization for each wind turbine, the double-layer optimization control mode proposed by the present application can more accurately depict the wake interaction between different wind turbines in the wind farm, improve the accuracy of control, and improve the solving speed through relatively simple operation logic in the machine control layer, which is more in line with the actual situation of frequent changes of offshore wind conditions and is more conducive to the application of field engineering.

[0054] 2) The present application also proposes a Gaussian curl-energy conservation wake superposition model GCEM wake model to represent the wake deflection caused by yaw and the wake superposition effect between multiple wind turbines. By introducing a curl term based on the principle of angular momentum conservation to describe the lateral deflection of the wake, and based on the wake superposition method based on the principle of energy conservation, the interaction between multiple wakes in the whole field is modeled, which can accurately capture the asymmetry caused by the deflection of the upstream unit. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 A flowchart of the active wake control method based on machine-field cooperation of the present application.

[0056] Figure 2 A comparison chart of the case where the unit is aligned with the wind direction; a) directly against the wind direction, b) not completely directly against the wind direction.

[0057] Figure 3 A data interaction schematic diagram of FLORIS used for wake area analysis of the present application.

[0058] Figure 4 Flow chart of the process of solving the MPC-based machine-control layer multi-objective control optimization problem. DETAILED DESCRIPTION

[0059] The application will be described in detail below with reference to the drawings and specific embodiments. The embodiments are implemented on the premise of the technical solutions of the application, and detailed implementation modes and specific operation processes are given, but the protection scope of the application is not limited to the following embodiments.

[0060] Embodiment 1

[0061] The application proposes a machine-field collaborative active wake control method, which is collaboratively operated by a field control layer and a machine control layer. The field control layer solves the yaw angle under the condition of optimal power of the whole field based on the whole field wake interaction, and the machine control layer considers the influence of the fatigue load of the wind turbine on the basis of dynamic tracking and adjustment of the yaw instruction of the field control.

[0062] S1, calculate the real-time wind speed of each wind turbine in the wind farm based on the wake superposition model considering the wake asymmetry.

[0063] The application proposes a Gaussian curl-energy conservation wake superposition model GCEM (Gaussian-Curl Energy-Conserving wake model), which considers the wake deflection caused by yaw and the wake superposition effect between multiple wind turbines. The GCEM model includes two parts: one is a Gaussian-Curl model for describing the single wake deflection of a wind turbine; the other is a multi-wake interaction superposition principle based on energy conservation.

[0064] To improve the power generation capacity of the downstream wind turbine, the upstream wind turbine deflects the wake area by yawing, so that the wake center deviates to the leeward side of the wind turbine, sacrificing part of the power generation capacity of the upstream wind turbine, reducing the wake effect, and effectively improving the power generation efficiency of the downstream wind turbine.

[0065] Let wind turbine j be the upstream wind turbine of wind turbine i. As shown in the figure, Figure 2 When wind turbine j deflects, the wake center line will deviate, thereby changing the axial distance x i of wind turbine i relative to j. To accurately capture the asymmetry caused by the deflection, the application introduces a curl term based on the principle of angular momentum conservation to describe the lateral deviation of the wake. The wake deflection amount can be expressed as 0.3C T α j , where C T is the thrust coefficient, and α j is the yaw angle of wind turbine j. The axial distance x i of wind turbine i is calculated as follows:

[0066] x i = L cos(|θ i - θ + 0.3C T α j |) (1)

[0067] where L is the distance between wind turbine i and j, θ is the wind direction angle, and θ i is the angle between the line connecting wind turbine i and j and the y-axis.

[0068] Considering the wake deflection, the wind speed received by wind turbine i can be expressed as

[0069]

[0070] where v0is the far-field incoming wind speed, D is the diameter of the wind turbine rotor, k a is the wake depth coefficient, k b is the wake diffusion coefficient, δ is the wake diffusion width, and y i is the radial distance of wind turbine i. According to engineering experience, k a and k b are taken as 0.38 and 0.04, respectively.

[0071] The wind speed loss of wind turbine i under the influence of a single upstream wake is:

[0072]

[0073] The above formula constitutes the wind speed model of wind turbines considering wake deflection. In a wind farm, some wind turbines may only be affected by a single wake area, while others may be affected by multiple wakes, resulting in the accumulation of wind speed loss. Therefore, this paper uses the wake superposition method based on the principle of energy conservation to model the interaction between multiple wakes.

[0074] Let J i be the set of all upstream wind turbines of wind turbine i, and j ∈ J i represent any one of them. Assuming that the total energy loss of wind turbine i is the superposition result of the influence of each upstream wind turbine, the actual wind speed v i can be calculated as follows:

[0075]

[0076] where Γ is the set of wind turbines affected by only a single wake.

[0077] S2, considering the field control layer optimization for maximum power in the whole field, obtaining the optimal yaw angle under different wind condition combinations.

[0078] The present application realizes the maximization of total power of wind farm by adjusting the yaw angle of each wind turbine on the basis of satisfying the yaw angle limit and wake effect model, using gradient descent method and traversal optimization.

[0079] An offshore wind farm containing N wind turbines is built, and the wind turbines are numbered from upstream to downstream according to the wind direction. The optimization problem of the field control layer can be described as follows:

[0080] The objective function is set to maximize the total output power of the field:

[0081]

[0082] The constraint conditions include:

[0083] The output power of wind turbine i:

[0084]

[0085] The yaw angle should satisfy the mechanical limit:

[0086] -α max ≤α i ≤α max (7)

[0087] The calculation of wind speed v i is based on the aforementioned GCEM model:

[0088] (1),(3),(4) (8)

[0089] Where P i is the output power of wind turbine i, α i is the yaw angle of the wind turbine, ρ is the air density, A is the swept area of the wind wheel, f p is the power attenuation coefficient, and C P is the power coefficient.

[0090] The field control layer optimization model built for the above formula faces the following two main challenges in solving:

[0091] First, due to the irregularity and time-varying nature of wake propagation, the identification of the upstream wind turbine set affected by the wake becomes complex. To address this problem, the present application establishes a data interaction mechanism between the proposed field control strategy and FLORIS, as shown in Figure 3 The interaction is carried out under discrete wind speed and wind direction conditions. FLORIS updates the wake area according to the current yaw setting, and the optimization model adjusts the control variables accordingly, and iterates continuously until the control target is met.

[0092] Second, considering that a typical wind farm usually contains nearly one hundred wind turbines, the optimization problem has nonlinear characteristics in constraints (3), (4) and (6), and the overall problem has large size, high dimension, and complex coupling relationship between variables, which brings significant challenges to the solving process. To improve the solving efficiency and the quality of the optimization results, this paper proposes a hybrid optimization method combining gradient descent and local traversal, as shown in Algorithm 1. This method first uses the gradient descent algorithm to quickly obtain a feasible solution, and then performs a performance-based local traversal search around this solution. This strategy ensures the efficiency of the solution while improving the robustness of the optimization results and the quality of the yaw setting.

[0093]

[0094]

[0095] S3, Multi-objective control of the machine-control layer based on MPC.

[0096] In the machine-control layer control, the MPC strategy is adopted to dynamically adjust the yaw angle of each wind turbine according to the yaw angle command provided by the field-control layer. This method reduces the tracking error while considering the suppression of the unit fatigue load, thereby ensuring the smoothness and efficiency of the wind turbine yaw action.

[0097] MPC determines the optimal yaw control sequence by predicting the future state within a limited time domain and performing rolling optimization. At any time t, the system state vector X i,t includes the yaw angle α i,t of wind turbine i, the yaw angle speed i,t , the generator speed ω i,t , and the yaw motor output torque τ i,t , where α i,t is also the control input u i,t+k+1 . Within the prediction time domain T, the state update of wind turbine i can be described by the following state space model:

[0098] X s = A i,t+k X s + B i,t+k u i,t+k + ζ S , k = 0,..., T-1 (9)

[0099] where A S is the state transition matrix, B i,t+k is the control input matrix, and ζ i,t+k is the external disturbance term such as wind speed and direction change.

[0100] The output equation is:

[0101] Y s = Cs X i,t+k +n i,t+k ,k=0,…,T-1 (10)

[0102] where Y i,t+k is the output vector, containing the predicted yaw angle C S is the output matrix, n i,t+k is the measurement noise.

[0103] 1) Objective function

[0104] The objective function J of the machine control layer takes into account both the yaw angle tracking error and the fatigue load of the yaw bearing, and the reference trajectory is given by the optimal yaw angle generated by the field control layer:

[0105]

[0106] where:

[0107]

[0108] where λ is the weight coefficient of multi-objective control, e i,t+k and L i,t+k are the yaw angle tracking error and the fatigue load of the yaw bearing of the machine group i at time t+k, K yaw is the fatigue load stiffness coefficient, and β is the gain coefficient used to consider the influence of wind speed v on load estimation.

[0109] The above problem can be converted into the following quadratic form:

[0110]

[0111] where Q and R are weight matrices, is the yaw angle, is the state phasor of .

[0112] 2) Constraint conditions

[0113] To ensure the reliability of the yaw system, the yaw angle and yaw angle velocity are subject to mechanical safety constraints:

[0114]

[0115] To avoid frequent yawing due to small wind direction disturbances, thereby generating additional energy consumption and fatigue load, a yaw start threshold constraint is introduced:

[0116] |α i,t +1-α i,t |≥α high (16)

[0117] According to IEC standards, the fatigue load caused by yaw must not exceed the fatigue load caused by wind shear, therefore the following must be met:

[0118]

[0119] To maintain the economic viability of a wind farm, energy consumption during yaw should not exceed the power generation revenue generated during the same period:

[0120] E t:t+T ≤P t:t+T (18)

[0121] Among them, E t:t+T Let P be the energy consumption during time period T. t:t+T This represents the amount of electricity generated during the corresponding time period T.

[0122]

[0123] Among them, C yaw Energy consumption per unit yaw angle, Δα i,t+k This refers to the yaw angle adjustment range. The power generation is the power generated without yaw, and Δt is the time step.

[0124] 3) Solution process

[0125] like Figure 4 As shown, the machine control layer control process based on MPC includes four modules: predictive model, rolling optimization, control execution, and feedback correction based on disturbance observer (DOB).

[0126] 3.1 Prediction Model:

[0127] At time t, the system first completes disturbance compensation, and then predicts the system state at time t+1 based on the state-space equations (9) and (10). After optimization, the disturbance error is compensated and the predicted state is corrected, thereby obtaining a more accurate state estimate and providing a data basis for subsequent rolling optimization.

[0128] 3.2 Scrolling optimization:

[0129] MPC minimizes the objective function J by solving an optimization problem, while introducing constraints such as yaw rate, yaw angle, load, and energy consumption (15)–(20) to ensure the feasibility of the solution. The optimization time window covers from t to t+T and rolls over time to continuously update the optimal control sequence.

[0130] 3.3 Control Execution:

[0131] The controller executes only the first control action in the current optimized sequence and sends it to the yaw actuator. Simultaneously, it collects real-time actuator response data for subsequent state estimation and disturbance observer updates.

[0132] 3.4 Feedback correction based on DOB:

[0133] The system uses DOB to estimate external disturbance, and then corrects the prediction model to improve the anti-disturbance ability of the system. Disturbance estimation can be expressed as:

[0134]

[0135] where L d is the disturbance estimation gain, is the predicted yaw angle, Y i,t+k is the actual measured value.

[0136] The estimated disturbance is used to correct the control input to obtain the compensated input can be expressed as:

[0137]

[0138] The prediction model updates the state at the next time according to the above formula:

[0139]

[0140] Embodiment 2

[0141] As a second aspect of the present application, the present application also provides an electronic device, comprising: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned wind farm active wake control method based on machine-field cooperation. In addition to the above-mentioned processor, memory and interface, any data processing capable device in which the device in the embodiment is generally according to the actual function of the any data processing capable device, and can also include other hardware, and this will not be described again.

[0142] Embodiment 3

[0143] As a third aspect of the present invention, this application also provides a computer-readable storage medium storing computer instructions thereon, which, when executed by a processor, implement the above-described wind farm active wake control method based on machine-farm cooperation. The computer-readable storage medium can be an internal storage unit of any data processing device as described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0144] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0145] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A wind farm active wake control method based on turbine-farm cooperation, characterized in that, The method is operated collaboratively by the field control layer and the machine control layer; The real-time wind speed of each wind turbine in the wind farm is calculated based on the Gaussian curl-energy conservation wake superposition model that characterizes the asymmetric characteristics of the wake and the interaction and superposition of multiple wakes in the entire wind farm. In the field control layer, the goal is to maximize the total output power of the wind farm and obtain the optimal yaw angle under different wind condition combinations. In the aforementioned machine control layer, considering the yaw angle tracking error and the fatigue load of the yaw bearing, model predictive control is applied over a continuous time range to dynamically adjust the yaw angle of each wind turbine based on the optimal yaw angle obtained from the field control layer.

2. The active wake control method for wind farms based on turbine-farm cooperation according to claim 1, characterized in that, The field control layer employs a hybrid optimization method combining gradient descent and local traversal. First, a feasible solution is obtained using the gradient descent algorithm. Then, a performance-based local traversal search is performed around this feasible solution. The specific steps are as follows: The wind direction is divided into multiple discrete intervals, and the wind speed is divided into multiple intervals; For each wind speed-wind direction combination, the wake region is iteratively updated using wind farm simulation tools, and the actual wind speed of the wind turbine is calculated based on the Gaussian-curl energy conservation wake model. Calculate the total field output power based on the objective function and constraints of the field control layer; Set a maximum number of iterations, calculate the wind turbine's gradient in each iteration, and update the yaw angle; If the difference between the yaw angles of the two consecutive iterations is less than the convergence threshold, then output the current yaw angle and exit the iteration. Based on the yaw angle obtained through iteration, a local traversal optimization is performed within a set local search range, and the optimal yaw angle corresponding to the maximum power across the entire field is selected. The optimal yaw angle for each wind speed-direction combination is obtained by iterative analysis, and then used for real-time yaw angle control in the engine control layer.

3. The active wake control method for wind farms based on turbine-farm cooperation according to claim 1, characterized in that, The objective function of the field control layer is to maximize the total output power P of the wind farm. In the formula, v i (α i ,θ) represents the yaw angle α of wind turbine i. i The actual wind speed at the wind direction angle θ; α i P is the yaw angle of wind turbine i; N is the total number of wind turbines in the wind farm; i The output power of fan i: In the formula, ρ is the air density; A is the swept area of ​​the wind turbine; C P v is the power coefficient; i f represents the actual wind speed of fan i; p This is the power attenuation coefficient.

4. The active wake control method for wind farms based on turbine-farm cooperation according to claim 1, characterized in that, The actual wind speed of the wind turbine is calculated using a Gaussian-curl energy conservation wake model, which introduces a curl term based on the principle of angular momentum conservation. The total energy loss of wind turbine i is the superposition result of the influence of each upstream wind turbine. The actual wind speed v i The calculation is as follows: Wherein, Γ represents a collection of wind turbines that are only affected by a single wake; The wind speed received by fan i: Where v0 is the incoming wind speed from a distance, D is the diameter of the fan impeller, and k a K is the wake depth coefficient. b Let α be the wake diffusion coefficient. j Let y be the yaw angle of wind turbine j, δ be the wake spread width, and y be the yaw angle of wind turbine j. o x is the radial distance of fan i. o The axial distance of fan i: x i =Lcos(|θ i -θ+0.3C T a j |) Where L is the distance between wind turbines i and j; θ is the wind direction angle; θ i C is the angle between the line connecting fan i and j and the y-axis. T This is the thrust coefficient.

5. The active wake control method for wind farms based on turbine-farm cooperation according to claim 1, characterized in that, The machine control layer employs a model predictive control strategy, which predicts future states within a finite time domain and performs rolling optimization to determine the optimal yaw control sequence. The specific control process is as follows: The disturbance estimation fed back by the disturbance observer is used to correct the control input to complete the disturbance compensation. Then, the system state at the next moment is predicted based on the wind turbine state update and the system output state space equation according to the corrected control input. Minimize the objective function of the rolling solution optimization problem within the optimization time window, and continuously update the optimal control sequence; The controller executes only the first control action in the current optimization sequence and sends the control action to the yaw actuator; at the same time, it collects the actuator's real-time response data for the next step of state estimation and disturbance observer update; The disturbance observer estimates the external disturbance based on the deviation between the predicted yaw angle and the actual measured value, and feeds the disturbance estimate back to the prediction model for disturbance compensation.

6. The active wake control method for wind farms based on turbine-farm cooperation according to claim 1, characterized in that, The objective function of the machine control layer comprehensively considers the yaw angle tracking error and the fatigue load of the yaw bearing, and the reference trajectory is given by the optimal yaw angle generated by the field control layer. Where λ is the weighting coefficient of multi-objective control; e i,t+k and L i,t+k These are the yaw angle tracking error and yaw bearing fatigue load of unit i at time t+k; Where, α i,t+k α i,t+k Yaw angle of unit i at time t+k and yaw angle reference input; K yaw β is the fatigue load stiffness coefficient; β is the gain coefficient used to account for the influence of wind speed v on load estimation.

7. The active wake control method for wind farms based on turbine-farm cooperation according to claim 1, characterized in that, The state-space model for updating the state of wind turbine i is represented as follows: X i,t+k+1 =A s X i,t+k +B s u i,t+k +ζ i,t+k ,k=0,…,T-1 Among them, A S State transition matrix, B S To control the input matrix, ζ i,t+k For external disturbance terms; The system output equation is expressed as: Y i,t+k =C s X i,t+k +n i,t+k ,k=0,…,T-1 Among them, Y i,t+k The output vector contains the predicted yaw angle. C S For the output matrix, n i,t+k For measuring noise.

8. The active wake control method for wind farms based on turbine-farm cooperation according to claim 1, characterized in that, The constraints of the machine control layer include: mechanical safety constraints on yaw angle and yaw angular velocity, yaw start threshold constraints, fatigue load constraints caused by yaw, and energy consumption constraints during yaw.

9. A wind farm active wake control device based on turbine-farm collaboration, comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the active wake control method for wind farms based on farm-machine collaboration as described in any one of claims 1-8.

10. A storage medium having a program stored thereon, characterized in that, When the program is executed, it implements the active wake control method for wind farms based on machine-farm cooperation as described in any one of claims 1-8.

Images