Hybrid control method for output power of offshore wind plant under wake flow influence

Abstract

The invention discloses a hybrid control method for output power of an offshore wind plant under wake flow influence. The hybrid control method comprises the following steps: S1, acquiring operation parameters V0 and P of the offshore wind plant in real time; s2, inputting V0 and P into a controller, and outputting an axial induction factor vector a of a fan in the whole offshore wind plant by the controller; s3, adjusting each fan in the offshore wind plant according to the vector a; and repeating the steps S1 to S3. According to the hybrid optimization method for axial induction factor control, the output power can be remarkably improved in the actual operation of a wind power plant, real-time optimization and updating of the fan control quantity can be achieved under the dynamic wind condition and the uncertain wake flow condition, the adaptive capacity of a control strategy to environment disturbance is enhanced, and the wind power generation efficiency is improved. And the total power generation efficiency and the operation robustness of the wind power plant are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of wind farms, and in particular to a hybrid control method for output power of an offshore wind farm under the influence of wake. Background Art

[0002] With the development of wind power technology and the ongoing transformation of the global energy mix, the scale of offshore wind farms continues to expand, and their share of total installed renewable energy capacity is increasing. However, the large-scale integration of wind farms also presents numerous operational and control challenges. First, the densely packed layout of wind turbines creates a significant wake effect, which degrades wind conditions for downstream turbines. This not only reduces wind energy efficiency but also exacerbates fatigue load accumulation on the turbines, shortening their lifespan. Second, the high volatility and measurement uncertainty of actual wind speeds further increase the unpredictability of wind farm operations, necessitating robust and adaptable control methods. To address these challenges, researchers have recently focused their control efforts on regulating the axial induction factor (AIF) of each wind turbine in a wind farm. By rationally allocating AIF to each turbine, this approach improves overall power generation performance, load distribution, and system economics under conditions of wake interference and inaccurate wind conditions, becoming a key path to achieving efficient offshore wind farm operation.

[0003] Currently, in order to improve the power generation capacity of offshore wind farms, wind turbine axial induction factor control methods can be mainly divided into three categories: model-based methods, data-based methods, and hybrid control methods that integrate the two. Model-based methods rely on accurate modeling of the wind farm wake propagation and power output mechanism. Introducing system structure characteristics in the controller design phase can achieve good convergence speed and optimization performance. Some studies use the alternating direction multiplier method to achieve distributed optimization and control of the axial induction factor, and some use collaborative optimization or model predictive control methods to improve the overall power generation and alleviate the fatigue load of the unit. However, the model method is highly sensitive to wind parameters. Modeling errors or inaccurate sensor measurements may lead to deviations in system behavior, limiting its adaptability in actual operating scenarios.

[0004] To overcome these limitations, some research has turned to exploring data-driven axial induction factor control methods. These methods do not rely on prior models of the wind farm. Instead, they update strategies based on online wind turbine operating data through mechanisms such as reinforcement learning, Markov decision processes, or machine learning. These methods exhibit a certain degree of robustness and exploration capabilities. Some studies have used deep reinforcement learning to obtain optimal axial induction factor setting strategies, or have used genetic algorithms to achieve a trade-off between power generation efficiency and equipment health. However, these methods typically have high requirements for data volume and exploration cycles, and suffer from slow convergence and large fluctuations in control performance.

[0005] In order to balance convergence speed and environmental adaptability, hybrid control methods have gradually become a research hotspot in recent years. This type of method attempts to embed data-driven policy search into a model-driven optimization framework, improving policy robustness while maintaining optimization efficiency. Some studies have proposed first obtaining a high-quality initial policy through free model exploration, and then optimizing the wind farm output power with the help of an accurate model. Although hybrid methods have good overall performance, most current studies still only verify their effectiveness under static or ideal wind conditions, and do not fully consider the impact of wind speed measurement errors or wind fluctuations on control performance during actual operation. As a result, the applicability of the method in real wind farms is still insufficient.

[0006] In summary, the existing technology has the following problems:

[0007] 1) While existing offshore wind farm axial induction factor control methods perform well under ideal wind conditions, single-model driven methods are susceptible to model errors and can fall into local optimality in real-world operating situations, such as wind speed measurement errors and wake uncertainty. While purely data-driven strategies offer some robustness, they suffer from slow convergence and low optimization efficiency. Existing methods have poor adaptability, easily leading to system optimization failure or unstable performance, making it difficult to achieve efficient wind farm operation.

[0008] 2) Traditional control has a single objective, focusing only on maximizing power generation, without considering the structural fatigue problems that may be caused by frequent adjustments to the axial induction factor. It is difficult to ensure the healthy operation of the wind turbine while increasing power.

[0009] Therefore, those skilled in the art are committed to developing a hybrid control method for output power of an offshore wind farm under the influence of wake. Summary of the Invention

[0010] To achieve the above object, the present invention provides a hybrid control method for output power of an offshore wind farm under the influence of wake, comprising the following steps:

[0011] S1: Real-time acquisition of offshore wind farm operating parameters V0 and P, where V0 is the free wind speed of the wind farm and P is the active power vector of all wind turbines in the farm obtained by the SCADA system;

[0012] S2: Input V0 and P into the controller, and the controller outputs the axial induction factor vector a of the wind turbines in the entire offshore wind farm;

[0013] S3: Regulate each wind turbine in the offshore wind farm according to vector a;

[0014] Repeat S1-S3.

[0015] Furthermore, the objective function of the controller can be specifically:

[0016]

[0017] In the formula, the wind turbines in the entire offshore wind farm are is the axial induction factor vector of the wind turbines in the entire offshore wind farm, is the initial axial induction factor vector set by the wind farm control center, μ is the regulation cost coefficient, and λ is the wind power generation price. is the active power vector of all wind turbines in the field, is the feasible region of the axial induction factor.

[0018] Furthermore, the control strategy of the controller includes a gradient estimation stage and a momentum update stage.

[0019] Furthermore, the gradient estimation stage includes joint estimation in both model-driven and data-driven directions.

[0020] Furthermore, the specific iterative process of the gradient estimation stage is:

[0021]

[0022] Where k represents the kth iteration, a k is the vector of axial induction factors of wind turbines in the entire offshore wind farm at the kth iteration, P k is the power vector of all wind turbines at the kth iteration, is the model driving direction obtained based on the power sensitivity matrix, is the data-driven direction based on residual feedback, is the sensitivity matrix calculated from the current wind speed and induction factor at the kth iteration, v k is a random vector uniformly sampled on the unit sphere, satisfying ||v||=1, β k is the weighting coefficient of model-driven and data-driven directions.

[0023] Furthermore, the weight of the model-driven direction is higher in the initial stage of iteration, and the weight of the data-driven direction gradually increases in the later stages of iteration.

[0024] Furthermore, the momentum update stage uses the heavy ball method to update the momentum.

[0025] Furthermore, the specific process of the heavy ball method is:

[0026] w k =a k

[0027]

[0028] Where η is the step size, is the momentum factor, ζv k+1 To explore perturbations,

[0029] First update the candidate solution wk , and then use the integrated gradient direction Perform intermediate variable z k Update and use the momentum term to correct the candidate solution trajectory w k+1 , add a certain exploration disturbance to avoid the solution falling into the local optimum, and finally project the candidate solution to the feasible region to obtain a new axial induction factor value a k+1 .

[0030] In a second aspect, the present invention further discloses a device for controlling output power of an offshore wind farm under the influence of wake, comprising:

[0031] Data acquisition module: used to obtain the operating parameters V0 and P of offshore wind farms in real time;

[0032] Control module: used to input offshore wind farm operating parameters V0 and P and output the axial induction factor vector a of the wind turbines in the entire offshore wind farm;

[0033] Execution module: used to adjust each wind turbine according to the axial induction factor vector a of the wind turbines in the entire offshore wind farm output by the controller module.

[0034] In a third aspect, the present invention further discloses a computer device, comprising:

[0035] Memory;

[0036] one or more processors coupled to the memory;

[0037] One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by one or more processors, and the one or more applications are configured to execute the control method according to any one of claims 1 to 8.

[0038] Technical Effects

[0039] 1) This paper addresses the common issues of wind speed inaccuracy and dynamic wake interference in wind farm operation by proposing a hybrid optimization method for axial induction factor (AIF) control. This method integrates three technical mechanisms: first, a model-driven stochastic gradient descent method is used to construct approximate guidance information, providing a fast convergence infrastructure; second, a data-driven zero-order residual feedback mechanism is introduced to enhance the ability to explore unknown perturbations and local extrema; and third, a heavy ball acceleration strategy that incorporates momentum information effectively improves the algorithm's convergence speed and stability. This paper combines the rapid optimization capabilities of a model-driven approach with the exploration capabilities of a data-driven approach to design a convergence-guaranteed hybrid control framework. It also introduces a momentum acceleration mechanism for iterative updates, thereby improving the robustness and convergence efficiency of the controller under uncertain wind conditions. This method can significantly improve output power during actual wind farm operation. Under dynamic wind conditions and uncertain wake conditions, it enables real-time optimization and updating of wind turbine control variables, enhances the control strategy's adaptability to environmental perturbations, and effectively improves the wind farm's overall power generation efficiency and operational robustness. Furthermore, it reduces the algorithm's computational burden, facilitating project deployment and long-term operation.

[0040] 2) The present invention designs a wind farm control objective function that takes into account both power generation benefits and mechanical degradation costs, achieving an optimal balance between power generation efficiency and unit life. The present invention explicitly introduces the fatigue cost of wind turbines into the control objective function, and by weighing the power benefits and regulation costs during the optimization process, guides the rational distribution of axial induction factors, and avoids excessive power pursuit and excessive regulation of local wind turbines. The present invention can extend the service life of wind turbines, reduce maintenance costs, and improve the economic efficiency and safety of wind farm operations throughout their entire life cycle without significantly affecting the power generation level, and has good engineering applicability prospects.

[0041] It is understandable that the beneficial effects of the second and third aspects mentioned above can be found in the above related descriptions and will not be repeated here.

[0042] The concept, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose, characteristics and effects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a flow chart of a hybrid control method for offshore wind farm output power under the influence of wake in a preferred embodiment of the invention;

[0044] Figure 2 This is a schematic diagram of a single fan wake model according to a preferred embodiment of the present invention;

[0045] Figure 3 This is a schematic diagram of a simulation of the wake field of an 8-wind turbine scenario according to a preferred embodiment of the present invention;

[0046] Figure 4This is a diagram showing the convergence effect of the axial induction factor under standard wind conditions of a preferred embodiment of the present invention.

[0047] Figure 5 This is a diagram showing the improvement effect of the standard wind condition objective function of a preferred embodiment of the present invention;

[0048] Figure 6 This is a diagram showing the improvement effect of the actual wind condition objective function of a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0049] The following describes several preferred embodiments of the present invention with reference to the accompanying drawings to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0050] In the drawings, components with identical structures are denoted by the same reference numerals, and components with similar structures or functions are denoted by similar reference numerals. The size and thickness of each component shown in the drawings are arbitrary and are not limited by the present invention. For clarity, the thickness of components in some places in the drawings is appropriately exaggerated.

[0051] like Figure 1 As shown, the present invention provides a hybrid control method for output power of an offshore wind farm under the influence of wake, comprising the following steps:

[0052] S1: Real-time acquisition of offshore wind farm operating parameters V0 and P, where V0 is the free wind speed of the wind farm and P is the active power vector of all wind turbines in the farm obtained by the SCADA system;

[0053] S2: Input V0 and P into the controller, and the controller outputs the axial induction factor vector a of the wind turbines in the entire offshore wind farm;

[0054] S3: Regulate each wind turbine in the offshore wind farm according to vector a;

[0055] Repeat S1-S3.

[0056] Furthermore, the objective function of the controller can be specifically:

[0057]

[0058] In the formula, the wind turbines in the entire offshore wind farm are is the axial induction factor vector of the wind turbines in the entire offshore wind farm, is the initial axial induction factor vector set by the wind farm control center, μ is the regulation cost coefficient, and λ is the wind power generation price. is the active power vector of all wind turbines in the field, is the feasible region of the axial induction factor.

[0059] Furthermore, the control strategy of the controller includes a gradient estimation stage and a momentum update stage.

[0060] Furthermore, the gradient estimation stage includes joint estimation in both model-driven and data-driven directions.

[0061] Furthermore, the specific iterative process of the gradient estimation stage is:

[0062]

[0063] Where k represents the kth iteration, a k is the vector of axial induction factors of wind turbines in the entire offshore wind farm at the kth iteration, P k is the power vector of all wind turbines at the kth iteration, is the model driving direction obtained based on the power sensitivity matrix, is the data-driven direction based on residual feedback, is the sensitivity matrix calculated from the current wind speed and induction factor at the kth iteration, v k is a random vector uniformly sampled on the unit sphere, satisfying ||v||=1, β k is the weighting coefficient of model-driven and data-driven directions.

[0064] Furthermore, the weight of the model-driven direction is higher in the initial stage of iteration, and the weight of the data-driven direction gradually increases in the later stages of iteration.

[0065] Furthermore, the momentum update stage uses the heavy ball method to update the momentum.

[0066] Furthermore, the specific process of the heavy ball method is:

[0067] w k =a k

[0068]

[0069] Where η is the step size, is the momentum factor, ζv k+1 To explore perturbations,

[0070] First update the candidate solution w k , and then use the integrated gradient direction Perform intermediate variable z k Update and use the momentum term to correct the candidate solution trajectory w k+1 , add a certain exploration disturbance to avoid the solution falling into the local optimum, and finally project the candidate solution to the feasible region to obtain a new axial induction factor value a k+1 .

[0071] (The following is a further description of the above specific implementation method)

[0072] 1. Overall description of the control scheme

[0073] The active power output of a wind farm is affected by the interaction between the operating status of its wind turbines and their wakes. Especially in offshore wind farms with densely distributed turbines, the wake effect can worsen wind resource conditions for downstream turbines, reducing overall wind energy utilization and power generation, and increasing mechanical damage to equipment during turbine adjustments. Therefore, in wind farm control, it is necessary to actively regulate the wake effect by coordinating the axial induction factors of each turbine, thereby improving overall power output and reducing the risk of structural degradation.

[0074] Traditional control schemes often optimize the wind turbine's axial induction factor (AIF) based on model-driven approaches. However, these approaches often rely on precise modeling and struggle to adapt to real-world operating conditions, such as wind speed measurement errors and wake propagation uncertainties. Some studies have attempted to use data-driven approaches for policy learning. While these approaches offer some robustness, they often suffer from slow convergence and unstable control performance, making them difficult to deploy in real-world wind farms over the long term. To address these issues, this embodiment proposes a hybrid control framework that combines model-driven and data-driven approaches. Specifically, the system estimates directional gradients based on a wind farm wake-power model and introduces a real-time output feedback correction mechanism to adapt to actual wind disturbances, thereby dynamically adjusting the AIF of the wind turbine. To improve the controller's convergence performance, a heavy ball acceleration mechanism is further introduced. In each iteration, the current search direction is corrected based on historical search trajectories, resulting in faster convergence of the objective function. This control scheme can achieve stable wind farm operation and target power improvement even under conditions of inaccurate wind speed measurements and incomplete wake modeling, demonstrating excellent engineering adaptability and potential for widespread adoption.

[0075] 2. Introduction to offshore wind farm wake-power model

[0076] (1) Wind farm wake model

[0077] like Figure 2 As shown, consider an offshore wind farm with n wind turbines, where the set of wind turbines is N = {1, 2, ..., n}, where turbine j is located upstream of turbine i. To describe the wake interference relationship between wind turbines in an offshore wind farm, this embodiment uses the Park model to model wake velocity loss. The wake gradually expands with downstream propagation distance in the horizontal plane, and its impact range is determined by the downstream distance x and the lateral distance r of the wake centerline.

[0078] The velocity loss caused by fan i to a point (x, r) downstream is defined as

[0079]

[0080] Among them, a i is the axial induction factor of the fan, D i is the rotor diameter of the wind turbine, and m is the expansion coefficient of the wake radius. If there are multiple upstream wind turbines in the wind farm, their wake intervals may overlap with each other. In this case, the wind speed loss at wind turbine i should comprehensively consider the wake contributions of all upstream wind turbines, and the linear superposition model can be used to express it as

[0081]

[0082] in, is the axial induction factor vector of the wind turbine in the entire wind farm. At this time, the effective wind speed of the wind turbine is

[0083] V i (a)=V0(1-δV i (a))

[0084] Among them, V0 is the free wind speed of the wind farm.

[0085] (2) Wind farm power model

[0086] Under the premise of known wind speed, the output power of the wind turbine can be calculated from its axial induction factor. Assuming that the wind turbines in the wind farm are of the same model, the output power of the i-th wind turbine can be expressed as

[0087]

[0088] Where ρ is the air density, A is the swept area of ​​the wind wheel calculated from the blade diameter, and C P (a i )=4a i (1-a i ) 2 Based on this, under the condition that the axial induction factor and wake flow are coupled with each other, a mapping relationship between the axial induction factor of the wind turbine and the power output is established.

[0089] 3. Introduction to offshore wind farm power optimization model

[0090] (1) Introduction to the comprehensive optimization problem of power generation revenue and adjustment cost

[0091] Since the adjustment of the axial induction factor will cause changes in the wind turbine pitch angle and torque controller, thus affecting the fatigue life of the unit, it is necessary to consider the impact on the wind turbine operation stability while improving the power generation performance. To this end, this embodiment constructs the following optimization problem

[0092]

[0093] in, is the initial axial induction factor vector set by the wind farm control center, μ is the regulation cost coefficient, and λ is the wind power generation price. is the power vector of all wind turbines in the field, is the feasible region of the axial induction factor. This control objective is to improve the overall power output of the wind farm without significantly deviating from the original settings of the wind turbines and causing structural damage.

[0094] (2) Introduction to the simplified objective function of axial induction factor control

[0095] In actual control, the wind turbine power output P can be expressed as a function of the axial induction factor a, so the optimization problem can be equivalently transformed into a constrained optimization problem only about the control quantity, that is, the axial induction factor. The simplified objective function is defined as

[0096]

[0097] C P (a i )=4a i (1-a i ) 2

[0098] V i (a)=V0(1-δV i (a))

[0099]

[0100] This problem has clear engineering physics implications: adjusting the axial induction factor under conditions of wake interference and uncertain wind conditions can improve the economic efficiency and system reliability of wind farms. The simplified objective function will be used in algorithm iteration and property verification.

[0101] 4. Design of hybrid controller for fan axial induction factor

[0102] (1) Introduction to hybrid controller structure

[0103] The wind farm controller proposed in this invention is based on a feedback control framework and dynamically adjusts the axial induction factor of each wind turbine through an iterative optimization method, aiming to improve the overall power generation of the wind farm and suppress structural damage.

[0104] The controller design combines model-driven and data-driven optimization approaches and introduces a momentum mechanism to improve the smoothness and convergence speed of the search trajectory. This controller, known as the "Heavy-ball Feedback Optimization (HBFO)" controller, operates in two phases: the gradient estimation phase and the momentum update phase. The iterative process is described below.

[0105] 1) Gradient estimation stage

[0106] This stage is used to determine the direction of decrease of the current axial induction factor. To enhance the robustness of the controller to wind condition uncertainty, a joint estimation method based on model-driven and data-driven approaches is introduced. First, based on the axial induction factor of each wind turbine in the current iteration, the wind speed and power generation under the influence of the wake are calculated. Then, the following iterative process is performed:

[0107]

[0108] in, It is the model driving direction obtained based on the power sensitivity matrix, reflecting the power change trend under the model prediction and has a good convergence speed; In the data-driven direction based on residual feedback, the difference between the current and previous iteration objective function values ​​is used to estimate the downward trend, which has stronger adaptability to model errors. is the sensitivity matrix calculated from the current wind speed and induction factor at the kth iteration, reflecting the impact of the change in axial induction factor on power output. k is a random vector uniformly sampled on the unit sphere, satisfying ||v||=1. k is the weighted coefficient of the model-driven and data-driven directions. In the initial stage of iteration, the model-driven direction has a higher weight to accelerate convergence; in the later stage, the data-driven direction is gradually strengthened to improve environmental adaptability.

[0109] 2) Momentum update phase

[0110] In this stage, the heavy-ball method is used to update the momentum to avoid the optimization falling into the local optimum and improve the iterative stability. The specific process is as follows:

[0111] w k =a k

[0112]

[0113] Where η is the step size, is the momentum factor, ζv k+1 To explore perturbations,

[0114] First update the candidate solution w k , and then use the integrated gradient direction Perform intermediate variable z k Update and use the momentum term to correct the candidate solution trajectory w k+1 , add a certain exploration disturbance to avoid the solution falling into the local optimum, and finally project the candidate solution to the feasible region to obtain a new axial induction factor value a k+1 .

[0115] In summary, in each iteration, the controller comprehensively utilizes the model sensitivity information and real-time performance feedback based on the current wind conditions and power generation status to update the axial induction factor of each wind turbine, thereby gradually approaching the optimal value of the objective function and achieving a comprehensive improvement in the overall output performance and operational reliability of the wind farm.

[0116] (2) Controller convergence analysis

[0117] In order to theoretically verify the effectiveness of the heavy ball feedback optimization (HBFO) controller proposed in this embodiment, the convergence of its simplified objective function is analyzed. First, it can be seen that The axial induction factor a is M-Lipschitz and L-smooth. In the entire feasible region is bounded, that is, there exists a constant G such that for any Both

[0118] On the basis of satisfying the above properties, we further analyze the error source of the gradient estimation in the controller and establish its upper bound. The error between the approximate sensitivity matrix and the true sensitivity matrix is ​​defined as

[0119]

[0120] Then the error between the integrated descent direction in the controller proposed by the present invention and the true gradient of the system satisfies the following limit

[0121]

[0122] in, is the overall upper bound of the mixed estimation error. The three terms on the right side of the inequality sign represent the model error, the model-independent estimation error, and the additional bias caused by the disturbance. Further, the square error bound is

[0123]

[0124] Where Q is a constant, σ 2 Estimate the upper bound of the variance for the controller's descent direction. The following conclusion about the overall convergence can be obtained: Assume that the controller runs T iterations with a step size of

[0125]

[0126] Then it satisfies the following asymptotic convergence property

[0127]

[0128] in, Represents the optimal objective function value. The convergence term on the right side indicates that the expected value of the square of the gradient norm is Sublinear velocity decay.

[0129] In summary, the controller proposed in this embodiment still has asymptotic convergence properties in the presence of model inaccuracies and disturbances, and can achieve the preset control objectives.

[0130] 5. Verify the control method using an actual offshore wind farm system

[0131] In order to verify the effectiveness of the proposed control method, this embodiment built an 8-wind turbine system based on the Horns RevI offshore wind farm layout on the MATLAB platform. The wind turbines are arranged at 7 times the rotor diameter (7D), and the wind turbines uniformly adopt the NREL-5MW model. The control variable is the axial induction factor of each wind turbine, and its value range is set to [0,0.5], and the nominal initial value is 1 / 3. The system takes into account the model uncertainty caused by the wake interference and wind speed measurement error, and uses the control effect under actual operating conditions as the evaluation basis. The wake field simulation of this verification scenario is as follows Figure 3 shown.

[0132] In order to verify the effectiveness of our proposed method, the following three sets of comparative experiments are carried out:

[0133] (1) Axial induction factor convergence comparison experiment: Under the jump wind condition where the wind speed is set to 8m / s initially and switched to 10m / s at the 1000th iteration, random Gaussian noise is superimposed every 10 steps to simulate the inaccuracy of the wind speed, and the axial induction factor convergence performance of the three control methods in this environment is compared. The methods involved in the comparison are: the heavy ball feedback optimization method (HBFO) proposed in this invention, the standard stochastic gradient descent method (SGD), and the unaccelerated hybrid control method (Hybrid). In order to obtain an accurate reference benchmark, the ideal model prior results under known wind conditions are used as a control;

[0134] (2) Comparative experiment on the improvement effect of objective function: Under the same wind condition jump and inaccuracy environment as Experiment 1, the improvement effect of different methods on the objective function of the wind farm is evaluated. The objective function comprehensively considers the wind power benefit and the mechanical degradation cost caused by the adjustment of the axial induction factor, and the optimal solution obtained by offline calculation using FLORIS software under known precise wind conditions is used as the theoretical upper limit benchmark. The methods involved in the comparison include: the HBFO method proposed in this invention, the non-accelerated hybrid method, and the Greedy control strategy;

[0135] (3) Verification of the objective function improvement effect under measured SCADA data: To evaluate the performance of the proposed controller under actual wind conditions, 30 sets of continuous wind speed samples were selected from the SCADA data of an offshore wind farm in China as input scenarios. Each set of samples corresponds to a 10-minute period, covering a total operating cycle of 300 minutes. For each wind condition sample, the objective function was calculated using the following three methods: the HBFO method proposed in this paper, the model-based control method (Model-Based), and the greedy method. The results of each method during the iterative process were compared.

[0136] Experiments were carried out under three different conditions and the following experimental results were obtained.

[0137] The results of experiment (1) are as follows Figure 4 The results show that the HBFO method can quickly converge to the reference result and maintain strong robustness and exploration capabilities in the presence of disturbances. In contrast, the traditional hybrid method significantly prolongs its convergence time under the influence of disturbances. The stochastic gradient descent method based on a static model has a sluggish response to AIF adjustments in the range of wind conditions and exhibits significant oscillation, making it difficult to converge and lacking adaptability.

[0138] The results of experiment (2) are as follows Figure 5 As shown in Figure 2 . Simulation results show that at wind speeds of 8 m / s, the HBFO method only differs from the FLORIS results by approximately 20.02%, significantly outperforming both the Hybrid and Greedy methods. At wind speeds of 10 m / s, the difference narrows to 12.27%. The proposed HBFO controller effectively adjusts the control strategy in real time under the periodic wind speed disturbances present in actual wind data, ensuring that the cluster power trajectory remains close to the static optimal power level obtained using offline FLORIS optimization. Especially during periods of significant wind fluctuations and disturbances, the HBFO method demonstrates excellent dynamic response, rapidly adjusting the AIF control strategy to follow the optimal power trend and avoiding significant power offsets and lags. In contrast, the average power output of other control methods remains significantly lower than the offline optimal decision during wind speed disturbances. Overall, the HBFO controller demonstrates excellent optimality approximation and energy capture efficiency under dynamic disturbances. It not only exhibits good convergence performance but also achieves real-time control results that are closer to the ideal optimization result.

[0139] The results of experiment (3) are as follows Figure 6 The HBFO method achieves higher objective function values ​​in most time periods, and its performance curve is significantly improved compared to the other two methods. This shows that it has strong environmental adaptability and robustness under actual wind fluctuation scenarios, can effectively optimize the overall power generation capacity of wind farms, and has good application potential in improving the power generation efficiency of offshore wind farms.

[0140] The preferred embodiments of the present invention have been described in detail above. It should be understood that numerous modifications and variations based on the concepts of the present invention are possible without inventive effort by those skilled in the art. Therefore, any technical solution that can be derived by one skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A hybrid control method for output power of an offshore wind farm under the influence of wake, characterized in that: The following steps are involved: S1: Real-time acquisition of offshore wind farm operating parameters V0 and P, where V0 is the free wind speed of the wind farm and P is the active power vector of all wind turbines in the farm obtained by the SCADA system; S2: Input V0 and P into the controller, and the controller outputs the axial induction factor vector a of the wind turbines in the entire offshore wind farm; S3: Regulate each wind turbine in the offshore wind farm according to vector a; Repeat S1-S3.

2. The control method according to claim 1, wherein: The objective function of the controller is: Where, the set of wind turbines in the entire offshore wind farm is N = {1, 2, ..., n}, is the axial induction factor vector of the wind turbines in the entire offshore wind farm, is the initial axial induction factor vector set by the wind farm control center, μ is the adjustment cost coefficient, λ is the wind power generation price, is the active power vector of all wind turbines in the field, is the feasible region of the axial induction factor.

3. The control method according to claim 1, wherein: The control strategy of the controller includes a gradient estimation stage and a momentum update stage.

4. The control method according to claim 3, wherein: The gradient estimation stage includes joint estimation in two directions: model-driven and data-driven.

5. The control method according to claim 4, wherein: The specific iterative process of the gradient estimation stage is: Where k represents the kth iteration, a k is the vector of axial induction factors of wind turbines in the entire offshore wind farm at the kth iteration, P k is the power vector of all wind turbines at the kth iteration, is the model driving direction obtained based on the power sensitivity matrix, is the data-driven direction based on residual feedback, ,i,j∈N is the sensitivity matrix calculated by the current wind speed and induction factor at the kth iteration, v k is a random vector uniformly sampled on the unit sphere, satisfying ||v||=1, β k is the weighting coefficient of model-driven and data-driven directions.

6. The control method according to claim 5, wherein: The weight of the model-driven direction is higher in the initial stage of iteration, and the weight of the data-driven direction gradually increases in the later stage of iteration.

7. The control method according to claim 3, wherein: The momentum updating stage adopts the heavy ball method to perform momentum updating.

8. The control method according to claim 7, wherein: The specific process of the heavy ball method is: w k =a k In k+1 =z k +θ(z k -With k-1 ) Where η is the step size, θ is the momentum factor, ζv k+1 To explore perturbations, First update the candidate solution w k , and then use the integrated gradient direction Perform intermediate variable z k Update and use the momentum term to correct the candidate solution trajectory w k+1 , add a certain exploration disturbance to avoid the solution falling into the local optimum, and finally project the candidate solution to the feasible region to obtain the new axial induction factor value a k+1 .

9. An output power control device for an offshore wind farm under the influence of wake, characterized in that: include: Data acquisition module: used to obtain real-time operating parameters of offshore wind farms V0, V i and P; Control module: used to input offshore wind farm operating parameters V0, V i and P and output the axial induction factor vector a of the wind turbines in the entire offshore wind farm; Execution module: used to adjust each wind turbine according to the axial induction factor vector a of the wind turbines in the entire offshore wind farm output by the controller module.

10. A computer device, characterized in that: include: Memory; one or more processors coupled to the memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by one or more processors, and the one or more applications are configured to execute the control method according to any one of claims 1 to 8.

Images