Multi-time scale combined wind power plant operation control method, device and equipment

Abstract

The invention discloses a multi-time-scale combined wind power plant operation control method, device and equipment, and belongs to the field of wind power plant operation control, and the method comprises the steps: obtaining the wind condition measurement data of a target wind power plant, carrying out the wind power prediction, obtaining the predicted wind condition data of the target wind power plant in a prediction period, and obtaining the predicted wind condition data of the target wind power plant; inputting the wind regime measurement data and the predicted wind regime data into a multi-time-scale combined control model, so that the multi-time-scale combined control model determines a yaw reference angle of each fan according to a first optimization function corresponding to a large-time-scale control layer and the predicted wind regime data, generates an initial yaw angle control scheme, and outputs the initial yaw angle control scheme; and according to the second optimization function corresponding to the small time scale control layer and the current measurement data, adjusting the initial yaw control scheme, outputting the target yaw angle control scheme, and controlling the target wind power plant according to the target yaw angle control scheme, so that by implementing the method and the system, the mechanical loss can be reduced, and the wind power generation efficiency can be improved. Therefore, the service life of the fan is prolonged.

Description

Technical Field

[0001] The present invention relates to the field of wind farm operation control, and in particular to a wind farm operation control method, device and equipment combining multiple time scales. Background Art

[0002] With the acceleration of the global energy transition, the proportion of wind power installed capacity has increased from 9.03% to 15.55%, and wind power is becoming increasingly important as a clean energy source. However, the wake effect, which is prevalent in wind farms, can reduce the power generation capacity of downstream wind turbines by 60%, resulting in a loss of up to 4% of the total power generation of the entire wind farm. It also increases fatigue loads on wind turbines and shortens equipment life. Furthermore, power grids with high wind power penetration rely on wind farms to provide auxiliary services such as secondary frequency regulation. However, actual wind conditions are subject to significant uncertainty and volatility due to weather and turbulence, further exacerbating the complexity of multi-objective optimization control of wind farms.

[0003] Currently, the mainstream control strategies for wind farms include maximum power point tracking (MPPT) and yaw control based on a lookup table (LUT). The MPPT strategy increases power generation by maximizing the power coefficient of a single machine, but its non-yaw operation mode exacerbates the wake effect, resulting in power generation losses for downstream wind turbines. The LUT method reduces the computational burden by pre-storing the offline optimized yaw angle, but suffers from poor robustness and inability to adapt to dynamic changes in the wind farm, and frequent yaw actions accelerate mechanical wear. In recent years, model predictive control (MPC) has been introduced into wind farm optimization, such as maximizing power generation or power tracking based on computational fluid dynamics (CFD) or data-driven dynamic wake models. However, the computational cost of existing methods based on the Navier-Stokes equations or neural networks is too high to meet real-time control requirements; while MPC based on analytical models such as FLORIS improves efficiency, it generally ignores the multi-objective optimization of fatigue loads and the ability to predict wind conditions, resulting in frequent control strategies in fluctuating environments, exacerbating mechanical losses.

[0004] Therefore, how to reduce the mechanical loss of the wind turbine during the yaw angle adjustment process is a technical problem that needs to be solved at present. Summary of the Invention

[0005] The present invention provides a wind farm operation control method, device and equipment combining multiple time scales, which can solve the problem of how to reduce the mechanical loss of the wind turbine during the control process to increase the service life of the wind turbine.

[0006] The present invention provides a wind farm operation control method combining multiple time scales, comprising:

[0007] Obtaining wind condition measurement data of a target wind farm, and performing wind power forecasting based on the wind condition measurement data to obtain forecasted wind condition data of the target wind farm within a forecast period; wherein the target wind farm includes a plurality of wind turbines, and the wind condition measurement data includes historical measurement data and current measurement data;

[0008] The wind condition measurement data and the predicted wind condition data are input into a control model combined with multiple time scales, so that the control model combined with multiple time scales determines the yaw reference angle of each wind turbine according to the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, generates an initial yaw angle control scheme, and adjusts the initial yaw control scheme according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and outputs a target yaw angle control scheme; wherein the large time scale is a time scale determined according to the wind condition stability corresponding to the current measurement data, and the small time scale is a time scale determined according to the wind turbine arrangement and control requirements of the target wind farm, and the initial yaw angle control scheme is used to characterize the yaw angle reference angle of each wind turbine within the prediction period;

[0009] The target wind farm is controlled according to the target yaw angle control scheme.

[0010] In an embodiment of the present invention, in a control layer corresponding to a large time scale determined according to the degree of wind stability, an initial yaw angle control scheme is generated based on predicted wind condition data to determine the long-term reference angle of the yaw angle of each wind turbine in a future prediction period, and in a control layer corresponding to a small time scale determined according to the wind turbine arrangement and control requirements, a target yaw angle control scheme is determined based on current wind condition data to fine-tune the yaw angle of each wind turbine in real time according to the target yaw angle control scheme, thereby reducing the mechanical wear and vibration of each wind turbine during the overall operation process, reducing mechanical losses, and thus extending the service life of the wind turbine.

[0011] Furthermore, after controlling the target wind farm, the method further includes:

[0012] Obtaining a real-time grid dispatch instruction and determining a target output power of the target wind farm, and obtaining the actual output power corresponding to the target wind farm at the current moment;

[0013] Inputting the target output power and the actual output power into the multi-timescale combined control model, so that the multi-timescale combined control model outputs an adjustment instruction for the thrust coefficient of each of the wind turbines according to a third optimization function corresponding to a micro-timescale control layer, with the goal of minimizing the difference between the target output power and the actual output power; wherein the time scale corresponding to the micro-timescale control layer is determined according to the rotational speed of the pitch angle of each of the wind turbines;

[0014] Power control is performed on the target wind farm according to the adjustment instruction of the thrust coefficient of each wind turbine.

[0015] In this way, by obtaining grid instructions in real time and adjusting the thrust coefficient of the wind turbine to minimize the difference between the target output power and the actual output power, the system can respond to grid dispatch instructions quickly and accurately.

[0016] Furthermore, the wind power forecast is performed based on the wind condition measurement data to obtain the predicted wind condition data of the target wind farm within the forecast period, specifically:

[0017] Decomposing the wind condition measurement data by an adaptive noise complete set empirical mode decomposition method to obtain an initial modal component set, wherein the initial modal component set includes a high-frequency noise component and a low-frequency trend component;

[0018] Adding the low-frequency trend components in the initial modal component set to obtain the target modal component;

[0019] The target modal component is input into a wind condition prediction module, and the predicted wind condition data is output, wherein the wind condition prediction model is a model constructed based on the Hankel dynamic modal decomposition algorithm.

[0020] This can separate the high-frequency noise components in the wind measurement data and obtain the basic trend of future wind conditions, greatly reducing the complexity of subsequent data-driven wind power prediction and enhancing the reliability of the prediction results.

[0021] Furthermore, the initial yaw control scheme is adjusted according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and a target yaw angle control scheme is output, specifically:

[0022] Based on the current measurement data, calculating a wake interaction matrix between each wind turbine, and sorting each wind turbine according to the wake interaction matrix to obtain a wind turbine sequence;

[0023] Based on the wind turbine sequence and the second optimization function, the yaw angle of each wind turbine in the initial yaw control scheme is adjusted in sequence to maximize the actual output power of the wind farm, thereby obtaining the target yaw angle control scheme.

[0024] This iterative adjustment of the wind turbine yaw angle maximizes the wind farm's power output under the wake effect. Leveraging the hierarchical indexing of the wind turbine cluster, the algorithm prioritizes upstream turbines to ensure the greatest impact on downstream turbines, increasing overall farm power generation while ensuring real-time adaptability and computational efficiency.

[0025] Furthermore, the wind condition measurement data includes wind speed values, wind direction values, and yaw angles corresponding to each wind turbine. The control model combining multiple time scales determines the yaw reference angle of each wind turbine based on the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, and generates an initial yaw angle control scheme, specifically:

[0026] Under the condition that the operation constraints and load constraints are met, the initial yaw angle control scheme is obtained by maximizing the weighted power output value of all wind turbines, wherein the weighted power output value is obtained by weighted calculation based on the wind speed value, wind direction value, yaw angle and corresponding weight value corresponding to each wind turbine.

[0027] In this way, by constraining fatigue loads and mechanical limits, over-limit operation is avoided, the balance between wind farm safety and economy is improved, and power generation efficiency during high wind speed periods is prioritized through weighted power calculation.

[0028] Furthermore, the first optimization function includes a first objective function and a first constraint function, specifically:

[0029] The expression of the first objective function is as follows:

[0030]

[0031] The expression of the first constraint function is as follows:

[0032]

[0033] γ min ≤y i ≤y max , i=1,2,…,n

[0034]

[0035] Among them, P i represents the power of the i-th fan, w t Represents the weight coefficient at different times, γ=[γ1,γ2,...,γ n ] represents the yaw angle of all wind turbines, γ max and γ min Represent the maximum yaw angle and the minimum yaw angle, v t ,θ t They represent the predicted results of wind speed and wind direction at time t, respectively. WF represents the maximum value of wind turbine load in the target wind farm, represents the maximum load when all wind turbines in the target wind farm are facing the wind, δ F Represents the coefficient.

[0036] In this way, by clarifying the multi-objective weights of power generation, load, and yaw angle, the power generation efficiency during high wind speed periods is prioritized and the operating costs of wind farms are improved.

[0037] Furthermore, the second optimization function includes a second objective function and a second constraint function, specifically:

[0038] The expression of the second objective function is as follows:

[0039]

[0040] The expression of the second constraint function is as follows:

[0041] -Δγ min ≤Δy i ≤Δy max , i=1,2,…,n

[0042]

[0043] γ min ≤y i ≤y max ;

[0044] Among them, P i represents the power of the i-th wind turbine, Δγ=[Δγ1,Δγ2,...,Δγ n ] represents the yaw adjustment angle of all wind turbines, p k Represents different wind direction scenes θ k The probability under the condition, a represents the number of scenes, γ max and γ min Represent the maximum yaw angle and the minimum yaw angle respectively, represents the yaw reference for large-time-scale model predictive control calculations, Δγ max Indicates the maximum yaw adjustment angle.

[0045] This iterative algorithm allows for dynamic evaluation and adjustment of turbine yaw based on real-time conditions, enabling the wind farm to respond to changing wind direction and turbine performance. Each optimization step is performed in real time, ensuring that the yaw angle is adjusted promptly to maximize total power output, improving real-time control.

[0046] In this way, the thrust coefficient of the wind turbine is adjusted to track the dispatch instruction curve issued by the power grid, thereby increasing the flexibility of wind farm power generation.

[0047] Another embodiment of the present invention further provides a wind farm operation control device combining multiple time scales, comprising: a data processing module, a yaw angle adjustment module, and a control module;

[0048] The data processing module is configured to obtain wind condition measurement data of a target wind farm, and perform wind power forecasting based on the wind condition measurement data to obtain forecasted wind condition data of the target wind farm within a forecast period; wherein the target wind farm includes a plurality of wind turbines, and the wind condition measurement data includes historical measurement data and current measurement data;

[0049] The yaw angle adjustment module inputs the wind condition measurement data and the predicted wind condition data into a multi-time scale combined control model, so that the multi-time scale combined control model determines the yaw reference angle of each wind turbine according to the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, generates an initial yaw angle control scheme, and adjusts the initial yaw control scheme according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and outputs a target yaw angle control scheme; wherein the large time scale is a time scale determined according to the wind condition stability corresponding to the current measurement data, and the small time scale is a time scale determined according to the wind turbine arrangement and control requirements of the target wind farm, and the initial yaw angle control scheme is used to characterize the yaw angle reference angle of each wind turbine within the prediction period;

[0050] The control module is configured to control the target wind farm according to the target yaw angle control scheme.

[0051] In an embodiment of the present invention, in a control layer corresponding to a large time scale determined according to the degree of wind stability, an initial yaw angle control scheme is generated based on predicted wind condition data to determine the long-term reference angle of the yaw angle of each wind turbine in a future prediction period, and in a control layer corresponding to a small time scale determined according to the wind turbine arrangement and control requirements, a target yaw angle control scheme is determined based on current wind condition data to fine-tune the yaw angle of each wind turbine in real time according to the target yaw angle control scheme, thereby reducing the mechanical wear and vibration of each wind turbine during the overall operation process, reducing mechanical losses, and thus extending the service life of the wind turbine.

[0052] Another embodiment of the present invention further provides a terminal device, comprising: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps of the wind farm operation control method combining multiple time scales of the present invention are implemented.

[0053] Another embodiment of the present invention further provides a computer-readable storage medium item, comprising: a stored computer program, which, when the computer program is executed, controls the device where the computer-readable storage medium is located to execute the steps of the wind farm operation control method combining multiple time scales of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0055] Figure 1 This is a flow chart of a wind farm operation control method combining multiple time scales provided by an embodiment of the present invention;

[0056] Figure 2 1 is a flow chart of a model predictive control method provided by an embodiment of the present invention;

[0057] Figure 3 Schematic diagram of the coordination relationship of the model predictive control framework provided by an embodiment of the present invention;

[0058] Figure 4 is a schematic diagram of a small time scale model predictive control provided by an embodiment of the present invention;

[0059] Figure 5 Schematic diagram of the principle of the wind turbine sorting algorithm provided by an embodiment of the present invention;

[0060] Figure 6 is a schematic diagram of a wind farm provided by an embodiment of the present invention;

[0061] Figure 7 Schematic diagram of the structure of a wind farm operation control device combining multiple time scales provided by an embodiment of the present invention;

[0062] Figure 8 It is a schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0063] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions in this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0064] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0065] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0066] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0067] In the description of the embodiments of this application, the term "and / or" is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0068] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0069] In the description of the embodiments of the present application, unless otherwise expressly specified or limited, technical terms such as "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integration; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components or interactions between two components. Those skilled in the art can understand the specific meanings of the above terms in the embodiments of the present application based on specific circumstances.

[0070] In order to solve the problem of how to reduce the mechanical loss of wind turbines during the control process in the prior art, an embodiment of the present invention provides a wind farm operation control method combining multiple time scales, such as Figure 1 As shown, the method includes steps S101-S103, which are specifically as follows:

[0071] Step S101: Obtain wind condition measurement data of a target wind farm, and perform wind power forecasting based on the wind condition measurement data to obtain forecasted wind condition data of the target wind farm within a forecast period; wherein the target wind farm includes multiple wind turbines, and the wind condition measurement data includes historical measurement data and current measurement data.

[0072] In this embodiment, if Figure 2 The figure shows a flow chart of the model predictive control method provided by the present invention, in which wind condition measurement data of the target wind farm is collected through SCADA, and a data-driven method based on the Koopman operator is used to predict future short-term wind condition data. SCADA (Supervisory Control And Data Acquisition) is a computer system used for industrial automation control and is used for real-time monitoring, data collection and remote control of industrial equipment or infrastructure. The Koopman operator is a linear operator that captures the evolution of observable quantities that describe possible nonlinear dynamic systems.

[0073] As another example of an embodiment of the present invention, wind power prediction is performed based on the wind condition measurement data to obtain predicted wind condition data of the target wind farm within the prediction period, specifically: the wind condition measurement data is decomposed by an adaptive noise complete set empirical mode decomposition method to obtain an initial modal component set, wherein the initial modal component set includes a high-frequency noise component and a low-frequency trend component; the low-frequency trend components in the initial modal component set are added together to obtain a target modal component; the target modal component is input into a wind condition prediction module to output the predicted wind condition data, wherein the wind condition prediction model is a model constructed based on the Hankel dynamic modal decomposition algorithm.

[0074] In this embodiment, turbulence is prevalent in the ambient wind at a wind farm, resulting in a significant amount of high-frequency noise in the wind measurement data. This poses significant challenges to subsequent data-driven predictions. Furthermore, turbulent noise in wind farms is highly random. Therefore, before performing wind predictions, this embodiment first decomposes the wind measurement data using the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) method. The resulting low-frequency components are then summed to determine the wind trend.

[0075] CEEMDAN is a signal decomposition method based on CEEMD that introduces an adaptive control strategy and adds random noise. This overcomes the problem of white noise shifting from high to low frequencies in CEEMD. The CEEMDAN decomposition of an original data sequence is represented by m modal components and a residual term. The decomposition formula is as follows:

[0076]

[0077] After component decomposition using the CEEMDAN method, dynamic mode decomposition (DMD) is applied to the existing forecast analysis system for wind forecasting. DMD was originally developed to identify spatiotemporal coherent structures from high-dimensional time series data. It provides a linear reduced-order representation for the dynamics of possible nonlinear systems through a set of modes with related oscillation frequencies and attenuation or growth rates. When the analyzed system is linear, the modes obtained by DMD correspond to the linear normal modes of the system. When the analyzed system is nonlinear, the reason why DMD can be applied to nonlinear analysis systems is its close relationship with the spectral analysis of the Koopman operator. Specifically, DMD analysis approximates the eigenmodes and eigenvalues of the infinite-dimensional linear Koopman operator, which is related to the evolution of the system over time, thereby providing a linearized finite-dimensional representation for the dynamics of the nonlinear original system. The analysis system can be represented by the following model formula:

[0078]

[0079] in, represents the state of the system at time t, γ represents the parameters of the system, and f() represents its dynamics. The discrete-time dynamic system can be represented by the following expression:

[0080] x k+1 =F(x k );

[0081] Koopman introduced the perspective of operator theory and proved that there exists an infinite-dimensional linear operator K that acts on all measurement functions g such that:

[0082]

[0083] The Koopman operator describes the evolution of the value of the observable g along the phase space trajectory:

[0084] g(x k+1 )=Kg(x k );

[0085] In DMD theory, the infinite-dimensional Koopman operator K can be approximated by a finite-dimensional linear matrix A. In a nonlinear system, observations of the system state can be obtained. For each time step j, the snapshot of the system is defined as a column vector x that collects the complete state of the system. j . The two matrices can be obtained as follows:

[0086] X=[x1,x2,...,x m-1 ],X′=[x2,x3,...,xm ];

[0087] The evolution of the system's state variables is expressed as follows:

[0088] X′=AX;

[0089] in, is the pseudo-inverse of X.

[0090] The value of A satisfies the following optimization problem:

[0091]

[0092] Therefore, the following expression is used to approximately construct the matrix A:

[0093]

[0094] in,‖·‖ F represents the Frobenius norm.

[0095] According to the singular value decomposition X = Φ∑W * , the matrix A is expressed as follows:

[0096] A=X′WΣ -1 φ * ;

[0097] Then project it onto Φ * On the column space of , construct a low-order approximate matrix of A. The matrix is expressed as follows:

[0098]

[0099] The eigenvalues and eigenvectors of are expressed as:

[0100]

[0101] The corresponding DMD pattern is expressed as:

[0102]

[0103] The Hankel-DMD method is applied to the prediction of future wind speed and direction changes. Hankel-DMD involves extending the original state by embedding a set of d time-delayed copies in the state vector. One time step separates the two subsequent shifted copies, and the resulting modified data matrix is organized in the form of two Hankel matrices:

[0104]

[0105] Only the eigenvalue and eigenvector identification of the algorithm are modified, so the original equation X=[x1,x2,...,x m-1 ],X′=[x2,x3,...,x m ] and the results derived subsequently. The use of time-delayed replicas as additional observables in the system is related to the Koopman operator as a universal linearization basis. When the rank of the data matrix (usually the length of its column vectors) is small, time-delayed replicas are useful for increasing its rank and thus the number of SVD modes that are dynamically modeled, which can improve the accuracy of the predictions.

[0106] The linear analysis and prediction system constructed based on the above method performs wind condition prediction to obtain predicted wind condition data of the target wind farm within the prediction period, wherein the wind condition includes wind direction and wind speed.

[0107] This embodiment uses the CEEMDAN method to ensure that the original signal is uncontaminated, and the signal decomposition error is extremely small or even negligible. Therefore, the CEEMDAN method is an accurate and efficient sequence decomposition method. Using this method, high-frequency noise components in wind farm measurement data can be separated to obtain the basic trend of future wind conditions. This significantly reduces the complexity of subsequent data-driven wind power forecasting and enhances the reliability of the forecast results.

[0108] Step S102: input the wind condition measurement data and the predicted wind condition data into a multi-time scale combined control model, so that the multi-time scale combined control model determines the yaw reference angle of each wind turbine according to the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, generates an initial yaw angle control scheme, and adjusts the initial yaw control scheme according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and outputs a target yaw angle control scheme; wherein the large time scale is a time scale determined according to the wind condition stability corresponding to the current measurement data, and the small time scale is a time scale determined according to the wind turbine arrangement and control requirements of the target wind farm, and the initial yaw angle control scheme is used to characterize the yaw angle reference angle of each wind turbine within the prediction period.

[0109] In this embodiment, a multi-timescale model predictive control framework is proposed. Based on this framework, the multi-timescale control model proposed in this invention is constructed to achieve maximum power optimization of the target wind farm. Model predictive control (MPC) is a rolling-horizon optimization strategy. Its current control action is obtained by solving a finite-horizon open-loop optimal control problem at each sampling instant. The current state of the process serves as the initial state of the optimal control problem, and the solved optimal control sequence only implements the first control action. Table 1 shows a two-stage model predictive control (MPC) framework, which demonstrates the design focus of MPC at different time scales and involves wind stability, control frequency, wind prediction, environmental factors, and other factors that affect control effectiveness. The first stage of MPC corresponds to the large-timescale control layer, which is used to determine the yaw reference angle within the prediction period. The second stage of MPC corresponds to the small-timescale control layer, which is used to fine-tune the yaw reference angle based on the current wind measurement data.

[0110] Table 1 Two-stage model predictive control framework

[0111]

[0112] like Figure 3 As shown in FIG, it is a schematic diagram of the coordination relationship of the model predictive control framework provided by an embodiment of the present invention, wherein the coordination relationship refers to the various components in the MPC framework, such as the predictive analysis system, the optimizer, the feedback mechanism, and the multi-level control strategy, such as the collaborative working mechanism between the large / small time scale control strategies, aiming to achieve global goals, such as maximum power generation and minimum load, and Δt1 and Δt2 are the control steps of the first stage and the second stage, respectively.

[0113] The predicted wind condition measurement data is input into a multi-time scale combined control model, wherein the control model is constructed based on the above-mentioned model predictive control framework. Corresponding to the first stage of MPC, a first optimization problem of the wind turbine yaw angle on a large time scale is established, and the first optimization function of the large time scale control layer is determined based on the first optimization problem. In the large time scale control layer, the yaw reference angle of each wind turbine is determined based on the first optimization function and the input predicted wind condition data, and an initial yaw angle control scheme is generated. Figure 4 The figure shows a schematic diagram of the small time scale model predictive control provided by an embodiment of the present invention. After the large time scale control layer performs yaw optimization, the yaw action benchmark of the second stage MPC within the prediction period T is obtained by linear interpolation according to Δt2. Since there are often errors in the prediction of wind conditions, and the changes in wind conditions are highly volatile and random, when the actual wind turbine performs each yaw action, it is necessary to fine-tune the yaw angle obtained by the first-stage MPC according to the actual wind condition measurement and fluctuation degree at the current moment. Based on this, corresponding to the second stage of MPC, a second optimization problem on a small time scale is established through the wind farm yaw control optimization algorithm, and the second optimization function of the large time scale control layer is determined based on the second optimization problem. In the small time scale control layer, the initial yaw control scheme is adjusted according to the second optimization function and the current measurement data, and the target yaw angle control scheme is output.

[0114] As another example of an embodiment of the present invention, the wind condition measurement data includes wind speed values, wind direction values and yaw angles corresponding to each wind turbine, and the control model combining multiple time scales determines the yaw reference angle of each wind turbine according to the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, and generates an initial yaw angle control scheme, specifically: under the condition of satisfying operating constraints and load constraints, the initial yaw angle control scheme is obtained by maximizing the weighted power output values of all wind turbines, wherein the weighted power output value is obtained by weighted calculation based on the wind speed values, wind direction values, yaw angles and corresponding weight values corresponding to each wind turbine.

[0115] In this embodiment, the first optimization function is used as the objective function for solving the first optimization problem. Under the condition of satisfying the operating constraints and fatigue load limit conditions, the Nomad solver is used to maximize the sum of the weighted power outputs of all wind turbines in a discrete time period to obtain the yaw reference angle of each wind turbine in the prediction period.

[0116] As another example of an embodiment of the present invention, the first optimization function includes a first objective function and a first constraint function, specifically:

[0117] The expression of the first objective function is as follows:

[0118]

[0119] The expression of the first constraint function is as follows:

[0120]

[0121] γ min ≤y i ≤y max , i=1,2,…,n;

[0122]

[0123] Among them, P irepresents the power of the i-th fan, w t Represents the weight coefficient at different times, γ=[γ1,γ2,...,γ n ] represents the yaw angle of all wind turbines, γ max and γ min Represent the maximum yaw angle and the minimum yaw angle, v t ,θ t They represent the predicted results of wind speed and wind direction at time t, respectively. WF represents the maximum value of wind turbine load in the target wind farm, represents the maximum load when all wind turbines in the target wind farm are facing the wind, δ F Represents the coefficient.

[0124] In this embodiment, a first optimization problem corresponding to each large-time-scale MPC included in the prediction period is modeled, and a first optimization function is constructed as follows:

[0125] Objective function:

[0126]

[0127] Constraint function:

[0128]

[0129] γ min ≤y i ≤y max , i=1,2,…,n;

[0130]

[0131] and:

[0132]

[0133] F WT =f P +f t ;

[0134] Among them, P i represents the power of the i-th wind turbine at time t, w t Represents the weight coefficient at different times, γ=[γ1,γ2,...,γ n ] represents the yaw angle of all wind turbines, γ max and γ min Represent the maximum yaw angle and the minimum yaw angle, v t ,θ t They represent the predicted results of wind speed and wind direction at time t, respectively. WF represents the maximum value of wind turbine load in the target wind farm, represents the maximum load when all wind turbines in the target wind farm are facing the wind, δ F It represents the coefficient, which is usually set to 0.05. rated Indicates the rated power of the fan, T life represents the design life of the fan, r represents the compensation coefficient, D and I represent the turbulence disturbance coefficient and turbulence intensity respectively.

[0135] As another example of an embodiment of the present invention, the second optimization function includes a second objective function and a second constraint function, specifically:

[0136] The expression of the second objective function is as follows:

[0137]

[0138] The expression of the second constraint function is as follows:

[0139] -Δγ min ≤Δy i ≤Δy max , i=1,2,…,n

[0140]

[0141] γ min ≤y i ≤y max ;

[0142] Among them, P i represents the power of the i-th wind turbine, Δγ=[Δγ1,Δγ2,...,Δγ n ] represents the yaw adjustment angle of all wind turbines, p k Represents different wind direction scenes θ k The probability under the condition, a represents the number of scenes, γ max and γ min Represent the maximum yaw angle and the minimum yaw angle respectively, represents the yaw reference for large-time-scale model predictive control calculations, Δγ max Indicates the maximum yaw adjustment angle.

[0143] In this embodiment, since the wind speed has little effect on the results of the optimization problem, only the random optimization of the wind direction is considered. The change of wind direction usually has the characteristics of Gaussian distribution on both sides of the average value. Therefore, this embodiment uses the Gaussian model to fit the probability distribution of the high-frequency component p(Δθ)~N(μ,σ θ 2 ), where μ = 0, from ±σ θSeveral typical scenarios are selected within the interval as representative inputs for solving the stochastic programming problem. The selection of these typical scenarios can effectively cover the main change modes of wind direction and provide a reliable data basis for subsequent optimization and control analysis. Figure 2 , the second optimization problem of the second stage MPC is established through the wind farm yaw control optimization algorithm, and the second optimization function is determined. The yaw control optimization algorithm aims to maximize the power output of the wind farm under the wake effect by iteratively adjusting the yaw angle of the wind turbine. The second optimization function is expressed as follows:

[0144] The expression of the second objective function is as follows:

[0145]

[0146] The expression of the second constraint function is as follows:

[0147] -Δγ min ≤Δy i ≤Δy max , i=1,2,…,n;

[0148]

[0149] γ min ≤y i ≤y max ;

[0150] Among them, P i represents the power of the i-th wind turbine, Δγ=[Δγ1,Δγ2,...,Δγ n ] represents the yaw adjustment angle of all wind turbines, p k Represents different wind direction scenes θ k The probability under the condition, a represents the number of scenes, γ max and γ min Represent the maximum yaw angle and the minimum yaw angle respectively, represents the yaw reference for large-time-scale model predictive control calculations, Δγ max Indicates the maximum yaw adjustment angle.

[0151] As an example of an embodiment of the present invention, the initial yaw control scheme is adjusted according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and a target yaw angle control scheme is output, specifically: based on the current measurement data, the wake interaction matrix between each wind turbine is calculated, and the wind turbines are sorted according to the wake interaction matrix to obtain a wind turbine sequence; based on the wind turbine sequence and the second optimization function, the yaw angle of each wind turbine in the initial yaw control scheme is adjusted in turn, with the goal of maximizing the actual output power of the wind farm, to obtain the target yaw angle control scheme.

[0152] In this embodiment, since the wake effect in a wind farm can reduce the effective wind speed and power generation of downstream wind turbines, a model of the wind farm wake effect is usually established by considering the interaction between upstream and downstream wind turbines. This embodiment proposes a wind turbine sorting algorithm based on calculating the mutual influence relationship between wind turbines, and further proposes a yaw control optimization algorithm for solving the second optimization problem. Figure 5 The figure shows a schematic diagram of the principle of the wind turbine sorting algorithm provided by an embodiment of the present invention. In the wind turbine sorting algorithm, the wake interaction matrix R is introduced to represent the velocity attenuation influence relationship between wind turbines. For a target wind farm including n wind turbines, the size of R is n*n. FLORIS is an engineering-level tool for wind farm flow field modeling and power prediction. By simplifying fluid dynamics equations, it can quickly simulate the wake effect and the interaction between wind turbines. In the FLORIS model, the wind turbine effective wind speed u i The calculation formula is as follows:

[0153]

[0154] Where Δu ij It represents the speed attenuation of the j-th fan to the i-th fan, and R can be expressed as:

[0155]

[0156] Where Δu set Indicates the set speed attenuation impact judgment threshold. When the impact of the current exhaust fan is greater than the set value, it is considered that there is an impact relationship between the two fans.

[0157] Therefore, in the wind turbine ranking algorithm, the wake interaction matrix R is calculated based on the above calculation formula and the current measurement data. Based on the iterative calculation of the wake interaction matrix R, the hierarchical index of each wind turbine is obtained, that is, the wind turbine sequence. The update method of this hierarchical index is expressed as follows:

[0158]

[0159] H(t+1)←H(t)+R;

[0160]

[0161] Based on the turbine sequence of each wind turbine in the target wind farm, the wind farm's yaw control optimization algorithm is used, with the second optimization function as the objective function of the small-time-scale control layer. The initial yaw control scheme is dynamically fine-tuned to generate the target yaw control scheme. Specifically, using the hierarchical index of each wind turbine, the algorithm prioritizes upstream wind turbines to ensure the greatest impact on downstream wind turbines, and adjusts the yaw angle in real time with each iteration. In this algorithm setting, the motion of the yaw bearing is assumed to be linear, and the optimization process follows the following steps:

[0162] (1) Initialization: Set the initial yaw angle γ = [γ1,γ2,...,γ n ] and calculate the reference power output P0.

[0163] (2) Identify the hierarchical index H of the wind turbines from high to low, and start the calculation from the turbine with the highest hierarchical index.

[0164] (3) Iterative adjustment: For each optimization step k:

[0165] For each turbine i in the hierarchy, two yaw adjustments are calculated:

[0166] γ1=γ k +Δγ,γ2=γ k -Δγ;

[0167] Calculate the yaw gain gradient:

[0168] P adjusted,1 =f(γ1),P adjusted,2 =f(γ2);

[0169]

[0170] Where f(·) is the wind farm power calculation function of FLORIS, and gradient is the current yaw gain gradient.

[0171] Calculate optimal yaw:

[0172] Δγ k =[Δy1,Δy1,...,Δy n ]=α*gradient;

[0173] Among them, α is a setting parameter.

[0174] Update the yaw angle:

[0175] γ k+1 =γ k +Δγ k ;

[0176] Moves to the fan with the next level index. If the current fan is at the most upstream level, returns to the highest level index.

[0177] (4) Convergence check: Repeat the adjustment until all turbines are optimized or the maximum number of iterations K is reached. opt .

[0178] (5) Power growth assessment: Calculate the total power output P total , and evaluate improvements.

[0179]

[0180] The wind farm yaw control optimization algorithm proposed in this embodiment demonstrates several key advantages that enhance its effectiveness in maximizing power output under wake effects while ensuring real-time adaptability and computational efficiency. First, the algorithm follows a greedy optimization approach. Through iterative operation, it allows the yaw angle of the wind turbine to be dynamically evaluated and adjusted based on real-time conditions. This enables the wind farm to respond to changing wind direction and wind turbine performance. As the system evolves over time, each optimization step is executed in real time, ensuring that the yaw angle is adjusted promptly to maximize total power output. This makes the algorithm suitable for operating environments with constantly changing operating conditions, such as wind speed fluctuations and wind turbine failure conditions, making it suitable for immediate adjustments during operation. Second, to handle dynamic node changes, the wake interaction matrix R represents the impact of one wind turbine's wake on another. When a wind turbine is connected to or disconnected from the network, the matrix R is recalculated to account for the new configuration. At the same time, the hierarchy index H, which prioritizes the wind turbines based on their impact on downstream wind turbines, is recalculated. This recalculation ensures that newly connected wind turbines are correctly integrated into the hierarchy while disconnected wind turbines are excluded. Therefore, when wind turbines are temporarily offline due to maintenance or failure, the algorithm can seamlessly adjust to maintain optimal performance. Third, the algorithm also respects the physical limitations imposed by the mechanical characteristics of the wind turbine. Specifically, the step size Δγ is limited by the yaw bearing speed ω. max and control duration t l The constraint of each iteration yaw adjustment Δγ is expressed as follows: Δγ=ωl maxThis ensures that the algorithm adheres to the physical constraints of wind turbine operation, preventing over-adjustments or mechanical stress on the turbine's yaw system. Fourth, the algorithm also improves efficiency through hierarchical prioritization. Using cluster group indices to prioritize upstream turbines ensures that the most influential turbines are optimized first, followed by downstream turbines. This hierarchical approach reduces computational complexity by focusing on turbines with the greatest impact on the wind farm's total output. By optimizing turbines with higher hierarchical indices first, the algorithm effectively calculates the cascading impact of yaw adjustments for each turbine, ensuring significant and measurable improvements in power output in fewer iterations. Fifth, the algorithm's iterative nature allows it to continuously improve power output through small, incremental yaw adjustments. These adjustments are evaluated based on real-time power forecasts calculated using the FLORIS model. By comparing power output under different yaw configurations, the algorithm selects the optimal orientation for each turbine, resulting in an increase in overall power production.

[0181] Step S103: Control the target wind farm according to the target yaw angle control scheme.

[0182] In this embodiment, based on the target yaw control scheme, the yaw angle of each wind turbine in the target wind farm is controlled and adjusted to achieve the target yaw angle of each wind turbine in the target yaw control scheme, thereby realizing operation control of the target wind farm.

[0183] As another example of an embodiment of the present invention, after the target wind farm is controlled, it also includes: obtaining real-time grid dispatching instructions and determining the target output power of the target wind farm, and obtaining the actual output power corresponding to the target wind farm at the current moment; inputting the target output power and the actual output power into the multi-time scale combined control model, so that the multi-time scale combined control model outputs an adjustment instruction for the thrust coefficient of each wind turbine according to the third optimization function corresponding to the micro-time scale control layer, with the goal of minimizing the difference between the target output power and the actual output power; wherein the time scale corresponding to the micro-time scale control layer is determined according to the rotational speed of the pitch angle of each wind turbine; and according to the adjustment instruction of the thrust coefficient of each wind turbine, the target wind farm is power controlled.

[0184] In this embodiment, the MPC framework is a three-stage MPC framework. As shown in Table 2, it is a three-stage model predictive control (MPC) framework. A control model combining multiple time scales is constructed based on the three-stage MPC framework. In the model, the third stage of MPC corresponds to the micro-time scale control layer, which is used to implement the automatic generation control (AGC) function, that is, by adjusting the thrust coefficient of the wind turbine, tracking the dispatch instruction curve issued by the power grid in real time, and increasing the flexibility of wind farm power generation.

[0185] Table 2 Three-stage model predictive control framework

[0186]

[0187]

[0188] like Figure 3 As shown, Δt3 is the control step length of the third stage.

[0189] Continue to see Figure 2 In order to realize the automatic generation control (AGC) function, the target wind farm should have the ability to track the grid dispatching instructions, where the grid dispatching instructions (Pref) are the reference power signals issued by the grid operator to coordinate the wind farm's power generation with the grid demand. When the wind farm power tracking control based on axial induction is carried out, there is no need to consider the wake effect between wind turbines, and the wind farm can be modeled as several decoupled subsystems. In the third stage of MPC, the target wind farm obtains the grid dispatching instructions in real time through the SCADA system. Combined with the grid dispatching instructions, the third optimization problem of the micro-time scale control layer is established through the Gurobi solver, and the third optimization function corresponding to the third optimization problem is determined. The third optimization function includes the third objective function and the third constraint function. The function is expressed as follows:

[0190] The expression of the third objective function is as follows:

[0191]

[0192] The expression of the third constraint function is as follows:

[0193]

[0194] Among them, P ref It represents the power tracking instruction issued by the power grid to the target wind farm, P WF represents the total power of the target wind farm, n represents the number of fans, Indicates the power factor of the fan, and its relationship with the thrust coefficient is: c p represents the proportionality coefficient, is the thrust coefficient of the fan at the current moment, Indicates the thrust coefficient adjustment, t cost represents the time required for the micro-time scale model predictive control to complete, which is determined by adjusting the ΔC′ of the fan with the largest angle. T,i and adjust speed OK, r represents the weight coefficient.

[0195] The actual output power of the target wind farm at the current moment is obtained, and the target output power is determined according to the obtained real-time grid dispatch instruction. The actual output power and the target output power are both input into a multi-time scale combined control model, so that the multi-time scale combined control model outputs an adjustment instruction for the thrust coefficient of each of the wind turbines according to the third optimization function corresponding to the micro-time scale control layer, with the goal of minimizing the difference between the target output power and the actual output power, to achieve power control of the target wind farm.

[0196] In an embodiment of the present invention, in a control layer corresponding to a large time scale determined according to the degree of wind stability, an initial yaw angle control scheme is generated based on predicted wind condition data to determine the long-term reference angle of the yaw angle of each wind turbine in a future prediction period, and in a control layer corresponding to a small time scale determined according to the wind turbine arrangement and control requirements, a target yaw angle control scheme is determined based on current wind condition data to fine-tune the yaw angle of each wind turbine in real time according to the target yaw angle control scheme, thereby reducing the mechanical wear and vibration of each wind turbine during the overall operation process, reducing mechanical losses, and thus extending the service life of the wind turbine.

[0197] Based on the above embodiments, in this embodiment, the wind farm operation model prediction and control method combined with multiple time scales provided by the present invention is described with reference to specific examples. Figure 6 Schematic diagram of the wind farm provided by the present invention. A wind farm with six wind turbines was selected as the test wind farm to verify the effectiveness of the three-stage multi-time-scale MPC yaw control method proposed in the present invention.

[0198] It should be noted that Δt1 in the multi-time-scale MPC framework proposed in this embodiment needs to be determined according to the stability of the actual environmental wind conditions, while Δt2 needs to be determined according to the arrangement and control requirements of the wind farm and the rotation speed of the pitch angle, which is generally 5s.

[0199] 1. Analysis of wind condition prediction examples.

[0200] Actual wind farm wind data often contains significant high-frequency noise due to turbulence. CEEMDAN is first used to decompose the wind speed and direction data. The large low-frequency components of the decomposed data are then summed to reconstruct the wind speed and direction data. Once the processed wind data is obtained, the Hankel-DMD method is used for prediction. In this example, the Koopman operator is fitted using data from the past 120 hours. The data from the past hour is then used to predict wind speed and direction changes for the next hour. Extensive testing has shown that this method has consistently achieved prediction errors of around 2% for future wind conditions.

[0201] 2. Multi-time scale MPC control case test.

[0202] The power boost and power tracking effects of multi-timescale MPC control were tested at the aforementioned wind speeds and directions. Based on the known wind direction stability, the large-timescale control step size Δt1 was set to 30 minutes; the small-timescale control step size was set to 2 minutes, consistent with conventional MPPT strategies; and the micro-timescale control step size was set to 5 seconds.

[0203] In terms of power improvement effect, two benchmark methods are considered for comparison:

[0204] (1) MPPT strategy: In the greedy strategy, the yaw direction of each wind turbine is aligned with the wind direction, and each wind turbine operates in the local maximum power point tracking mode, i.e., γ i =0, The control step of the MPPT strategy is consistent with the small time scale strategy, which is 2 minutes.

[0205] (2) Direct optimization: According to the current wind conditions, the optimal yaw angle is directly solved. The constraints are consistent with the small time scale strategy, and the control step size is also 2 minutes.

[0206] Table 3 Optimization results

[0207] MPPT Direct optimization Large time scale control layer Small time scale control layer Average power / WM 23.9 28.2 27.8 28.1 Total yaw degree / ° 488 820 301 480

[0208] According to the results in Table 3, it is clear that the multi-timescale MPC strategy proposed in this embodiment has significant advantages over the traditional MPPT strategy in terms of increasing wind farm power. Specifically, compared with the MPPT strategy, the proposed MPC strategy increases power generation by more than 15%, which is comparable to the improvement achieved by the direct optimization strategy. This result shows that the use of a multi-timescale MPC strategy can not only effectively improve the power output of a wind farm, but also ensure the efficient operation of the wind farm under variable wind conditions, avoiding the performance bottlenecks that may exist in the MPPT strategy. In particular, in scenarios with large wind speed fluctuations or complex wind conditions, the MPC strategy can more accurately adjust wind turbine operation, thereby better responding to environmental changes.

[0209] Furthermore, compared to a direct optimization strategy, the total yaw angle control strategy generated by the dual-layer MPC strategy is only about half that of a direct optimization strategy, and even lower than that of an MPPT strategy. This feature significantly reduces mechanical losses in wind turbines during the control process, especially in complex wind conditions. It reduces the fatigue burden on the mechanical system caused by frequent yaw angle adjustments, thereby extending the service life of the wind turbines. Reduced mechanical losses in wind farms not only improve the economic benefits of equipment but also reduce repair and maintenance costs, improving the operational efficiency of the entire wind farm.

[0210] Under the control of the multi-timescale MPC strategy, the yaw angle adjustment of the wind turbines is smoother and more coordinated, and the operating status of the wind turbine group can reflect changes in wind speed and direction in real time, avoiding energy loss and mechanical wear caused by excessive yaw angle adjustment. Based on the MPPT strategy, the two-layer MPC optimization strategy effectively controls the fatigue load of the wind farm to the lowest range, significantly reducing the operational fatigue loss of the wind farm. Through reasonable optimization control, the two-layer MPC strategy proposed in this embodiment can optimize the operation of the wind turbine while ensuring power generation efficiency, extend its service life, and slow down the fatigue process of the wind turbine structure.

[0211] In this experiment, the automatic generation control (AGC) function based on wind turbine thrust coefficient adjustment proposed in this embodiment demonstrated its excellent performance in tracking dispatch instructions. The experimental results show that this control strategy can quickly and accurately track dispatch instructions, and the response time for each adjustment is maintained within a very low range, demonstrating a good dispatch effect. This not only ensures the wind farm's ability to quickly respond to changes in grid demand, but also avoids the negative impact of untimely or excessive adjustments while ensuring power output stability, further improving the wind farm's operational flexibility and dispatch efficiency.

[0212] In summary, the multi-time-scale MPC strategy proposed in the embodiment of the present invention not only performs well in improving the power generation of wind farms, reducing mechanical losses, and controlling fatigue loads, but also can effectively improve the overall operating performance and economy of wind farms through precise power tracking and rapid response scheduling, and has broad application prospects.

[0213] Based on the above method embodiment, a corresponding device embodiment is provided, such as Figure 7 As shown, an embodiment of the present invention provides a wind farm operation control device combining multiple time scales, including: a data processing module 701, a yaw angle adjustment module 702 and a control module 703;

[0214] The data processing module 701 is configured to obtain wind condition measurement data of a target wind farm and perform wind power forecasting based on the wind condition measurement data to obtain forecasted wind condition data of the target wind farm within a forecast period; wherein the target wind farm includes a plurality of wind turbines, and the wind condition measurement data includes historical measurement data and current measurement data;

[0215] The yaw angle adjustment module 702 inputs the wind condition measurement data and the predicted wind condition data into a multi-time scale combined control model, so that the multi-time scale combined control model determines the yaw reference angle of each wind turbine according to the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, generates an initial yaw angle control scheme, and adjusts the initial yaw control scheme according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and outputs a target yaw angle control scheme; wherein the large time scale is a time scale determined according to the wind condition stability corresponding to the current measurement data, and the small time scale is a time scale determined according to the wind turbine arrangement and control requirements of the target wind farm, and the initial yaw angle control scheme is used to characterize the yaw angle reference angle of each wind turbine within the prediction period;

[0216] The control module 703 is configured to control the target wind farm according to the target yaw angle control scheme.

[0217] It can be understood that the above-mentioned device embodiment corresponds to the method embodiment of the present invention, which can implement the wind farm operation control method combining multiple time scales provided by any of the above-mentioned method embodiments of the present invention.

[0218] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. Furthermore, in the drawings of the device embodiments provided by the present invention, the connection relationship between modules indicates that they have a communication connection, which may be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement the present invention without inventive effort.

[0219] Based on the above-mentioned embodiment of the wind farm operation control method combining multiple time scales, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the wind farm operation control method combining multiple time scales according to any embodiment of the present invention is implemented.

[0220] For example, in this embodiment, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to implement the present invention. The one or more module elements may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0221] The terminal device may be a computing device such as a desktop computer, a notebook computer, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0222] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, connecting various parts of the entire terminal device using various interfaces and lines.

[0223] Based on the above method embodiments, another embodiment of the present invention provides a computer-readable storage medium, including a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the multi-time scale combined wind farm operation control method described in any one of the above method embodiments of the present invention.

[0224] like Figure 8 FIG. 8 is a schematic diagram of the physical structure of an electronic device, which may be a robot or other electronic device. The electronic device may include: a processor 810, a communications interface 820, a memory 830, and a communication bus 840. The processor 810, the communications interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute a wind farm operation control method combining multiple time scales, including:

[0225] Obtaining wind condition measurement data of a target wind farm, and performing wind power forecasting based on the wind condition measurement data to obtain forecasted wind condition data of the target wind farm within a forecast period; wherein the target wind farm includes a plurality of wind turbines, and the wind condition measurement data includes historical measurement data and current measurement data;

[0226] The wind condition measurement data and the predicted wind condition data are input into a control model combined with multiple time scales, so that the control model combined with multiple time scales determines the yaw reference angle of each wind turbine according to the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, generates an initial yaw angle control scheme, and adjusts the initial yaw control scheme according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and outputs a target yaw angle control scheme; wherein the large time scale is a time scale determined according to the wind condition stability corresponding to the current measurement data, and the small time scale is a time scale determined according to the wind turbine arrangement and control requirements of the target wind farm, and the initial yaw angle control scheme is used to characterize the yaw angle reference angle of each wind turbine within the prediction period;

[0227] The target wind farm is controlled according to the target yaw angle control scheme.

[0228] Wherein, the module / unit integrated in the device / terminal equipment, if implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0229] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0230] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A wind farm operation control method combining multiple time scales, characterized in that: include: Obtaining wind condition measurement data of a target wind farm, and performing wind power forecasting based on the wind condition measurement data to obtain forecasted wind condition data of the target wind farm within a forecast period; wherein the target wind farm includes a plurality of wind turbines, and the wind condition measurement data includes historical measurement data and current measurement data; The wind condition measurement data and the predicted wind condition data are input into a control model combined with multiple time scales, so that the control model combined with multiple time scales determines the yaw reference angle of each wind turbine according to the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, generates an initial yaw angle control scheme, and adjusts the initial yaw control scheme according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and outputs a target yaw angle control scheme; wherein the large time scale is a time scale determined according to the wind condition stability corresponding to the current measurement data, and the small time scale is a time scale determined according to the wind turbine arrangement and control requirements of the target wind farm, and the initial yaw angle control scheme is used to characterize the yaw angle reference angle of each wind turbine within the prediction period; The target wind farm is controlled according to the target yaw angle control scheme.

2. The wind farm operation control method combining multiple time scales according to claim 1, characterized in that: After controlling the target wind farm, the method further includes: Obtaining a real-time grid dispatch instruction and determining a target output power of the target wind farm, and obtaining the actual output power corresponding to the target wind farm at the current moment; Inputting the target output power and the actual output power into the multi-timescale combined control model, so that the multi-timescale combined control model outputs an adjustment instruction for the thrust coefficient of each of the wind turbines according to a third optimization function corresponding to a micro-timescale control layer, with the goal of minimizing the difference between the target output power and the actual output power; wherein the time scale corresponding to the micro-timescale control layer is determined according to the rotational speed of the pitch angle of each of the wind turbines; Power control is performed on the target wind farm according to the adjustment instruction of the thrust coefficient of each wind turbine.

3. The wind farm operation control method combining multiple time scales according to claim 1, characterized in that: The wind power prediction is performed based on the wind condition measurement data to obtain the predicted wind condition data of the target wind farm within the prediction period, specifically: Decomposing the wind condition measurement data by an adaptive noise complete set empirical mode decomposition method to obtain an initial modal component set, wherein the initial modal component set includes a high-frequency noise component and a low-frequency trend component; Adding the low-frequency trend components in the initial modal component set to obtain the target modal component; The target modal component is input into a wind condition prediction module, and the predicted wind condition data is output, wherein the wind condition prediction model is a model constructed based on the Hankel dynamic modal decomposition algorithm.

4. The wind farm operation control method combining multiple time scales according to claim 1, characterized in that: The initial yaw control scheme is adjusted according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and a target yaw angle control scheme is output, specifically: Based on the current measurement data, calculating a wake interaction matrix between each wind turbine, and sorting each wind turbine according to the wake interaction matrix to obtain a wind turbine sequence; Based on the wind turbine sequence and the second optimization function, the yaw angle of each wind turbine in the initial yaw control scheme is adjusted in sequence to maximize the actual output power of the wind farm, thereby obtaining the target yaw angle control scheme.

5. The wind farm operation control method combining multiple time scales according to claim 1, characterized in that: in, The wind condition measurement data includes wind speed values, wind direction values, and yaw angles corresponding to each wind turbine. The control model combining multiple time scales determines the yaw reference angle of each wind turbine based on the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, and generates an initial yaw angle control scheme, specifically: Under the condition that the operation constraints and load constraints are met, the initial yaw angle control scheme is obtained by maximizing the weighted power output value of all wind turbines, wherein the weighted power output value is obtained by weighted calculation based on the wind speed value, wind direction value, yaw angle and corresponding weight value corresponding to each wind turbine.

6. The wind farm operation control method combining multiple time scales according to claim 5, characterized in that: The first optimization function includes a first objective function and a first constraint function, specifically: The expression of the first objective function is as follows: The expression of the first constraint function is as follows: Among them, P i represents the power of the i-th fan, w t Represents the weight coefficient at different times, γ=[γ1,γ2,...,γ n ] represents the yaw angle of all wind turbines, γ max and γ min Represent the maximum yaw angle and the minimum yaw angle, v t ,θ t They represent the predicted results of wind speed and wind direction at time t, respectively. WF represents the maximum value of wind turbine load in the target wind farm, represents the maximum load when all wind turbines in the target wind farm are facing the wind, δ F Represents the coefficient.

7. The wind farm operation control method combining multiple time scales according to claim 4, characterized in that: The second optimization function includes a second objective function and a second constraint function, specifically: The expression of the second objective function is as follows: The expression of the second constraint function is as follows: -Dg min ≤Δy i ≤Δy max ,i=1,2,…,n c min ≤y i ≤y max ; Among them, P i represents the power of the i-th wind turbine, Δγ=[Δγ1,Δγ2,...,Δγ n ] represents the yaw adjustment angle of all wind turbines, p k Represents different wind direction scenes θ k The probability under the condition, a represents the number of scenes, γ max and γ min Represent the maximum yaw angle and the minimum yaw angle respectively, represents the yaw reference for large-time-scale model predictive control calculations, Δγ max Indicates the maximum yaw adjustment angle.

8. A wind farm operation control device combining multiple time scales, characterized in that: include: Data processing module, yaw angle adjustment module and control module; The data processing module is configured to obtain wind condition measurement data of a target wind farm, and perform wind power forecasting based on the wind condition measurement data to obtain forecasted wind condition data of the target wind farm within a forecast period; wherein the target wind farm includes a plurality of wind turbines, and the wind condition measurement data includes historical measurement data and current measurement data; The yaw angle adjustment module inputs the wind condition measurement data and the predicted wind condition data into a multi-time scale combined control model, so that the multi-time scale combined control model determines the yaw reference angle of each wind turbine according to the first optimization function corresponding to the large time scale control layer and the predicted wind condition data, generates an initial yaw angle control scheme, and adjusts the initial yaw control scheme according to the second optimization function corresponding to the small time scale control layer and the current measurement data, and outputs a target yaw angle control scheme; wherein the large time scale is a time scale determined according to the wind condition stability corresponding to the current measurement data, and the small time scale is a time scale determined according to the wind turbine arrangement and control requirements of the target wind farm, and the initial yaw angle control scheme is used to characterize the yaw angle reference angle of each wind turbine within the prediction period; The control module is configured to control the target wind farm according to the target yaw angle control scheme.

9. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for controlling the operation of a wind farm combining multiple time scales according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that include: A stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the wind farm operation control method combining multiple time scales according to any one of claims 1 to 7.

Images