A wind farm optimal power control method, system, device and storage medium

Abstract

The application discloses a kind of wind farm optimal power control method, system, equipment and storage medium, comprising: obtaining the position of each wind turbine in wind farm, wind speed and wind direction information;The position of each wind turbine in wind farm, wind speed and wind direction information are input into the intelligent agent after training, and the yaw angle, axial induction coefficient and inclination angle information of each wind turbine in wind farm are obtained, the wind turbine in wind farm is controlled according to the yaw angle, axial induction coefficient and inclination angle information of each wind turbine in wind farm, so that the total power generation of wind farm is maximized, and the optimal power control of wind farm is completed, the method, system, equipment and storage medium can realize the maximization of total power generation of wind farm.

Description

Technical Field

[0001] This invention belongs to the field of new energy power generation technology, and relates to a method, system, equipment and storage medium for optimal power control of wind farms. Background Technology

[0002] With the large-scale development and utilization of wind power, the density and capacity of wind turbines in wind farms are constantly increasing. This leads to a more significant decrease in wind speed after the upstream wind turbines pass through, resulting in reduced output power of downstream turbines and increased fatigue loads on the turbines, thus increasing operation and maintenance costs. Due to the wake effect, the overall power generation of a wind farm can be reduced by up to 54%. Furthermore, the wake effect causes an annual economic loss of 20-30%. Therefore, reducing the impact of the wake effect to increase the total power output of the wind farm has become a hot topic in wind farm research.

[0003] Currently, some scholars have conducted research on improving wind farm efficiency. Some studies aim to optimize wind turbine location. Busby, RL et al., in "Wind Power: The Industry Grows Up," addressed wind turbine positioning, arguing that larger turbine spacing ensures the wake returns to free-flow pressure levels. Developers also need to consider other factors such as electrical connection costs and maintenance procedures when designing wind farms, primarily adopting a grid layout with appropriate turbine spacing in the prevailing wind direction. In current research on wind farm output power optimization, active control strategies can be divided into axial induction control and wake yaw strategies. Johnson, KE et al., in "Wind farm control: addressing the aerodynamic interaction among wind turbines," proposed axial induction control. Reducing the rated output power of upstream turbines lowers the axial induction coefficient, which decreases momentum in the wake region, exposing downstream turbines to higher incident wind speeds and allowing them to obtain more electrical energy at higher rotor speeds. Fleming, P. et al. proposed wake yaw in “Simulation comparison of wake mitigation control strategies for a two-turbine case.” The aim is to adjust the misalignment between the rotor plane and the incident wind so that the upper and lower parts of the rotor plane are subjected to different aerodynamic loads. This imbalance causes the wind to gain momentum in the crosswind direction, that is, the wake is deflected, which in turn affects the total power generation of the wind farm. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method, system, device and storage medium for optimal power control of wind farms, which can maximize the total power generation of wind farms.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] In one aspect, the present invention provides a method for optimal power control of a wind farm, comprising:

[0007] Obtain information on the location, wind speed, and wind direction of each wind turbine in the wind farm;

[0008] The location, wind speed, and wind direction information of each wind turbine in the wind farm are input into the trained agent to obtain the yaw angle, axial sensing coefficient, and tilt angle information of each wind turbine in the wind farm. Based on the yaw angle, axial sensing coefficient, and tilt angle information of each wind turbine in the wind farm, the wind turbines in the wind farm are controlled to maximize the total power generation of the wind farm and complete the optimal power control of the wind farm.

[0009] Also includes:

[0010] Construct training samples, wherein the training samples include the power generation P of wind turbine i. i Location (X) i ,Y i Wind speed (wv), wind direction (wd), yaw angle (yaw) i axial inductance coefficient ai i and tilt angle i ;

[0011] Constructing intelligent agents;

[0012] The agent is trained using the training samples to obtain the trained agent.

[0013] The reward / punishment function used in the training process of the agent using the training samples is as follows:

[0014]

[0015] Where r is the reward for the agent, I is the number of wind turbines in the wind farm, and N is the number of agents.

[0016] The agent is trained using the training samples based on the MADDPG algorithm.

[0017] Before constructing the training samples, the following steps are also included:

[0018] Obtain the free flow velocity, turbulence intensity, and yaw angle of the wind turbines in the wind field;

[0019] The wind speed and output power of the wind turbine are calculated based on the free flow velocity, turbulence intensity, and yaw angle of the wind turbine in the wind farm.

[0020] Based on the free flow velocity, turbulence intensity, and yaw angle of the wind turbine in the wind farm, the wind speed at the location of the wind turbine and the output power of the wind turbine are calculated using the Floris model.

[0021] The Floris model is as follows:

[0022] C γ =P γ / P0=(-0.0003γ 3 -0.000025γ 2 +0.997γ)

[0023] C β =P β / P0=-0.185+0.285cos(0.105β)+0.014sin(0.105β)

[0024] -0.144cos(0.209β)-0.017sin(0.209β)

[0025] +0.06cos(0.314β)+0.011sin(0.314β)

[0026] -0.016cos(0.419β)-0.005sin(0.419β)

[0027]

[0028] In a second aspect, the present invention provides an optimal power control system for a wind farm, comprising:

[0029] The acquisition module is used to acquire information on the location, wind speed, and wind direction of each wind turbine in the wind farm.

[0030] The control module is used to input the position, wind speed and wind direction information of each wind turbine in the wind farm into the trained agent to obtain the yaw angle, axial sensing coefficient and tilt angle information of each wind turbine in the wind farm. Based on the yaw angle, axial sensing coefficient and tilt angle information of each wind turbine in the wind farm, the module controls the wind turbines in the wind farm to maximize the total power generation of the wind farm and complete the optimal power control of the wind farm.

[0031] In three aspects, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the optimal power control method for the wind farm.

[0032] In four aspects, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the wind farm optimal power control method.

[0033] The present invention has the following beneficial effects:

[0034] In practical operation, the wind farm optimal power control method, system, device, and storage medium described in this invention maximize the total power generation of the wind farm based on a trained intelligent agent. Specifically, the position, wind speed, and wind direction information of each wind turbine in the wind farm are input into the trained intelligent agent to obtain the yaw angle, axial sensing coefficient, and tilt angle information of each wind turbine in the wind farm. Based on the yaw angle, axial sensing coefficient, and tilt angle information of each wind turbine in the wind farm, the wind turbines in the wind farm are controlled to maximize the total power generation of the wind farm, which meets the requirements of distributed operation control of wind farms. Attached Figure Description

[0035] The accompanying drawings, which form part of this specification, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0036] Figure 1 Schematic diagram of wake effect

[0037] Figure 2a This is a schematic diagram of the wind turbine's yaw angle;

[0038] Figure 2b This is a schematic diagram of the wind turbine tilt angle;

[0039] Figure 3 This is a diagram of the computational framework of the Floris model.

[0040] Figure 4 A schematic diagram of fitting the power-yaw angle curve to the experimental data;

[0041] Figure 5 This is a schematic diagram for fitting the power-tilt angle curve;

[0042] Figure 6 A schematic diagram for training an intelligent agent;

[0043] Figure 7 This is a framework diagram of the MADDPG method;

[0044] Figure 8 This is a schematic diagram of the reward curve. Detailed Implementation

[0045] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0046] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0047] The present invention will now be described in further detail with reference to the accompanying drawings:

[0048] Example 1

[0049] The optimal power control method for wind farms described in this invention includes the following steps:

[0050] like Figure 1 As shown, the wake effect refers to the viscous interaction between the blades of the upstream wind turbine and the incident free flow when two wind turbines are arranged in the same direction in the wind. This interaction results in a local area behind the upstream wind turbine with low wind speed and high turbulence. This effect is called the wake effect, and the local area affected by the wake effect is called the wake region. The overlap between the wake region and the downstream wind turbine will lead to a decrease in the velocity of the incident flow received by the downstream wind turbine, a decrease in output power, and a deterioration in performance.

[0051] Active wind farm control strategies can be broadly classified into two categories. The first category is axial induction control strategy. The wake effect is generated by the viscous interaction between the wind turbine and the free flow. Axial induction control means adjusting the power extraction rate of the upstream wind turbine, which will simultaneously affect the intensity of the interaction between the wind turbine and the free flow, thereby changing the velocity deficit in the wake region. Let the axial induction coefficient α be the ratio of the decrease in velocity from the free flow in front of the turbine to the wake velocity behind the turbine.

[0052] Let the tip speed ratio (TSR) be... For the pitch angle β, the turbine's thrust coefficient and power coefficient are both:

[0053]

[0054]

[0055] Where ρ is the air density, A is the area of ​​the rotor's rotating plane, U is the free-flow velocity in front of the turbine, and C... T and C P Let P and T be the thrust coefficient and power coefficient, respectively, and P and T be the output power and thrust against the free flow, respectively. The thrust of the turbine against the free flow determines the rate of speed decrease after the rotor, i.e., the axial inductance coefficient, which can be written as a formula related to the thrust coefficient.

[0056] Based on the above analysis, adjusting and reducing the power extraction rate of the upstream wind turbine will reduce the wake deficit behind the turbine, thereby increasing the incident flow velocity of the downstream wind turbine and increasing the output power of the downstream wind turbine. By setting the axial induction coefficient of the upstream wind turbine, the overall output power of the wind farm will be improved.

[0057] The second type of active wind farm control strategy is wake redirection. A conventional wake redirection method involves resetting the yaw angle. This method deflects the wake of the upstream turbine in the horizontal plane, reducing the overlap between the wake region and the rotation plane of the downstream turbine, thereby increasing the overall power output of the wind farm. Another feasible method is to change the tilt angle of the upstream turbine, causing the wake to be redirected in the vertical direction, reducing the overlap between the wake region and the downstream turbine. The settings for the yaw angle and tilt angle are shown in Figure 2.

[0058] In this invention, the parametric model Floris for yaw angle redirection wind fields is first expanded. Actual data collected in a wind field in India is used to fit a power curve, replacing the empirical power formula in the original model. An interface for tilt angle redirection control strategies is added to this model, and the tilt angle-output power formula is fitted using the results of a CFD simulation model to improve the interface. This invention constructs a wind field simulation model that includes interfaces for all active control strategies and replaces inaccurate empirical formulas with formulas fitted from actual data, laying a solid foundation for multi-strategy control.

[0059] Floris simulation of the effect of yaw angle on wake effect and output power calculation process as follows: Figure 3 As shown, Figure 3 As shown, the Floris model first simulates the wake region and the wake velocity deficit within the region. Then, it superimposes the velocity deficit with the free flow of the wind field to obtain the actual wind speed at the corresponding location of each turbine, and calculates the corresponding turbine output power P based on this wind speed. u (γ), that is:

[0060]

[0061] Where γ is the wind turbine yaw angle, P p As an empirical parameter, it is generally taken as 3. In this invention, wind field experimental data from Howland et al. in India over a period of two years were collected. The experimental data was filtered, and the filtered data was used to fit a power-yaw angle curve, which replaced the above empirical formula in the Floris model. The fitted curve is shown below. Figure 4 As shown.

[0062] The wind field environment model used in this study is:

[0063] C γ =P γ / P0=(-0.0003γ 3 -0.000025γ 2 +0.997γ)

[0064] C β =P β / P0=-0.185+0.285cos(0.105β)+0.014sin(0.105β)

[0065] -0.144cos(0.209β)-0.017sin(0.209β)

[0066] +0.06cos(0.314β)+0.011sin(0.314β)

[0067] -0.016cos(0.419β)-0.005sin(0.419β)

[0068]

[0069] in, This represents the rated output power of the turbine at a wind speed u. 'a' is the axial inductance coefficient, 'β' is the turbine tilt angle, 'γ' is the turbine yaw angle, 'ρ' is the air density (typically taken as 1.25), and 'C' is the turbine speed. P For the power factor, this paper uses the NERL-15MW standard wind turbine. The value of this parameter can be obtained from... Figure 3 We obtain A as the rotor plane area and u as the effective wind speed at the turbine.

[0070] Yaw angle redirects the wake in the horizontal direction, while tilt angle redirects the wake in the vertical direction. Adjusting the tilt angle also helps increase power. In this invention, the SOWFA fluid simulation model is used to simulate a three-dimensional wind field model, read the output power corresponding to tilt angle adjustment, and fit a power-tilt angle curve, such as... Figure 5 As shown.

[0071] By combining multiple active control strategies such as yaw control, tilt control, and axial sensing control, and updating empirical formulas in the model using experimental data, a multi-interface wind field model is formed. The axial sensing coefficient, tilt angle, and yaw angle are input into this environmental model, which outputs the wind field power. Reinforcement learning is then introduced to train and obtain the optimal operating point of the wind field.

[0072] Taking the maximization of total wind farm power generation as the optimization objective, the multi-agent deep deterministic policy gradient algorithm in multi-agent deep reinforcement learning is used to solve for the wind turbine deflection angle, tilt angle and axial induction coefficient.

[0073] In the context of multi-agent deep reinforcement learning, multiple computerized agents learn to take individual actions at discrete time steps t. These actions constitute a joint action to maximize the total reward from the environment. The rewards of each agent are related in a game-like manner, and state transitions follow a Markov process. If, in a given state, an agent chooses an operation that yields a lower reward, then in subsequent operations in that state, the agent can choose an operation that produces a higher reward. Within this framework, there are multiple learning and decision-making entities—the multi-agent group—and an environment, which is everything outside the agents, such as… Figure 6 As shown.

[0074] The actions performed not only affect the immediate reward signal, but also all subsequent reward signals. Initially, the agent receives the environmental state s. t And based on this, perform an action that prompts the environment state to change to s. t+1 In addition to state, the environment also uses reward signals r. t+1 This provides the agent with behavior-related feedback in the form of [something]. The state is changed from [something]... t Move to s t+1 The transition function typically contains a random component and cannot be determined solely by a. t The role of π is determined. The set of rules that an agent follows when mapping states to probability distributions of actions is called a policy (π). The goal of multi-agent reinforcement learning is to provide policies for multiple agents that maximize the total future reward of the agents.

[0075] The MADDPG algorithm employs a framework of centralized training and distributed execution, specifically as follows: Figure 7 As shown, the MADDPG algorithm uses global state and global action for centralized training. After training, the agent performs actions based on its own observations (partial observations) to maximize the total reward.

[0076] Let the power generation capacity of wind turbine i be P. i For the position of fan i (X) i ,Y iWind speed (wv), wind direction (wd), deflection angle (yaw) i Axial inductance coefficient ai i and tilt angle i The function is as follows:

[0077] P i =f(X) i ,Y i ,wv,wd,yaw i ,ai i ,tilt i (4)

[0078] To avoid sparse rewards during the agent's learning process, a reward and penalty are added after each action. The agents are in a cooperative relationship, and the reward and penalty function for each step is as follows:

[0079]

[0080] Where r is the reward for the agent, I is the number of wind turbines in the wind farm, and N is the number of agents.

[0081] After training, the trained agent is obtained. During actual control, it acquires the position, wind speed, and wind direction information of each wind turbine in the wind farm. The acquired position, wind speed, and wind direction information of each wind turbine in the wind farm is then input into the trained agent to obtain the yaw angle, axial sensing coefficient, and tilt angle information of each wind turbine in the wind farm. Based on the yaw angle, axial sensing coefficient, and tilt angle information of each wind turbine in the wind farm, the agent controls the wind turbines in the wind farm to maximize the total power generation of the wind farm and complete the optimal power control of the wind farm.

[0082] Simulation Experiment

[0083] To verify the correctness of the strategy described in this invention, a single agent can control four wind turbines, i.e., a wind turbine cluster. In this invention, there are two agents, controlling a total of seven wind turbines. In agent 1, wind turbine 1 has coordinates (0,0), wind turbine 2 has coordinates (400,0), wind turbine 3 has coordinates (800,0), and wind turbine 4 has coordinates (1200,0), with a wind direction of 270° and a wind speed of 8 m / s. In agent 2, wind turbine 5 has coordinates (1200,0), wind turbine 6 has coordinates (0,-400), wind turbine 7 has coordinates (0,400), and wind turbine 8 has coordinates (0,800), with a wind direction of 270° and a wind speed of 8 m / s. It should be noted that wind turbine 4 in agent 1 and wind turbine 5 in agent 2 are the same wind turbine, used for information exchange between agents. The scheduling result is not affected by the position or number of wind turbines in the cluster. The hyperparameter settings of the MADDPG algorithm are shown in Table 2. It should be noted that the state dimension of a single agent is 10, which includes the coordinate values ​​of 4 wind turbines (X-axis and Y-axis), wind direction, wind speed, and a total of 10 parameters.

[0084] Table 1

[0085]

[0086] As can be seen from the reward curve, this invention exhibits good convergence, with the reward value rising rapidly and eventually converging to 1.95*10. 8 In the vicinity, the algorithm's reward value initially fluctuated greatly. The low reward value was due to the agent's continuous exploration, which is a normal phenomenon.

[0087] The specific control parameters for the maximum power output of the wind turbine cluster are shown in Table 2.

[0088] Table 2

[0089]

[0090] As shown in Table 2, in Agent 1, the coordinates of wind turbine 1 are (0,0), wind turbine 2 is (400,0), wind turbine 3 is (800,0), and wind turbine 4 is (1200,0), with a wind direction of 270° and a wind speed of 8 m / s. In Agent 2, the coordinates of wind turbine 5 are (1200,0), wind turbine 6 is (0,-400), wind turbine 7 is (0,400), and wind turbine 8 is (0,800), with a wind direction of 270° and a wind speed of 8 m / s. The optimal operating point of the wind turbine cluster after training with the MAPPO algorithm is shown in Table 2. The total power of the wind turbines is 8.28 MW, which is a significant improvement compared to the total power of 7.63 MW when only the yaw angle is controlled.

[0091] Example 2

[0092] The wind farm optimal power control system of the present invention includes:

[0093] The acquisition module is used to acquire information on the location, wind speed, and wind direction of each wind turbine in the wind farm.

[0094] The control module is used to input the position, wind speed and wind direction information of each wind turbine in the wind farm into the trained agent to obtain the yaw angle, axial sensing coefficient and tilt angle information of each wind turbine in the wind farm. Based on the yaw angle, axial sensing coefficient and tilt angle information of each wind turbine in the wind farm, the module controls the wind turbines in the wind farm to maximize the total power generation of the wind farm and complete the optimal power control of the wind farm.

[0095] Example 3

[0096] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a wind farm optimal power control method. The memory may include main memory, such as high-speed random access memory (RAM), and may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industry-standard architecture bus, a peripheral component interconnection standard bus, an extended industry-standard architecture bus, etc. The bus may be categorized as an address bus, a data bus, a control bus, etc. The memory stores the program; specifically, the program may include program code, which includes computer operation instructions. The memory may include main memory and non-volatile memory, and provides instructions and data to the processor.

[0097] Example 4

[0098] A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the wind farm optimal power control method. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.

[0099] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0100] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.

[0101] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0102] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0103] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for optimal power control in a wind farm, characterized in that, include: Obtain information on the location, wind speed, and wind direction of each wind turbine in the wind farm; The location, wind speed, and wind direction information of each wind turbine in the wind farm are input into the trained agent to obtain the yaw angle, axial sensing coefficient, and tilt angle information of each wind turbine in the wind farm. Based on the yaw angle, axial sensing coefficient, and tilt angle information of each wind turbine in the wind farm, the wind turbines in the wind farm are controlled to maximize the total power generation of the wind farm and complete the optimal power control of the wind farm. Also includes: Construct training samples, which include the power generation of wind turbine i. ,Location( , Wind speed (wv), wind direction (wd), yaw angle axial inductance coefficient and tilt angle ; Constructing intelligent agents; The agent is trained using the training samples to obtain the trained agent; Before constructing the training samples, the following steps are also included: Obtain the free flow velocity, turbulence intensity, and yaw angle of the wind turbines in the wind field; The wind speed and output power of the wind turbines at their locations are calculated based on the free flow velocity, turbulence intensity, and yaw angle in the wind farm. Based on the free flow velocity, turbulence intensity, and yaw angle of the wind turbine in the wind field, the wind speed and output power of the wind turbine at its location are calculated using the Floris model. The Floris model is as follows: in, air density, For power factor, The rotor plane area, This represents the effective wind speed at the turbine in the wind farm. To correspond to wind speed The rated output power of the turbine, The axial inductance coefficient. For turbine tilt angle, This is the turbine yaw angle.

2. The optimal power control method for a wind farm according to claim 1, characterized in that, The reward / punishment function used in the training process of the agent using the training samples is as follows: （5） in, The reward for the agent is I, where I is the number of wind turbines in the wind farm, and N is the number of agents.

3. The optimal power control method for a wind farm according to claim 1, characterized in that, The agent is trained using the training samples based on the MADDPG algorithm.

4. A wind farm optimal power control system, characterized in that, include: The acquisition module is used to acquire information on the location, wind speed, and wind direction of each wind turbine in the wind farm. The control module is used to input the position, wind speed and wind direction information of each wind turbine in the wind farm into the trained agent to obtain the yaw angle, axial sensing coefficient and tilt angle information of each wind turbine in the wind farm. Based on the yaw angle, axial sensing coefficient and tilt angle information of each wind turbine in the wind farm, the module controls the wind turbines in the wind farm to maximize the total power generation of the wind farm and complete the optimal power control of the wind farm. Also includes: Construct training samples, which include the power generation of wind turbine i. ,Location( , Wind speed (wv), wind direction (wd), yaw angle axial inductance coefficient and tilt angle ; Constructing intelligent agents; The agent is trained using the training samples to obtain the trained agent; Before constructing the training samples, the following steps are also included: Obtain the free flow velocity, turbulence intensity, and yaw angle of the wind turbines in the wind field; The wind speed and output power of the wind turbines at their locations are calculated based on the free flow velocity, turbulence intensity, and yaw angle in the wind farm. Based on the free flow velocity, turbulence intensity, and yaw angle of the wind turbine in the wind field, the wind speed and output power of the wind turbine at its location are calculated using the Floris model. The Floris model is as follows: in, air density, For power factor, The rotor plane area, This represents the effective wind speed at the turbine in the wind farm. To correspond to wind speed The rated output power of the turbine, The axial inductance coefficient. For turbine tilt angle, This is the turbine yaw angle.

5. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the wind farm optimal power control method as described in any one of claims 1-3.

6. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the wind farm optimal power control method as described in any one of claims 1-3.

Images