Wind power plant frequency cooperative control method and device
By combining the Koopman operator and the consensus algorithm, frequency coordinated control of wind farms is achieved, which solves the frequency stability problem caused by the weak inertia of wind turbines and improves the frequency regulation response accuracy and system stability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SOUTH CHINA UNIV OF TECH
- Filing Date
- 2025-12-25
- Publication Date
- 2026-04-21
AI Technical Summary
Wind turbines are connected to the power grid through power electronic devices. The rotor is decoupled from the system frequency and has weak inertia support capacity. After large-scale access, the system inertia is reduced, and frequency stability is challenged. Existing frequency regulation schemes have poor parameter adaptability, inaccurate evaluation, and insufficient coordination.
The Koopman operator theory and consensus algorithm are used for wind farm frequency coordinated control. By upgrading the state space model, dynamic adaptive correction of the droop control coefficient is achieved, thereby improving the frequency regulation response accuracy of the wind farm and the frequency stability of the power system.
Precise quantification of the frequency regulation potential of wind turbine clusters and dynamic matching of the real-time frequency regulation capability of wind turbines improve the frequency regulation response accuracy and system frequency stability of wind farms, and reduce the frequency fluctuation amplitude.
Smart Images
Figure CN121906500A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power control technology, specifically to a method and device for coordinated frequency control of wind farms. Background Technology
[0002] As a core component of clean and renewable energy, wind power continues to increase its share in the power system. However, wind turbines are connected to the grid through power electronic devices, and the rotor is decoupled from the system frequency. This results in weak inertia support capabilities, and large-scale integration will lead to a decrease in system inertia, posing a serious challenge to frequency stability.
[0003] To address this issue, existing technologies often employ droop control within the active power control of wind turbines to participate in the primary frequency regulation of the system. However, traditional frequency regulation schemes have significant drawbacks: First, droop control parameters are mostly fixed values, failing to consider the dynamic operational differences among wind turbines within the wind farm (such as uneven wind speed distribution, wake effects, turbine malfunctions, or speed exceeding limits leading to frequency regulation exit), resulting in poor parameter adaptability. Second, there is a lack of accurate assessment of the dynamic frequency regulation capability of wind turbine groups, and parameter settings lack quantitative basis, easily leading to over-regulation or insufficient response. Third, the lack of a collaborative optimization mechanism among wind turbines within the group, with independent control of each turbine resulting in low overall frequency regulation efficiency and large system frequency fluctuations.
[0004] Therefore, there is an urgent need for a wind farm frequency coordination control technology that can improve the frequency regulation performance of wind farms and solve the problems of parameter fixation, inaccurate evaluation, and insufficient coordination in existing technologies. Summary of the Invention
[0005] To overcome the technical shortcomings of existing wind farm frequency regulation, such as poor adaptability of droop control parameters, inaccurate frequency regulation capability assessment, and insufficient intra-group coordination, this invention provides a wind farm frequency collaborative control method and device. This invention is based on Koopman operator theory and consensus algorithms for wind farm frequency collaborative control. Koopman operator theory transforms nonlinear dynamic characteristics into high-dimensional linear relationships through state space dimensionality enhancement, making it suitable for modeling and identifying complex wind farm dynamic characteristics. The consensus algorithm enables collaborative optimization of multi-agent systems, providing technical support for adaptive correction of wind turbine parameters within the group, thereby achieving dynamic adaptive correction of the droop control coefficient and improving the frequency regulation response accuracy of the wind farm and the frequency stability of the power system.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides a wind farm frequency coordinated control method, comprising the following steps: A frequency regulation capability evaluation model for wind turbine clusters is constructed based on the Koopman operator, and the limit droop coefficient of each wind turbine cluster is obtained. The droop control coefficient of each wind turbine in the group is adaptively corrected based on the consensus algorithm to obtain the corrected dynamic droop control coefficient of each wind turbine. When a fluctuation in the frequency of a wind farm is detected, each wind turbine adjusts its output active power according to the dynamic droop control coefficient. When the frequency decreases, the active power output increases, and when the frequency increases, the active power output decreases. Through the coordinated optimization of the wind turbines in the group, the overall frequency regulation response of the wind farm is achieved.
[0007] As a preferred technical solution, a frequency regulation capability evaluation model for wind turbine clusters is constructed based on the Koopman operator to obtain the limiting droop coefficient of each wind turbine cluster, specifically including: Offline training is conducted using historical or simulation data to obtain key data of wind farms participating in frequency regulation, including: minimum speed of wind turbine group, droop control coefficient, number of wind turbines and wind speed; The droop control coefficient, number of wind turbines, and wind speed are upgraded to a higher dimension, mapping the state space to a higher dimension space. Based on the Koopman operator, a high-dimensional linear relationship is constructed between the droop control coefficient, the number of wind turbines, and the wind speed and minimum speed to obtain a frequency regulation capability evaluation model. Based on the real-time clustering results of the wind farm, the real-time number of wind turbines and the equivalent wind speed of each wind turbine group are input into the frequency regulation capability evaluation model to calculate the limit droop coefficient of each wind turbine group. The limit droop coefficient is the maximum droop control coefficient corresponding to the extreme value of the wind turbine speed when it reaches the limit value during the frequency regulation process.
[0008] As a preferred technical solution, the collected historical data is trained offline by extending the dynamic pattern decomposition algorithm.
[0009] As a preferred technical solution, the droop control coefficient, number of wind turbines, and wind speed are upgraded, specifically including: The observation function is constructed based on Hermitian polynomials, and the droop control coefficient, number of wind turbines, and wind speed are increased in dimension and expressed as follows: ; in, For the input variable vector, Represents the observation function, This is the droop control coefficient. Number of wind turbines This refers to wind speed.
[0010] As a preferred technical solution, a high-dimensional linear relationship between the droop control coefficient, the number of wind turbines, wind speed, and minimum rotational speed is constructed based on the Koopman operator, specifically expressed as follows: ; in, This represents the minimum operating speed of the wind turbine group during frequency regulation. This is the droop control coefficient. Number of wind turbines For wind speed, For the observation function, This is the Koopman operator.
[0011] As a preferred technical solution, the droop control coefficients of each wind turbine in the group are adaptively corrected based on a consensus algorithm, specifically including: Real-time collection of rotational speed data for each fan, and calculation of consistency status indicators for each fan; Calculate the differences in the status indicators of each wind turbine within the same wind turbine group; The correction amount for the droop control coefficient is generated through a proportional-integral control strategy. The limit droop coefficient is corrected based on the correction amount to obtain the corrected dynamic droop control coefficient for each wind turbine.
[0012] As a preferred technical solution, the correction amount is expressed as follows: ; in, , respectively wind turbine The ratio and integral coefficient of the state deviation from other units in the same group , respectively wind turbine Japanese wind machine Consistency status index This indicates the relationship between two wind turbines. When, it indicates the wind turbine Japanese wind machine Belonging to the same aircraft group, when When, it indicates the fan. Japanese wind machine They do not belong to the same aircraft group.
[0013] As a preferred technical solution, the consistency status index is expressed as follows: ; in, The initial speed of the fan before it participates in frequency regulation. This is the speed limit value. This indicates the offset of the wind turbine's available frequency regulation capability at this moment relative to the initial moment.
[0014] As a preferred technical solution, the corrected dynamic sag control coefficient for each wind turbine is expressed as follows: ; in, For the first The corrected dynamic sag control coefficient for typhoon turbines. For wind turbine groups The assessed limit sag coefficient, For wind turbine groups The average droop coefficient of each unit.
[0015] The present invention also provides a wind farm frequency coordinated control device for implementing the above-mentioned wind farm frequency coordinated control method, comprising: an evaluation module, a correction module, and a coordinated control module; The evaluation module is used to construct a frequency regulation capability evaluation model for wind turbine groups based on the Koopman operator, and to obtain the limit droop coefficient of each wind turbine group. The correction module is used to adaptively correct the droop control coefficient of each wind turbine in the group based on the consensus algorithm, so as to obtain the corrected dynamic droop control coefficient of each wind turbine. The collaborative control module is used to achieve the overall frequency regulation response of the wind farm through the collaborative optimization of the wind turbines in the group. When it is determined that the frequency of the wind farm fluctuates, each wind turbine adjusts its output active power according to the dynamic droop control coefficient. When the frequency decreases, the active power output increases, and when the frequency increases, the active power output decreases.
[0016] Compared with the prior art, the present invention has the following advantages and beneficial effects: (1) This invention establishes a frequency regulation capability evaluation model based on the Koopman operator theory, accurately characterizes the nonlinear frequency regulation characteristics of wind turbine groups through a data-driven approach, and provides a quantitative basis for parameter correction by calculating the limit droop coefficient, thus avoiding the blindness of traditional fixed parameters.
[0017] (2) The present invention uses a consensus algorithm to realize the adaptive correction of the droop control coefficient of the wind turbine in the group, dynamically matches the real-time frequency adjustment capability of the wind turbine, solves the problem of poor parameter adaptability in dynamic scenarios such as wind speed change and unit shutdown, and improves the coordination of wind turbines in the group.
[0018] (3) By using dynamic droop control coefficient and collaborative control strategy, the present invention significantly improves the frequency regulation response accuracy of wind farms, reduces the frequency fluctuation amplitude of the system, and effectively enhances the frequency stability of the power system under high wind power penetration. Attached Figure Description
[0019] Figure 1 This is a schematic diagram of the overall process of the wind farm frequency coordinated control method of the present invention; Figure 2 A flowchart illustrating the frequency coordination control of a wind farm that takes into account wind turbine operating conditions; Figure 3 This is a flowchart of the wind farm frequency regulation capability evaluation based on Koopman theory in this invention; Figure 4 A comparison chart of training results for evaluating the frequency regulation capability of wind farms based on Koopman theory under a sudden load surge scenario; Figure 5 A comparison chart of training results for evaluating the frequency regulation capability of wind farms based on Koopman theory under a sudden load reduction scenario; Figure 6 A comparison chart of the changes in the speed of each fan before and after the correction of the droop control coefficient; Figure 7 A comparison chart of active power output of wind farms under different control strategies; Figure 8 This is a comparison chart of the system frequency response under different control strategies. Detailed Implementation
[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0021] This embodiment uses a wind farm with 24 doubly-fed induction generators (DFIGs) as an example. The rated capacity of each wind turbine is 2MW, and the speed limit is... Initial rotational speed The range is 0.635~0.829 pu; the wind farm is connected to the three-machine nine-node power system after being collected by the collection line, and the system base capacity is 100MW; the simulation step size is set to 0.001s, the simulation duration is 50s, and the disturbance condition is: when the system load suddenly increases by 50MW (scenario 1) or suddenly decreases by 50MW (scenario 2) after running for 10s, the wind farm is triggered to participate in the primary frequency regulation; like Figure 1 As shown, this embodiment provides a wind farm frequency coordinated control method, including the following steps: S1: As Figure 2 As shown, the wind turbines in the wind farm are grouped according to their operating conditions. Based on the Koopman operator, a frequency regulation capability evaluation model for the wind turbine group is established through a data-driven approach to obtain the limit droop coefficient of each wind turbine group. like Figure 3 As shown, this embodiment constructs a frequency regulation capability assessment model through a data-driven approach to accurately quantify the maximum frequency regulation potential of a wind turbine cluster, specifically including: (1) Based on the operating conditions of each wind turbine in the wind farm, the wind farm is grouped and offline training is performed using historical data or simulation data to obtain key data after the wind farm participates in frequency regulation, including: the minimum speed of the wind turbine group. Initial droop control coefficient Number of wind turbines Equivalent wind speed taking wake effect into account The equivalent wind speed needs to take into account the wake effect of the upstream fan on the downstream fan, and is calculated using the Jessen model. In this embodiment, a clustering method based on gap measure is adopted to divide the 24 wind turbines into 4 groups. The clustering results are shown in Table 1 below, which ensures that the dynamic characteristics of wind turbines within the same group are similar, laying the foundation for subsequent group-level evaluation and coordinated control.
[0022] Table 1. Results of wind turbine clustering
[0023] In this embodiment, historical operating data of each turbine group was obtained through Matlab / Simulink simulation, including the minimum turbine speed under different wind speeds and different droop control coefficients. A total of 5,000 sets of data were collected, of which 3,000 sets were used as the training set and 2,000 sets were used as the test set. (2) Upgrade the droop control coefficient, number of fans and wind speed to map the state space to a higher dimension; In this embodiment, since the frequency regulation characteristics of the wind turbine group (the relationship between the minimum speed and the droop control coefficient, the number of wind turbines, and the wind speed) are nonlinear, an observation function is introduced. An orthogonal basis is constructed using Hermitian polynomials to ensure the linear separability of variable relationships in high-dimensional space. The expression is as follows: ; in, For the input variable vector; In this embodiment, a third-order Hermitian polynomial is used to construct the observation function. For input variables The dimension is increased, and the spatial dimension becomes 10. (3) Using the Koopman operator, a high-dimensional linear relationship between the droop control coefficient, the number of wind turbines, and the wind speed and minimum rotational speed is constructed to obtain the frequency regulation capability evaluation model, which is expressed as: ; in, This represents the minimum operating speed of the wind turbine group during frequency regulation. This is the droop control coefficient. Number of wind turbines For wind speed, For the observation function, For the Koopman operator; In this embodiment, the Extended Dynamic Mode Decomposition (EDMD) algorithm is used to train and solve the collected historical data offline, thereby achieving a linearized characterization of the nonlinear frequency modulation characteristics. The Koopman operator matrix is solved, a frequency modulation capability evaluation model is established, and Koopman operator theory is used to fit the model. The results are as follows: Figure 4 , Figure 5 As shown, the test set validation results indicate that the maximum fitting error is 0.5% (load surge scenario) and 1.52% (load decrease scenario), and the model accuracy meets engineering requirements.
[0024] (4) Based on the real-time clustering results of the wind farm (a clustering method based on gap measure can be used to group wind turbines with similar dynamic characteristics into the same group), input the real-time number of wind turbines and equivalent wind speed of each wind turbine group into the frequency regulation capability evaluation model, and calculate the limit droop coefficient of each wind turbine group. The limit droop coefficient is the maximum droop control coefficient that ensures the fan does not trigger speed over-limit exit during frequency regulation, providing an upper limit constraint for subsequent parameter correction; In this embodiment, the real-time equivalent wind speed (taking wake effect) and the number of wind turbines of each wind turbine group are input into the evaluation model, and the limiting droop coefficient of each wind turbine group is obtained by back-calculation. The results are shown in Tables 2 and 3 below: Table 2 Comparison of Average Errors in Dynamic Response of Equivalent Units (Load Surge Scenario)
[0025] Table 3 Comparison of Average Errors in Dynamic Response of Equivalent Units (Sudden Load Reduction Scenario)
[0026] The above calculation method evaluates the ultimate droop coefficient of the entire wind turbine group. However, there are certain differences among the wind turbines in the group. If all wind turbines in the group adopt this control parameter, the result will have a certain deviation. The Koopman evaluation model evaluates the droop control coefficient corresponding to the extreme speed of the wind turbine when it participates in frequency regulation. Some deviations may lead to the speed exceeding the limit. Therefore, it is necessary to further subdivide the frequency regulation control parameters of each wind turbine. Based on the ultimate droop coefficient result obtained by the Koopman method, the frequency regulation parameters should be further corrected considering the state differences between each wind turbine. S2: Based on the limit droop coefficient, the consensus algorithm is used to adaptively correct the droop control coefficient of each wind turbine in the group, so as to obtain the corrected dynamic droop control coefficient of each wind turbine. like Figure 2 As shown, a consensus algorithm is used to adaptively correct the droop control coefficients of each wind turbine in the group. The consensus algorithm achieves coordinated balancing of the frequency regulation capabilities of the wind turbines within the group, specifically including: (1) Real-time acquisition of rotational speed data of each fan, calculation of consistency status index of each fan, used to measure the frequency regulation capability of the fan, specifically expressed as: ; in, The initial speed of the fan before it participates in frequency regulation. This is the speed limit value. This indicates the offset of the wind turbine's available frequency regulation capability at this moment relative to the initial moment; (2) Calculate the difference in the status indicators of each wind turbine within the same wind turbine group, expressed as: ; in, , respectively wind turbine Japanese wind machine Consistency status indicators; (3) Based on the difference value, the correction amount of the droop control coefficient is generated through the proportional-integral control strategy. Specifically, it is expressed as: ; in, , respectively wind turbine The ratio and integral coefficient of the state deviation from other units in the same group , respectively wind turbine Japanese wind machine Consistency status index This indicates the relationship between two wind turbines. When, it indicates the wind turbine Japanese wind machine Belonging to the same aircraft group, when When, it indicates the fan. Japanese wind machine They do not belong to the same aircraft group; In this embodiment, the communication weight is set to 1 (i.e., full connectivity communication among all wind turbines in the group), the proportional coefficient is 0.3, and the integral coefficient is 0.05. The correction amount for the droop control coefficient of each wind turbine is calculated. .
[0027] (4) Combining the correction amount and the adaptive component, the limit droop coefficient is corrected to obtain the corrected dynamic droop control coefficient for each wind turbine, expressed as: ; in, For the first The corrected dynamic sag control coefficient for typhoon turbines. For wind turbine groups The assessed limit sag coefficient, For wind turbine groups The average sag coefficient of each unit can be used to calculate the sag coefficient of each wind turbine in the entire wind farm. S3: Based on the dynamic droop control coefficient, achieve coordinated control of wind farm participation in the primary frequency regulation of the system.
[0028] In this embodiment, the dynamic droop control coefficient of each wind turbine is input into the active power control module of the wind turbine group, replacing the traditional fixed droop coefficient. When the system frequency fluctuates, each wind turbine adjusts its output active power according to the real-time dynamic droop control coefficient: when the frequency decreases, the active power output is increased; when the frequency increases, the active power output is decreased. Through the coordinated optimization of wind turbines within the group, the overall frequency regulation response of the wind farm is accurately matched with the system frequency requirements, thereby improving the frequency regulation accuracy and system stability.
[0029] The control effect of this invention is verified by comparing it with the traditional fixed parameter method through simulation. The simulation results are as follows: Table 4 Comparison of Control Effects (Sudden Load Increase Scenario) Table 5 Comparison of Control Effects (Sudden Load Reduction Scenario)
[0030] The results show that the method of the present invention has a lower minimum frequency value (in the case of a sudden increase in load) or a higher frequency value (in the case of a sudden decrease in load) and a smaller frequency change rate compared with the traditional fixed parameter method, indicating that the dynamic droop control coefficient and the cooperative control strategy effectively improve the frequency modulation accuracy and system stability.
[0031] After adaptively controlling the droop control coefficients of all wind turbines in the wind farm, the effect is as follows: Figure 6 As shown in the figures (left: load surge scenario, right: load decrease scenario), it can be seen that after the droop coefficient correction, the state indicators of wind turbines within the same cluster tend to be the same. Under this control strategy, each wind turbine achieves adaptive control of the droop coefficient according to its frequency regulation capability. None of the wind turbines in the wind farm reach the minimum limit, avoiding the situation where they are forced to exit frequency regulation due to exceeding the speed limit.
[0032] like Figure 7 and Figure 8 The figures shown are comparisons of the frequency regulation effects of wind farms not participating in frequency regulation, based on the Koopman evaluation method (non-adaptive), and the strategy of this invention (the left figure shows a scenario of sudden load increase, and the right figure shows a scenario of sudden load decrease).
[0033] Under the scenario of sudden load surge, the highest active power output of the wind farm in the strategy of this invention is 126.62MW, which is higher than that of the Koopman evaluation method. This is because the wind farm has a strong frequency regulation capability in the early stage of frequency regulation, and the strategy of this invention adjusts the droop control coefficient according to the frequency regulation capability, thus releasing more frequency regulation energy in the early stage of frequency regulation, which is more conducive to providing frequency support for the system. However, in the later stage of frequency regulation, it can be seen that the total active power output of the wind farm in the Koopman evaluation method drops significantly. This is because some units exit frequency regulation due to speed exceeding the limit, resulting in a large power deficit, which also leads to a secondary frequency drop (see...). Figure 8 In this invention, the droop control coefficient adaptively decreases as the unit speed decreases (i.e., the frequency regulation capability decreases). Therefore, all units can remain above the minimum limit throughout the entire frequency regulation process, resulting in a smoother power output.
[0034] In scenarios involving sudden load reduction, the total active power output curve of the wind farm in this invention's strategy is generally lower than that obtained using the Koopman evaluation method. This is because the wind turbines in this invention adjust their droop coefficients according to their own conditions, ensuring that the rotor speed remains within a reasonable range, thus sustainably and smoothly reducing active power output and providing frequency support for the system. In contrast, the Koopman evaluation method assesses the ultimate droop coefficient of the entire wind turbine group. Some turbines in the group may have actual frequency regulation capabilities lower than the assessed value, leading to excessive release of regulation capabilities and causing them to exit frequency regulation due to rotor speed exceeding limits. This significant power deficit results in a secondary frequency increase, exacerbating the severity of the accident. This invention's strategy can adjust the droop coefficient according to its own conditions, releasing greater frequency regulation energy in the initial stage of frequency regulation. Therefore, compared to the other two control strategies, it improves the maximum frequency change rate, which is more conducive to the safe and stable operation of the system.
[0035] This embodiment verifies the feasibility and superiority of the method of the present invention through actual simulation. Compared with the traditional fixed parameter control method, the frequency regulation capability evaluation model based on Koopman operator theory of the present invention can accurately quantify the dynamic frequency regulation potential of wind turbines, the consistency algorithm realizes the adaptive correction of the droop control coefficient, and the cooperative control strategy improves the overall frequency regulation performance of the wind farm, effectively solving the defects of the prior art.
[0036] Example 2 This embodiment provides a wind farm frequency coordinated control device for implementing the wind farm frequency coordinated control method of Embodiment 1. The device includes: an evaluation module, a correction module, and a coordinated control module. The evaluation module is used to build a frequency regulation capability evaluation model for wind turbine groups based on the Koopman operator. It receives real-time wind farm grouping results, number of wind turbines and equivalent wind speed data, calculates the limit droop coefficient of each wind turbine group, and outputs it to the correction module. The correction module receives the limit droop coefficient output by the evaluation module, collects the real-time speed data of each wind turbine in the group, calculates the consistency status index and the state difference within the group, generates the correction amount of the droop control coefficient through the consistency algorithm (PI control), and finally outputs the dynamic droop control coefficient of each wind turbine to the control module. The collaborative control module receives the dynamic droop control coefficient output by the correction module, embeds it into the active power control loop of the wind turbine, and controls each wind turbine to adjust its active power output in real time according to the system frequency deviation signal, so as to realize the collaborative control of the wind farm participating in the primary frequency regulation of the system.
[0037] This invention accurately assesses frequency regulation capability through a data-driven approach and dynamically corrects droop parameters, solving the problems of poor parameter adaptability and insufficient coordination in traditional control. It can prevent wind turbine speed from exceeding the limit and exiting frequency regulation, significantly improving the frequency regulation response accuracy of wind farms and the frequency stability of power systems. It is suitable for primary frequency regulation optimization of power systems with high wind power penetration.
[0038] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.
Claims
1. A wind farm frequency coordinated control method, characterized in that, Includes the following steps: A frequency regulation capability evaluation model for wind turbine clusters is constructed based on the Koopman operator, and the limit droop coefficient of each wind turbine cluster is obtained. The droop control coefficient of each wind turbine in the group is adaptively corrected based on the consensus algorithm to obtain the corrected dynamic droop control coefficient of each wind turbine. When a fluctuation in the frequency of a wind farm is detected, each wind turbine adjusts its output active power according to the dynamic droop control coefficient. When the frequency decreases, the active power output increases, and when the frequency increases, the active power output decreases. Through the coordinated optimization of the wind turbines in the group, the overall frequency regulation response of the wind farm is achieved.
2. The wind farm frequency coordinated control method according to claim 1, characterized in that, A frequency regulation capability assessment model for wind turbine clusters is constructed based on the Koopman operator, and the limiting droop coefficient of each wind turbine cluster is obtained, specifically including: Offline training is conducted using historical or simulation data to obtain key data of wind farms participating in frequency regulation, including: minimum speed of wind turbine group, droop control coefficient, number of wind turbines and wind speed; The droop control coefficient, number of wind turbines, and wind speed are upgraded to a higher dimension, mapping the state space to a higher dimension space. Based on the Koopman operator, a high-dimensional linear relationship is constructed between the droop control coefficient, the number of wind turbines, and the wind speed and minimum speed to obtain a frequency regulation capability evaluation model. Based on the real-time clustering results of the wind farm, the real-time number of wind turbines and the equivalent wind speed of each wind turbine group are input into the frequency regulation capability evaluation model to calculate the limit droop coefficient of each wind turbine group. The limit droop coefficient is the maximum droop control coefficient corresponding to the extreme value of the wind turbine speed when it reaches the limit value during the frequency regulation process.
3. The wind farm frequency coordinated control method according to claim 2, characterized in that, Offline training was performed on the collected historical data using an extended dynamic pattern decomposition algorithm.
4. The wind farm frequency coordinated control method according to claim 2, characterized in that, The sag control coefficient, number of wind turbines, and wind speed are upgraded, specifically including: The observation function is constructed based on Hermitian polynomials, and the droop control coefficient, number of wind turbines, and wind speed are increased in dimension and expressed as follows: ； in, For the input variable vector, Represents the observation function, This is the droop control coefficient. Number of wind turbines This refers to wind speed.
5. The wind farm frequency coordinated control method according to claim 2, characterized in that, Based on the Koopman operator, a high-dimensional linear relationship is constructed between the droop control coefficient, the number of wind turbines, wind speed, and minimum rotational speed, specifically expressed as follows: ； in, This represents the minimum operating speed of the wind turbine group during frequency regulation. This is the droop control coefficient. Number of wind turbines For wind speed, For the observation function, This is the Koopman operator.
6. The wind farm frequency coordinated control method according to claim 1, characterized in that, The droop control coefficients of each wind turbine in the group are adaptively corrected based on the consensus algorithm, specifically including: Real-time collection of rotational speed data for each fan, and calculation of consistency status indicators for each fan; Calculate the differences in the status indicators of each wind turbine within the same wind turbine group; The correction amount for the droop control coefficient is generated through a proportional-integral control strategy. The limit droop coefficient is corrected based on the correction amount to obtain the corrected dynamic droop control coefficient for each wind turbine.
7. The wind farm frequency coordinated control method according to claim 6, characterized in that, The correction amount is expressed as: ； in, , respectively wind turbine The ratio and integral coefficient of the state deviation from other units in the same group , respectively wind turbine Japanese-style fan Consistency status index This indicates the relationship between two wind turbines. When, it indicates the wind turbine Japanese-style fan Belonging to the same aircraft group, when When, it indicates the fan. Japanese-style fan They do not belong to the same aircraft group.
8. The wind farm frequency coordinated control method according to claim 7, characterized in that, The consistency status index is expressed as: ； in, The initial speed of the fan before it participates in frequency regulation. This is the speed limit value. This indicates the offset of the wind turbine's available frequency regulation capability at this moment relative to the initial moment.
9. The wind farm frequency coordinated control method according to claim 8, characterized in that, The corrected dynamic sag control coefficient for each wind turbine is expressed as follows: ； in, For the first The corrected dynamic sag control coefficient for typhoon turbines. For wind turbine groups The assessed limit sag coefficient, For wind turbine groups The average droop coefficient of each unit.
10. A wind farm frequency coordinated control device, characterized in that, The method for implementing the wind farm frequency coordinated control method according to any one of claims 1-9 includes: an evaluation module, a correction module, and a coordinated control module; The evaluation module is used to construct a frequency regulation capability evaluation model for wind turbine groups based on the Koopman operator, and to obtain the limit droop coefficient of each wind turbine group. The correction module is used to adaptively correct the droop control coefficient of each wind turbine in the group based on the consensus algorithm, so as to obtain the corrected dynamic droop control coefficient of each wind turbine. The collaborative control module is used to achieve the overall frequency regulation response of the wind farm through the collaborative optimization of the wind turbines in the group. When it is determined that the frequency of the wind farm fluctuates, each wind turbine adjusts its output active power according to the dynamic droop control coefficient. When the frequency decreases, the active power output increases, and when the frequency increases, the active power output decreases.