Doubly-fed fan modeling method and system

Abstract

The invention discloses a doubly-fed fan modeling method and system, and relates to the technical field of fan modeling, and the method comprises the steps: selecting a clustering index, and collecting data; clustering the wind power plant data by adopting an improved K-means algorithm to construct an equivalent station model; and identifying the obtained equivalent station model. According to the method, the wind speed of the wind power plant is selected as a clustering index, meanwhile, the wind speed is closely related to control over the pitch angle of the doubly-fed wind turbine, and according to different wind speeds, the pitch angle is increased or decreased to reduce the power output of the wind turbine, so that the wind turbine is kept at the rated power level or more kinetic energy is captured, and starting of the wind turbine is facilitated; actual measurement data of an actual wind power plant can be well fitted, accurate modeling, clustering and identification optimization of the wind power plant are achieved, and the actual operation condition of the wind power plant can be accurately reflected by errors within an allowable range. Accuracy and consistency of data are ensured, and a reliable basis is provided for operation control of the wind power plant.

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbine modeling, and in particular to a doubly-fed wind turbine modeling method and system. Background Art

[0002] The scale of new energy sources continues to expand, and wind power has become an important part of the global energy structure. However, the output volatility of wind turbines has brought new challenges to the stability and reliability of the power system. Therefore, in order to effectively plan the power grid and adjust the output of wind turbines, it is necessary to perform equivalent modeling of large-scale wind turbines. At this stage, equivalent methods are roughly divided into two categories: single-machine equivalence and multi-machine equivalence. Single-machine equivalence is to abstract the entire wind farm into a single functional unit. The equivalent wind farm is often compared with the actual one, so a modeling method that can better fit the characteristics of the actual wind farm is needed. Summary of the invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] Therefore, the technical problem solved by the present invention is: how to better fit the characteristics of the actual wind field and reduce the error.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a doubly-fed wind turbine modeling method, including: standardized collection, storage, and management of energy data; multi-party computing participants use a trusted SDK to perform homomorphic data encryption, and transmit the encrypted data to an energy big data platform for data calculation; and centralize fused computing data from different sources for data sharing.

[0006] As a preferred scheme of the doubly-fed wind turbine modeling method described in the present invention, wherein: a clustering index is selected and data is collected; the wind speed of the wind farm is selected using the clustering index, and according to different wind speeds, the pitch angle is increased or decreased to reduce the power output of the wind turbine to keep it at the rated power level or capture more kinetic energy to facilitate the start-up of the wind turbine.

[0007] As a preferred solution of the doubly-fed wind turbine modeling method described in the present invention, data processing is performed before and during clustering. It is assumed that the wind farm is composed of N×M wind turbines of the same model. After the wind farm is divided into P groups, each group with similar wind speed is classified into a class equivalent to a large wind turbine, wherein the capacity of the equivalent wind turbine is the product of the number of wind turbines in the group and the capacity of a single wind turbine. The processing formula for the samples is expressed as follows:

[0008]

[0010] Among them, x′ i For processed data, is the data average, is the standard deviation of the data.

[0011] As a preferred solution of the doubly-fed wind turbine modeling method described in the present invention, the construction of an equivalent station model includes: inputting the wind speed of the wind turbine measured in the wind farm; using the silhouette coefficient SC as the K-means scoring indicator, and using the improved data as the initial clustering center point; applying the K-means algorithm to iterate, calculate SC, and select the number of classifications of the wind farm according to the degree to which its value is close to 1; and finally obtaining n classified equivalent wind turbines.

[0012] As a preferred solution of the doubly-fed wind turbine modeling method described in the present invention, the improved K-means algorithm includes selecting a sample from the processed data as the center of the initial cluster, calculating the shortest distance D(x′) between each sample and the current existing cluster center. i ), calculate the probability P of each sample being selected as the next cluster center, expressed as:

[0013]

[0014] The next cluster center is selected according to the roulette method until K cluster centers that meet the requirements are selected.

[0015] As a preferred solution of the doubly-fed wind turbine modeling method described in the present invention, for the evaluation index of K-means, the concept of silhouette coefficient SC is used to quantify the consistency in the cluster to which each data point belongs and the separation between clusters. For each point m in the data set, the calculation formula of the silhouette coefficient s(m) is expressed as:

[0016]

[0017] As a preferred solution of the doubly-fed wind turbine modeling method described in the present invention, the equivalent station model identification includes, after equivalent modeling, data preprocessing, denoising the wind speed and pitch angle time series data, and eliminating abnormal values. At the same time, the parameter set is initialized, the error tolerance of the identification is set, and the initial weight coefficient set of the target parameter in the identification is defined;

[0018] Select the key dynamic response parameters of the wind farm, collect the dynamic response data of the above parameters in time series in real time for equivalent groups in different wind speed ranges, and link the wind speed data with the pitch angle adjustment of the doubly fed wind turbine to ensure the response consistency of the model under different wind speed conditions and provide diversified sampling input data for identification;

[0019] Based on the clustering results and the characteristics of the wind turbine model, a state space model is generated for each type of equivalent cluster. This model contains the dynamic response equations of the electrical and mechanical structures of the wind turbines. The state space is divided into multiple sub-intervals so that the dynamic response characteristics of the equivalent cluster of wind farms can be mapped to different state intervals to ensure that the dynamic response of the wind farm under the condition of rapid wind speed changes can be accurately identified.

[0020] Multi-level parameter optimization is performed on each equivalent cluster, and the genetic algorithm is used to adjust the key parameters in the state space model. An adaptive weight allocation mechanism is introduced in the optimization, and different weights are gradually adjusted according to the error of the identified target parameters to focus on the main response of the wind turbine under different working conditions. Through multiple iterations, the objective function converges within the preset error range.

[0021] Based on the identification model, the same input conditions as the actual wind speed changes are applied in the simulation environment to obtain the dynamic output sequence of the simulation model. The simulation results are fitted with the measured data, and the mean square error (MSE) and mean absolute error (MAE) of each set of data are calculated. Based on this, the model parameters are further adjusted to ensure that the error between the simulation data and the measured data is gradually reduced to the set range.

[0022] The identification error is compared with the target accuracy. For the interval with large error, the data is resampled and the model parameters are updated. The above process is repeated for iteration. The number of iterations is controlled according to the score of the silhouette coefficient SC until the model reaches the optimal classification and identification accuracy.

[0023] When the identification error meets the preset requirements, the integrity and robustness of the final equivalent site model are tested. The robustness of the model under different wind speed conditions is verified through multiple tests to ensure that the model can accurately predict the dynamic response of the wind farm in practical applications and use the model as the equivalent site model of the wind farm.

[0024] In the second aspect, another object of the present invention is to provide a doubly fed wind turbine modeling system, comprising: a data acquisition module, a clustering modeling module and an identification and optimization module; the data acquisition module is responsible for selecting and collecting wind farm data; the clustering modeling module is used to cluster wind farm data based on an improved K-means algorithm to construct an equivalent station model; the identification and optimization module is used to identify the equivalent station model to ensure the accuracy and dynamic response characteristics of the model.

[0025] In a third aspect, a computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the doubly-fed wind turbine modeling method as described above when executing the computer program.

[0026] In a fourth aspect, a computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the doubly-fed wind turbine modeling method as described above.

[0027] Beneficial effects of the present invention: The doubly-fed wind turbine modeling method and system provided by the present invention selects the wind speed of the wind farm as a clustering index. At the same time, the wind speed is closely related to the pitch angle control of the doubly-fed wind turbine. According to different wind speeds, the pitch angle increases or decreases the angle to reduce the power output of the wind turbine so that it remains at the rated power level or captures more kinetic energy to facilitate the start-up of the wind turbine. Therefore, selecting wind speed as a clustering index facilitates real-time optimization of the pitch angle at different wind speeds and adjustment of the controller gain to ensure that the power output of the wind turbine is as close to the maximum power point as possible.

[0028] It can fit the measured data of the actual wind farm very well, and can accurately model, cluster and identify the wind farm, and its error is within the allowable range, which can accurately reflect the actual operation of the wind farm. It ensures the accuracy and consistency of the data and provides a reliable basis for the operation control of the wind farm. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0030] Figure 1 An overall flow chart of a doubly-fed wind turbine modeling method provided by one embodiment of the present invention;

[0031] Figure 2 A wind field clustering diagram of a doubly-fed wind turbine modeling method provided by an embodiment of the present invention;

[0032] Figure 3 A comparison diagram of measured data and simulation data of wind farm A in a doubly-fed wind turbine modeling method provided by one embodiment of the present invention;

[0033] Figure 4 Comparison between measured data and simulation data of wind farm B of a doubly-fed wind turbine modeling method provided by one embodiment of the present invention;

[0034] Figure 5 A comparison between measured data and simulation data of a wind farm C of a doubly-fed wind turbine modeling method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0036] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0037] Example 1

[0038] Reference Figure 1 , is an embodiment of the present invention, and provides a doubly-fed wind turbine modeling method, comprising:

[0039] S1: Select clustering indicators and collect data;

[0040] Furthermore, before multi-machine equivalent modeling, it is necessary to select a clustering index. In order to facilitate the acquisition of indicators and improve clustering efficiency, this paper selects the wind speed of the wind farm as the clustering index. At the same time, the wind speed is closely related to the pitch angle control of the doubly fed wind turbine. Depending on the wind speed, the pitch angle increases or decreases the angle to reduce the power output of the wind turbine to keep it at the rated power level or capture more kinetic energy to help start the wind turbine. Therefore, selecting wind speed as the clustering index is convenient for real-time optimization of the pitch angle at different wind speeds and adjustment of the controller gain to ensure that the power output of the wind turbine is as close to the maximum power point as possible.

[0041] S2: Use the improved K-means algorithm to cluster wind farm data and construct an equivalent station model;

[0042] Furthermore, data processing is required during clustering. The clustering method selected in this paper is to use improved K-means for clustering. Assuming that a wind farm consists of N×M wind turbines of the same model, the above method is used to divide the wind farm into P groups, and then each group with similar wind speed is classified into a class equivalent to a large wind turbine, where the capacity of the equivalent wind turbine is the product of the number of wind turbines in the group and the capacity of a single wind turbine.

[0043] It should be noted that the traditional K-means algorithm is sensitive to initial values, so the improved K-means will first process the samples before clustering, as shown in formula (1):

[0044]

[0045] Among them, x′ i For processed data, is the data average, is the data standard deviation. The improved K-means algorithm is as follows:

[0046] Select a sample from the processed data as the center of the initial cluster'

[0047] Calculate the shortest distance D(x′) between each sample and the current cluster center i );

[0048] Calculate the probability P of each sample being selected as the next cluster center, as shown in formula (2):

[0049]

[0050] Select the next cluster center according to the roulette method until K cluster centers that meet the requirements are selected

[0051] For the evaluation index of K-means, some statisticians proposed the concept of Silhouette Coefficient (SC). SC can quantify the consistency of each data point in the cluster and the separation between clusters. The value range of the silhouette coefficient is [-1, 1]. The larger the value, the better the clustering effect. For each point m in the data set, the calculation formula of the silhouette coefficient s(m) is shown in formula (3):

[0052]

[0053] This paper uses the data selected based on the improved k-means as the initial cluster center, which can obtain more accurate cluster centers than the traditional K-means. The specific steps are as follows:

[0054] 1. Input the wind speed of the fan measured at the wind site;

[0055] 2. Use the silhouette coefficient SC as the K-means scoring indicator and the improved data as the initial cluster center point;

[0056] 3. Apply the K-means algorithm to iterate and calculate SC, and select the number of wind farm classifications according to the degree to which its value is close to 1;

[0057] 4. Finally, we get n classified and equivalent wind turbines.

[0058] S3: Identify the obtained equivalent station model.

[0059] Furthermore, (1) data preprocessing and parameter initialization: After equivalent modeling, in order to further improve the accuracy of model identification, data preprocessing is first performed to denoise the time series data such as wind speed and pitch angle to eliminate abnormal values. At the same time, the parameter set is initialized, the error tolerance of identification is set, and the initial weight coefficient set of the target parameters in the identification is defined so as to adjust the wind turbine response under different wind speeds.

[0060] (2) Model parameter determination and dynamic response sampling: Select the key dynamic response parameters of the wind farm, such as voltage U, reactive current I, reactive power Q, and active power P. For equivalent groups in different wind speed ranges, collect the dynamic response data of the above parameters in time series in real time. By linking the wind speed data with the pitch angle adjustment of the doubly fed wind turbine, the response consistency of the model under different wind speed conditions is ensured, and a variety of sampling input data is provided for identification.

[0061] (3) Model structure generation and state space partitioning: Based on the clustering results and wind turbine model characteristics, a state space model is generated for each type of equivalent cluster. This model contains the dynamic response equations of the electrical and mechanical structures of the wind turbine. Then, the state space is divided into multiple sub-intervals so that the dynamic response characteristics of the equivalent cluster of wind farms can be mapped to different state intervals to ensure that the dynamic response of the wind farm under rapid wind speed changes can be accurately identified.

[0062] (4) Multi-level parameter optimization: Perform multi-level parameter optimization on each equivalent cluster, and use an improved genetic algorithm to adjust key parameters in the state space model (such as damping coefficient, reactance, rotor impedance, etc.). The specific approach is:

[0063] An adaptive weight allocation mechanism is introduced in the optimization process to gradually adjust different weights according to the error of the identified target parameters to focus on the main response of the wind turbine under different working conditions. Through multiple iterations, the objective function converges within the preset error range.

[0064] (5) Dynamic fitting of simulation and real data: Based on the identification model, the same input conditions as the real wind speed change are applied in the simulation environment to obtain the dynamic output sequence of the simulation model. Then, the simulation results are fitted with the measured data, and the mean square error (MSE) and mean absolute error (MAE) of each set of data are calculated. Based on this, the model parameters are further adjusted to ensure that the error between the simulation data and the measured data is gradually reduced to within the set range.

[0065] (6) Error feedback and identification accuracy improvement: The identification error is compared with the target accuracy. For the interval with large error, the data is resampled and the model parameters are updated. The process of steps (3) to (5) is repeated for iteration. The number of iterations is controlled according to the score of the silhouette coefficient SC until the model reaches the optimal classification and identification accuracy.

[0066] (7) Verification of identification results and model update: When the identification error meets the preset requirements, the integrity and robustness test of the final equivalent site model is carried out. The robustness of the model under different wind speed conditions is verified through multiple tests to ensure that the model can accurately predict the dynamic response of the wind farm in practical applications. This model is used as the equivalent site model of the wind farm.

[0067] Example 2

[0068] According to an embodiment of the present invention, a doubly-fed wind turbine modeling system is provided, comprising:

[0069] 1. Data acquisition module

[0070] Function: Responsible for selecting and collecting key dynamic response data of wind farms, such as wind speed, pitch angle, voltage, current, and power.

[0071] process:

[0072] Wind speed and related operating data are collected in real time from multiple wind turbines in a wind farm through a sensor network.

[0073] Data preprocessing, including data cleaning, outlier removal, and denoising, is performed to ensure data accuracy and consistency.

[0074] The data is stored in a database, providing high-quality original data sources for clustering and identification.

[0075] 2. Clustering Modeling Module

[0076] Function: Cluster wind farm data based on the improved K-means algorithm and build an equivalent station model.

[0077] process:

[0078] Wind speed is selected as the clustering indicator, and data clustering is performed based on the distribution of wind speed.

[0079] The improved K-means algorithm is used to select the processed data as the initial cluster center point, and the silhouette coefficient (SC) is used as the scoring standard to optimize the clustering effect and ensure the accuracy of the clustering results.

[0080] Wind turbines with similar wind speeds are grouped into the same equivalent cluster, and each cluster is equivalent to a large wind turbine. The total capacity is calculated and the corresponding equivalent site model is constructed to provide a basic model for the subsequent identification module.

[0081] 3. Identification and optimization module

[0082] Function: Identify the equivalent station model to ensure the accuracy and dynamic response characteristics of the model.

[0083] process:

[0084] State space generation and parameter initialization: Generate a state space model based on the clustering model, divide the response interval under different wind speed conditions, and initialize the identification error tolerance and weight coefficient.

[0085] Multi-level parameter optimization: Genetic algorithm is used to perform multi-level optimization on key parameters in the state space model, and an adaptive weight allocation mechanism is introduced to gradually converge the error to within the preset tolerance.

[0086] Simulation verification and iterative optimization: Place the identification model in a simulation environment, fit it with real data, calculate the mean square error (MSE) and mean absolute error (MAE), and further optimize the parameters based on error feedback to ensure the accuracy of the model.

[0087] Model verification and testing: Conduct integrity and robustness tests on models that meet the identification error conditions, verify their robustness under different wind speed conditions, and ultimately determine the accuracy of the equivalent site model to ensure that it can be actually applied to the dynamic response prediction of wind farms.

[0088] Through the mutual cooperation of the above modules, the system can realize accurate modeling, clustering and identification optimization of wind farms, ensure that the response of the model under dynamic wind speed conditions is consistent with the measured data, and provide a reliable basis for the operation and control of wind farms.

[0089] Example 3

[0090] An embodiment of the present invention is different from the first two embodiments in that:

[0091] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of 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, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0092] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, which can be embodied in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor or other system that can fetch instructions from an instruction execution system, apparatus or device and execute instructions), or used in conjunction with these instruction execution systems, apparatuses or devices. For the purpose of this specification,

[0093] A "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

[0094] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0095] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0096] Example 4

[0097] See also Figure 2-Figure 5 An embodiment of the present invention provides a doubly-fed wind turbine modeling method. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0098] This paper uses the wind speed measured at a wind farm in the southwest as the original data for clustering. The obtained wind speed is shown in Table 1:

[0099] Table 1 Wind speed in wind farm

[0100]

[0101]

[0102] Note: No fan is installed at *

[0103] The calculated silhouette coefficient scores are shown in Table 2:

[0104] Table 2 Silhouette coefficient score table

[0105]

[0106] In summary, when the number of types is 3, the silhouette coefficient is closest to 1, and the clustering effect is better. The wind farms after clustering are as follows: Figure 2 .

[0107] Then, the A, B, and C equivalent wind farms are identified respectively, that is, the voltage, active power, reactive power, and active current data of the A, B, and C wind farms after disturbance are compared between the actual measured scene and the equivalent modeled scene. Figure 3 , 4 , 5. It can be seen that the wind farm equivalent modeling proposed in this paper using wind speed as the clustering index can well fit the measured data of the actual wind farm, and its error is within the allowable range and can accurately reflect the actual operation of the wind farm. Therefore, the clustering method proposed in this paper verifies that this strategy is applicable to the equivalent modeling of large-scale wind farms.

Claims

1. A doubly-fed wind turbine modeling method, characterized in that: include: Select clustering indicators and collect data; The improved K-means algorithm is used to cluster the wind farm data and construct an equivalent station model; The obtained equivalent station model is identified.

2. The doubly-fed wind turbine modeling method according to claim 1, characterized in that: The clustering index selects the wind speed of the wind farm. According to different wind speeds, the pitch angle is increased or decreased to reduce the power output of the wind turbine to keep it at the rated power level or capture more kinetic energy to facilitate the start-up of the wind turbine.

3. The doubly-fed wind turbine modeling method according to claim 2, characterized in that: Data processing is performed before and during clustering. Assuming that the wind farm consists of N×M wind turbines of the same model, after the wind farm is divided into P groups, each group with similar wind speed is classified as a large wind turbine, where the capacity of the equivalent wind turbine is the product of the number of wind turbines in the group and the capacity of a single wind turbine. The processing formula for the samples is expressed as: Among them, x′ i For processed data, is the data average, is the standard deviation of the data.

4. The doubly-fed wind turbine modeling method according to claim 3, characterized in that: The constructing of the equivalent station model includes inputting the wind speed of the wind turbine measured at the wind farm; The silhouette coefficient SC is used as the K-means scoring index, and the improved data is used as the initial cluster center point; The K-means algorithm is applied iteratively to calculate SC, and the number of wind farm classifications is selected according to the degree to which its value is close to 1; ultimately, n classified and equivalent wind turbines are obtained.

5. The doubly-fed wind turbine modeling method according to claim 4, characterized in that: The improved K-means algorithm includes selecting a sample from the processed data as the center of the initial cluster, calculating the shortest distance D′(x i ), calculate the probability P of each sample being selected as the next cluster center, expressed as: The next cluster center is selected according to the roulette method until K cluster centers that meet the requirements are selected.

6. The doubly-fed wind turbine modeling method according to claim 5, characterized in that: For the evaluation index of K-means, the concept of silhouette coefficient SC is used to quantify the consistency of each data point in the cluster and the separation between clusters. For each point m in the data set, the calculation formula of silhouette coefficient s(m) is expressed as:

7. The doubly-fed wind turbine modeling method according to claim 6, characterized in that: The equivalent station model identification includes, after equivalent modeling, data preprocessing, denoising the wind speed and pitch angle time series data, and eliminating abnormal values. At the same time, the parameter set is initialized, the error tolerance of the identification is set, and the initial weight coefficient set of the target parameters in the identification is defined; Select the key dynamic response parameters of the wind farm, collect the dynamic response data of the above parameters in time series in real time for equivalent groups in different wind speed ranges, and link the wind speed data with the pitch angle adjustment of the doubly fed wind turbine to ensure the response consistency of the model under different wind speed conditions and provide diversified sampling input data for identification; Based on the clustering results and the characteristics of the wind turbine model, a state space model is generated for each type of equivalent cluster. This model contains the dynamic response equations of the electrical and mechanical structures of the wind turbines. The state space is divided into multiple sub-intervals so that the dynamic response characteristics of the equivalent cluster of wind farms can be mapped to different state intervals to ensure that the dynamic response of the wind farm under the condition of rapid wind speed changes can be accurately identified. Multi-level parameter optimization is performed on each equivalent cluster, and the genetic algorithm is used to adjust the key parameters in the state space model. An adaptive weight allocation mechanism is introduced in the optimization, and different weights are gradually adjusted according to the error of the identified target parameters to focus on the main response of the wind turbine under different working conditions. Through multiple iterations, the objective function converges within the preset error range. Based on the identification model, the same input conditions as the actual wind speed changes are applied in the simulation environment to obtain the dynamic output sequence of the simulation model. The simulation results are fitted with the measured data, and the mean square error (MSE) and mean absolute error (MAE) of each set of data are calculated. Based on this, the model parameters are further adjusted to ensure that the error between the simulation data and the measured data is gradually reduced to the set range. The identification error is compared with the target accuracy. For the interval with large error, the data is resampled and the model parameters are updated. The above process is repeated for iteration. The number of iterations is controlled according to the score of the silhouette coefficient SC until the model reaches the optimal classification and identification accuracy. When the identification error meets the preset requirements, the integrity and robustness of the final equivalent site model are tested. The robustness of the model under different wind speed conditions is verified through multiple tests to ensure that the model can accurately predict the dynamic response of the wind farm in practical applications and use the model as the equivalent site model of the wind farm.

8. A system using the doubly-fed wind turbine modeling method according to any one of claims 1 to 7, characterized in that: include: Data acquisition module, clustering modeling module and identification optimization module; The data acquisition module is responsible for selecting and collecting wind farm data; The clustering modeling module is used to cluster wind farm data based on an improved K-means algorithm to construct an equivalent station model; The identification and optimization module is used to identify the equivalent station model to ensure the accuracy and dynamic response characteristics of the model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the doubly-fed wind turbine modeling method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the doubly-fed wind turbine modeling method according to any one of claims 1 to 7 are implemented.

Images