A method and system for identifying parameter data of a doubly-fed wind turbine

Abstract

The application provides a doubly-fed fan parameter data identification method and system, which comprises the following steps: obtaining measured data of a doubly-fed fan by setting faults with different voltage amplitudes; constructing a doubly-fed fan model according to a doubly-fed fan control system, wherein the control parameters in the model are identification targets; performing rough estimation on the control parameters by using a recursive least square algorithm, and updating the control parameters by using random training samples in the measured data one by one, and stopping the updating when the control parameter adjustment amount is lower than a first threshold value, and then switching to a genetic algorithm; performing fine estimation on the control parameters by using the genetic algorithm, verifying the accuracy of the control parameters by using samples in a verification sample library in the measured data after each iteration, and stopping the iteration when the root mean square error between the output of the doubly-fed fan model and the verification sample is lower than a second threshold value, and taking the corresponding control parameters as optimal results, and taking the optimal results as the estimated values of the fan parameters.

Description

Technical Field

[0001] This invention belongs to the field of wind turbine parameter data identification technology, and particularly relates to a method and system for identifying doubly fed wind turbine parameter data. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] As wind power penetration continues to increase, the dynamic processes of wind turbines under AC faults have a more significant impact on the system. Using time-domain simulation to assess system safety and stability is an effective method for verifying the grid-connected characteristics of wind turbines. However, due to insufficient parameter information provided by wind turbine manufacturers, setting parameters in the wind turbine simulation model is quite difficult, which limits the accurate simulation of the dynamic characteristics of wind turbines.

[0004] Existing methods for identifying control parameters of doubly-fed induction generator (DFIG) wind turbines mainly fall into two categories. One approach employs mathematical optimization methods, such as least squares and sensitivity analysis. These methods can calculate parameters based on data distribution from a mathematical optimization perspective, exhibiting good robustness against bad data. However, their individual calculation processes are lengthy, and the required sample size cannot be precisely determined, leading to reduced identification efficiency. The other approach uses heuristic search methods, such as particle swarm optimization and evolutionary computation. These methods achieve parameter identification and optimization by randomly trying the evaluation values ​​of various solutions within the solution space. Because mathematical optimization methods partially simplify the model, heuristic search can achieve better results than mathematical optimization. However, these methods are highly dependent on the initial population distribution and are prone to getting trapped in local optima.

[0005] Genetic algorithms are heuristic algorithms, typically requiring extensive sampling of the optimization decision space to find the optimal value. Due to the high dimensionality and wide distribution of wind turbine parameters, the solution space for parameter identification is large. In this case, choosing a large population size for the genetic algorithm leads to a sharp increase in computation; conversely, choosing a small population size can cause the optimization results to easily enter local optima, resulting in parameter identification errors. Therefore, directly using a genetic algorithm for parameter identification reduces efficiency. A method combining genetic algorithms with recursive least squares is proposed to address this problem. Summary of the Invention

[0006] To overcome the shortcomings of the prior art, the present invention provides a method for identifying doubly fed wind turbine parameter data, which can obtain the optimal result as an estimate of the wind turbine parameters.

[0007] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0008] In the first aspect, a method for identifying parameter data of a doubly-fed wind turbine is disclosed, including:

[0009] By setting faults with different voltage amplitudes, measured data of the doubly fed wind turbine were obtained.

[0010] A doubly fed fan model is constructed based on the doubly fed fan control system, where the control parameters in the doubly fed fan model are the identification targets.

[0011] The recursive least squares algorithm is used to roughly estimate the control parameters. The control parameters are updated successively using random training samples from the measured data. When the adjustment of the control parameters is lower than the first threshold, the update stops and the genetic algorithm step is started.

[0012] A genetic algorithm is used to estimate the control parameters in detail. After each iteration, the accuracy of the control parameters is verified by samples from the verification sample library in the measured data. The iteration stops when the root mean square error between the output of the doubly fed wind turbine model and the verification sample is lower than the second threshold. The corresponding control parameters are taken as the optimal results, and the optimal results are used as the estimated values ​​of the wind turbine parameters.

[0013] As a further technical solution, by setting faults with different voltage amplitudes, measured data of the doubly-fed wind turbine can be obtained. The specific process is as follows:

[0014] By setting faults with different voltage amplitudes, the dynamic process of the fan under fault conditions is simulated, the parameters of the doubly fed fan are recorded, and the required input and output measured data are estimated.

[0015] As a further technical solution, after obtaining the measured data of the doubly fed wind turbine, the following steps are also taken in sequence: preprocessing the measured data using Kalman filtering;

[0016] The measured data were standardized based on the fan capacity and operating voltage parameters.

[0017] The measured data is divided into two parts: a training sample library and a validation sample library.

[0018] As a further technical solution, a doubly fed wind turbine model is constructed based on the doubly fed wind turbine control system. The doubly fed wind turbine model includes an asynchronous motor model, a converter control model, and a fault voltage ride-through protection control model.

[0019] As a further technical solution, when using the recursive least squares algorithm to roughly estimate the control parameters, the input feature is the voltage feature during a fault, the output feature is the stator-side power of the doubly-fed induction generator (DFIG), and the parameter estimate is the parameter that the DFIG needs to identify.

[0020] As a further technical solution, when using a genetic algorithm to perform fine estimation of control parameters, according to the requirements for identifying doubly fed wind turbine parameters, the encoding rule of genetic information is to use the parameters to be identified as genes of chromosomes, and the wind turbine parameters are standardized to variables in the range of [0,1] as the feature values ​​of genes, with each chromosome representing a complete set of wind turbine parameters.

[0021] As a further technical solution, when generating the initial population of the genetic algorithm, the output of each step in the recursive least squares update process is used as the initial population.

[0022] Secondly, a doubly fed wind turbine parameter data identification system is disclosed, including:

[0023] The measured data acquisition module for the doubly-fed induction generator (DFIG) is configured to acquire measured data of the DFIG by setting faults with different voltage amplitudes.

[0024] The doubly fed wind turbine model building module is configured to: build a doubly fed wind turbine model based on the doubly fed wind turbine control system, wherein the control parameters in the model are the identification targets;

[0025] The recursive least squares parameter estimation module is configured to: use the recursive least squares algorithm to roughly estimate the control parameters, update the control parameters one by one using random training samples from the measured data, stop updating when the adjustment amount of the control parameters is lower than the first threshold, and switch to the genetic algorithm.

[0026] The genetic algorithm parameter optimization module is configured to: use a genetic algorithm to perform fine estimation of control parameters; after each iteration, use samples from the verification sample library in the measured data to verify the accuracy of the control parameters; when the root mean square error between the output of the doubly fed wind turbine model and the verification sample is lower than the second threshold, the iteration stops, and the corresponding control parameters are taken as the optimal result, and the optimal result is taken as the estimated value of the wind turbine parameters.

[0027] The above one or more technical solutions have the following beneficial effects:

[0028] The technical solution of this invention uses a recursive least squares algorithm to roughly estimate the parameters. When the parameter adjustment is lower than a threshold, the update stops and a genetic algorithm is used. The genetic algorithm is then used to finely estimate the parameters. The result of each update of the recursive least squares method is used as the initial population. A verification sample library is used to verify the accuracy of the parameters. When the root mean square error between the parameter model output and the verification sample is lower than a threshold, the iteration stops. The optimal result is used as the estimated value of the wind turbine parameters, thus avoiding the problem of getting trapped in local optima due to using a single method to obtain control parameters.

[0029] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0030] The accompanying drawings, which form part of this invention, 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 improper limitation of the invention.

[0031] Figure 1 This is a structural diagram of a doubly fed fan according to an embodiment of the present invention;

[0032] Figure 2 This is a flowchart of the parameter identification method according to an embodiment of the present invention;

[0033] Figure 3 This is a simulation interface diagram of an embodiment of the present invention. Detailed Implementation

[0034] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0035] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0036] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0037] Example 1

[0038] This embodiment discloses a method for identifying parameter data of a doubly fed wind turbine. (See attached document) Figure 2 As shown, it includes:

[0039] S1. Set different voltage faults and obtain the measured data of the doubly fed wind turbine. The measured data specifically includes the active current, reactive current, active power, reactive power, wind turbine speed, and DC bus voltage of the wind turbine. Perform Kalman filtering on the wind turbine data and standardize the wind turbine data. Divide the data into two parts: training sample library and verification sample library.

[0040] The specific process for obtaining measured data of a doubly-fed induction generator (DFIG) is as follows: set the fault voltage of the fan to 0.05, 0.2, 0.35, 0.5, 0.75, 0.9, 1.2, 1.25, and 1.3 times the rated value, and measure the active current, reactive current, active power, and reactive power of the fan.

[0041] Since the turbine parameters determine the response characteristics of its dynamic process, it is necessary to estimate the turbine parameters through these response characteristics. Because the control parameters are typically functions of the turbine terminal voltage, fault tests are required for various voltages to fit the control curve.

[0042] The more samples collected using the above method, the closer the fitting results for the control characteristics and other parameters will be to the actual values.

[0043] S2. Construct a doubly fed wind turbine model based on the doubly fed wind turbine control system. The control parameters in the model are the identification targets. The input of the model is the port voltage, and the output is the active and reactive power.

[0044] S3 uses a recursive least squares algorithm to roughly estimate the parameters, and updates the parameters with random training samples one by one. When the parameter adjustment is lower than the threshold, the update stops and the genetic algorithm is switched.

[0045] S4 uses a genetic algorithm to perform fine estimation of parameters. After each iteration, the accuracy of the parameters is verified using samples in the verification sample library. The iteration stops when the root mean square error between the parameter model output and the verification sample is lower than the threshold, and the optimal result is used as the estimated value of the wind turbine parameters.

[0046] In step S1, hardware-in-the-loop (HIL) simulation techniques are used to simulate the dynamic process of the wind turbine under fault conditions by setting faults with different voltage amplitudes, and the measured input and output data required for parameter estimation of the doubly-fed induction generator (DFIG) are recorded. To reduce the impact of noise and equipment measurement errors on parameter estimation, Kalman filtering is used to preprocess the measured data. The measured data is also standardized according to parameters such as wind turbine capacity and operating voltage. To verify that the predicted parameters meet the accuracy requirements, the measured data is divided into a training sample library and a validation sample library.

[0047] In step S2, for the doubly-fed induction generator (DFIG) with the required parameters, a turbine model is constructed considering its existing control loops, where the control parameters in the model are the identification targets. A typical DFIG model is constructed as follows:

[0048] The wind turbine model mainly includes an asynchronous motor model, a converter control model, and a fault voltage ride-through protection control model. Wind turbine parameter identification primarily aims to construct the dynamic response characteristics of the wind turbine under fault conditions. These dynamic response characteristics, i.e., the active and reactive power output during a fault, are determined by the fault voltage ride-through protection control, while the transient process is determined by the motor and converter control models. Using this model can effectively reflect the dynamic response characteristics of the wind turbine during fault processes, constructing a more accurate equivalent model of the wind turbine. (Further details about the wind turbine model are omitted as they are not relevant to the main text.) Figure 1 As shown.

[0049] The expression for the asynchronous motor model is:

[0050]

[0051] Where u, i, and ψ are the voltage, current, and flux linkage phasors of the wind turbine, respectively, and the subscripts s and r represent the stator and rotor, respectively, ωs and ω sl These represent the stator and slip angular velocities, respectively. The subscripts d and q represent the d-axis phasor and q-axis phasor, respectively. P s and Q s These are the stator active and reactive power, respectively.

[0052] The converter control model is as follows:

[0053] When faults such as AC short circuits cause the grid connection voltage of wind power to deviate significantly from the rated value, current limiting control is typically implemented to prevent damage to the converter within the wind turbine due to overvoltage and overcurrent. Simultaneously, to meet the requirement of providing a certain amount of reactive power support to the system during wind power fault ride-through, the low-voltage ride-through (LVRT) control strategy is as follows:

[0054] The input to the current limiting circuit is

[0055]

[0056]

[0057] Among them, I Ac and I Re P represents the active and reactive current inputs for the current limiting circuit, respectively. Tref and Q Tref These are the active and reactive power references, respectively, V Tref For voltage reference, V T For voltage, K vc This is the reactive current regulation coefficient.

[0058] The output of the current limiting circuit is determined by the terminal voltage amplitude. When the voltage is higher than the low voltage threshold, the active current priority control mode is used; otherwise, the reactive current priority control mode is used. Specifically, the active current priority control mode is as follows:

[0059]

[0060] Among them, I max I is the current limiting value. rd,ref and I Qrq,ref These are the reference values ​​for active current and reactive current, respectively.

[0061] Reactive current priority control mode is

[0062]

[0063] The current reference value is converted into the converter output voltage via a PI circuit.

[0064]

[0065] Among them, K p and Ki These are the control parameters for the PI controller.

[0066] As can be seen from the above model, the input to the doubly fed wind turbine model is the stator voltage U. sd and U sq The output is active power P. s and reactive power Q s The parameter to be identified is [K] p ,K i ,K vc ,I max V Tref ].

[0067] In practice, the fan model can be modified based on the additional control module inside the doubly fed fan to increase the number of identification parameters.

[0068] In step S3, during the fault ride-through process of the doubly-fed induction generator (DFIG), its output active and reactive currents are typically linearly related to the fault voltage. The recursive least squares method can quickly fit this linear relationship and update the results based on each new data collected. The specific implementation process of the recursive least squares method is as follows:

[0069] Recursive least squares is an iterative method for parameter estimation, and its mathematical expression for parameter update is:

[0070]

[0071] in, Let K be the parameter estimate for the k-th update. k Let X be the gain matrix. k Let ε be the input feature of the k-th sample. k Let be the prediction error for the k-th sample.

[0072] The gain matrix and prediction error are calculated as follows:

[0073]

[0074]

[0075] Among them, y k Let be the output feature of the k-th sample.

[0076] Since matrix inversion is involved in calculating the gain matrix, the matrix inverse theorem is used to simplify the gain matrix.

[0077]

[0078] In this example, the input characteristic is the voltage characteristic during a fault, the output characteristic is the stator-side power of the doubly-fed induction generator (DFIG), and the parameter estimates are the parameters that the DFIG needs to identify.

[0079] In step S3, to address the nonlinear factors introduced by the converter's maximum current limit and fault ride-through trigger voltage, a genetic algorithm is used for heuristic global optimization. This improves the accuracy of the recursive least squares method results and correctly considers the nonlinear factors in the model. The specific implementation process of the genetic algorithm is as follows:

[0080] 3-1) Encoding

[0081] According to the requirements for identifying doubly fed wind turbine parameters, the encoding rule for genetic information is to use the parameters to be identified as genes of chromosomes, and the wind turbine parameters are standardized to variables in the range of [0,1] as the characteristic values ​​of genes, with each chromosome representing a complete set of wind turbine parameters.

[0082] 3-2) Initial Population Generation

[0083] To improve the evaluation value and distribution characteristics of the initial population, the output results of each update process using the recursive least squares method are used as the initial population. This ensures that high-quality wind turbine parameters are included, improving the evaluation value of the population, while also guaranteeing a wider distribution of wind turbine parameters and avoiding the problem of local optima.

[0084] The input to the recursive least squares method is the measured electrical quantities during wind turbine fault ride-through, such as current and power; the output is the parameters in the wind turbine model, such as voltage reference values ​​and reactive current regulation coefficients. The recursive least squares method updates the parameters iteratively, meaning that the output wind turbine parameters are updated once for each input sample. Therefore, when a large amount of measured data is used as input for fitting using the recursive least squares method, an equal number of wind turbine parameter results will be generated. These results gradually approach the actual values ​​of the wind turbine parameters.

[0085] 3-3) Fitness Assessment

[0086] Wind turbine parameter identification requires the use of measured data samples. One hundred samples are randomly selected from the training sample library as the training set. The fitness of each individual is evaluated as the root mean square error of that parameter set across all samples in the training set.

[0087]

[0088] Where, x j and y j Let F be the input and output features of the j-th sample, respectively, ξ be the identification parameter, and F be the input and output features of the j-th sample. i Let i be the fitness of individual i.

[0089] 3-4) Selection, crossover, and mutation

[0090] Crossover is an operation that randomly selects two individuals and swaps one of their genes, thereby enriching the diversity of gene combinations in the population. This operation does not add new genes. Its expression is:

[0091]

[0092] Where ξ(t+1) represents an individual in the (t+1)th generation of the population, ξ k For the k-th gene in an individual, Let K be the k-th gene in individual a, where K is the gene position of the crossover operation.

[0093] Mutation is a random modification to a gene in an individual, thereby enriching the genetic diversity of the population. This operation adds new genes to the population. Its expression is:

[0094]

[0095] Where random[0,1] is a random number between 0 and 1, and K is the position of the mutation operation.

[0096] The selection operation selects superior individuals based on their fitness with a certain probability to construct the next generation of the population. The relationship between the probability of an individual being selected and its fitness satisfies...

[0097]

[0098] Where p i Let M be the probability that individual i is selected for the next generation, and M be the population size.

[0099] Finally, after three operations—crossover, mutation, and selection—a new population is generated.

[0100] 3-5) Termination Conditions

[0101] One hundred samples were randomly selected from the validation sample library as the validation set. A wind turbine model was constructed using the optimal parameters. The model was simulated under the operating conditions corresponding to the voltage input of the test samples. The difference between the output power of the wind turbine model and the output of the test samples was used as the error to evaluate the fit of the wind turbine characteristics under these parameters. If the root mean square error was less than 1%, the parameter update ended, and the individual with the highest fitness was selected as the parameter identification result; otherwise, a new round of genetic algorithm iteration began. The individual information in the genetic algorithm is the wind turbine parameter information, specifically including current limit value, voltage reference value, reactive current regulation coefficient, and wind turbine generator stator and rotor impedance parameters.

[0102] The simulation of the model under optimal parameters can be performed using the PSCAD electromagnetic transient simulation program. In the constructed wind turbine model (i.e., the model from step 2), the optimal parameters are set, and the simulation interface is shown below. Figure 3 As shown, the power grid fault is set according to the input voltage characteristics of the detection sample. The parameter estimation error is obtained by comparing the simulation results with the output power characteristics of the detection sample.

[0103] Based on the above recursive least squares coarse estimation and genetic algorithm fine optimization, the key control parameters of the doubly fed wind turbine rotor side control can be identified, and the simulation error of the corresponding parameters can be ensured to be less than 1%.

[0104] Example 2

[0105] The purpose of this embodiment is to provide 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 program to implement the steps of the above-described method.

[0106] Example 3

[0107] The purpose of this embodiment is to provide a computer-readable storage medium.

[0108] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the above method.

[0109] Example 4

[0110] The purpose of this embodiment is to provide a doubly fed wind turbine parameter data identification system, including:

[0111] The measured data acquisition module for the doubly-fed induction generator (DFIG) is configured to acquire measured data of the DFIG by setting faults with different voltage amplitudes.

[0112] The doubly fed wind turbine model building module is configured to: build a doubly fed wind turbine model based on the doubly fed wind turbine control system, wherein the control parameters in the model are the identification targets;

[0113] The recursive least squares parameter estimation module is configured to: use the recursive least squares algorithm to roughly estimate the control parameters, update the control parameters one by one using random training samples from the measured data, stop updating when the adjustment amount of the control parameters is lower than the first threshold, and switch to the genetic algorithm.

[0114] The genetic algorithm parameter optimization module is configured to: use a genetic algorithm to perform fine estimation of control parameters; after each iteration, use samples from the verification sample library in the measured data to verify the accuracy of the control parameters; when the root mean square error between the output of the doubly fed wind turbine model and the verification sample is lower than the second threshold, the iteration stops, and the corresponding control parameters are taken as the optimal result, and the optimal result is taken as the estimated value of the wind turbine parameters.

[0115] The steps and methods involved in the apparatuses of Embodiments 2, 3, and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0116] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0117] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for identifying parameter data of a doubly-fed induction generator (DFIG), characterized in that, include: By setting faults with different voltage amplitudes, measured data of the doubly fed wind turbine were obtained. A doubly fed fan model is constructed based on the doubly fed fan control system, wherein the control parameters in the doubly fed fan model are the identification targets; The recursive least squares algorithm is used to roughly estimate the control parameters. The input feature is the voltage feature during the fault, the output feature is the stator power of the doubly fed wind turbine, and the parameter estimate is the parameter that the doubly fed wind turbine needs to identify. The control parameters are updated successively using random training samples from the measured data. When the adjustment of the control parameters is lower than the first threshold, the update stops and the genetic algorithm is switched. A genetic algorithm is used to estimate the control parameters in detail. After each iteration, the accuracy of the control parameters is verified by samples from the verification sample library in the measured data. The iteration stops when the root mean square error between the output of the doubly fed wind turbine model and the verification sample is lower than the second threshold. The corresponding control parameters are taken as the optimal result, and the optimal result is used as the estimated value of the wind turbine parameters. When using a genetic algorithm to perform fine estimation of control parameters, according to the requirements for identifying doubly fed wind turbine parameters, the encoding rule of genetic information is to use the parameters to be identified as genes of chromosomes, and the wind turbine parameters are standardized to variables in the range of [0, 1] as the feature values ​​of genes, with each chromosome representing a complete set of wind turbine parameters.

2. The method for identifying parameter data of a doubly-fed wind turbine as described in claim 1, characterized in that, By setting faults with different voltage amplitudes, measured data of the doubly-fed induction generator (DFIG) was obtained. The specific process is as follows: By setting faults with different voltage amplitudes, the dynamic process of the fan under fault conditions is simulated, the parameters of the doubly fed fan are recorded, and the required input and output measured data are estimated.

3. The method for identifying parameter data of a doubly-fed wind turbine as described in claim 1, characterized in that, After obtaining the measured data of the doubly fed wind turbine, the process also includes: preprocessing the measured data using Kalman filtering; The measured data were standardized based on the fan capacity and operating voltage parameters. The measured data is divided into two parts: a training sample library and a validation sample library.

4. The method for identifying parameter data of a doubly fed wind turbine as described in claim 1, characterized in that, A doubly fed wind turbine model is constructed based on the doubly fed wind turbine control system. The doubly fed wind turbine model includes an asynchronous motor model, a converter control model, and a fault voltage ride-through protection control model.

5. The method for identifying parameter data of a doubly-fed wind turbine as described in claim 1, characterized in that, When generating the initial population for the genetic algorithm, the output of each step in the recursive least squares update process is used as the initial population.

6. A doubly-fed induction generator (DFIG) wind turbine parameter data identification system, characterized in that, include: The measured data acquisition module for the doubly-fed induction generator (DFIG) is configured to acquire measured data of the DFIG by setting faults with different voltage amplitudes. The doubly fed wind turbine model building module is configured to: build a doubly fed wind turbine model based on the doubly fed wind turbine control system, wherein the control parameters in the model are the identification targets; The recursive least squares parameter estimation module is configured to: use the recursive least squares algorithm to roughly estimate the control parameters, the input feature is the voltage feature during the fault, the output feature is the stator side power of the doubly fed wind turbine, and the parameter estimate is the parameter that the doubly fed wind turbine needs to identify. The control parameters are updated successively using random training samples from the measured data. When the adjustment of the control parameters is lower than the first threshold, the update stops and the genetic algorithm is switched. The genetic algorithm parameter optimization module is configured to: use a genetic algorithm to perform fine estimation of control parameters; after each iteration, use samples from the verification sample library in the measured data to verify the accuracy of the control parameters; when the root mean square error between the output of the doubly fed wind turbine model and the verification sample is lower than the second threshold, the iteration stops, and the corresponding control parameters are taken as the optimal result, and the optimal result is taken as the estimated value of the wind turbine parameters. When using a genetic algorithm to perform fine estimation of control parameters, according to the requirements for identifying doubly fed wind turbine parameters, the encoding rule of genetic information is to use the parameters to be identified as genes of chromosomes, and the wind turbine parameters are standardized to variables in the range of [0, 1] as the feature values ​​of genes, with each chromosome representing a complete set of wind turbine parameters.

7. 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 program, it implements the steps of the method described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it performs the steps of the method described in any one of claims 1-5.

Images