A method and system for wind turbine generator parameter identification based on deep reinforcement learning

Abstract

This invention discloses a method and system for wind turbine generator parameter identification based on deep reinforcement learning. The method includes: acquiring an initial dataset; extracting mechanical, electrical, and control features from the initial data; establishing corresponding mechanical subsystem models, electrical subsystem models, and control subsystem models based on the mechanical, electrical, and control features; analyzing the dynamic coupling relationship between the mechanical, electrical, and control subsystem models to obtain a coupling matrix; coupling the mechanical, electrical, and control subsystem models based on the coupling matrix to obtain a comprehensive mathematical model; and identifying wind turbine generator parameters based on the comprehensive mathematical model to obtain the identification result. This invention achieves accurate modeling and parameter identification of wind turbine generators, providing important support for optimizing generator performance and predictive maintenance.

Description

Technical Field

[0001] This invention belongs to the field of data analysis technology, and in particular relates to a method and system for identifying wind turbine generator parameters based on deep reinforcement learning. Background Technology

[0002] In the process of wind turbine parameter identification, establishing an accurate and comprehensive mathematical model of the wind turbine is a key technical problem. A wind turbine is a complex system, comprising multiple subsystems such as mechanical, electrical, and control systems, each with its specific parameters and dynamic characteristics. These subsystems are interconnected and influence each other, making the mathematical model of the entire system extremely complex. Furthermore, the wind turbine's operating environment is highly variable; changes in external factors such as wind speed and direction affect the system's parameters and performance, further increasing the difficulty of modeling. Establishing an accurate and robust mathematical model of the wind turbine, considering the dynamic characteristics and inter-system coupling relationships, and capable of adapting to changes in the external environment, is a pressing technical challenge that needs to be addressed. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes a method and system for identifying wind turbine generator parameters based on deep reinforcement learning. This method can establish an accurate and robust mathematical model of the wind turbine generator, and the model can adapt to changes in the external environment.

[0004] This invention provides a method for identifying wind turbine generator parameters based on deep reinforcement learning, comprising:

[0005] Obtain the initial dataset;

[0006] Extract the mechanical, electrical, and control features of the initial data;

[0007] Based on the mechanical, electrical, and control characteristics, establish corresponding mechanical subsystem models, electrical subsystem models, and control subsystem models;

[0008] Analyze the dynamic coupling relationship between the mechanical subsystem model, electrical subsystem model, and control subsystem model, and obtain the coupling matrix;

[0009] Based on the coupling matrix, the mechanical subsystem model, electrical subsystem model, and control subsystem model are coupled to obtain a comprehensive mathematical model;

[0010] The parameters of the wind turbine are identified based on the comprehensive mathematical model, and the identification results are obtained.

[0011] Optionally, obtaining the initial dataset includes:

[0012] Obtain historical operating data of wind turbines;

[0013] The historical operating data is preprocessed, and key features related to the operating status and performance of the wind turbine are extracted from the preprocessed data.

[0014] The key features are fused to obtain the initial dataset.

[0015] Optionally, based on the mechanical, electrical, and control characteristics, establishing corresponding mechanical subsystem models, electrical subsystem models, and control subsystem models includes:

[0016] A system identification method based on a physical model is adopted to extract mechanical vibration frequency and damping coefficient from the initial dataset, and to estimate mechanical parameters using the least squares method to construct a mechanical subsystem model.

[0017] Using data-driven identification technology, generator speed, output voltage, and current are extracted from the initial dataset. A neural network algorithm is then used to fit the generator characteristics to obtain an electrical subsystem model.

[0018] A state-space based modeling method is adopted to extract the controller input and output signals from the initial dataset, and the Kalman filter algorithm is used to estimate the system state to establish a control subsystem model.

[0019] Optionally, constructing a mechanical subsystem model includes:

[0020] The least squares method is used to estimate the vibration frequency and damping coefficient of the mechanical system to obtain estimated values ​​of the mechanical parameters. If there is a large deviation between the estimated values ​​of the mechanical parameters and the actual system, the mechanical parameters are continuously adjusted through an iterative optimization algorithm to obtain the best estimated values.

[0021] Based on the optimal estimate, construct the dynamic equations of the mass-spring-damped system;

[0022] Based on the dynamic equations of the mass-spring-damped system, a physical model of the mechanical subsystem is established.

[0023] Optionally, obtaining the electrical subsystem model includes:

[0024] Build a neural network model and optimize the hyperparameters of the neural network using grid search;

[0025] Using training set data, a backpropagation algorithm is employed to train the neural network model, continuously updating the model parameters so that the model can fit the voltage-current-speed relationship of the generator, thereby obtaining the electrical subsystem model. The training set consists of generator speed, output voltage, and current.

[0026] Optionally, establishing a control subsystem model includes:

[0027] The state-space modeling method is used to construct the state equations and output equations of the control subsystem;

[0028] Based on the aforementioned state equation and output equation, the Kalman filter algorithm is used to estimate the system state of the control subsystem.

[0029] Based on the system state, determine whether the control characteristics of the control subsystem meet the preset conditions. If they do, determine the currently established control subsystem model.

[0030] If the estimated system state does not meet the preset conditions, adjust the parameters of the state equation and output equation, and re-estimate the system state until the preset conditions are met.

[0031] Optionally, the dynamic coupling relationship between the mechanical subsystem model, electrical subsystem model, and control subsystem model is analyzed to obtain the coupling matrix, including:

[0032] Based on the mechanical subsystem model, electrical subsystem model, and control subsystem model, a system identification method is used to obtain the dynamic characteristic parameters of each subsystem and the dynamic coupling relationship parameters between subsystems.

[0033] Based on the obtained dynamic characteristic parameters and dynamic coupling relationship parameters, a state space model is constructed, and a mapping relationship matrix between state variables and subsystem input and output variables is established.

[0034] Cross-correlation analysis is used to calculate the correlation coefficient matrix between variables of each subsystem, quantify the degree of mutual influence between subsystems, and obtain the coupling matrix.

[0035] Optionally, obtaining the comprehensive mathematical model includes:

[0036] The coupling matrix is ​​combined with the models of each subsystem. Through matrix transformation and equation merging, the state equations and output equations of each subsystem are integrated with the coupling matrix to form the comprehensive state equations and output equations of the wind turbine generator set, i.e., the comprehensive mathematical model.

[0037] Optionally, wind turbine parameters are identified based on the comprehensive mathematical model, and the identification results include:

[0038] The Kalman filter algorithm is used to filter the noise in the comprehensive mathematical model.

[0039] The recursive least squares method is used to identify parameters in the filtered comprehensive mathematical model and to estimate the values ​​of unknown parameters in the model online.

[0040] Based on the parameter identification results, the mechanical parameters, electrical parameters, and control parameters in the integrated mathematical model are updated to obtain the optimized integrated mathematical model.

[0041] The optimized integrated mathematical model is applied to the real-time control system of the wind turbine generator set to identify the wind turbine generator parameters and obtain the identification results.

[0042] The present invention also provides a wind turbine generator parameter identification system based on deep reinforcement learning, comprising: a data acquisition module, a mechanical characteristic identification module, an electrical characteristic identification module, a control characteristic identification module, a coupling analysis module, a model integration module, and a parameter identification module;

[0043] The data acquisition module is used to acquire historical operating data of the wind turbine generator set, including mechanical vibration, electrical output, control signals, and environmental parameters such as wind speed and wind direction, to form an initial dataset.

[0044] The mechanical characteristic identification module is used to extract mechanical vibration frequency and damping coefficient from the initial dataset and establish a mechanical subsystem model based on a physical model-based system identification method for mechanical characteristics.

[0045] The electrical characteristic identification module is used to extract generator speed, output voltage, and current parameters from the initial dataset using data-driven identification technology, and to establish an electrical subsystem model.

[0046] The control characteristic identification module is used to extract controller input and output signals from the initial dataset and establish a control subsystem model based on a state-space modeling method for the control characteristics.

[0047] The coupling analysis module is used to analyze the dynamic coupling relationship between the subsystems based on the mechanical subsystem model, electrical subsystem model, and control subsystem model, and to establish a coupling matrix.

[0048] The model integration module is used to integrate the mechanical subsystem model, electrical subsystem model, control subsystem model and coupling matrix to form a comprehensive mathematical model of the wind turbine generator set;

[0049] The parameter identification module is used to identify the parameters of the wind turbine generator set based on a comprehensive mathematical model and obtain the identification result data.

[0050] Compared with the prior art, the present invention has the following advantages and technical effects:

[0051] This invention first acquires historical operating data of the wind turbine generator set, including mechanical, electrical, control, and environmental parameters, forming an initial dataset. Then, subsystem models are established for the mechanical, electrical, and control characteristics respectively, and system identification is performed using physical models, data-driven approaches, and state-space methods. Next, the dynamic coupling relationships between the subsystems are analyzed, and a coupling matrix is ​​established. Finally, the subsystem models and coupling matrices are integrated to form a comprehensive mathematical model of the wind turbine generator set, and parameter identification is performed based on this model. This invention, by comprehensively considering the mechanical, electrical, and control characteristics and their coupling relationships, achieves accurate modeling and parameter identification of the wind turbine generator set, providing important support for optimizing unit performance and predictive maintenance. Attached Figure Description

[0052] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0053] Figure 1 This is a flowchart of a wind turbine generator parameter identification method based on deep reinforcement learning according to an embodiment of the present invention. Detailed Implementation

[0054] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0055] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0056] This embodiment proposes a method for identifying wind turbine generator parameters based on deep reinforcement learning, such as... Figure 1 As shown, the specific steps include:

[0057] Obtain the initial dataset;

[0058] Extract the mechanical, electrical, and control characteristics of the initial data;

[0059] Based on the mechanical, electrical, and control characteristics, establish corresponding mechanical subsystem models, electrical subsystem models, and control subsystem models;

[0060] Analyze the dynamic coupling relationship between the mechanical subsystem model, the electrical subsystem model, and the control subsystem model, and obtain the coupling matrix;

[0061] Based on the coupling matrix, the mechanical subsystem model, electrical subsystem model, and control subsystem model are coupled to obtain a comprehensive mathematical model.

[0062] The parameters of the wind turbine are identified based on the comprehensive mathematical model, and the identification results are obtained.

[0063] Furthermore, obtaining the initial dataset includes:

[0064] Obtain historical operating data of wind turbines;

[0065] Historical operating data is preprocessed, and key features related to the operating status and performance of wind turbines are extracted from the preprocessed data;

[0066] Key features are fused to obtain an initial dataset.

[0067] Specifically, based on historical operating data of wind turbines, mechanical vibration, electrical output, control signals, and environmental parameters such as wind speed and direction are acquired. For the various heterogeneous data obtained, data cleaning and preprocessing techniques are employed to remove missing values, outliers, and other noise data, improving data quality and usability. Through feature engineering methods, key features related to the operating status and performance of wind turbines, including time-domain and frequency-domain features, are extracted from the preprocessed historical operating data to form standardized feature vectors. Data fusion technology is then used to fuse the extracted multi-source heterogeneous data, including mechanical vibration, electrical output, control signals, and environmental parameters, to construct a comprehensive wind turbine operating dataset.

[0068] Furthermore, based on the mechanical, electrical, and control characteristics, corresponding mechanical subsystem models, electrical subsystem models, and control subsystem models are established, including:

[0069] A system identification method based on a physical model is adopted to extract mechanical vibration frequency and damping coefficient from the initial dataset, and to estimate mechanical parameters using the least squares method to construct a mechanical subsystem model.

[0070] Using data-driven identification technology, generator speed, output voltage, and current are extracted from the initial dataset. A neural network algorithm is then used to fit the generator characteristics to obtain an electrical subsystem model.

[0071] A state-space based modeling method is adopted to extract the controller input and output signals from the initial dataset, and the Kalman filter algorithm is used to estimate the system state to establish a control subsystem model.

[0072] Furthermore, constructing the mechanical subsystem model includes:

[0073] The least squares method is used to estimate the vibration frequency and damping coefficient of the mechanical system to obtain estimated values ​​of the mechanical parameters. If there is a large deviation between the estimated values ​​of the mechanical parameters and the actual system, the mechanical parameters are continuously adjusted through an iterative optimization algorithm to obtain the best estimated values.

[0074] Based on the best estimate, construct the dynamic equations of the mass-spring-damped system;

[0075] Based on the dynamic equations of the mass-spring-damped system, a physical model of the mechanical subsystem is established.

[0076] Specifically, this involves improving the system's dynamic response performance and anti-interference capability; continuously accumulating operational data of the mechanical system; estimating key parameters such as vibration frequency and damping coefficient of the mechanical system using the least squares method based on the initial dataset to obtain estimated values ​​of mechanical parameters; constructing the dynamic equations of the mass-spring-damped system based on the estimated mechanical parameters to establish a physical model of the mechanical subsystem; if there is a large deviation between the estimated mechanical parameters and the actual system, continuously adjusting the mechanical parameters using iterative optimization algorithms such as gradient descent to make the dynamic model output fit the measured data and improve model accuracy; based on the constructed dynamic model of the mechanical subsystem, using system identification methods such as recursive least squares to estimate the state of the mechanical system's vibration frequency, damping coefficient, etc. in real time to achieve online identification of mechanical characteristics; inputting the identified mechanical characteristic parameters into the controller, using adaptive control algorithms such as model reference adaptive control to adjust the controller parameters in real time to achieve robust control of the mechanical system; and constructing a closed-loop control model of the mechanical system through the combination of mechanical characteristic identification and adaptive control, using incremental learning algorithms to update the mechanical model and controller parameters to achieve continuous optimization and adaptive control of mechanical characteristics, ensuring long-term stable operation of the system.

[0077] Furthermore, obtaining the electrical subsystem model includes:

[0078] Build a neural network model and optimize the hyperparameters of the neural network using grid search;

[0079] Using the training set data, the backpropagation algorithm is used to train the neural network model and continuously update the model parameters so that the model can fit the voltage-current-speed relationship of the generator and obtain the electrical subsystem model. The training set consists of generator speed, output voltage, and current.

[0080] Specifically, based on the initial dataset, data preprocessing techniques are employed to clean and standardize electrical parameter data such as generator speed, output voltage, and current, removing outliers and noise to obtain a high-quality electrical characteristic dataset. For the preprocessed electrical characteristic dataset, feature engineering methods are used to extract key features reflecting the generator's electrical characteristics, such as average speed, peak voltage, and active power, forming an optimized feature dataset. The optimized feature dataset is randomly divided into training and testing sets. The training set is used to train the neural network model, and the testing set is used to evaluate model performance. If the partition ratio is unreasonable, it is adjusted, and the dataset is re-partitioned. The network structure of the neural network model is designed according to the complexity of the electrical model, including the number of neurons in the input, hidden, and output layers, as well as the selection of activation functions. Hyperparameters of the neural network are optimized using methods such as grid search. Using the training set data, the backpropagation algorithm is employed to train the neural network model, continuously updating the model parameters to enable the model to fit the voltage-current-speed relationship of the generator. During training, the model's loss function and accuracy metrics are monitored. The performance of the trained neural network model is evaluated on the test set, calculating evaluation metrics such as mean squared error and coefficient of determination. If the model performance does not meet the requirements, the network structure and hyperparameters are adjusted, and the model is retrained. The trained neural network model is deployed to the online system to acquire real-time parameters such as generator speed, voltage, and current, which are then input into the model to predict the generator's electrical characteristics and obtain the real-time voltage-current-speed relationship, providing a basis for generator monitoring and optimized control.

[0081] Furthermore, establishing the control subsystem model includes:

[0082] The state-space modeling method is used to construct the state equations and output equations of the control subsystem;

[0083] Based on the state equation and output equation, the Kalman filter algorithm is used to estimate the system state of the control subsystem;

[0084] Based on the system state, determine whether the control characteristics of the control subsystem meet the preset conditions. If they do, determine the currently established control subsystem model.

[0085] If the estimated system state does not meet the preset conditions, adjust the parameters of the state equation and output equation, and re-estimate the system state until the preset conditions are met.

[0086] Specifically, for control characteristics, the input and output signals of the controller are extracted. Based on the extracted input and output signals, a state-space modeling method is used to construct the state equations and output equations of the control subsystem. For the established state and output equations, a Kalman filter algorithm is used to estimate the system state of the control subsystem. Based on the estimated system state, it is determined whether the control characteristics of the control subsystem meet preset conditions. If they do, the currently established control subsystem model is finalized. If the estimated system state does not meet the preset conditions, the parameters of the state and output equations are adjusted, and the system state is re-estimated until the preset conditions are met. The established control subsystem model is integrated with other subsystem models to construct a complete system model. Based on the constructed system model, machine learning algorithms such as support vector machines and neural networks are used to train and optimize the system, resulting in the final control system model.

[0087] Furthermore, the dynamic coupling relationships between the mechanical subsystem model, the electrical subsystem model, and the control subsystem model are analyzed, and the coupling matrix is ​​obtained, including:

[0088] Based on the mechanical subsystem model, electrical subsystem model, and control subsystem model, a system identification method is used to obtain the dynamic characteristic parameters of each subsystem and the dynamic coupling relationship parameters between subsystems.

[0089] Based on the obtained dynamic characteristic parameters and dynamic coupling relationship parameters, a state space model is constructed, and a mapping relationship matrix between state variables and subsystem input and output variables is established.

[0090] Cross-correlation analysis is used to calculate the correlation coefficient matrix between variables of each subsystem, quantify the degree of mutual influence between subsystems, and obtain the coupling matrix.

[0091] Specifically, based on the models of the mechanical, electrical, and control subsystems, a system identification method is used to obtain the dynamic characteristic parameters of each subsystem and the dynamic coupling relationship parameters between them. Based on these parameters, a state-space model is constructed, establishing a mapping matrix between state variables and subsystem input / output variables. Cross-correlation analysis is used to calculate the correlation coefficient matrix between subsystem variables, quantifying the degree of mutual influence and obtaining the coupling matrix. Based on the mutual influence coefficients in the coupling matrix, the coupling strength between subsystems is determined, identifying the primary and secondary coupling relationships. For subsystems with primary coupling relationships, a decoupling control algorithm is employed to decouple them and reduce mutual interference. For subsystems with secondary coupling relationships, an adaptive filtering algorithm is used to reduce the impact of coupling interference on control performance. The results of decoupling control and adaptive filtering are used as feedforward compensation and superimposed with the feedback controller output of the subsystems to form a composite control quantity, improving the system's dynamic performance and robustness.

[0092] Furthermore, obtaining the comprehensive mathematical model includes:

[0093] By combining the coupling matrix with the models of each subsystem, and through matrix transformation and equation merging, the state equations and output equations of each subsystem are integrated with the coupling matrix to form the comprehensive state equations and output equations of the wind turbine generator, i.e., the comprehensive mathematical model.

[0094] Specifically, mathematical models are established for the mechanical, electrical, and control subsystems of the wind turbine generator set. Using state-space representation, the state equations and output equations of each subsystem are obtained. The coupling relationships between subsystems are analyzed, and a coupling matrix reflecting their interactions is constructed. This coupling matrix is ​​then combined with the models of each subsystem. Through matrix transformation and equation merging, the state equations and output equations of each subsystem are integrated with the coupling matrix to form the comprehensive state equations and output equations of the wind turbine generator set. Dimensionality reduction is applied to the comprehensive state equations and output equations to remove redundant state variables, simplify the model structure, and improve computational efficiency. For the comprehensive mathematical model, a state observer and a state feedback controller are designed to achieve state estimation and feedback control of the wind turbine generator set. System identification algorithms, such as least squares or Kalman filtering, are used to identify and optimize the parameters of the comprehensive mathematical model, improving its accuracy and adaptability. Based on the comprehensive mathematical model, a simulation platform for the wind turbine generator set is developed to analyze and evaluate the system's dynamic characteristics and control performance, providing guidance for practical engineering applications.

[0095] Furthermore, based on the comprehensive mathematical model, wind turbine parameters are identified, and the identification results include:

[0096] The Kalman filter algorithm is used to filter noise in the comprehensive mathematical model;

[0097] The recursive least squares method is used to identify parameters in the filtered comprehensive mathematical model and to estimate the values ​​of unknown parameters in the model online.

[0098] Based on the parameter identification results, the mechanical parameters, electrical parameters, and control parameters in the integrated mathematical model are updated to obtain the optimized integrated mathematical model.

[0099] The optimized integrated mathematical model is applied to the real-time control system of the wind turbine generator set to identify the wind turbine generator parameters and obtain the identification results.

[0100] Specifically, a comprehensive mathematical model is established based on real-time operating data of the wind turbine generator set. This model includes unknown parameters such as mechanical, electrical, and control parameters. A Kalman filter algorithm is used to filter noise in the comprehensive mathematical model, improving its accuracy and robustness. Recursive least squares is then used to identify parameters in the filtered model, estimating the values ​​of unknown parameters online. Based on the parameter identification results, the mechanical, electrical, and control parameters in the comprehensive mathematical model are updated, resulting in an optimized model. This optimized model is then applied to the real-time control system of the wind turbine generator set for optimized control. The optimization effect and control performance improvement are evaluated by comparing the wind turbine generator set operating data before and after optimization. If the optimization effect does not meet the expected goals, the wind turbine generator set operating data is reacquired, and the comprehensive mathematical model and control system are iteratively optimized until the performance requirements are met.

[0101] This embodiment also provides a wind turbine generator parameter identification system based on deep reinforcement learning, including: a data acquisition module, a mechanical characteristic identification module, an electrical characteristic identification module, a control characteristic identification module, a coupling analysis module, a model integration module, and a parameter identification module;

[0102] The data acquisition module is used to acquire historical operating data of the wind turbine generator set, including mechanical vibration, electrical output, control signals, and environmental parameters such as wind speed and wind direction, to form an initial dataset;

[0103] The mechanical characteristic identification module is used to identify mechanical characteristics using a physical model-based system identification method, extracting mechanical vibration frequency and damping coefficient from the initial dataset, and establishing a mechanical subsystem model.

[0104] The electrical characteristic identification module is used to extract generator speed, output voltage, and current parameters from the initial dataset using data-driven identification technology, and to establish an electrical subsystem model.

[0105] The control characteristic identification module is used to extract the controller input and output signals from the initial dataset and establish a control subsystem model based on the state space modeling method for the control characteristics.

[0106] The coupling analysis module is used to analyze the dynamic coupling relationships between the mechanical subsystem model, electrical subsystem model, and control subsystem model, and to establish the coupling matrix.

[0107] The model integration module is used to integrate the mechanical subsystem model, electrical subsystem model, control subsystem model, and coupling matrix to form a comprehensive mathematical model of the wind turbine generator set.

[0108] The parameter identification module is used to identify the parameters of wind turbine generators based on a comprehensive mathematical model and obtain the identification result data.

[0109] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for identifying wind turbine generator parameters based on deep reinforcement learning, characterized in that, include: Obtain the initial dataset; Extract the mechanical, electrical, and control features from the initial dataset; Based on the mechanical, electrical, and control characteristics, establish corresponding mechanical subsystem models, electrical subsystem models, and control subsystem models; Based on the aforementioned mechanical, electrical, and control characteristics, the corresponding mechanical subsystem model, electrical subsystem model, and control subsystem model are established, including: A system identification method based on a physical model is adopted to extract mechanical vibration frequency and damping coefficient from the initial dataset, and to estimate mechanical parameters using the least squares method to construct a mechanical subsystem model. Using data-driven identification technology, generator speed, output voltage, and current are extracted from the initial dataset. A neural network algorithm is then used to fit the generator characteristics to obtain an electrical subsystem model. A state-space based modeling method is adopted to extract the controller input and output signals from the initial dataset, and the Kalman filter algorithm is used to estimate the system state to establish a control subsystem model. Analyze the dynamic coupling relationship between the mechanical subsystem model, electrical subsystem model, and control subsystem model, and obtain the coupling matrix; Analyze the dynamic coupling relationships between the mechanical subsystem model, electrical subsystem model, and control subsystem model, and obtain the coupling matrix, including: Based on the mechanical subsystem model, electrical subsystem model, and control subsystem model, a system identification method is used to obtain the dynamic characteristic parameters of each subsystem and the dynamic coupling relationship parameters between subsystems. Based on the obtained dynamic characteristic parameters and dynamic coupling relationship parameters, a state space model is constructed, and a mapping relationship matrix between state variables and subsystem input and output variables is established. The correlation coefficient matrix between variables of each subsystem is calculated by using cross-correlation analysis to quantify the degree of mutual influence between subsystems and obtain the coupling matrix. Based on the coupling matrix, the mechanical subsystem model, electrical subsystem model, and control subsystem model are coupled to obtain a comprehensive mathematical model; The parameters of the wind turbine are identified based on the comprehensive mathematical model, and the identification results are obtained.

2. The method for identifying wind turbine generator parameters based on deep reinforcement learning according to claim 1, characterized in that, Obtaining the initial dataset includes: Obtain historical operating data of wind turbines; The historical operating data is preprocessed, and key features related to the operating status and performance of the wind turbine are extracted from the preprocessed data. The key features are fused to obtain the initial dataset.

3. The method for identifying wind turbine generator parameters based on deep reinforcement learning according to claim 1, characterized in that, Constructing a mechanical subsystem model includes: The least squares method is used to estimate the vibration frequency and damping coefficient of the mechanical system to obtain estimated values ​​of the mechanical parameters. If there is a large deviation between the estimated values ​​of the mechanical parameters and the actual system, the mechanical parameters are continuously adjusted through an iterative optimization algorithm to obtain the best estimated values. Based on the optimal estimate, construct the dynamic equations of the mass-spring-damped system; Based on the dynamic equations of the mass-spring-damped system, a physical model of the mechanical subsystem is established.

4. The method for identifying wind turbine generator parameters based on deep reinforcement learning according to claim 1, characterized in that, Obtaining the electrical subsystem model includes: Build a neural network model and optimize the hyperparameters of the neural network using grid search; Using training set data, a backpropagation algorithm is employed to train the neural network model, continuously updating the model parameters so that the model can fit the voltage-current-speed relationship of the generator, thereby obtaining the electrical subsystem model. The training set consists of generator speed, output voltage, and current.

5. The method for identifying wind turbine generator parameters based on deep reinforcement learning according to claim 1, characterized in that, Establishing a control subsystem model includes: The state-space modeling method is used to construct the state equations and output equations of the control subsystem; Based on the aforementioned state equation and output equation, the Kalman filter algorithm is used to estimate the system state of the control subsystem. Based on the system state, determine whether the control characteristics of the control subsystem meet the preset conditions. If they do, determine the currently established control subsystem model. If the estimated system state does not meet the preset conditions, adjust the parameters of the state equation and output equation, and re-estimate the system state until the preset conditions are met.

6. The method for identifying wind turbine generator parameters based on deep reinforcement learning according to claim 1, characterized in that, Obtaining the comprehensive mathematical model includes: The coupling matrix is ​​combined with the models of each subsystem. Through matrix transformation and equation merging, the state equations and output equations of each subsystem are integrated with the coupling matrix to form the comprehensive state equations and output equations of the wind turbine generator set, i.e., the comprehensive mathematical model.

7. The method for identifying wind turbine generator parameters based on deep reinforcement learning according to claim 6, characterized in that, Based on the comprehensive mathematical model, wind turbine parameters are identified, and the identification results are obtained, including: The Kalman filter algorithm is used to filter the noise in the comprehensive mathematical model. The recursive least squares method is used to identify parameters in the filtered comprehensive mathematical model and to estimate the values ​​of unknown parameters in the model online. Based on the parameter identification results, the mechanical parameters, electrical parameters, and control parameters in the integrated mathematical model are updated to obtain the optimized integrated mathematical model. The optimized integrated mathematical model is applied to the real-time control system of the wind turbine generator set to identify the wind turbine generator parameters and obtain the identification results.

8. A wind turbine generator parameter identification system based on deep reinforcement learning implemented according to any one of claims 1-7, characterized in that, include: The system includes a data acquisition module, a mechanical characteristic identification module, an electrical characteristic identification module, a control characteristic identification module, a coupling analysis module, a model integration module, and a parameter identification module. The data acquisition module is used to acquire historical operating data of the wind turbine generator set, including mechanical vibration, electrical output, control signals, and environmental parameters such as wind speed and wind direction, to form an initial dataset. The mechanical characteristic identification module is used to extract mechanical vibration frequency and damping coefficient from the initial dataset and establish a mechanical subsystem model based on a physical model-based system identification method for mechanical characteristics. The electrical characteristic identification module is used to extract generator speed, output voltage, and current parameters from the initial dataset using data-driven identification technology, and to establish an electrical subsystem model. The control characteristic identification module is used to extract controller input and output signals from the initial dataset and establish a control subsystem model based on a state-space modeling method for the control characteristics. The coupling analysis module is used to analyze the dynamic coupling relationship between the subsystems based on the mechanical subsystem model, electrical subsystem model, and control subsystem model, and to establish a coupling matrix. The model integration module is used to integrate the mechanical subsystem model, electrical subsystem model, control subsystem model and coupling matrix to form a comprehensive mathematical model of the wind turbine generator set; The parameter identification module is used to identify the parameters of the wind turbine generator set based on a comprehensive mathematical model and obtain the identification result data.

Images