Wind power system adaptive control method based on time sequence prediction and wind speed data fusion

Abstract

The invention discloses a wind power system adaptive control method based on time sequence prediction and wind speed data fusion, and relates to the technical field of wind power system control, and the method comprises the steps: obtaining the long-term historical data of a wind power plant, and constructing a wind speed and wind direction initial time sequence prediction model; acquiring real-time wind speed and wind direction data based on a multi-source heterogeneous sensor, and calculating actual generating capacity and virtual generating capacity; based on the actual power generation amount and the virtual power generation amount, continuously calculating a time sequence difference degree between the actual power generation amount and the virtual power generation amount to obtain an accumulated efficiency difference value; if the accumulated efficiency difference value is greater than 0, selecting a predicted value as an optimal control reference, and if the accumulated efficiency difference value is less than or equal to 0, selecting a sensor measurement value; and power generation capacity maximization and mechanical load minimization serve as a multi-objective optimization function, optimal yaw angle and pitch angle instructions are solved in real time, and accurate self-adaptive control over the wind turbine generator is achieved. The method has the beneficial effects that the overall energy efficiency of the wind power plant is optimized, and the wind power industry is promoted to be intelligentized and refined.

Description

Technical Field

[0001] This invention relates to the field of wind power system control technology, specifically to an adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data. Background Technology

[0002] Current adaptive control of wind power systems mostly relies on traditional physical models and simple control strategies, lacking in-depth analysis and prediction of time-series data such as wind speed. Although some studies have attempted to improve control performance using wind speed prediction, most methods have failed to effectively integrate time-series data with real-time wind speed information, resulting in insufficient prediction accuracy and an inability to cope with complex and dynamically changing wind conditions. In addition, existing technologies have limited capabilities in multi-objective optimization and real-time adjustment, often neglecting the fine-tuning of wind turbines under different environmental conditions, which affects the power generation efficiency and equipment lifespan of wind power systems. Summary of the Invention

[0003] To address the aforementioned technical issues, this paper presents an adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data. This solution overcomes the limitations of existing technologies in multi-objective optimization and real-time adjustment, which often neglect the fine-tuning of wind turbine units under different environmental conditions, thus affecting the power generation efficiency and equipment lifespan of wind power systems.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An adaptive control method for wind power systems based on the fusion of time-series forecasting and wind speed data includes: Obtain long-term historical data of wind farms, extract dynamic correlation features between wind speed and wind direction, and construct an initial time series prediction model for wind speed and wind direction; Based on multi-source heterogeneous sensors deployed in different geographical locations of the wind farm, real-time wind speed and direction data are acquired. Combined with the initial time-series prediction model of wind speed and direction, the sensor and predicted measurement values ​​are used as control inputs to calculate the actual power generation and virtual power generation. Based on actual power generation and virtual power generation, an adaptive sliding window technique is used to continuously calculate the time-series difference between actual power generation and virtual power generation to obtain the cumulative efficiency difference. If the cumulative performance difference is greater than 0, the predicted value is selected as the optimal control reference; if the cumulative performance difference is less than or equal to 0, the sensor measurement value is selected as the optimal control reference. Using the optimal control benchmark as the input to the main controller of the wind turbine, and taking maximizing power generation and minimizing mechanical load as multi-objective optimization functions, the optimal yaw angle and pitch angle commands are solved in real time and sent to the actuators to achieve precise adaptive control of the wind turbine.

[0005] Preferably, long-term historical data of wind farms are acquired, data preprocessing is performed, and time series data are constructed. Based on time series data, wind speed and wind direction at a specific future time step are used as the target variables for prediction; Acquire data from several wind turbines, and extract the wind speed and wind direction values ​​of the wind turbine data for n consecutive time steps before time t, as local time series features; Based on the spatial layout of the wind farm and the prevailing wind direction, several upstream wind turbines are identified, and the real-time wind speed and wind direction values ​​of the upstream wind turbines at time t are recorded as spatial influence characteristics. By splicing local temporal features and spatial influence features, a feature set relating wind speed and wind direction is constructed. Using historical wind speed and direction characteristics over a period of time as input, mean squared error as the loss function, the AdamW optimizer iteratively optimizes the model parameters using gradient descent until the training loss converges. The predicted wind speed and direction values ​​at a future time are used as output to construct an initial time-series prediction model for wind speed and direction.

[0006] Preferably, real-time wind speed and direction data are acquired based on multi-source heterogeneous sensors deployed in different geographical locations of the wind farm, and the data is preprocessed. Based on multi-source heterogeneous sensors, real-time wind speed and the current operating status of the wind turbine are obtained. Combined with the wind turbine's own power characteristic curve, the actual power generation of the real-time sensors is calculated according to the aerodynamic power model. The aerodynamic power model formula is as follows: in, The total mechanical power captured by the wind turbine rotor from the wind. air density, The circular area swept by the rotor blades of the wind turbine. The wind energy utilization coefficient, The tip velocity ratio is the ratio of the linear velocity at the blade tip to the incoming air velocity. The pitch angle is the angle between the blade cross-section chord length and the rotor's plane of rotation. The wind speed acting on the rotor plane; Based on the initial time series prediction model of wind speed and direction, real-time wind speed and direction data are used as input to obtain future wind speed and direction prediction data. Based on the predicted future wind speed and direction data, combined with the optimal wind energy capture curve provided by the wind turbine manufacturer, the optimal blade pitch angle and tip speed ratio setpoints of the wind turbine are calculated. These are then packaged and substituted into the aerodynamic power model to calculate the theoretical maximum power generation under the predicted wind speed and direction, thus obtaining the virtual power generation.

[0007] Preferably, based on the actual power generation and the virtual power generation, a control period of k is set, and the difference between the actual power generation and the virtual power generation within the control period k is calculated; Based on real-time wind speed and direction data, the window length is dynamically adjusted according to the fluctuations in actual and virtual power generation, and the actual total power generation and virtual total power generation within the time window are calculated. Based on the actual total power generation and the virtual total power generation within the time window, the difference between the actual power generation and the virtual power generation within the entire window is calculated. The time-series difference between the actual power generation and the virtual power generation is continuously calculated to obtain the cumulative efficiency difference, as shown in the following formula: in, In time A given window length, To control the minimum and maximum window length within a period, This represents the standard deviation of wind speed over a given interval around the current time. Standard deviation of wind direction within a certain range around the current time. This is the sensitivity adjustment constant. A small constant set to prevent the denominator from being zero; For a moment The cumulative performance difference Let be the actual power generation at time i. Let i be the virtual power generation at time i.

[0008] Preferably, based on the obtained cumulative performance difference, the relationship between the cumulative performance difference and a preset threshold is determined; The relationship between the cumulative performance difference and the preset threshold can be intuitively represented by the following formula: in, As the reference wind speed, For indicator functions, when When =1, switch to prediction mode. When =0, switch to sensor mode. Let r be the predicted wind speed. To predict the wind direction at time r, Let r be the actual wind speed at time r. Let r be the actual wind direction at time r. Used as the baseline wind direction.

[0009] Preferably, based on the LSTM long short-term memory network, a multi-layer LSTM unit is designed with a time series as the input layer to learn the long-term dependence of wind speed and wind direction in the time series, and a fully connected layer is constructed to predict the trend of wind speed and wind direction changes in the future as the output layer. A dynamic model of a wind turbine is constructed using the physical characteristics and environmental factors of the wind turbine as constraints.

[0010] Preferably, the dynamic model of the output layer and wind turbine is combined with the predicted wind speed and wind direction change trend in the future period. The optimization objectives are to maximize power generation and minimize mechanical load. The MPC model predictive control method is used to continuously optimize and solve the optimal yaw angle and pitch angle commands in each control cycle. The optimal yaw angle and pitch angle commands are calculated and updated in real time. Based on the obtained optimal yaw angle and pitch angle commands, the wind turbine actuators are sent to achieve precise adaptive control of the wind turbine.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes an adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data. This scheme achieves precise adaptive control of the wind power system by fusing time-series predictions of wind speed and direction with real-time sensor data. By introducing adaptive sliding window technology, the weights of predicted values ​​and sensor measurements can be dynamically adjusted, significantly improving control accuracy and system adaptability. This method optimizes yaw and pitch angle commands in real time within each control cycle, ensuring maximum power generation and minimum mechanical load, improving the operating efficiency and reliability of wind turbines, reducing mechanical losses, and extending equipment lifespan. Simultaneously, the system can cope with changes in wind conditions, optimize the overall energy efficiency of the wind farm, and promote the development of the wind power industry towards intelligent and refined control. Attached Figure Description

[0012] Figure 1 This is a flowchart of an adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data. Detailed Implementation

[0013] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0014] Reference Figure 1 As shown, the adaptive control method for wind power systems based on the fusion of time-series forecasting and wind speed data includes: S1. Obtain long-term historical data of the wind farm, extract the dynamic correlation features between wind speed and wind direction, and construct an initial time series prediction model for wind speed and wind direction; the long-term historical data of the wind farm includes: timestamp, wind speed, wind direction and ambient temperature for each wind turbine; Step S1 includes the following: Acquire long-term historical data of wind farms, perform data preprocessing, and construct time-series data. Based on time series data, wind speed and wind direction at a specific future time step are used as the target variables for prediction; Acquire data from several wind turbines, and extract the wind speed and wind direction values ​​of the wind turbine data for n consecutive time steps before time t, as local time series features; Based on the spatial layout of the wind farm and the prevailing wind direction, several upstream wind turbines are identified, and the real-time wind speed and wind direction values ​​of the upstream wind turbines at time t are recorded as spatial influence characteristics. By splicing local temporal features and spatial influence features, a feature set relating wind speed and wind direction is constructed. Using historical wind speed and direction characteristics over a period of time as input, mean squared error as the loss function, the AdamW optimizer iteratively optimizes the model parameters using gradient descent until the training loss converges. The predicted wind speed and direction values ​​at a future time are used as output to construct an initial time-series prediction model for wind speed and direction.

[0015] When using it, refer to the content of step S1 above: Existing wind speed and direction prediction models typically rely on traditional time series methods, but these methods suffer from drawbacks such as low computational efficiency, difficulty in handling long-term temporal dependencies, and poor parallelization capabilities. Furthermore, traditional methods are also lacking in the integration of spatial features and cannot fully consider the spatiotemporal relationships between wind turbines in a wind farm, thus limiting prediction accuracy. In contrast, this step can more efficiently capture long-term temporal dependencies and has good parallelization capabilities, significantly improving training efficiency. This step not only improves the performance of wind speed and direction prediction models but also demonstrates significant computational advantages when dealing with real-time prediction of large-scale wind farms.

[0016] S2. Based on multi-source heterogeneous sensors deployed in different geographical locations of the wind farm, real-time wind speed and wind direction data are obtained. Combined with the initial time series prediction model of wind speed and wind direction, the sensor and predicted measurement values ​​are used as control inputs to calculate the actual power generation and virtual power generation. Step S2 includes the following: Based on multi-source heterogeneous sensors deployed in different geographical locations of the wind farm, real-time wind speed and wind direction data are acquired and data preprocessing is performed. Based on multi-source heterogeneous sensors, real-time wind speed and the current operating status of the wind turbine are obtained. Combined with the wind turbine's own power characteristic curve, the actual power generation of the real-time sensors is calculated according to the aerodynamic power model. The aerodynamic power model formula is as follows: in, The total mechanical power captured by the wind turbine rotor from the wind. air density, The circular area swept by the rotor blades of the wind turbine. The wind energy utilization coefficient, The tip velocity ratio is the ratio of the linear velocity at the blade tip to the incoming air velocity. The pitch angle is the angle between the blade cross-section chord length and the rotor's plane of rotation. The wind speed acting on the rotor plane; Based on the initial time series prediction model of wind speed and direction, real-time wind speed and direction data are used as input to obtain future wind speed and direction prediction data. Based on the predicted future wind speed and direction data, combined with the optimal wind energy capture curve provided by the wind turbine manufacturer, the optimal blade pitch angle and tip speed ratio setpoints of the wind turbine are calculated. These are then packaged and substituted into the aerodynamic power model to calculate the theoretical maximum power generation under the predicted wind speed and direction, thus obtaining the virtual power generation.

[0017] When using it, refer to the content of step S2 above: Current research in the wind power field mainly focuses on predicting the power generation of a single wind turbine or a small area, ignoring the differences in wind speed and direction at different geographical locations within the wind farm, leading to inaccurate power generation predictions. Although real-time data acquisition by sensors is available, most methods fail to combine multi-source sensors with time-series prediction models for dynamic optimization, and the control strategies do not fully consider the adaptability of wind turbine operating conditions, easily resulting in inefficient power generation. This step improves the accuracy of power generation prediction and optimizes wind turbine operating parameters by combining real-time data from multi-source sensors with wind speed and direction prediction models, thereby increasing power generation efficiency. Compared with traditional methods, this method can better cope with the dynamic changes in the wind farm, reduce energy waste, and provide accurate decision support for wind power operation and management.

[0018] S3. Based on actual power generation and virtual power generation, an adaptive sliding window technique is used to continuously calculate the time-series difference between actual power generation and virtual power generation to obtain the cumulative efficiency difference. Step S3 includes the following: Based on the actual power generation and the virtual power generation, a control period of k is set, and the difference between the actual power generation and the virtual power generation within the control period k is calculated. Based on real-time wind speed and direction data, the window length is dynamically adjusted according to the fluctuations in actual and virtual power generation, and the actual total power generation and virtual total power generation within the time window are calculated. Based on the actual total power generation and the virtual total power generation within the time window, the difference between the actual power generation and the virtual power generation within the entire window is calculated. The time-series difference between the actual power generation and the virtual power generation is continuously calculated to obtain the cumulative efficiency difference, as shown in the following formula: in, In time A given window length, To control the minimum and maximum window length within a period, This represents the standard deviation of wind speed over a given interval around the current time. Standard deviation of wind direction within a certain range around the current time. This is the sensitivity adjustment constant. A small constant set to prevent the denominator from being zero; For a moment The cumulative performance difference Let be the actual power generation at time i. Let i be the virtual power generation at time i.

[0019] When using it, refer to the content of step S3 above: Current domestic and international technologies generally use fixed time windows to calculate power generation variance, neglecting dynamic factors such as wind speed and direction, resulting in inaccurate calculation results and difficulty in coping with wind condition fluctuations. In addition, although some methods consider wind speed and direction, they lack adaptive adjustment mechanisms, resulting in slow adjustment responses and affecting power generation efficiency. Adopting adaptive sliding window technology can dynamically adjust the window length according to real-time changes in wind speed and direction, improving the accuracy of variance calculation and thus optimizing wind farm operation. By using the standard deviation of wind speed and direction as adjustment factors, the real-time adaptability of the system is improved, enhancing power generation efficiency and system stability.

[0020] S4. If the cumulative performance difference is greater than 0, the predicted value is selected as the optimal control reference; if the cumulative performance difference is less than or equal to 0, the sensor measurement value is selected as the optimal control reference. Step S4 includes the following: Based on the obtained cumulative performance difference, determine the relationship between the cumulative performance difference and the preset threshold; The relationship between the cumulative performance difference and the preset threshold can be intuitively represented by the following formula: in, As the reference wind speed, For indicator functions, when When =1, switch to prediction mode. When =0, switch to sensor mode. Let r be the predicted wind speed. To predict the wind direction at time r, Let r be the actual wind speed at time r. Let r be the actual wind direction at time r. Used as the baseline wind direction.

[0021] When using it, refer to the content of step S4 above: This decision-making mechanism based on cumulative performance difference enhances the system's adaptability, enabling it to dynamically adjust control strategies under different environmental conditions, thereby improving the overall efficiency and stability of the system. Through this method, the system can more intelligently switch between predicted and actual measured values ​​to optimize control performance.

[0022] S5. Using the optimal control benchmark as the input to the main controller of the wind turbine, and taking the maximization of power generation and the minimization of mechanical load as the multi-objective optimization function, the optimal yaw angle and pitch angle commands are solved in real time and sent to the actuator to achieve precise adaptive control of the wind turbine. Step S5 includes the following: Based on the LSTM (Long Short-Term Memory) network, a multi-layer LSTM unit is designed with a time series as the input layer to learn the long-term dependence of wind speed and wind direction in the time series. A fully connected layer is constructed to predict the trend of wind speed and wind direction changes in the future as the output layer. A dynamic model of a wind turbine is constructed using the physical characteristics and environmental factors of the wind turbine as constraints.

[0023] Step S5 also includes the following: Combining the predicted wind speed and direction trends over a future period as the dynamic model of the output layer and wind turbine, with the optimization objectives of maximizing power generation and minimizing mechanical load, the MPC model predictive control method is used to continuously optimize and solve for the optimal yaw angle and pitch angle commands in each control cycle, calculate the optimal yaw angle and pitch angle commands, and update the optimal yaw angle and pitch angle commands in real time. Based on the obtained optimal yaw angle and pitch angle commands, the wind turbine actuators are sent to achieve precise adaptive control of the wind turbine.

[0024] When using it, refer to the content of step S5 above: Currently, most wind turbine control systems employ simple control strategies, making it difficult to cope with complex environmental changes such as wind speed and direction fluctuations. Traditional methods do not fully utilize machine learning techniques such as LSTM (Long Short-Term Memory) networks for deep time-series analysis, and their multi-objective optimization performance is poor, with limited real-time prediction and dynamic adjustment capabilities, thus affecting power generation efficiency and equipment lifespan. By predicting wind speed and direction changes using LSTM networks and combining them with MPC (Multi-Purpose Control) model predictive control methods for rolling optimization, yaw angle and pitch angle commands can be precisely adjusted to achieve multi-objective optimization that maximizes power generation and minimizes mechanical load. This method improves the efficiency and reliability of wind turbines, reduces mechanical losses, extends equipment lifespan, and effectively copes with complex wind conditions.

[0025] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. An adaptive control method for wind power systems based on the fusion of time-series forecasting and wind speed data, characterized in that, include: S1. Obtain long-term historical data of wind farms, extract the dynamic correlation features between wind speed and wind direction, and construct an initial time series prediction model for wind speed and wind direction. S2. Based on multi-source heterogeneous sensors deployed in different geographical locations of the wind farm, real-time wind speed and wind direction data are obtained. Combined with the initial time series prediction model of wind speed and wind direction, the sensor and predicted measurement values ​​are used as control inputs to calculate the actual power generation and virtual power generation. S3. Based on actual power generation and virtual power generation, an adaptive sliding window technique is used to continuously calculate the time-series difference between actual power generation and virtual power generation to obtain the cumulative efficiency difference. S4. If the cumulative performance difference is greater than 0, the predicted value is selected as the optimal control reference; if the cumulative performance difference is less than or equal to 0, the sensor measurement value is selected as the optimal control reference. S5. Using the optimal control benchmark as the input to the main controller of the wind turbine, and taking the maximization of power generation and the minimization of mechanical load as the multi-objective optimization function, the optimal yaw angle and pitch angle commands are solved in real time and sent to the actuator to achieve precise adaptive control of the wind turbine.

2. The adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data as described in claim 1, characterized in that, S1 includes: Acquire long-term historical data of wind farms, perform data preprocessing, and construct time-series data. Based on time series data, wind speed and wind direction at a specific future time step are used as the target variables for prediction; Acquire data from several wind turbines, and extract the wind speed and wind direction values ​​of the wind turbine data for n consecutive time steps before time t, as local time series features; Based on the spatial layout of the wind farm and the prevailing wind direction, several upstream wind turbines are identified, and the real-time wind speed and wind direction values ​​of the upstream wind turbines at time t are recorded as spatial influence characteristics. By splicing local temporal features and spatial influence features, a feature set relating wind speed and wind direction is constructed. Using historical wind speed and direction characteristics over a period of time as input, mean squared error as the loss function, the AdamW optimizer iteratively optimizes the model parameters using gradient descent until the training loss converges. The predicted wind speed and direction values ​​at a future time are used as output to construct an initial time-series prediction model for wind speed and direction.

3. The adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data as described in claim 1, characterized in that, S2 includes: Based on multi-source heterogeneous sensors deployed in different geographical locations of the wind farm, real-time wind speed and wind direction data are acquired and data preprocessing is performed. Based on multi-source heterogeneous sensors, real-time wind speed and the current operating status of the wind turbine are obtained. Combined with the wind turbine's own power characteristic curve, the actual power generation of the real-time sensors is calculated according to the aerodynamic power model. The aerodynamic power model formula is as follows: in, The total mechanical power captured by the wind turbine rotor from the wind. air density, The circular area swept by the rotor blades of the wind turbine. The wind energy utilization coefficient, The tip velocity ratio is the ratio of the linear velocity at the blade tip to the incoming air velocity. The pitch angle is the angle between the blade cross-section chord length and the rotor's plane of rotation. The wind speed acting on the rotor plane; Based on the initial time series prediction model of wind speed and direction, real-time wind speed and direction data are used as input to obtain future wind speed and direction prediction data. Based on the predicted future wind speed and direction data, combined with the optimal wind energy capture curve provided by the wind turbine manufacturer, the optimal blade pitch angle and tip speed ratio setpoints of the wind turbine are calculated. These are then packaged and substituted into the aerodynamic power model to calculate the theoretical maximum power generation under the predicted wind speed and direction, thus obtaining the virtual power generation.

4. The adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data as described in claim 3, characterized in that, S3 includes: Based on the actual power generation and the virtual power generation, a control period of k is set, and the difference between the actual power generation and the virtual power generation within the control period k is calculated. Based on real-time wind speed and direction data, the window length is dynamically adjusted according to the fluctuations in actual and virtual power generation, and the actual total power generation and virtual total power generation within the time window are calculated. Based on the actual total power generation and the virtual total power generation within the time window, the difference between the actual power generation and the virtual power generation within the entire window is calculated. The time-series difference between the actual power generation and the virtual power generation is continuously calculated to obtain the cumulative efficiency difference, as shown in the following formula: in, In time A given window length, To control the minimum and maximum window length within a period, This represents the standard deviation of wind speed over a given interval around the current time. Standard deviation of wind direction within a certain range around the current time. This is the sensitivity adjustment constant. A small constant set to prevent the denominator from being zero; For a moment The cumulative performance difference Let be the actual power generation at time i. Let i be the virtual power generation at time i.

5. The adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data according to claim 4, characterized in that, S4 includes: Based on the obtained cumulative performance difference, determine the relationship between the cumulative performance difference and the preset threshold; The relationship between the cumulative performance difference and the preset threshold can be intuitively represented by the following formula: in, As the reference wind speed, For indicator functions, when When =1, switch to prediction mode. When =0, switch to sensor mode. Let r be the predicted wind speed. To predict the wind direction at time r, Let r be the actual wind speed at time r. Let r be the actual wind direction at time r. Used as the baseline wind direction.

6. The adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data according to claim 5, characterized in that, S5 includes: Based on the LSTM (Long Short-Term Memory) network, a multi-layer LSTM unit is designed with a time series as the input layer to learn the long-term dependence of wind speed and wind direction in the time series. A fully connected layer is constructed to predict the trend of wind speed and wind direction changes in the future as the output layer. A dynamic model of a wind turbine is constructed using the physical characteristics and environmental factors of the wind turbine as constraints.

7. The adaptive control method for wind power systems based on the fusion of time-series prediction and wind speed data as described in claim 6, characterized in that, The S5 also includes: Combining the predicted wind speed and direction trends over a future period as the dynamic model of the output layer and wind turbine, with the optimization objectives of maximizing power generation and minimizing mechanical load, the MPC model predictive control method is used to continuously optimize and solve for the optimal yaw angle and pitch angle commands in each control cycle, calculate the optimal yaw angle and pitch angle commands, and update the optimal yaw angle and pitch angle commands in real time. Based on the obtained optimal yaw angle and pitch angle commands, the wind turbine actuators are sent to achieve precise adaptive control of the wind turbine.

Images