Wind turbine generator load reduction method based on laser radar

Abstract

The invention discloses a wind turbine generator load reduction method based on a laser radar. The method comprises the following steps: step 1, measuring wind field parameters by using the laser radar; the wind field parameters comprise wind speed, wind direction and turbulence intensity; 2, establishing an RBF neural network model, and analyzing and processing the wind field parameters by using the RBF neural network model; step 3, training the RBF neural network model; 4, predicting wind field parameters; 5, establishing a fan blade physical model; and step 6, performing variable pitch adjustment on the fan based on the PID controller. According to the method, a complete wind turbine generator set load reduction adjusting system is formed in the whole process of wind field parameter measurement, prediction, fan blade physical model establishment and PID controller variable pitch adjustment. All links are matched with one another and work cooperatively, and the load reduction target of the wind turbine unit can be effectively achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind power, and in particular relates to a wind turbine load reduction method based on laser radar. Background Art

[0002] During operation, wind turbines are subject to a variety of complex loads, including aerodynamic loads, gravity loads, inertial loads, etc. Among them, aerodynamic loads are the main source of loads, which change continuously with changes in wind speed, wind direction and turbulence intensity. Under severe wind conditions such as strong winds and turbulence, the load on the wind turbine will increase sharply, which may cause damage to components such as blades and towers, affecting the safe operation and service life of the wind turbine. For example, under strong wind conditions, the blades may be subjected to excessive bending stress and torsional stress, resulting in fatigue damage to the blades; the tower may be subjected to excessive lateral force and bending moment, causing the tower to tilt or collapse. In addition, the increase in turbulence intensity will also intensify the vibration of the wind turbine, affecting the stability and reliability of the unit.

[0003] In order to solve the load problem of wind turbines, traditional methods mainly include passive load reduction and active load reduction. The passive load reduction method mainly improves the load resistance of wind turbines by optimizing the structural design of wind turbines, such as increasing the strength of blades and improving the stability of towers. However, this method often requires increasing the cost and weight of wind turbines, and has limited adaptability to complex wind conditions.

[0004] In view of the difficulties of existing technologies, many people are studying the wind turbine load reduction technology based on LiDAR. At present, there are still some problems with the wind turbine load reduction method based on LiDAR, such as the imperfect processing and analysis methods of LiDAR data, and the accuracy and response speed of the load reduction control algorithm need to be improved. Therefore, it is urgent to further study and develop the wind turbine load reduction method based on LiDAR to improve the safe operation and service life of wind turbines. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a wind turbine load reduction method based on laser radar. The present invention forms a complete wind turbine load reduction adjustment system from the whole process of wind field parameter measurement and prediction to the establishment of wind turbine blade physical model and then to PID controller pitch adjustment. Each link cooperates with each other and works collaboratively to effectively achieve the load reduction target of the wind turbine, which helps to improve the safe operation and service life of the wind turbine.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: A method for reducing load of a wind turbine based on laser radar, comprising the following steps: Step 1: Use laser radar to measure wind field parameters; the wind field parameters include wind speed, wind direction, and turbulence intensity; Step 2: Establish an RBF neural network model and use the RBF neural network model to analyze and process wind field parameters; Step 3: Train the RBF neural network model; Step 4: predict wind field parameters; Step 5: Establish a physical model of the fan blade; Step 6: Regulate the pitch of the wind turbine based on the PID controller.

[0007] Preferably, in step 1, the method for measuring the wind field parameters is as follows: (1) Wind speed: The laser radar measures wind speed in the horizontal direction (0°, 90°, 180°, 270°) and vertical direction (at specific height intervals) to obtain the wind speed distribution in different directions and heights. At the same time, the instantaneous value, maximum value, minimum value and average value of the wind speed are measured. Among them, the instantaneous wind speed Reflects the wind speed at the current moment, the average wind speed The average wind speed over a certain period of time. and minimum wind speed They represent the extreme wind speed in that time period.

[0008] (2) Wind direction: The laser radar determines the wind direction by measuring the direction of the wind and recording the frequency and magnitude of wind direction changes. The wind direction is measured in degrees and rotates clockwise from 0° (due north). .

[0009] (3) Turbulence intensity: The standard deviation, variance and coefficient of variation of wind speed measured by lidar reflect the turbulence intensity. Turbulence intensity is expressed as standard deviation, denoted as , and gives the variance and coefficient of variation The measurement of turbulence intensity helps to understand the instability of the wind field, which is of great significance to the operational stability and load control of wind turbines.

[0010] Preferably, in step 2, the method for analyzing and processing the wind field parameters using the RBF neural network model is as follows: (1) the RBF neural network model includes: an input layer, a hidden layer and an output layer; Input layer: receives input parameters, including historical wind speed data, i.e. , , , ; Wind direction data ; Turbulence intensity data, i.e. , , ; Hidden layer: consists of multiple radial basis function (RBF) neurons; the output of each RBF neuron is a function of the distance between the input vector and the center vector, using the Gaussian function as the activation function: , in is the input vector, It is The center vector of each RBF neuron, It is The width parameter of each RBF neuron; Output layer: Output prediction parameters, i.e. predicted wind speed ( , , , ), predicted wind direction ( ) and predicted turbulence intensity ( , , ).

[0011] (2) Output layer calculation formula: The output of the output layer is a linear combination of the outputs of the hidden layers: For predicted wind speed: , in is the number of neurons in the hidden layer, Is the connection The weights of the hidden layer neurons and the instantaneous wind speed output node, is the bias term of the instantaneous wind speed output node.

[0012] Similarly, , , .

[0013] For predicted wind direction: , in Is the connection The weights of the hidden layer neurons and the wind direction output node, is the bias term for the wind direction output node.

[0014] For predicting turbulence intensity: , , , in , , They are connected The weights of the hidden layer neurons and the standard deviation, variance, and coefficient of variation output nodes, , , They are the bias terms of the standard deviation, variance, and coefficient of variation output nodes respectively.

[0015] When establishing the RBF neural network model, historical wind speed data is an important input parameter. The wind speed data measured by the lidar in different directions and heights, including instantaneous wind speed, average wind speed, maximum wind speed and minimum wind speed, provide the RBF neural network with rich wind field information, enabling it to learn the changing laws of the wind field more accurately, thereby improving the prediction accuracy of future wind field parameters (predicted wind speed, predicted wind direction, predicted turbulence intensity).

[0016] Preferably, in step 3, the process of training the RBF neural network model is as follows: (1) Create a data set: historical wind speed, wind direction, turbulence intensity data, and corresponding timestamps. Divide the data set into a training set, a validation set, and a test set according to a certain ratio, for example, a ratio of 7:2:1.

[0017] (2) Optimization objective: The goal is to minimize the mean square error (MSE) and mean absolute error (MAE) between the predicted value and the actual value. The MSE calculation formula is: ,in is the sample size, is the actual value, is the predicted value; the MAE calculation formula is .

[0018] (3) Parameter adjustment: The stochastic gradient descent method is used to adjust the weights from the hidden layer to the output layer, and the K-Means clustering algorithm is used to select the center of the basis function. At the same time, the number of hidden layer neurons and the activation function of the RBF neural network are adjusted to improve the performance and generalization ability of the model.

[0019] Preferably, in step 4, the method for predicting wind field parameters is as follows: (1) Input parameters: Predicted wind speed: RBF neural network predicts wind speed, including instantaneous wind speed , average wind speed , Maximum wind speed and minimum wind speed .

[0020] Predicted wind direction: The predicted wind direction of the RBF neural network is denoted as .

[0021] Predicted turbulence intensity: RBF neural network predicted turbulence intensity, including standard deviation ,variance and coefficient of variation .

[0022] (2) Output parameters: Impeller center wind speed: predicted wind speed at the center of the impeller, including instantaneous wind speed , average wind speed , Maximum wind speed and minimum wind speed .

[0023] Impeller center wind direction: The predicted wind direction at the impeller center, denoted as .

[0024] Impeller Center Turbulence Intensity: Predicted turbulence intensity at the impeller center, including standard deviation ,variance and coefficient of variation .

[0025] Preferably, in step 5, the method for establishing the physical model of the fan blade is as follows: (1) Based on the principles of fluid mechanics and aerodynamics, combined with the wind speed distribution model, a mathematical model of the wind speed, wind direction, and turbulence intensity at the center of the impeller is established. The factors such as the rotation effect of the impeller, the blocking effect of the wind wheel, and the wake effect are fully considered to improve the accuracy and practicality of the model.

[0026] The wind speed distribution model formula is: , in, is the hub center wind speed. Since the wind shear factor is deleted, only part. is the wind shear component, It is the tower shadow effect component.

[0027] (2) Parameter adjustment: Calibrate model parameters using historical data, including impeller radius, number of blades, blade length, airfoil parameters, hub center height, etc. At the same time, adjust the empirical coefficient and correction factor of the model to ensure the accuracy and stability of the model.

[0028] (3) Establish an aerodynamic performance model: Based on the geometric parameters and aerodynamic characteristics of the blade and combined with the blade element momentum theory, an aerodynamic performance model is established. The relationship between the lift coefficient and drag coefficient of the blade and the blade length, airfoil, angle of attack and tip speed ratio is considered.

[0029] The formula for calculating the torque in blade element momentum theory is: , , in, is the air density, is the incoming flow velocity at the blade element, is the blade element equivalent area, and are the lift and drag coefficients of the airfoil, It is The pitch angle of the blades.

[0030] (4) Input parameters: Blade length: The initial length and design length range of the blade.

[0031] Airfoil: Different types of airfoils (such as NACA airfoil, Selig airfoil, etc.), including the name, parameters and performance curves of the airfoil.

[0032] Angle of attack: the initial angle of attack and the design angle of attack range of the blade.

[0033] Tip speed ratio: the initial tip speed ratio and design tip speed ratio range of the blade.

[0034] Air Density: The actual measured air density.

[0035] The incoming flow velocity at the blade element is calculated based on the wind speed distribution model and the wind speed at the center of the impeller, and is recorded as .

[0036] Leaf element equivalent area: calculated based on the leaf geometric parameters, denoted as .

[0037] (5) Output parameters: Optimized blade parameters: optimized blade length, airfoil, angle of attack and tip speed ratio.

[0038] Preferably, in step 6, the method for adjusting the pitch of the wind turbine based on the PID controller is as follows: (1) Input parameters: Optimized impeller center wind speed: After being processed by the Kalman filter, data including instantaneous wind speed, average wind speed, maximum wind speed and minimum wind speed are obtained.

[0039] Optimized impeller center wind direction: The optimized impeller center wind direction obtained by Kalman filter processing is recorded as .

[0040] Optimized impeller center turbulence intensity: The optimized impeller center turbulence intensity obtained by Kalman filter processing includes standard deviation, variance and coefficient of variation.

[0041] (2) Output parameters: Pitch angle: The pitch angle of each blade, accurate to a certain precision.

[0042] (3) Control methods: PID control: Based on the predicted wind speed, wind direction, and turbulence intensity, the pitch angle of each blade is adjusted to achieve optimal power generation efficiency while minimizing mechanical loads. The proportional (P), integral (I), and derivative (D) parameters of the PID controller are adjusted by trial and error or the Ziegler-Nichols method.

[0043] Parameter adjustment: Through simulation and actual machine testing, adjust the proportional (P), integral (I) and differential (D) parameters of the PID controller to ensure the robustness and response speed of the control system. At the same time, adjust the sampling time and control cycle of the PID controller to adapt to different wind speed changes and system response requirements.

[0044] Preferably, in step 1, when measuring wind speed, the laser radar measures angles in the horizontal direction at least including 0°, 90°, 180°, and 270°. When measuring in the vertical direction, the laser radar measures the wind speed at every interval of a fixed unit height to obtain the wind speed distribution in different directions and heights.

[0045] A wind turbine load reduction regulation system based on laser radar, comprising: Laser radar: used to measure wind field parameters; the wind field parameters include wind speed, wind direction, and turbulence intensity; Data analysis and processing unit: used to establish an RBF neural network model, use the RBF neural network model to analyze and process wind field parameters; and train the RBF neural network model at the same time; Early warning and prediction unit: used to predict wind field parameters; Modeling unit: used to establish the physical model of the wind turbine blades; PID controller-based actuator: used to adjust the pitch of the wind turbine based on the PID controller. A computer device comprising: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, the laser radar-based wind turbine load reduction method is implemented.

[0046] The present invention can achieve the following beneficial effects: 1. The present invention forms a complete wind turbine load reduction adjustment system from the whole process of wind field parameter measurement and prediction to the establishment of wind turbine blade physical model and then to the PID controller pitch adjustment. Each link cooperates with each other and works in coordination to effectively achieve the wind turbine load reduction target.

[0047] 2. By measuring wind speed in multiple horizontal and vertical directions, a more comprehensive wind speed distribution can be obtained. This helps to more accurately capture the complexity and changing trends of the wind field, making the predicted parameters such as wind direction and turbulence intensity more reliable. For example, the difference in wind speed in different directions can reflect the change in wind direction, while the change in wind speed in the vertical direction helps to analyze the change in turbulence intensity.

[0048] 3. After predicting the wind field parameters, it is necessary to further predict parameters such as wind speed, wind direction and turbulence intensity at the center of the impeller. The wind speed distribution data at different directions and heights measured by the lidar, combined with subsequent mathematical models and algorithms, can more accurately deduce the wind field conditions at the center of the impeller. For example, by considering factors such as the rotation effect of the impeller, the blocking effect of the wind wheel and the wake effect, the data measured by the lidar can more accurately predict the wind speed, wind direction and turbulence intensity at the center of the impeller.

[0049] 4. Since LiDAR can provide wind speed data in multiple directions and heights, it is possible to more comprehensively consider the factors affecting the wind field when establishing a prediction model for parameters such as wind speed at the center of the impeller. For example, changes in wind speed in the vertical direction can affect the turbulence intensity at the center of the impeller, while differences in wind speed in different horizontal directions can affect the wind direction at the center of the impeller. By comprehensively considering these factors, the accuracy of the prediction of the impeller center parameters can be improved, providing more reliable input data for subsequent physical models of wind turbine blades and pitch adjustment.

[0050] 5. When establishing the physical model of wind turbine blades, the geometric parameters and aerodynamic characteristics of the blades need to be considered. The wind field data measured by LiDAR can provide information such as the actual incoming flow velocity, wind direction and turbulence intensity. By analyzing the wind speed distribution in different directions and heights, the wind load on the blades at different positions can be determined, thereby optimizing the design of the blades, improving the efficiency of wind energy capture and reducing the mechanical load. The accuracy and stability of the physical model of wind turbine blades depends on the quality and reliability of the input data. The wind field data measured by LiDAR can provide more accurate boundary conditions and initial conditions for the model, thereby improving the accuracy and stability of the model.

[0051] 6. Since the laser radar can measure the changes in the wind field in real time, the PID controller can respond to the changes in the wind field more quickly and adjust the pitch angle of the blades in time. At the same time, accurate wind field data can improve the adjustment accuracy of the PID controller, so that the wind turbine can better adapt to different wind conditions and improve power generation efficiency and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0053] The preferred solution is Figure 1 As shown, a method for reducing load of a wind turbine based on laser radar comprises the following steps: Step 1: Use laser radar to measure wind field parameters; the wind field parameters include wind speed, wind direction, and turbulence intensity; Step 2: Establish an RBF neural network model and use the RBF neural network model to analyze and process wind field parameters; Step 3: Train the RBF neural network model; Step 4: predict wind field parameters; Step 5: Establish a physical model of the fan blade; Step 6: Regulate the pitch of the wind turbine based on the PID controller.

[0054] Furthermore, in step 1, the wind field parameters are measured as follows: (1) Wind speed: The laser radar measures wind speed in the horizontal direction (0°, 90°, 180°, 270°) and vertical direction (at specific height intervals) to obtain the wind speed distribution in different directions and heights. At the same time, the instantaneous value, maximum value, minimum value and average value of the wind speed are measured. Among them, the instantaneous wind speed Reflects the wind speed at the current moment, the average wind speed The average wind speed over a certain period of time. and minimum wind speed They represent the extreme wind speed in that time period.

[0055] (2) Wind direction: The laser radar determines the wind direction by measuring the direction of the wind and recording the frequency and magnitude of wind direction changes. The wind direction is measured in degrees and rotates clockwise from 0° (due north). .

[0056] (3) Turbulence intensity: The standard deviation, variance and coefficient of variation of wind speed measured by lidar reflect the turbulence intensity. Turbulence intensity is expressed as standard deviation, denoted as , and gives the variance and coefficient of variation The measurement of turbulence intensity helps to understand the instability of the wind field, which is of great significance to the operational stability and load control of wind turbines.

[0057] Furthermore, in step 2, the method for analyzing and processing the wind field parameters using the RBF neural network model is as follows: (1) the RBF neural network model includes: an input layer, a hidden layer, and an output layer; Input layer: receives input parameters, including historical wind speed data, i.e. , , , ; Wind direction data ; Turbulence intensity data, i.e. , , ; Hidden layer: consists of multiple radial basis function (RBF) neurons; the output of each RBF neuron is a function of the distance between the input vector and the center vector, using the Gaussian function as the activation function: , in is the input vector, It is The center vector of each RBF neuron, It is The width parameter of each RBF neuron; Output layer: Output prediction parameters, i.e. predicted wind speed ( , , , ), predicted wind direction ( ) and predicted turbulence intensity ( , , ).

[0058] (2) Output layer calculation formula: The output of the output layer is a linear combination of the outputs of the hidden layers: For predicted wind speed: , in is the number of neurons in the hidden layer, Is the connection The weights of the hidden layer neurons and the instantaneous wind speed output node, is the bias term of the instantaneous wind speed output node.

[0059] Similarly, , , .

[0060] For predicted wind direction: , in Is the connection The weights of the hidden layer neurons and the wind direction output node, is the bias term for the wind direction output node.

[0061] For predicting turbulence intensity: , , , in , , They are connected The weights of the hidden layer neurons and the standard deviation, variance, and coefficient of variation output nodes, , , They are the bias terms of the standard deviation, variance, and coefficient of variation output nodes respectively.

[0062] When establishing the RBF neural network model, historical wind speed data is an important input parameter. The wind speed data measured by the lidar in different directions and heights, including instantaneous wind speed, average wind speed, maximum wind speed and minimum wind speed, provide the RBF neural network with rich wind field information, enabling it to learn the changing laws of the wind field more accurately, thereby improving the prediction accuracy of future wind field parameters (predicted wind speed, predicted wind direction, predicted turbulence intensity).

[0063] Furthermore, in step 3, the process of training the RBF neural network model is as follows: (1) Create a data set: historical wind speed, wind direction, turbulence intensity data, and corresponding timestamps. Divide the data set into a training set, a validation set, and a test set according to a certain ratio, for example, a ratio of 7:2:1.

[0064] (2) Optimization objective: The goal is to minimize the mean square error (MSE) and mean absolute error (MAE) between the predicted value and the actual value. The MSE calculation formula is: ,in is the sample size, is the actual value, is the predicted value; the MAE calculation formula is .

[0065] (3) Parameter adjustment: The stochastic gradient descent method is used to adjust the weights from the hidden layer to the output layer, and the K-Means clustering algorithm is used to select the center of the basis function. At the same time, the number of hidden layer neurons and the activation function of the RBF neural network are adjusted to improve the performance and generalization ability of the model.

[0066] Furthermore, in step 4, the method for predicting wind field parameters is as follows: (1) Input parameters: Predicted wind speed: RBF neural network predicts wind speed, including instantaneous wind speed , average wind speed , Maximum wind speed and minimum wind speed .

[0067] Predicted wind direction: The predicted wind direction of the RBF neural network is denoted as .

[0068] Predicted turbulence intensity: RBF neural network predicted turbulence intensity, including standard deviation ,variance and coefficient of variation .

[0069] (2) Output parameters: Impeller center wind speed: predicted wind speed at the center of the impeller, including instantaneous wind speed , average wind speed , Maximum wind speed and minimum wind speed .

[0070] Impeller center wind direction: The predicted wind direction at the impeller center, denoted as .

[0071] Impeller Center Turbulence Intensity: Predicted turbulence intensity at the impeller center, including standard deviation ,variance and coefficient of variation .

[0072] Furthermore, in step 5, the method for establishing the physical model of the wind turbine blade is as follows: (1) Based on the principles of fluid mechanics and aerodynamics, combined with the wind speed distribution model, a mathematical model of the wind speed, wind direction, and turbulence intensity at the center of the impeller is established. The rotation effect of the impeller, the blocking effect of the wind wheel, the wake effect, and other factors are fully considered to improve the accuracy and practicality of the model.

[0073] The wind speed distribution model formula is: , in, is the hub center wind speed. Since the wind shear factor is deleted, only part. is the wind shear component, It is the tower shadow effect component.

[0074] (2) Parameter adjustment: Calibrate model parameters using historical data, including impeller radius, number of blades, blade length, airfoil parameters, hub center height, etc. At the same time, adjust the empirical coefficient and correction factor of the model to ensure the accuracy and stability of the model.

[0075] (3) Establish an aerodynamic performance model: Based on the geometric parameters and aerodynamic characteristics of the blade and combined with the blade element momentum theory, an aerodynamic performance model is established. The relationship between the lift coefficient and drag coefficient of the blade and the blade length, airfoil, angle of attack and tip speed ratio is considered.

[0076] The formula for calculating the torque in blade element momentum theory is: , , in, is the air density, is the incoming flow velocity at the blade element, is the blade element equivalent area, and are the lift and drag coefficients of the airfoil, It is The pitch angle of the blades.

[0077] The blade element can be understood as a micro-element segment on the blade. The blade is divided into many tiny segments along the span direction, and each such tiny segment is called a blade element. When analyzing the aerodynamic performance of the blade, the blade is usually regarded as consisting of a series of blade elements. By performing a force analysis on each blade element, the aerodynamic characteristics of the blade as a whole, such as lift, drag, and torque, can be obtained. Considering the parameters such as the incoming flow velocity at the blade element, air density, lift coefficient and drag coefficient of the airfoil, combined with methods such as blade element momentum theory, the performance of the blade and the output power of the wind turbine can be calculated.

[0078] (4) Input parameters: Blade length: The initial length and design length range of the blade.

[0079] Airfoil: Different types of airfoils (such as NACA airfoil, Selig airfoil, etc.), including the name, parameters and performance curves of the airfoil.

[0080] Angle of attack: the initial angle of attack and the design angle of attack range of the blade.

[0081] Tip speed ratio: the initial tip speed ratio and design tip speed ratio range of the blade.

[0082] Air Density: The actual measured air density.

[0083] The incoming flow velocity at the blade element is calculated based on the wind speed distribution model and the wind speed at the center of the impeller, and is recorded as .

[0084] Leaf element equivalent area: calculated based on the leaf geometric parameters, denoted as .

[0085] (5) Output parameters: Optimized blade parameters: optimized blade length, airfoil, angle of attack and tip speed ratio.

[0086] Furthermore, in step 6, the method for adjusting the pitch of the wind turbine based on the PID controller is as follows: (1) Input parameters: Optimized impeller center wind speed: After being processed by the Kalman filter, data including instantaneous wind speed, average wind speed, maximum wind speed and minimum wind speed are obtained.

[0087] Optimized impeller center wind direction: The optimized impeller center wind direction obtained by Kalman filter processing is recorded as .

[0088] Optimized impeller center turbulence intensity: The optimized impeller center turbulence intensity obtained by Kalman filter processing includes standard deviation, variance and coefficient of variation.

[0089] The Kalman filter is an algorithm that uses the linear system state equation to optimally estimate the system state through system input and output observation data. The input of the Kalman filter generally includes the predicted values ​​of parameters such as the wind speed at the center of the impeller, wind direction, and turbulence intensity, as well as the corresponding measured values. After processing by the Kalman filter, the following content is output: Instantaneous wind speed: The wind speed value at the center of the impeller at a certain moment obtained after filtering is closer to the true value and reduces the influence of measurement noise and prediction error.

[0090] Average wind speed: The average wind speed at the center of the impeller over a period of time reflects the overall level of wind speed at the center of the impeller over that period of time more accurately.

[0091] Maximum wind speed and minimum wind speed: respectively represent the maximum and minimum wind speed values ​​that may occur at the center of the impeller within a specific time period after being processed by the Kalman filter, which helps to evaluate the operating status and load conditions of the wind turbine under different wind conditions.

[0092] (2) Output parameters: Pitch angle: The pitch angle of each blade, accurate to a certain precision.

[0093] (3) Control methods: PID control: Based on the predicted wind speed, wind direction, and turbulence intensity, the pitch angle of each blade is adjusted to achieve optimal power generation efficiency while minimizing mechanical loads. The proportional (P), integral (I), and derivative (D) parameters of the PID controller are adjusted by trial and error or the Ziegler-Nichols method.

[0094] Parameter adjustment: Through simulation and actual machine testing, adjust the proportional (P), integral (I) and differential (D) parameters of the PID controller to ensure the robustness and response speed of the control system. At the same time, adjust the sampling time and control cycle of the PID controller to adapt to different wind speed changes and system response requirements.

[0095] When adjusting the pitch angle based on the PID controller, the optimized wind speed, wind direction and turbulence intensity at the center of the impeller are required as input. After a series of processing and prediction, the wind field data measured by the lidar can provide accurate input data for the PID controller, enabling it to more accurately adjust the pitch angle of the blades to achieve the best power generation efficiency while minimizing the mechanical load.

[0096] Furthermore, in step 1, when measuring wind speed, the laser radar measures angles in the horizontal direction at least including 0°, 90°, 180°, and 270°. When measuring in the vertical direction, the laser radar measures the wind speed at every interval of a fixed unit height to obtain the wind speed distribution in different directions and heights.

[0097] A wind turbine load reduction regulation system based on laser radar, comprising: Laser radar: used to measure wind field parameters; the wind field parameters include wind speed, wind direction, and turbulence intensity; Data analysis and processing unit: used to establish an RBF neural network model, use the RBF neural network model to analyze and process wind field parameters; and train the RBF neural network model at the same time; Early warning and prediction unit: used to predict wind field parameters; Modeling unit: used to establish the physical model of the wind turbine blades; PID controller-based actuator: used to adjust the pitch of the wind turbine based on the PID controller. A computer device comprising: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, the laser radar-based wind turbine load reduction method is implemented.

[0098] Embodiment 1: 1. Implementation Scenario Assume that there is a wind farm, and one of the wind turbines needs to be load-reduced. The model of the wind turbine is XYZ-1000, with a rated power of 1000 kW, a hub height of 80 meters, a blade length of 40 meters, and a three-blade design. The average air density in the area where the wind farm is located is 1.2 kg / m3.

[0099] 2. Implementation steps 1. Wind field parameter measurement The wind field is measured using LiDAR. The LiDAR measures wind speed in four directions, 0°, 90°, 180°, and 270°, horizontally, and every 10 meters vertically. At the same time, the instantaneous value, maximum value, minimum value, and average value of wind speed, as well as wind direction (in degrees, clockwise from due north) and turbulence intensity (determined by measuring the standard deviation, variance, and coefficient of variation of wind speed) are recorded.

[0100] 2. RBF neural network model establishment and training Collect wind field data from the past week, including wind speed, wind direction, and turbulence intensity data, as well as corresponding timestamps. Divide these data into training set, validation set, and test set in a ratio of 7:2:1.

[0101] The RBF neural network model is established, including input layer, hidden layer and output layer. The input layer receives historical wind speed data (instantaneous wind speed, average wind speed, maximum wind speed, minimum wind speed), wind direction data and turbulence intensity data. The hidden layer is composed of multiple RBF neurons, using Gaussian function as activation function. The output layer outputs predicted wind speed (instantaneous wind speed, average wind speed, maximum wind speed, minimum wind speed), predicted wind direction and predicted turbulence intensity.

[0102] With the goal of minimizing the mean square error (MSE) and mean absolute error (MAE) between the predicted value and the actual value, the stochastic gradient descent method is used to adjust the weights from the hidden layer to the output layer, and the K-Means clustering algorithm is used to select the center of the basis function. At the same time, the number of hidden layer neurons and the activation function of the RBF neural network are adjusted. After many experiments, the number of hidden layer neurons is determined to be 50, and the activation function is the Gaussian function.

[0103] 3. Wind field parameter prediction The trained RBF neural network model is used to predict the wind field parameters for a period of time in the future. The input parameters are the predicted wind speed (instantaneous wind speed, average wind speed, maximum wind speed, minimum wind speed), predicted wind direction and predicted turbulence intensity of the RBF neural network. The output parameters are the wind speed at the center of the impeller (instantaneous wind speed, average wind speed, maximum wind speed, minimum wind speed), the wind direction at the center of the impeller and the turbulence intensity at the center of the impeller (standard deviation, variance and coefficient of variation).

[0104] 4. Establishment of physical model of wind turbine blades Based on the principles of fluid mechanics and aerodynamics, combined with the wind speed distribution model, a mathematical model of wind speed, wind direction, and turbulence intensity at the center of the impeller is established. Factors such as the rotation effect of the impeller, the blocking effect of the wind wheel, and the wake effect are considered. The wind speed distribution model formula is: ,in is the hub center wind speed.

[0105] The model parameters were calibrated through historical data, including the impeller radius of 20 meters, the number of blades of 3, the blade length of 40 meters, the airfoil parameters of NACA 63-418 airfoil, the hub center height of 80 meters, etc. At the same time, the empirical coefficients and correction factors of the model were adjusted to ensure the accuracy and stability of the model.

[0106] The aerodynamic performance model is established based on the geometric parameters and aerodynamic characteristics of the blades, combined with the blade element momentum theory. The torque calculation formula in the blade element momentum theory is: , ,in The air density is 1.2 kg / m3. is the incoming flow velocity at the blade element, is the blade element equivalent area, and are the lift and drag coefficients of the airfoil, It is The pitch angle of the blades.

[0107] The input parameters are blade length (initial length is 40 meters, design length range is 38-42 meters), airfoil (NACA 63-418 airfoil, including airfoil name, parameters and performance curve), angle of attack (initial angle of attack is 5°, design angle of attack range is 3°-7°), tip speed ratio (initial tip speed ratio is 8, design tip speed ratio range is 7-9), air density is 1.2 kg / m3, the incoming flow velocity at the blade element is calculated based on the wind speed distribution model and the wind speed at the center of the impeller, and the blade element equivalent area is calculated based on the blade geometry parameters. The output parameters are the optimized blade parameters, optimized blade length, airfoil, angle of attack and tip speed ratio.

[0108] 5. PID controller pitch control The input parameters are the optimized impeller center wind speed (data including instantaneous wind speed, average wind speed, maximum wind speed and minimum wind speed are obtained after Kalman filter processing), the optimized impeller center wind direction (the optimized impeller center wind direction is obtained by Kalman filter processing), and the optimized impeller center turbulence intensity (the optimized impeller center turbulence intensity is obtained by Kalman filter processing including standard deviation, variance and coefficient of variation).

[0109] The output parameter is the pitch angle of each blade, accurate to 0.1 degree.

[0110] The PID control method is used to adjust the pitch angle of each blade according to the predicted wind speed, wind direction, and turbulence intensity to achieve the optimal power generation efficiency while minimizing the mechanical load. The proportional (P), integral (I), and differential (D) parameters of the PID controller are adjusted by trial and error. After multiple tests, P = 0.5, I = 0.1, and D = 0.05 are determined. Through simulation and actual machine testing, the sampling time of the PID controller is adjusted to 0.1 seconds and the control cycle is 1 second to adapt to different wind speed changes and system response requirements.

[0111] III. Implementation Effect Evaluation After implementing the wind turbine load reduction method, the implementation effect can be evaluated in the following ways: 1. Monitor the power generation and mechanical load of the wind turbine. Compare the power generation and mechanical load changes before and after implementation to evaluate the load reduction effect. If the power generation remains stable or increases slightly, and the mechanical load is significantly reduced, the load reduction method is effective.

[0112] 2. Analyze the operating status of the blades. Observe the vibration, fatigue life and other indicators of the blades to determine whether the blades run more smoothly and their life is extended under the optimized parameters.

[0113] 3. Compare performance under different wind conditions. Test the performance of wind turbines under different wind speeds, wind directions and turbulence intensities, and evaluate the adaptability and effectiveness of load reduction methods under various wind conditions.

[0114] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A method for reducing load of a wind turbine based on laser radar, characterized in that The following steps are involved: Step 1: Use laser radar to measure wind field parameters; the wind field parameters include wind speed, wind direction, and turbulence intensity; Step 2: Establish an RBF neural network model and use the RBF neural network model to analyze and process wind field parameters; Step 3: Train the RBF neural network model; Step 4: predict wind field parameters; Step 5: Establish a physical model of the fan blade; Step 6: Regulate the pitch of the wind turbine based on the PID controller.

2. The method for reducing load of a wind turbine based on laser radar according to claim 1, characterized in that: In step 1, the wind field parameters are measured as follows: (1) Wind speed: The laser radar measures wind speed in the horizontal and vertical directions to obtain the wind speed distribution in different directions and heights. At the same time, it measures the instantaneous value, maximum value, minimum value and average value of the wind speed. Reflects the wind speed at the current moment, the average wind speed The average wind speed over a certain period of time. and minimum wind speed They represent the extreme wind speed in this time period respectively; (2) Wind direction: The lidar determines the wind direction by measuring the direction of the wind and recording the frequency and amplitude of wind direction changes. The wind direction is measured in degrees, rotating clockwise from 0°, and is recorded as ; (3) Turbulence intensity: The standard deviation, variance and coefficient of variation of wind speed measured by lidar reflect the turbulence intensity. The turbulence intensity is expressed as standard deviation, denoted as , and gives the variance and coefficient of variation .

3. The method for reducing load of a wind turbine based on laser radar according to claim 1, characterized in that: In step 2, the method of analyzing and processing the wind field parameters using the RBF neural network model is as follows: (1) the RBF neural network model includes: an input layer, a hidden layer and an output layer; Input layer: receives input parameters, including historical wind speed data, i.e. , , , ; Wind direction data ; Turbulence intensity data, i.e. , , ; Hidden layer: composed of multiple radial basis function RBF neurons; the output of each RBF neuron is a function of the distance between the input vector and the center vector, using the Gaussian function as the activation function: ， in is the input vector, It is The center vector of each RBF neuron, It is The width parameter of each RBF neuron; Output layer: Output prediction parameters, including predicted wind speed, i.e. , , , ; Predict wind direction ; predict turbulence intensity , , ; (2) Output layer calculation formula: The output of the output layer is a linear combination of the outputs of the hidden layers: For predicted wind speed: ， in is the number of neurons in the hidden layer, Is the connection The weights of the hidden layer neurons and the instantaneous wind speed output node, is the bias term of the instantaneous wind speed output node; For predicted wind direction: ， in Is the connection The weights of the hidden layer neurons and the wind direction output node, is the bias term of the wind direction output node; For predicting turbulence intensity: ， ， ， in , , They are connected The weights of the hidden layer neurons and the standard deviation, variance, and coefficient of variation output nodes, , , They are the bias terms of the standard deviation, variance, and coefficient of variation output nodes respectively.

4. The method for reducing load of a wind turbine based on laser radar according to claim 1, characterized in that: In step 3, the process of training the RBF neural network model is as follows: (1) Establish a data set: historical wind speed, wind direction, turbulence intensity data, and corresponding timestamps; divide the data set into training set, validation set, and test set according to a certain ratio; (2) Optimization objective: The goal is to minimize the mean square error (MSE) and mean absolute error (MAE) between the predicted value and the actual value. The MSE calculation formula is: ,in is the sample size, is the actual value, is the predicted value; the MAE calculation formula is ; (3) Parameter adjustment: The stochastic gradient descent method is used to adjust the weights from the hidden layer to the output layer, and the K-Means clustering algorithm is used to select the center of the basis function. At the same time, the number of hidden layer neurons and the activation function of the RBF neural network are adjusted to improve the performance and generalization ability of the model.

5. The method for reducing load of a wind turbine based on laser radar according to claim 1, characterized in that: In step 4, the method for predicting wind field parameters is as follows: (1) Input parameters: Predicted wind speed: RBF neural network predicts wind speed, including instantaneous wind speed , average wind speed , Maximum wind speed and minimum wind speed ; Predicted wind direction: The predicted wind direction of the RBF neural network is denoted as ; Predicted turbulence intensity: RBF neural network predicted turbulence intensity, including standard deviation ,variance and coefficient of variation ; (2) Output parameters: Impeller center wind speed: predicted wind speed at the center of the impeller, including instantaneous wind speed , average wind speed , Maximum wind speed and minimum wind speed ; Impeller center wind direction: The predicted wind direction at the impeller center, denoted as ; Impeller Center Turbulence Intensity: Predicted turbulence intensity at the impeller center, including standard deviation ,variance and coefficient of variation .

6. The method for reducing load of a wind turbine based on laser radar according to claim 1, characterized in that: In step 5, the method for establishing the physical model of the wind turbine blade is as follows: (1) based on the principles of fluid mechanics and aerodynamics, combined with the wind speed distribution model, a mathematical model of the wind speed, wind direction, and turbulence intensity at the center of the impeller is established; The wind speed distribution model formula is: ， in, is the hub center wind speed. Due to the elimination of wind shear factors, only part; is the wind shear component, It is the tower shadow effect component; (2) Parameter adjustment: Calibrate model parameters using historical data, including impeller radius, number of blades, blade length, airfoil parameters, and hub center height; at the same time, adjust the model's empirical coefficients and correction factors to ensure the accuracy and stability of the model; (3) Establish an aerodynamic performance model: Based on the geometric parameters and aerodynamic characteristics of the blade and combined with the blade element momentum theory, an aerodynamic performance model is established; the relationship between the lift coefficient and drag coefficient of the blade and the blade length, airfoil, angle of attack and tip speed ratio is considered; The formula for calculating the torque in blade element momentum theory is: ， ， in, is the air density, is the incoming flow velocity at the blade element, is the blade element equivalent area, and are the lift and drag coefficients of the airfoil, It is The pitch angle of the blades; (4) Input parameters: Blade length: the initial length and design length range of the blade; Airfoil: Different types of airfoils, including their names, parameters and performance curves; Angle of attack: initial angle of attack and design angle of attack range of the blade; Tip speed ratio: the initial tip speed ratio of the blade and the design tip speed ratio range; Air density: the actual measured air density; The incoming flow velocity at the blade element is calculated based on the wind speed distribution model and the wind speed parameters at the impeller center, and is recorded as ; Leaf element equivalent area: calculated based on the leaf geometric parameters, denoted as ; (5) Output parameters: Optimized blade parameters: optimized blade length, airfoil, angle of attack and tip speed ratio.

7. The method for reducing load of a wind turbine based on laser radar according to claim 1, characterized in that: In step 6, the method for adjusting the pitch of the wind turbine based on the PID controller is as follows: (1) Input parameters: Optimized impeller center wind speed: After being processed by the Kalman filter, data including instantaneous wind speed, average wind speed, maximum wind speed and minimum wind speed are obtained; Optimized impeller center wind direction: The optimized impeller center wind direction obtained by Kalman filter processing is recorded as ; Optimized impeller center turbulence intensity: The optimized impeller center turbulence intensity obtained by Kalman filter processing includes standard deviation, variance and coefficient of variation; (2) Output parameters: Pitch angle: the pitch angle of each blade; (3) Control methods: PID control: According to the predicted wind speed, wind direction, and turbulence intensity, the pitch angle of each blade is adjusted to achieve the best power generation efficiency while minimizing the mechanical load. The proportional, integral, and differential parameters of the PID controller are adjusted by trial and error or the Ziegler-Nichols method. Parameter adjustment: Through simulation and actual machine testing, the proportional, integral and differential parameters of the PID controller are adjusted to ensure the robustness and response speed of the control system; at the same time, the sampling time and control cycle of the PID controller are adjusted to adapt to different wind speed changes and system response requirements.

8. The method for reducing load of a wind turbine based on laser radar according to claim 1, characterized in that: In step 1, when measuring wind speed, the laser radar measures angles in the horizontal direction at least including 0°, 90°, 180°, and 270°. When measuring in the vertical direction, the laser radar measures wind speed at every interval of a fixed unit height to obtain the wind speed distribution in different directions and heights.

9. A wind turbine load reduction regulation system based on laser radar, characterized in that: include: Laser radar: used to measure wind field parameters; the wind field parameters include wind speed, wind direction, and turbulence intensity; Data analysis and processing unit: used to establish an RBF neural network model, use the RBF neural network model to analyze and process wind field parameters; and train the RBF neural network model at the same time; Early warning and prediction unit: used to predict wind field parameters; Modeling unit: used to establish the physical model of the wind turbine blades; PID controller-based actuator: used to adjust the pitch of the wind turbine based on the PID controller.

10. A computer device, characterized in that: include: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, a laser radar-based wind turbine load reduction method as described in any one of claims 1 to 8 is implemented.

Images