Yaw-to-wind error compensation method and system based on wind measurement data

Abstract

The present invention discloses a yaw error compensation method and system based on wind measurement data, which relates to the field of wind power generation equipment control technology. The method includes: setting wind measurement points to collect original wind field data sets; calculating the incoming flow and the nacelle wind direction angle based on the original wind field data sets; performing deviation calculation to obtain the yaw error angle; triggering a dual-channel compensation module, accumulating errors to build an error queue and performing batch compensation to obtain compensation results; retrieving yaw control parameters, analyzing the nacelle attitude based on the compensation results, updating the yaw control parameters and realizing adaptive compensation of yaw error. The present invention solves the technical problem that during the operation of a wind turbine, due to environmental factors such as wake, terrain, turbulence, and the loss of the equipment itself, the traditional yaw system is unable to accurately detect wind direction and compensate for wind error, resulting in low power generation efficiency and great equipment safety risks. The present invention achieves the technical effect of accurately improving the wind rotor wind accuracy, significantly improving power generation efficiency, and enhancing equipment operation safety.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation equipment control, and in particular to a yaw-to-wind error compensation method and system based on wind measurement data. Background Art

[0002] During wind power generation, the yaw system needs to adjust the nacelle to align the wind turbine's rotor with the wind direction in order to improve power generation efficiency and ensure the safety of the unit. In the existing technology, traditional yaw wind control mostly uses single-point wind measurement and a single compensation strategy, which exposes obvious limitations in complex wind field environments. Since wind turbines are affected by factors such as wake, terrain, and turbulence during operation, traditional methods are unable to accurately process multi-source environmental data, resulting in a large deviation between the calculated incoming wind direction angle and the nacelle wind direction angle, delayed wind error compensation, and difficulty in adaptively updating control parameters. In addition, the yaw brake is prone to oil leakage and other faults during long-term operation, which further affects the wind accuracy, resulting in low power generation efficiency and increased maintenance costs. The existing technology is unable to effectively solve the problems of wind measurement errors and dynamic compensation in complex environments, making it difficult to meet the needs of efficient and safe operation of wind turbines. Summary of the Invention

[0003] The present application provides a yaw error compensation method and system based on wind measurement data, which is used to solve the technical problem that during the operation of wind turbines, due to environmental factors such as wake, terrain, turbulence and the equipment's own losses, the traditional yaw system is unable to accurately detect wind direction and compensate for wind error, resulting in low power generation efficiency and great equipment safety hazards.

[0004] A first aspect of the present application provides a yaw error compensation method based on wind measurement data, the method comprising: setting a wind measurement point based on a wind turbine and starting a wind measurement device to collect data to obtain an original wind field data set; performing wind field inflow calculation based on the original wind field data to determine the inflow wind direction angle, and performing nacelle wind direction calculation based on the original wind field data to determine the nacelle wind direction angle; performing deviation calculation based on the inflow wind direction angle and the nacelle wind direction angle to obtain a yaw error angle; triggering a dual-channel compensation module according to the yaw error angle, accumulating errors of the yaw error angle, constructing an error queue, synchronizing the error queue to the dual-channel compensation module for batch compensation to obtain compensation results; retrieving yaw control parameters, performing attitude analysis on the wind turbine according to the compensation results to obtain nacelle attitude data, and updating the yaw control parameters according to the nacelle attitude data to achieve adaptive compensation of yaw error.

[0005] In a possible implementation, a wake impact analysis is performed based on the wind turbine to obtain a wake length parameter; the hub height data of the wind turbine is retrieved, and a two-dimensional coordinate system is constructed with the hub center as the origin, the hub height as the vertical axis, and the horizon as the horizontal axis; the rotor rotation plane of the wind turbine is projected onto the two-dimensional coordinate system to determine the plane projection direction; and the wind measurement point is set according to the plane projection direction in combination with the wake length parameter.

[0006] In a possible implementation, a first wind measurement point and a second wind measurement point are extracted based on the wind measurement point, wherein the first wind measurement point includes a laser radar device and the second wind measurement point includes an ultrasonic anemometer; based on the first wind measurement point, the laser radar device is activated to perform spiral scanning to obtain first wind field data; based on the second wind measurement point, the ultrasonic anemometer is activated to perform high-frequency sampling to obtain second wind field data; the first wind field data and the second wind field data are coordinate-uniformed to generate a three-dimensional standard data set; and data filtering is performed on the three-dimensional standard data set to obtain the original wind field data set.

[0007] In a possible implementation, the direction vector of the main wind direction is calculated based on the two-dimensional coordinate system to determine the main wind direction projection line; a digital elevation grid is constructed with the wind turbine as the center, and calculation is performed according to the digital elevation grid in combination with the main wind direction projection line to obtain terrain gradient parameters; terrain offset compensation is performed based on the terrain gradient parameters to obtain a dynamic terrain correction amount; an error angle core calculation is performed based on the dynamic terrain correction amount to determine the error angle; high-frequency wind speed data is extracted based on the wind measurement point, and the turbulence three-dimensional component is solved based on the high-frequency wind speed data to obtain the turbulence intensity; the error angle is adaptively attenuated based on the turbulence intensity to obtain the yaw to wind error angle.

[0008] In a possible implementation, a dynamic analysis is performed based on the yaw error angle according to the wind speed, a preset dynamic threshold is constructed, and a channel activation rule is formulated according to the preset dynamic threshold; the dual-channel compensation module is triggered according to the channel activation rule, and the dual-channel compensation module includes an active compensation channel and a passive compensation channel; when the yaw error angle is greater than the preset dynamic threshold, the active compensation channel is triggered; dynamic compensation adjustment is performed through the active compensation channel to obtain a first compensation result; when the yaw error angle is less than or equal to the preset dynamic threshold, the yaw error angle is accumulated as an error to construct an error queue; the error queue is synchronized to the passive compensation channel for dynamic batch compensation to obtain a second compensation result.

[0009] In a possible implementation, the error queue is synchronized to the passive compensation channel for cumulative calculation to obtain queue cumulative energy; equivalent calculation is performed based on the queue cumulative energy to obtain an equivalent compensation angle; torque analysis is performed according to the equivalent compensation angle to generate an equivalent torque instruction; the equivalent torque instruction is executed to perform dynamic batch compensation to obtain the second compensation result.

[0010] In a possible implementation, attitude oscillation calculation is performed on the wind turbine according to the compensation result to obtain attitude oscillation energy data; attitude stability evaluation is performed based on the attitude oscillation energy data to generate an attitude stability score; and frequency band interference elimination is performed according to the attitude stability score to obtain the cabin attitude data.

[0011] In a possible implementation, the generated power of the wind turbine is introduced, and the increase is calculated based on the generated power to obtain the generated power improvement rate; a multi-objective analysis is performed based on the cabin attitude data combined with the generated power improvement rate to construct a multi-objective optimization expected value; according to the multi-objective optimization expected value, the yaw control parameters are updated through reinforcement learning to generate adaptive yaw control parameters, and adaptive dual-domain synchronous compensation of yaw to wind error is performed according to the adaptive yaw control parameters.

[0012] In a possible implementation, reinforcement learning is performed based on the multi-objective optimization expectation value to construct a reinforcement learning state space; power generation environment analysis is performed based on the wind turbine to construct an environmental reward function; state vector analysis is performed according to the environmental reward function based on the reinforcement learning state space to obtain parameter increments, and the yaw control parameters are updated according to the parameter increments to generate adaptive yaw control parameters; time-frequency domain compensation is performed according to the adaptive yaw control parameters to generate dual-domain synchronous compensation results; performance verification is performed on the dual-domain synchronous compensation results, and when the verification passes, data is rolled back to determine yaw control stability parameters to achieve adaptive compensation of yaw to wind error.

[0013] A second aspect of the present application provides a yaw wind error compensation system based on wind measurement data, the system comprising: an original data set acquisition module, which sets a wind measurement point on the wind turbine and starts the wind measurement equipment to collect data to obtain an original wind field data set; a nacelle wind direction angle acquisition module, which performs wind field inflow calculation based on the original wind field data to determine the inflow wind direction angle, and performs nacelle wind direction calculation based on the original wind field data to determine the nacelle wind direction angle; a wind error angle acquisition module, which performs deviation calculation based on the inflow wind direction angle and the nacelle wind direction angle to obtain a yaw wind error angle; a compensation result acquisition module, which triggers a dual-channel compensation module according to the yaw wind error angle, accumulates errors of the yaw wind error angle, constructs an error queue, synchronizes the error queue to the dual-channel compensation module for batch compensation, and obtains compensation results; an adaptive compensation execution module, which retrieves yaw control parameters, performs attitude analysis on the wind turbine according to the compensation result, obtains nacelle attitude data, updates the yaw control parameters according to the nacelle attitude data, and realizes adaptive compensation of yaw wind error.

[0014] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0015] This application sets up wind measurement points around the wind turbine, uses lidar and ultrasonic anemometer to collect wind field data, obtains wind direction angle data through coordinate unification, terrain and turbulence correction, calculates the yaw error angle to the wind and triggers the dual-channel compensation module, combines error accumulation with reinforcement learning to optimize control parameters, and thus accurately compensates for the yaw error of the wind turbine, significantly improving the wind rotor's wind accuracy and power generation efficiency, achieving the technical effect of accurately improving the wind rotor's wind accuracy, significantly improving power generation efficiency and enhancing equipment operation safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 It is a flow chart of a method for compensating for yaw-to-wind error based on wind measurement data provided in an embodiment of the present application.

[0018] Figure 2 It is a structural diagram of a yaw-to-wind error compensation system based on wind measurement data provided in an embodiment of the present application.

[0019] Description of reference numerals: original data set acquisition module 1, cabin wind direction angle acquisition module 2, wind error angle acquisition module 3, compensation result acquisition module 4, adaptive compensation execution module 5. DETAILED DESCRIPTION

[0020] The present application provides a yaw error compensation method and system based on wind measurement data, which is used to solve the technical problem that during the operation of wind turbines, due to environmental factors such as wake, terrain, turbulence and the equipment's own losses, the traditional yaw system is unable to accurately detect wind direction and compensate for wind error, resulting in low power generation efficiency and great equipment safety hazards.

[0021] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0022] It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices.

[0023] Example 1, as Figure 1 As shown, a yaw-to-wind error compensation method based on wind measurement data, wherein the method includes:

[0024] Step A100: Setting wind measurement points based on wind turbines and starting wind measurement equipment to collect data to obtain an original wind field data set.

[0025] Specifically, a wake impact analysis is performed on the wind turbine to obtain the wake length parameter. A two-dimensional coordinate system is constructed with the hub center as the origin, the hub height as the vertical axis, and the horizon as the horizontal axis. The rotor rotation plane is projected onto the coordinate system to determine the projection direction. The wind measurement point is then set based on the wake length parameter. The specific steps are detailed in A110-A140.

[0026] Extract the first wind measurement point containing the lidar and the second wind measurement point containing the ultrasonic anemometer from the wind measurement points. Activate the devices for spiral scanning and high-frequency sampling to obtain wind field data, respectively. Unify the coordinates of the two types of data to generate a three-dimensional standard data set and filter it to obtain the original wind field data set. The specific steps are detailed in A150-A190.

[0027] Step A200: performing wind field incoming flow calculation based on the original wind field data to determine the incoming flow wind direction angle, performing nacelle wind direction calculation based on the original wind field data to determine the nacelle wind direction angle.

[0028] In the embodiment of the present application, the wind field inflow refers to the natural wind flow state reaching the area where the wind turbine is located. Its core parameters include wind direction angle, wind speed and turbulence characteristics, which are the key basis for the yaw system to adjust the direction of the cabin to the wind.

[0029] Optionally, when calculating the wind field flow and cabin wind direction based on the original wind field data set that has been obtained, the wind direction angle needs to be accurately determined in two steps. First, for the wind field flow calculation, the three-dimensional wind field data obtained by the spiral scanning of the lidar and the high-frequency sampling data of the ultrasonic anemometer are used to calculate the wind direction angle by the vector synthesis method: the three-dimensional wind speed components are projected into a two-dimensional coordinate system with the hub center as the origin, with the vertical axis being the hub height and the horizontal axis being the horizon. Calculate the horizontal wind direction component angle, where u represents the wind speed component of the horizontal axis (horizontal direction), and v represents the wind speed component of the horizontal direction (vertical axis) perpendicular to the horizontal axis. Combined with the vertical wind speed component w, perform vector synthesis of the three-dimensional wind direction to obtain the initial value of the incoming wind direction angle. On this basis, introduce digital elevation grid data (resolution 10m×10m) to calculate the terrain gradient parameter, and use the terrain gradient to offset the wind direction. , where k is the terrain influence coefficient, α is the terrain slope angle, the initial wind direction angle is corrected, and the high-frequency wind speed data is extracted to solve the standard deviation of the three-dimensional turbulence component and calculate the turbulence intensity , σ is the standard deviation of wind speed, is the average wind speed, and the turbulence intensity is used to adaptively attenuate the wind direction angle. The attenuation coefficient , and finally determine the corrected incoming wind direction angle.

[0030] Next, the calculation of the cabin wind direction angle is based on the current cabin attitude data. The cabin yaw angle β relative to the true north direction is obtained through the yaw system encoder. Combined with the real-time measurement value of the wind vane sensor installed in the cabin, its sampling frequency is 10Hz, and a sliding window filter with a window length of 5s is used to eliminate short-term wind direction fluctuations to obtain the actual windward direction angle of the cabin. In addition, the cabin attitude oscillation energy data needs to be introduced. It can be calculated through the three-dimensional vibration data collected by the acceleration sensor. When the attitude oscillation energy exceeds the threshold (such as 0.5 ), dynamically correct the wind direction angle measurement value, the correction amount , λ is the correction coefficient, It is the attitude oscillation energy, ensuring the measurement accuracy of the cabin wind direction angle.

[0031] Through multi-source data fusion, quantitative correction of terrain and turbulence, and dynamic calibration of cabin attitude, the problems of insufficient wind measurement accuracy of a single sensor and failure to consider interference from environmental factors in existing technologies have been solved. The calculation accuracy of the incoming wind direction angle and the cabin wind direction angle has been improved, providing key data support for the accurate calculation of yaw wind error.

[0032] Step A300: Calculate the deviation between the incoming wind direction angle and the nacelle wind direction angle to obtain a yaw-to-wind error angle.

[0033] In one embodiment of the present application, the main wind direction vector is calculated based on a two-dimensional coordinate system and the projection line is determined. A digital elevation grid is constructed with the wind turbine as the center to calculate the terrain gradient parameters. After the dynamic correction amount is obtained through terrain offset compensation, the error angle core calculation is performed. The high-frequency wind speed data at the wind measuring point is used to solve the adaptive attenuation of the turbulence intensity to the error angle to obtain the yaw-to-wind error angle. The specific steps are described in detail in A310-A360.

[0034] Step A400: triggering a dual-channel compensation module according to the yaw-to-wind error angle, accumulating the error of the yaw-to-wind error angle, building an error queue, synchronizing the error queue to the dual-channel compensation module for batch compensation, and obtaining a compensation result.

[0035] In the embodiment of the present application, the dual-channel compensation module is a core functional module for realizing dynamic compensation of yaw-to-wind error, which includes an active compensation channel and a passive compensation channel.

[0036] Specifically, the dual-channel compensation module is triggered according to the comparison result of the yaw error angle to wind and the preset threshold value constructed based on the dynamic analysis of wind speed: when the error angle is greater than the threshold, the active compensation channel is activated to obtain the first compensation result; otherwise, the accumulated errors are queued and synchronized to the passive compensation channel to obtain the second compensation result. The specific steps are described in detail in A410-A460.

[0037] Step A500: Retrieve yaw control parameters, perform attitude analysis on the wind turbine according to the compensation result, obtain nacelle attitude data, update the yaw control parameters according to the nacelle attitude data, and implement adaptive compensation of yaw to wind error.

[0038] In the embodiment of the present application, the yaw control parameters are key control variables used to adjust the attitude of the wind turbine nacelle to the wind, including yaw angular velocity, braking torque, attitude adjustment threshold, etc.

[0039] Specifically, based on the compensation result, the attitude oscillation of the wind turbine is calculated to obtain energy data, and the attitude stability is evaluated to generate a score. Then, the nacelle attitude data is obtained through frequency band interference elimination processing. The specific steps are described in detail in A510-A530.

[0040] Then, the power generation improvement rate is introduced, and a multi-objective analysis is performed on the cabin attitude data to construct the optimization expectation value. The yaw control parameters are updated through reinforcement learning to generate adaptive parameters, realizing the time-frequency domain synchronous adaptive compensation of yaw to wind error. The specific steps are described in detail in A540-A560.

[0041] Furthermore, step A100 in the method provided in the embodiment of the present application includes:

[0042] A110: Perform wake impact analysis based on wind turbines and obtain wake length parameters.

[0043] A120: Retrieve the hub height data of the wind turbine, construct a two-dimensional coordinate system with the hub center as the origin, the hub height as the vertical axis, and the horizon as the horizontal axis.

[0044] A130: Projecting the rotor rotation plane of the wind turbine onto the two-dimensional coordinate system to determine the plane projection direction.

[0045] A140: Setting the wind measurement point according to the plane projection direction and the wake length parameter.

[0046] In the embodiment of the present application, the wake effect refers to the airflow disturbance area formed behind the wind rotor after it rotates to capture wind energy. The wind direction and wind speed distribution in this area will deviate from the original wind field characteristics. This disturbance will cause deviations in the data obtained by the traditional single-point wind measurement method, thereby affecting the yaw system's judgment of the wind rotor's wind angle.

[0047] Specifically, first, the wake impact of the wind turbine is analyzed using the formula Calculate the wake length, where is the wake length, D is the rotor diameter, is the thrust coefficient, k is the terrain influence coefficient. This formula is derived based on the classical theory of wake calculation in the field of wind power generation, and comprehensively considers the influence of wind rotor diameter, thrust coefficient and terrain factors on the wake length: the wind rotor diameter D directly determines the physical scale of the wake influence, and the thrust coefficient Reflects the momentum loss of the airflow when the wind wheel captures wind energy, through The term quantifies the air velocity attenuation, while The first part is derived from the principle of conservation of momentum and is used to characterize the nonlinear effect of the thrust coefficient on the wake extension length. The terrain influence coefficient k is empirically corrected by technicians in this field based on the actual terrain such as flat land and mountainous areas that hinder or accelerate the diffusion of wake. Finally, the wake length is calculated using this formula. This formula comprehensively considers the influence of rotor diameter, thrust coefficient and terrain influence coefficient on the wake length, thereby obtaining an accurate wake length. , providing key data support for the subsequent setting of wind measurement points.

[0048] Next, retrieve the hub height data of the wind turbine, that is, the vertical distance from the center of the wind rotor to the ground, take the hub center as the origin of the two-dimensional coordinate system, set the hub height direction as the longitudinal axis (Z axis), and its positive direction is vertically upward, which can accurately reflect the spatial position of the wind rotor in the vertical direction; at the same time, set the plane where the horizon is located as the horizontal axis (X axis), and its positive direction is to the left or right along the horizontal plane, which can be determined according to the conventional direction of the actual wind direction, and construct a two-dimensional coordinate system with the hub center as the reference, to ensure that the spatial positioning of the wind measurement point has an accurate geometric correspondence with the actual windward direction of the wind rotor.

[0049] When projecting the wind turbine's rotor rotation plane—the circular plane perpendicular to the rotor's main axis formed by the rotating blades—onto a two-dimensional coordinate system with the hub center as its origin, it's necessary to first determine the plane's spatial orientation in three-dimensional space: the rotor's main axis typically has a certain inclination angle with the horizontal plane. For example, the main axis of a conventional horizontal-axis wind turbine is perpendicular to the horizon, so the rotor's rotation plane can be considered a standard circular surface perpendicular to the main axis. In a two-dimensional coordinate system with the hub height as the vertical axis and the horizon as the horizontal axis, this three-dimensional plane is orthographically projected in a direction perpendicular to the coordinate system's plane (i.e., perpendicular to the paper, inward or outward). This straight line segment forms a straight line segment in the two-dimensional coordinate system, and the direction in which this straight line segment extends is the plane's projection direction. This direction essentially reflects the orientation of the rotor's main axis in the horizontal plane, i.e., the rotor's current theoretical upwind orientation. Through this projection step, the windward direction of the wind rotor in the three-dimensional space can be converted into a linear direction parameter in the two-dimensional coordinate system, thereby providing an accurate direction reference for the horizontal layout of the wind measuring points. The wind measuring points need to be arranged along this projection direction to ensure that the detection axis of the lidar and ultrasonic anemometer is parallel to the windward direction of the wind rotor, so as to accurately capture the wind speed and direction data of the incoming flow in front of the wind rotor and avoid wind measurement errors caused by azimuth deviation.

[0050] Finally, according to the determined plane projection direction, combined with the wake length Set the wind measurement point. The first wind measurement point is set at 0.3 to 0.5 times The second wind measuring point is set at 0.7 to 0.9 times The setting of these two points is based on the quantitative analysis of the wake length by those skilled in the art, 0.3 to 0.5 times It can avoid the interference of the strong wake area near the wind rotor and obtain relatively pure incoming flow data, while 0.7 to 0.9 times At the same time, it can take into account the monitoring of far-field wake effects and achieve coverage of different areas of the wind farm.

[0051] By quantitatively analyzing the impact of wake vortexes, constructing a standardized coordinate system, accurately projecting the plane direction of the wind rotor, and scientifically defining wind measurement points based on the wake length, the interference of wind turbine wake vortexes on wind measurement data is effectively avoided, ensuring the accuracy and representativeness of data obtained at wind measurement points, and laying a reliable data foundation for the subsequent precise calculation and compensation of yaw-to-wind errors.

[0052] Furthermore, step A100 in the method provided in the embodiment of the present application includes:

[0053] A150: Extracting a first wind measurement point and a second wind measurement point based on the wind measurement point, where the first wind measurement point includes a laser radar device and the second wind measurement point includes an ultrasonic anemometer.

[0054] A160: Activate the laser radar device based on the first wind measurement point to perform spiral scanning to obtain first wind field data.

[0055] A170: Activate the ultrasonic anemometer based on the second wind measurement point to perform high-frequency sampling to obtain second wind field data.

[0056] A180: Perform coordinate unification processing on the first wind field data and the second wind field data to generate a three-dimensional standard data set.

[0057] A190: Perform data filtering on the three-dimensional standard data set to obtain the original wind field data set.

[0058] Optionally, when the wind measurement equipment is activated for data collection, the first and second wind measurement points are first extracted based on the previously set wind measurement points. The first wind measurement point is deployed with a lidar device, and the second wind measurement point is equipped with an ultrasonic anemometer. This collaborative deployment of two devices overcomes the limitations of traditional single-point wind measurement, which relies on a single sensor. While lidar has the ability to scan a wide range of wind fields, it is not good at capturing high-frequency turbulence characteristics. While ultrasonic anemometers can sample at high frequencies, their scanning range is limited. The combination of the two forms complementary data on the macroscopic wind field structure and microscopic turbulence characteristics.

[0059] For the lidar equipment at the first wind measurement point, activate the equipment to perform a spiral scanning action. Specifically, with the hub height as the scanning center, the cone angle range of ±15° is covered with an angular resolution of 0.5° in the vertical plane. The setting of this scanning parameter is based on the actual windward angle range of the wind rotor. The ±15° cone angle can cover the main wind direction when the wind rotor is operating normally. The 0.5° angular resolution ensures the spatial accuracy of the three-dimensional data of the wind field. Compared with the angular resolution of more than 1° of traditional lidar, it can capture the wind field gradient changes more finely, thereby obtaining the first wind field data containing the three-dimensional components of wind speed and wind direction.

[0060] The ultrasonic anemometer at the second wind measurement point collects three-dimensional wind speed components at a high frequency of 100 Hz. The 100Hz sampling frequency is determined based on the characteristic time scale of turbulence. Typical atmospheric turbulence frequencies range from 0.1Hz to 100Hz. This sampling rate effectively captures the high-frequency components of turbulent fluctuations. Conventional sensors with sampling frequencies below 100Hz are prone to distorting turbulence data. Simultaneously collecting three-dimensional wind speed components and temperature data provides the foundation for subsequent turbulence intensity calculations and temperature-induced sound velocity corrections.

[0061] After obtaining the above two types of wind field data, coordinate unification processing is required to solve the problem of multi-source data fusion errors. Since the installation positions of the lidar and ultrasonic anemometer are respectively located at the first wind measurement point and the second wind measurement point, their respective coordinate systems have offset and rotation differences from the spatial coordinate system of the actual operation of the wind turbine. The lidar coordinate system uses the installation position of the equipment itself as the origin, and the ultrasonic anemometer coordinate system uses its deployment point as the reference. The X, Y, and Z axis directions of the two may deviate from the hub center coordinate system due to different installation angles. When establishing a three-dimensional coordinate transformation matrix, first construct a unified coordinate system with the hub center as the origin, the hub height as the Z axis, and the horizon as the X axis, and then implement data mapping through the following steps:

[0062] Step a: Align the reference points and measure the three-dimensional coordinate offset of the installation positions of the lidar and ultrasonic anemometer relative to the center of the hub .

[0063] Step b: Rotation matrix calculation: Based on the pitch angle, yaw angle, and roll angle of the equipment installation, the Euler angle rotation matrix R is constructed to rotate the coordinate systems of the lidar and ultrasonic anemometer axially to be consistent with the unified coordinate system.

[0064] Step c: Translation transformation, by translation vector Translate the origin of the device coordinate system to the center of the wheel hub.

[0065] Step d: Matrix integration, combining the rotation matrix R and the translation vector T into a homogeneous transformation matrix , coordinate transformation is performed on the spiral scanning data of the lidar and the high-frequency sampling data of the ultrasonic anemometer so that they are all mapped to a unified coordinate system with the hub center as the origin, and finally a three-dimensional standard data set with consistent spatial scale is generated.

[0066] Finally, the Kalman filter algorithm is used to filter the three-dimensional standard data set. The core of the algorithm is to construct a state space model of the wind farm data and achieve recursive optimal estimation:

[0067] Step e: Define the state vector as the three-dimensional wind speed component and wind direction angle, establish a state transfer equation to describe the dynamic changes of the wind field, such as considering the first-order autoregressive process of wind speed pulsation, and construct a measurement equation to correlate the data collected by the lidar and ultrasonic anemometer.

[0068] Step f: When initializing the algorithm, the prior mean and covariance of the state variables are set. The state estimation value at the next moment is inferred using the state transfer equation through the prediction step, and the uncertainty of the natural fluctuation of the wind field is characterized by combining the process noise covariance matrix Q.

[0069] Step g: In the update phase, the actual measured data (three-dimensional wind field data with unified coordinates) is compared with the predicted value, and the Kalman gain is used to calculate the wind speed. , dynamically adjust the weights to make the estimated value optimally balanced between the predicted value and the measured value, where is the prediction covariance matrix, H is the measurement matrix, R is the measurement noise covariance matrix, is the transposed matrix of the measurement matrix H.

[0070] Step h: Then update the state equation Correct the state estimate and update the covariance matrix To iteratively optimize the estimation accuracy of the next moment, It is the best estimate of the wind farm state. Based on the previous moment , the current state predicted by the wind field model. is the raw data collected by the sensor. H is the measurement matrix. P is the updated covariance matrix. I is the identity matrix. is the forecast covariance matrix.

[0071] The Kalman filter algorithm adaptively adjusts the noise covariance matrices Q and R, where Q is set based on the empirical value of turbulence intensity and R is calibrated based on the equipment noise characteristics. It recursively filters the Gaussian noise in the three-dimensional wind speed components sampled at a high frequency of 100 Hz and the lidar scanning data. While removing outliers such as electromagnetic interference, it utilizes the dynamic characteristics of the state-space model to retain the high-frequency components of wind speed pulsations, ultimately improving the data signal-to-noise ratio and generating a high-precision original wind field dataset.

[0072] Through dual-device collaborative data collection, differentiated scanning parameter setting, coordinate unification and intelligent filtering processing, a wind measurement system with macro-micro data fusion was constructed. This not only solves the one-sided problem of traditional single-point wind measurement data, but also improves the spatiotemporal resolution of wind field data through high-precision scanning and high-frequency sampling, providing reliable data support for the subsequent accurate calculation of yaw-to-wind error.

[0073] Furthermore, step A300 in the method provided in the embodiment of the present application includes:

[0074] A310: Calculate the direction vector of the main wind direction based on the two-dimensional coordinate system and determine the main wind direction projection line.

[0075] A320: Construct a digital elevation grid with the wind turbine as the center, and perform calculations based on the digital elevation grid and the main wind direction projection line to obtain terrain gradient parameters.

[0076] A330: Perform terrain offset compensation based on the terrain gradient parameter to obtain a dynamic terrain correction amount.

[0077] A340: Perform error angle core calculation based on the dynamic terrain correction amount to determine the error angle.

[0078] A350: extracting high-frequency wind speed data based on the wind measurement points, and calculating the three-dimensional turbulence component according to the high-frequency wind speed data to obtain turbulence intensity.

[0079] A360: Adaptively attenuate the error angle based on the turbulence intensity to obtain the yaw-to-wind error angle.

[0080] In the embodiment of the present application, the digital elevation grid is formed by regularly arranged two-dimensional coordinate points. and its corresponding elevation value (z) characterize the terrain undulations around the wind turbine and are the basic data structure for calculating terrain gradient parameters.

[0081] Specifically, based on the two-dimensional coordinate system, the original wind field data set after coordinate unification and Kalman filtering is used to calculate the direction vector of the main wind direction using the vector synthesis method: the three-dimensional wind speed component is projected into the two-dimensional coordinate system, and the direction vector of the main wind direction is calculated by The angle of the horizontal wind direction component is calculated, and then multiple sets of wind speed data are weighted averaged. The weight is positively correlated with the wind speed amplitude to eliminate the directional fluctuations caused by turbulent pulsation and obtain the unit direction vector of the main wind direction. Starting from the hub center, the main wind direction projection line is extended along the vector direction. This projection line quantifies the spatial orientation of the mainstream direction of the wind field and provides a reference direction for the subsequent calculation of terrain gradient parameters based on the digital elevation grid (10m×10m resolution).

[0082] Next, a digital elevation grid (10m x 10m resolution) was constructed with the wind turbine as the center, and the terrain gradient parameters were calculated based on the main wind direction projection line. Specifically, the terrain elevation data along the main wind direction projection line was extracted, and the ratio of the elevation difference between adjacent grid points to the horizontal distance was calculated to obtain the terrain slope angle α. The terrain influence coefficient k was then used to quantify the degree of terrain interference with wind direction, such as k = 0.8 for flat land and k = 1.5 for mountainous areas, to obtain the terrain gradient parameters.

[0083] When terrain offset compensation is performed based on terrain gradient parameters, the formula is used. Calculate the dynamic terrain correction, where k is the terrain influence coefficient, which is determined based on the actual terrain, and α is the terrain slope angle. For example, when the mountain terrain slope angle α=10° and k=1.5, ≈2.6°, the dynamic terrain correction is used to compensate for the wind direction deviation caused by the terrain.

[0084] During the core error angle calculation phase, the difference between the incoming wind direction angle and the nacelle wind direction angle is used as the base error angle, and a dynamic terrain correction is added to obtain the initial error angle. The incoming wind direction angle is improved to ±1.5° accuracy after terrain correction. The nacelle wind direction angle is calculated by fusing the yaw encoder and wind vane data (using a 5s sliding window filter), also achieving an accuracy of ±1.5°. The difference between the two calculations is kept within ±3°.

[0085] Further extract 100Hz high-frequency wind speed data from the wind measurement points and calculate the three-dimensional wind speed components The standard deviation σ of ( is the average wind speed) to calculate the turbulence intensity. When the turbulence intensity is 0.2, the wind speed pulsation is significant, and the attenuation coefficient is used =0.96, the error angle is adaptively attenuated, so that the original ±4° error angle fluctuation (total amplitude 8°) is attenuated to ±3.84° (total amplitude 7.68°), and the fluctuation amplitude is reduced by about 4%. In fact, the attenuation effect is more significant as the turbulence intensity increases, and finally an accurate yaw-to-wind error angle is obtained.

[0086] Through multi-dimensional corrections such as main wind direction vector analysis, quantitative compensation of terrain gradient, and adaptive attenuation of turbulence intensity, a complete system for dynamic compensation of environmental factors and precise calculation of error angles has been constructed. This provides an accurate error benchmark for subsequent dual-channel compensation modules and solves the problem of inaccurate wind error calculation caused by environmental factors in traditional yaw systems.

[0087] Furthermore, step A400 in the method provided in the embodiment of the present application includes:

[0088] A410: Perform dynamic analysis based on the yaw-to-wind error angle according to wind speed, construct a preset dynamic threshold, and formulate a channel activation rule according to the preset dynamic threshold.

[0089] A420: Triggering the dual-channel compensation module according to the channel activation rule, where the dual-channel compensation module includes an active compensation channel and a passive compensation channel.

[0090] A430: When the yaw-to-wind error angle is greater than the preset dynamic threshold, trigger the active compensation channel.

[0091] A440: Perform dynamic compensation adjustment through the active compensation channel to obtain a first compensation result.

[0092] A450: When the yaw-to-wind error angle is less than or equal to the preset dynamic threshold, the yaw-to-wind error angle is accumulated as an error to construct an error queue.

[0093] A460: Synchronize the error queue to the passive compensation channel to perform dynamic batch compensation to obtain a second compensation result.

[0094] Specifically, when compensating for the wind error angle based on yaw, it is first necessary to establish a preset dynamic threshold based on the correlation between the error angle and the wind speed, as shown in Table 1. Those skilled in the art can understand from analyzing historical operating data that the wind speed It is negatively correlated with the error compensation sensitivity, so the formula is used (in is the reference threshold, usually 3°, is the wind speed sensitivity coefficient, the value ) Dynamically calculate the threshold. For example, when the wind speed =10m / s, preset dynamic threshold Compared to traditional fixed thresholds, this threshold can more accurately match the compensation requirements under different wind conditions. Based on this threshold, a channel activation rule is formulated: active compensation is triggered when the yaw error angle θ>Th, otherwise the passive compensation process begins.

[0095] The dual-channel compensation module is then triggered according to the activation rules. This module contains two compensation channels, active and passive: when the error angle exceeds the preset dynamic threshold, the active compensation channel is immediately activated and the real-time yaw adjustment value is output through the PID controller. , where Kp=0.8, Ki=0.2, Kd=0.1 are control parameters (proportional, integral and differential coefficients respectively), It is the yaw error angle to the wind, which drives the yaw motor to make dynamic corrections at an adjustment speed of 1° / s. This channel can effectively deal with large error scenarios caused by sudden strong winds.

[0096] Furthermore, when the active compensation channel is triggered based on the yaw-to-wind error angle, the yaw adjustment amount Δβ calculated by the PID controller will directly drive the yaw motor to perform real-time adjustment at a speed of 1° / s. During this process, the system synchronously monitors the change in the error angle: when the error angle starts to decrease from the initial value (such as 5°) and stabilizes below the preset dynamic threshold (such as 3.5°), the compensation is deemed effective. At this time, the final error angle state recorded and the attitude parameters after the yaw system adjustment are the first compensation result.

[0097] For example, when the error angle is 4° and the wind speed is 10m / s, after the active compensation channel is triggered, the PID controller outputs After 4 seconds of adjustment, the error angle is reduced to 3.2°. The error state and motor action parameters after adjustment, such as an adjustment angle of 3.8° and a time of 3.8s, constitute the first compensation result. The error angle converges quickly through real-time dynamic adjustment, and the response efficiency is improved compared to the traditional fixed threshold compensation.

[0098] If the error angle is less than or equal to the preset dynamic threshold, the current error angle is stored in an error queue for accumulation, thus building an error queue. The queue uses a sliding window mechanism with a window length of 10 seconds, corresponding to 1000 sets of data sampled at 100Hz. When the queue is full, it is synchronized to the passive compensation channel.

[0099] The error queue is synchronized to the passive compensation channel for cumulative calculation to obtain the queue cumulative energy. Based on the energy equivalent calculation, the equivalent compensation angle is obtained. Then, the torque analysis is performed based on this to generate the equivalent torque instruction. The instruction is executed to complete the dynamic batch compensation to obtain the second compensation result. The specific steps are described in detail in A461-A464.

[0100] By dividing compensation scenarios by dynamic thresholds and combining the collaborative strategy of active real-time adjustment and passive batch processing, we can solve the problem of fixed threshold compensation lag in existing technologies and avoid equipment wear caused by frequent adjustment of small errors, thus achieving dual optimization of compensation efficiency and equipment reliability.

[0101] Table 1: Dual-channel compensation module threshold and wind speed mapping table

[0102]

[0103] Furthermore, step A460 in the method provided in the embodiment of the present application includes:

[0104] A461: Synchronize the error queue to the passive compensation channel for cumulative calculation to obtain queue cumulative energy.

[0105] A462: Perform equivalent calculation based on the accumulated energy of the queue to obtain an equivalent compensation angle.

[0106] A463: Perform torque analysis according to the equivalent compensation angle and generate an equivalent torque instruction.

[0107] A464: Execute the equivalent torque instruction to perform dynamic batch compensation to obtain the second compensation result.

[0108] Specifically, when synchronizing the error queue to the passive compensation channel for dynamic batch compensation, first, the error queue is accumulated and calculated to obtain the queue cumulative energy. The error queue adopts a 10-second sliding window mechanism, corresponding to 1000 sets of yaw-to-wind error angle data with a high-frequency sampling of 100Hz, and is calculated by the formula , where i = 1 to 1000, For each error angle in the queue, the cumulative energy is calculated. This calculation method uses the square sum of the errors as the energy quantification index, which can better reflect the weight of large errors than the traditional direct summation. For example, when the queue contains 1000 2° error angles, the cumulative energy , providing a quantitative basis for subsequent equivalent calculations.

[0109] Then, when performing equivalent calculation based on the queue accumulated energy, the energy-angle conversion formula is used Convert the accumulated energy into an equivalent compensation angle. The derivation of this formula is based on the principle of energy conservation, assuming that the average impact of 1000 errors is equivalent to the square of a single angle, as mentioned above hour, , which means the equivalent compensation angle is 2°. This step aggregates the scattered error energy into a single angle parameter, solving the problem in existing technologies where small errors cannot effectively trigger compensation.

[0110] Then, in the torque analysis stage, the mechanical characteristic parameters of the wind turbine, such as the yaw system moment of inertia, are combined , damping coefficient , through the formula Calculate the equivalent torque instruction, where angular acceleration (t is the compensation time, 5s), angular velocity .

[0111] when hour, ,

[0112] ,

[0113] Substitution , that is, generate The equivalent torque command can not only effectively compensate for the cumulative error, but also avoid the mechanical shock caused by excessive torque in traditional single compensation.

[0114] Finally, when the equivalent torque command is executed, the yaw system drives the nacelle to rotate with the torque, completing dynamic batch compensation and obtaining the second compensation result.

[0115] Through error energy accumulation, equivalent angle conversion and dynamic torque calculation, a batch compensation system of energy aggregation, equivalent conversion and torque adaptation is constructed. Compared with the existing technology that ignores small errors or frequently compensates single steps, this method reduces the adjustment frequency of the yaw system while ensuring the compensation accuracy, extending the maintenance cycle of the yaw motor, and improving the annual power generation, thus achieving dual optimization of equipment reliability and power generation efficiency.

[0116] Furthermore, step A500 in the method provided in the embodiment of the present application includes:

[0117] A510: Perform attitude oscillation calculation on the wind turbine generator set according to the compensation result to obtain attitude oscillation energy data.

[0118] A520: Perform attitude stability evaluation based on the attitude oscillation energy data to generate an attitude stability score.

[0119] A530: Perform frequency band interference elimination according to the attitude stability score to obtain the cabin attitude data.

[0120] In one embodiment, first, an acceleration sensor with a sampling frequency of 100 Hz is used to collect X / Y / Z axis vibration data, and the formula is used. , where ax / ay / az are the instantaneous acceleration values ​​of each axis, and the attitude oscillation energy data within a unit time (such as 10s) is calculated. For example, when the cabin yaw adjustment causes a vibration acceleration of 0.5g, the accumulated energy within 10s is , this energy value quantifies the strength of the mechanical oscillation caused by the yaw motion.

[0121] Next, when evaluating stability based on attitude oscillation energy data, an energy-score mapping model is established: the baseline energy is set , when the oscillation energy When ≤E0, the score is 100 points. The score is 50 points, more than The time score is 0, and the attitude stability score of 0-100 is generated by linear interpolation. hour, The score directly reflects the stability of the cabin attitude.

[0122] Finally, in the frequency band interference elimination stage, the attitude data is filtered using a fourth-order Butterworth bandpass filter with a passband frequency of 0.1-10 Hz. This frequency band covers the primary mechanical resonance frequency of the yaw system (typically 5-8 Hz) and low-frequency interference caused by turbulence (0.1-2 Hz). By filtering out high-frequency noise above 10 Hz (such as motor electromagnetic interference) and slow drift below 0.1 Hz, pure cabin attitude data is obtained.

[0123] Through quantitative calculation of oscillation energy, stability score mapping and precise frequency band filtering, an attitude data processing system for energy analysis, stability evaluation and interference elimination was constructed. Compared with the crude processing method in the existing technology that does not consider oscillation energy and frequency band interference, the accuracy of cabin attitude data is improved, and a reliable attitude reference is provided for the subsequent adaptive update of yaw control parameters, which effectively solves the yaw adjustment lag caused by inaccurate attitude analysis in traditional methods.

[0124] Furthermore, step A500 in the method provided in the embodiment of the present application includes:

[0125] A540: The generated power of the wind turbine is introduced, and an increase calculation is performed based on the generated power to obtain a generated power increase rate.

[0126] A550: Perform a multi-objective analysis based on the cabin attitude data and the power generation power improvement rate to construct a multi-objective optimization expectation value.

[0127] A560: According to the multi-objective optimization expectation value, the yaw control parameters are updated through reinforcement learning to generate adaptive yaw control parameters, and adaptive dual-domain synchronous compensation of yaw to wind error is achieved according to the adaptive yaw control parameters.

[0128] In one embodiment, when updating the yaw control parameters based on the nacelle attitude data, the real-time power generation data of the wind turbine is first introduced with a sampling frequency of 1 Hz, and the increase is calculated by comparing it with the theoretical power curve, where the theoretical power curve is constructed based on the IEC61400-1 standard, which is a wind turbine design standard developed by the International Electrotechnical Commission. The specific formula used is Calculate the power generation rate, where is the current 10-minute average power generation, For example, when the wind speed is 10m / s, the theoretical power is 2MW, and the measured power is 1.9MW, then the increase rate is , which quantifies the impact of current yaw control on power generation efficiency.

[0129] Then, a multi-objective analysis is performed based on the cabin attitude data and the power generation rate. A three-dimensional optimization objective function is constructed, including the yaw error angle θ, attitude stability score S, and power generation rate η. , where the weight coefficient 、 、 , which can be dynamically adjusted according to the equipment maintenance priority. The minimum value of the function is solved by the particle swarm optimization algorithm to generate the multi-objective optimization expected value. For example, assuming that the above three-dimensional parameters θ = 1.5°, S = 70 points, and η = -5%, After optimization, the expected value must meet F≤5, realizing the coordinated optimization of wind accuracy, attitude stability and power generation efficiency, overcoming the power generation efficiency loss or equipment loss problem caused by single-target optimization in the existing technology.

[0130] Finally, a reinforcement learning state space is constructed based on the expected value of multi-objective optimization, and an environmental reward function is constructed in combination with power generation environment analysis. The parameter increment is obtained through state vector analysis to update the yaw control parameters. After the adaptive parameters are generated, synchronous compensation in the time-frequency domain is performed. After the performance verification is passed, the data is rolled back to determine the yaw control stability parameters, realizing adaptive compensation of yaw to wind error. The specific steps are described in detail in A561-A565.

[0131] Through quantitative analysis of the power generation rate improvement and the construction of a multi-objective optimization model, the update of yaw control parameters is upgraded from traditional empirical adjustment to data-driven intelligent optimization, achieving dual-dimensional optimization of wind energy utilization and equipment reliability.

[0132] Furthermore, step A560 in the method provided in the embodiment of the present application includes:

[0133] A561: Perform reinforcement learning based on the multi-objective optimization expectation value and construct a reinforcement learning state space.

[0134] A562: Analyze the power generation environment based on wind turbines and construct an environmental reward function.

[0135] A563: Perform state vector analysis based on the reinforcement learning state space according to the environmental reward function to obtain parameter increments, update the yaw control parameters according to the parameter increments, and generate adaptive yaw control parameters.

[0136] A564: Perform time-frequency domain compensation according to the adaptive yaw control parameters to generate a dual-domain synchronous compensation result.

[0137] A565: Verify the effectiveness of the dual-domain synchronous compensation results. When the verification passes, roll back the data to determine the yaw control stability parameters to achieve adaptive compensation of yaw to wind error.

[0138] Optionally, when updating the yaw control parameters through reinforcement learning based on the multi-objective optimization expectation value, first, a reinforcement learning state space is constructed. The yaw error angle θ, attitude stability score S, and power generation rate η in the multi-objective optimization expectation value are used as core state variables, combined with wind speed v, turbulence intensity Environmental parameters such as , which contains the current and previous state values.

[0139] Then, an environmental reward function is constructed based on the analysis of the wind turbine power generation environment. , where the wind error penalty term is , Stable Posture Reward , Power generation reward items , weight coefficient =0.4, =0.3, =0.3. For example, when θ=0.8°, S=85 minutes, and η=2%, the following calculation process may be performed: ,The environmental reward function quantifies the synergistic benefits of power generation efficiency and equipment stability.

[0140] Next, a deep Q-network algorithm is used for iterative training of the state vector, analyzing the state space and reward function. An ε-greedy strategy is used to select actions, i.e., parameter adjustment directions, and calculate parameter increments ΔKp, ΔKi, and ΔKd. When the state vector is input into the network, the output layer generates a three-dimensional action value, corresponding to the adjustment step size of the PID control parameters, such as ΔKp = ±0.05. After 1000 rounds of training, the parameters converge to the optimal solution, generating adaptive yaw control parameters. For example, an initial Kp = 0.8 is incremented by ΔKp = 0.1 after training, and then updated to Kp = 0.9, which increases the convergence speed of wind error by 20%.

[0141] After generating the adaptive yaw control parameters, synchronous compensation is performed in both the time and frequency domains: in the time domain, the yaw angle is adjusted in real time by the updated PID parameters, and the speed is adjusted by 1° / s. In the frequency domain, a 4th-order Butterworth bandpass filter with a passband of 0.1-10Hz is used to process the high-frequency error components caused by turbulence, achieving coordinated compensation of real-time error correction and frequency domain fluctuation suppression.

[0142] Finally, the dual-domain compensation results were validated by comparing 24 hours of measured data: when the wind error was stable within ±0.8°, the power generation rate was ≥3%, and the attitude stability score was ≥85 points, the verification passed and a data rollback was performed, storing the current parameters as stable parameters; otherwise, the parameters were rolled back to the previous version for retraining, thereby achieving adaptive compensation for the yaw wind error.

[0143] By constructing a multi-dimensional state space, a power generation environment reward function, deep reinforcement learning iteration and a dual-domain compensation mechanism, an adaptive compensation system is formed, thereby achieving the simultaneous optimization of wind energy utilization efficiency and equipment reliability.

[0144] In summary, the yaw-to-wind error compensation method based on wind measurement data provided in the embodiments of the present application has the following technical effects:

[0145] The present application collects wind field data in the internal wind measurement area and the wake critical area around the wind turbine, obtains relevant data such as wind field flow velocity and wind direction through coordinate unification and data filtering, calculates the changing trend and deviation information of the incoming wind direction angle and the cabin wind direction angle, and performs dynamic correction based on the marking and calculation results of the wake critical area, thereby accurately measuring the yaw to wind error under the influence of various environmental factors in the wind farm, making the yaw to wind error compensation result more accurate and reliable, and achieving the technical effect of accurately improving the wind rotor wind accuracy, significantly improving the power generation efficiency and enhancing the equipment operation safety.

[0146] Example 2, as Figure 2 As shown, based on the same inventive concept as the aforementioned embodiment 1, the embodiment of the present application provides a yaw-to-wind error compensation system based on wind measurement data, the system comprising:

[0147] The original data set acquisition module 1 sets the wind measurement point based on the wind turbine and starts the wind measurement equipment to collect data, thereby obtaining the original wind field data set.

[0148] The cabin wind direction angle acquisition module 2 performs wind field incoming flow calculation based on the original wind field data to determine the incoming flow wind direction angle, and performs cabin wind direction calculation based on the original wind field data to determine the cabin wind direction angle.

[0149] The wind error angle acquisition module 3 is used to calculate the deviation between the incoming wind direction angle and the nacelle wind direction angle to obtain the yaw wind error angle.

[0150] The compensation result acquisition module 4 is used to trigger the dual-channel compensation module according to the yaw-to-wind error angle, accumulate the error of the yaw-to-wind error angle, build an error queue, synchronize the error queue to the dual-channel compensation module for batch compensation, and obtain the compensation result.

[0151] The adaptive compensation execution module 5 is used to retrieve yaw control parameters, perform attitude analysis on the wind turbine according to the compensation results, obtain nacelle attitude data, and update the yaw control parameters according to the nacelle attitude data to achieve adaptive compensation of yaw to wind error.

[0152] Furthermore, the original data set acquisition module 1 is used to perform the following steps:

[0153] A wake impact analysis is performed based on the wind turbine to obtain a wake length parameter; the hub height data of the wind turbine is retrieved, and a two-dimensional coordinate system is constructed with the hub center as the origin, the hub height as the vertical axis, and the horizon as the horizontal axis; the rotor rotation plane of the wind turbine is projected onto the two-dimensional coordinate system to determine the plane projection direction; and the wind measurement point is set according to the plane projection direction and the wake length parameter.

[0154] Furthermore, the original data set acquisition module 1 is used to perform the following steps:

[0155] Based on the wind measurement points, a first wind measurement point and a second wind measurement point are extracted, wherein the first wind measurement point includes a laser radar device and the second wind measurement point includes an ultrasonic anemometer; based on the first wind measurement point, the laser radar device is activated to perform spiral scanning to obtain first wind field data; based on the second wind measurement point, the ultrasonic anemometer is activated to perform high-frequency sampling to obtain second wind field data; the first wind field data and the second wind field data are coordinate-uniformed to generate a three-dimensional standard data set; and data filtering is performed on the three-dimensional standard data set to obtain the original wind field data set.

[0156] Furthermore, the wind error angle acquisition module 3 is configured to perform the following steps:

[0157] The method comprises the following steps: calculating the direction vector of the main wind direction based on the two-dimensional coordinate system, determining the main wind direction projection line, constructing a digital elevation grid with the wind turbine as the center, and performing calculation according to the digital elevation grid in combination with the main wind direction projection line to obtain terrain gradient parameters; performing terrain offset compensation based on the terrain gradient parameters to obtain a dynamic terrain correction amount; performing error angle core calculation based on the dynamic terrain correction amount to determine the error angle; extracting high-frequency wind speed data based on the wind measurement point, and solving the turbulence three-dimensional component based on the high-frequency wind speed data to obtain the turbulence intensity; and adaptively attenuating the error angle based on the turbulence intensity to obtain the yaw-to-wind error angle.

[0158] Furthermore, the compensation result acquisition module 4 is configured to perform the following steps:

[0159] Based on the yaw error angle, a dynamic analysis is performed according to the wind speed, a preset dynamic threshold is constructed, and a channel activation rule is formulated according to the preset dynamic threshold; the dual-channel compensation module is triggered according to the channel activation rule, and the dual-channel compensation module includes an active compensation channel and a passive compensation channel; when the yaw error angle is greater than the preset dynamic threshold, the active compensation channel is triggered; dynamic compensation adjustment is performed through the active compensation channel to obtain a first compensation result; when the yaw error angle is less than or equal to the preset dynamic threshold, the yaw error angle is accumulated as an error to construct an error queue; the error queue is synchronized to the passive compensation channel for dynamic batch compensation to obtain a second compensation result.

[0160] Furthermore, the compensation result acquisition module 4 is configured to perform the following steps:

[0161] The error queue is synchronized to the passive compensation channel for cumulative calculation to obtain queue cumulative energy; equivalent calculation is performed based on the queue cumulative energy to obtain an equivalent compensation angle; torque analysis is performed according to the equivalent compensation angle to generate an equivalent torque instruction; the equivalent torque instruction is executed to perform dynamic batch compensation to obtain the second compensation result.

[0162] Furthermore, the adaptive compensation execution module 5 is configured to perform the following steps:

[0163] The wind turbine generator set is subjected to attitude oscillation calculation according to the compensation result to obtain attitude oscillation energy data; attitude stability is evaluated based on the attitude oscillation energy data to generate an attitude stability score; and frequency band interference is eliminated according to the attitude stability score to obtain the nacelle attitude data.

[0164] Furthermore, the adaptive compensation execution module 5 is configured to perform the following steps:

[0165] The generated power of the wind turbine is introduced, and the increase is calculated based on the generated power to obtain the generated power improvement rate; a multi-objective analysis is performed based on the nacelle attitude data and the generated power improvement rate to construct a multi-objective optimization expected value; according to the multi-objective optimization expected value, the yaw control parameters are updated through reinforcement learning to generate adaptive yaw control parameters, and adaptive dual-domain synchronous compensation of yaw to wind error is achieved according to the adaptive yaw control parameters.

[0166] Furthermore, the adaptive compensation execution module 5 is configured to perform the following steps:

[0167] Reinforcement learning is performed based on the multi-objective optimization expectation value to construct a reinforcement learning state space; power generation environment analysis is performed based on the wind turbine to construct an environmental reward function; state vector analysis is performed based on the reinforcement learning state space and according to the environmental reward function to obtain parameter increments, the yaw control parameters are updated according to the parameter increments to generate adaptive yaw control parameters; time-frequency domain compensation is performed according to the adaptive yaw control parameters to generate dual-domain synchronous compensation results; performance verification is performed on the dual-domain synchronous compensation results, and when the verification passes, data is rolled back to determine yaw control stability parameters to achieve adaptive compensation of yaw to wind error.

[0168] The yaw-to-wind error compensation system based on wind measurement data provided by an embodiment of the present invention can execute the yaw-to-wind error compensation method based on wind measurement data provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects of the execution method.

[0169] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0170] The above specific embodiments do not constitute a limitation to the scope of protection of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of this application should be included in the scope of protection of this application. In some cases, the actions or steps recorded in this application can be performed in an order different from that in the embodiments and can still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A yaw-to-wind error compensation method based on wind measurement data, characterized in that: The method comprises: Set wind measurement points based on wind turbines and start wind measurement equipment to collect data to obtain the original wind field data set; Calculating the incoming wind flow based on the original wind field data to determine the incoming wind direction angle, and calculating the cabin wind direction based on the original wind field data to determine the cabin wind direction angle; Calculating the deviation between the incoming wind direction angle and the cabin wind direction angle to obtain a yaw-to-wind error angle; triggering a dual-channel compensation module according to the yaw-to-wind error angle, accumulating the yaw-to-wind error angle, building an error queue, synchronizing the error queue to the dual-channel compensation module for batch compensation, and obtaining a compensation result; Retrieving yaw control parameters, performing attitude analysis on the wind turbine according to the compensation result, obtaining nacelle attitude data, and updating the yaw control parameters according to the nacelle attitude data to achieve adaptive compensation of yaw to wind error; The dual-channel compensation module is triggered according to the yaw-to-wind error angle, the yaw-to-wind error angle is accumulated, an error queue is constructed, and the error queue is synchronized to the dual-channel compensation module for batch compensation to obtain a compensation result, the method comprising: Performing a dynamic analysis based on the yaw-to-wind error angle according to the wind speed, constructing a preset dynamic threshold, and formulating a channel activation rule according to the preset dynamic threshold; Triggering the dual-channel compensation module according to the channel activation rule, the dual-channel compensation module includes an active compensation channel and a passive compensation channel; The active compensation channel outputs a real-time yaw adjustment value through a PID controller when the yaw error angle to wind exceeds a preset dynamic threshold, and performs dynamic compensation adjustment through the active compensation channel to obtain a first compensation result; The passive compensation channel is to build an error queue when the yaw to wind error angle is less than or equal to a preset dynamic threshold, synchronize the error queue to the passive compensation channel for cumulative calculation to obtain queue cumulative energy, and obtain an equivalent compensation angle based on the equivalent calculation of the queue cumulative energy. Then, torque analysis is performed based on this to generate an equivalent torque instruction, and batch compensation is performed to obtain a second compensation result, wherein the queue cumulative energy is the sum of the squares of the errors as an energy quantification indicator, and the energy-angle conversion formula is used. Convert the accumulated energy into an equivalent compensation angle.

2. The yaw-to-wind error compensation method based on wind measurement data according to claim 1, characterized in that: The wind measurement points are set based on the wind turbines. The methods include: Perform wake impact analysis based on wind turbines to obtain wake length parameters; Retrieving the hub height data of the wind turbine, constructing a two-dimensional coordinate system with the hub center as the origin, the hub height as the vertical axis, and the horizon as the horizontal axis; Projecting the rotor rotation plane of the wind turbine onto the two-dimensional coordinate system to determine the plane projection direction; The wind measurement point is set according to the plane projection direction and the wake length parameter.

3. The yaw-to-wind error compensation method based on wind measurement data according to claim 2, characterized in that: Start the wind measurement equipment to collect data and obtain the original wind field data set. The method includes: Extracting a first wind measurement point and a second wind measurement point based on the wind measurement point, wherein the first wind measurement point includes a laser radar device and the second wind measurement point includes an ultrasonic anemometer; activating the laser radar device based on the first wind measurement point to perform spiral scanning to obtain first wind field data; activating the ultrasonic anemometer based on the second wind measurement point to perform high-frequency sampling to obtain second wind field data; Performing coordinate unification processing on the first wind field data and the second wind field data to generate a three-dimensional standard data set; Data filtering is performed on the three-dimensional standard data set to obtain the original wind field data set.

4. The yaw-to-wind error compensation method based on wind measurement data according to claim 2, characterized in that: Calculating the deviation between the incoming wind direction angle and the nacelle wind direction angle to obtain a yaw error angle to the wind, the method comprising: Calculate the direction vector of the main wind direction based on the two-dimensional coordinate system and determine the main wind direction projection line; Constructing a digital elevation grid with the wind turbine as the center, and performing calculations according to the digital elevation grid in combination with the main wind direction projection line to obtain terrain gradient parameters; Perform terrain offset compensation based on the terrain gradient parameter to obtain a dynamic terrain correction amount; performing an error angle core calculation based on the dynamic terrain correction amount to determine the error angle; Extracting high-frequency wind speed data based on the wind measurement points, and calculating the turbulence three-dimensional component according to the high-frequency wind speed data to obtain turbulence intensity; The error angle is adaptively attenuated based on the turbulence intensity to obtain the yaw-to-wind error angle.

5. The yaw-to-wind error compensation method based on wind measurement data according to claim 1, characterized in that: Synchronizing the error queue to the passive compensation channel for dynamic batch compensation to obtain a second compensation result, the method comprising: Synchronizing the error queue to the passive compensation channel for cumulative calculation to obtain queue cumulative energy; Performing equivalent calculation based on the queue accumulated energy to obtain an equivalent compensation angle; Perform torque analysis according to the equivalent compensation angle to generate an equivalent torque instruction; Execute the equivalent moment instruction to perform dynamic batch compensation to obtain the second compensation result.

6. The yaw-to-wind error compensation method based on wind measurement data according to claim 1, characterized in that: Retrieving yaw control parameters, performing attitude analysis on the wind turbine according to the compensation results, and obtaining nacelle attitude data, the method includes: The attitude oscillation of the wind turbine is calculated according to the compensation result to obtain attitude oscillation energy data. The attitude oscillation energy is obtained by collecting X / Y / Z axis vibration data using an acceleration sensor with a sampling frequency of 100 Hz. The following formula is used: 2 +ay 2 +az 2 ) is calculated, where ax / ay / az are the instantaneous values ​​of acceleration of each axis; performing posture stability evaluation based on the posture oscillation energy data to generate a posture stability score; Frequency band interference elimination is performed according to the attitude stability score to obtain the cabin attitude data.

7. The yaw-to-wind error compensation method based on wind measurement data according to claim 6, characterized in that: The yaw control parameters are updated according to the cabin attitude data to achieve adaptive compensation of yaw to wind error, the method comprising: The generated power of the wind turbine is introduced, and an increase calculation is performed based on the generated power to obtain a generated power increase rate; Performing a multi-objective analysis based on the cabin attitude data and the power generation rate improvement to construct a multi-objective optimization expected value; According to the multi-objective optimization expected value, the yaw control parameters are updated through reinforcement learning to generate adaptive yaw control parameters, and adaptive dual-domain synchronous compensation of yaw to wind error is achieved according to the adaptive yaw control parameters.

8. The yaw-to-wind error compensation method based on wind measurement data according to claim 7, characterized in that: According to the multi-objective optimization expected value, the yaw control parameters are updated through reinforcement learning to generate adaptive yaw control parameters, and adaptive dual-domain synchronous compensation of yaw-to-wind error is achieved according to the adaptive yaw control parameters. The method includes: Perform reinforcement learning based on the multi-objective optimization expectation value to construct a reinforcement learning state space; Analyze the power generation environment based on wind turbines and construct an environmental reward function; performing a state vector analysis based on the reinforcement learning state space according to the environmental reward function to obtain a parameter increment, updating the yaw control parameter according to the parameter increment to generate an adaptive yaw control parameter; Perform time-frequency domain compensation according to the adaptive yaw control parameters to generate a dual-domain synchronous compensation result; The dual-domain synchronous compensation result is verified for effectiveness. When the verification is passed, data is rolled back to determine the yaw control stability parameters and realize adaptive compensation of yaw to wind error.

9. The yaw-to-wind error compensation system based on wind measurement data is characterized in that: A system for implementing the yaw-to-wind error compensation method based on wind measurement data according to any one of claims 1 to 8, comprising: The original data set acquisition module sets the wind measurement points based on the wind turbine and starts the wind measurement equipment to collect data to obtain the original wind field data set; a cabin wind direction angle acquisition module, which performs wind field incoming flow calculation based on the original wind field data to determine the incoming flow wind direction angle, and performs cabin wind direction calculation based on the original wind field data to determine the cabin wind direction angle; a wind error angle acquisition module, configured to calculate a deviation between the incoming wind direction angle and the nacelle wind direction angle to obtain a yaw wind error angle; a compensation result acquisition module, configured to trigger a dual-channel compensation module according to the yaw-to-wind error angle, accumulate errors of the yaw-to-wind error angle, construct an error queue, synchronize the error queue to the dual-channel compensation module for batch compensation, and obtain compensation results; An adaptive compensation execution module is used to retrieve yaw control parameters, perform attitude analysis on the wind turbine according to the compensation results, obtain nacelle attitude data, and update the yaw control parameters according to the nacelle attitude data to achieve adaptive compensation of yaw to wind error; Furthermore, the compensation result acquisition module is used to perform the following steps: Performing a dynamic analysis based on the yaw-to-wind error angle according to the wind speed, constructing a preset dynamic threshold, and formulating a channel activation rule according to the preset dynamic threshold; Triggering the dual-channel compensation module according to the channel activation rule, the dual-channel compensation module includes an active compensation channel and a passive compensation channel; When the yaw-to-wind error angle is greater than the preset dynamic threshold, an active compensation channel is triggered; dynamic compensation adjustment is performed through the active compensation channel to obtain a first compensation result; When the yaw-to-wind error angle is less than or equal to the preset dynamic threshold, the yaw-to-wind error angle is accumulated as an error to construct an error queue; The error queue is synchronized to the passive compensation channel to perform dynamic batch compensation to obtain a second compensation result.

Images