Fan yaw error self-learning correction method and system

Abstract

The application discloses a wind turbine yaw error self-learning correction method and system, and relates to the technical field of wind turbine yaw control. The method comprises the following steps: collecting wind speed data, wind direction data and yaw angle data in real time when the wind turbine is running; obtaining the actual yaw error of the wind turbine according to the wind speed data, the wind direction data and the yaw angle data; determining an error feature set based on historical yaw errors; obtaining error evolution features according to the matching length of the actual yaw error and the error feature set; self-learning to generate yaw correction parameters based on the error evolution features and the yaw angle data; and controlling the wind turbine according to the yaw correction parameters to correct the yaw error. Through the above method, the technical problems of error correction lag and inaccuracy caused by the lack of dynamic self-learning ability of the wind turbine yaw system in the prior art are solved.

Description

Technical Field

[0001] This application relates to the field of yaw control technology for wind turbine generator sets, and in particular to a self-learning correction method and system for wind turbine yaw error. Background Technology

[0002] Traditional wind turbine yaw control relies heavily on fixed parameters or simple feedback adjustments, making it difficult to adapt to complex and changing wind conditions and environmental disturbances. During long-term operation, the yaw system is susceptible to mechanical wear, sensor drift, and other factors, leading to deviations between the actual yaw angle and the theoretical optimal value. This results in decreased wind energy capture efficiency and accelerated component fatigue. Existing technologies typically depend on periodic manual calibration or static error compensation, lacking the ability to adaptively learn from the dynamic evolution of errors and thus failing to achieve real-time accurate correction. Therefore, there is an urgent need for a method that can autonomously optimize the yaw strategy based on historical data and real-time operating conditions to improve the economy and reliability of wind turbine operation.

[0003] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a self-learning correction method and system for wind turbine yaw error, which aims to solve the technical problems of lag and inaccuracy in error correction caused by the lack of dynamic self-learning capability in the existing wind turbine yaw system.

[0005] To achieve the above objectives, this application provides a self-learning correction method for wind turbine yaw error, the method comprising: Real-time acquisition of wind speed, wind direction, and yaw angle data during wind turbine operation; The actual yaw error of the wind turbine is obtained based on the wind speed data, the wind direction data, and the yaw angle data. Based on historical yaw errors, an error feature set is determined, and the error evolution characteristics are obtained according to the matching length between the actual yaw error and the error feature set. Based on the error evolution characteristics and the yaw angle data, yaw correction parameters are generated through self-learning. The wind turbine is controlled according to the yaw correction parameters to correct the yaw error.

[0006] In one embodiment, the step of obtaining the actual yaw error of the wind turbine based on the wind speed data, the wind direction data, and the yaw angle data includes: Based on the wind turbine power data and the wind speed data, a wind speed-power mapping relationship is constructed; Based on the wind direction data, the dominant wind direction range is determined, and the target power output point corresponding to the dominant wind direction range is selected from the wind speed-power mapping relationship; The theoretical yaw angle corresponding to the target power output point is compared with the yaw angle data to obtain the yaw angle deviation. The yaw angle deviation is standardized to obtain the actual yaw error.

[0007] In one embodiment, the step of determining an error feature set based on historical yaw errors and obtaining error evolution features according to the matching length between the actual yaw error and the error feature set includes: Historical yaw error samples in different wind direction and wind speed ranges are extracted from historical operational data to form an error feature set. Determine the sequence of actual yaw errors and each feature sequence in the error feature set, and calculate the matching degree between the sequence of actual yaw errors and each feature sequence in the error feature set; The sequence of actual yaw errors with matching degrees exceeding a preset threshold is selected as the target sequence, and the matching length is determined based on the number of consecutive data points matched between the target sequence and each feature sequence in the error feature set. Based on the matching length and the corresponding historical error evolution trend, the error evolution characteristics are obtained.

[0008] In one embodiment, the step of generating yaw correction parameters by self-learning based on the error evolution characteristics and the yaw angle data includes: The error evolution characteristics are fused with the yaw angle data to obtain the correction decision input vector; The correction decision input vector is input into the parameter generation model, and a preliminary yaw correction parameter set is output. Based on the current operating conditions of the wind turbine, the preliminary yaw correction parameter set is modified to generate yaw correction parameters.

[0009] In one embodiment, the step of fusing the error evolution characteristics with the yaw angle data to obtain the correction decision input vector includes: The error evolution characteristics are quantized and encoded, transforming them from a trend description into a numerical feature vector that includes the direction and rate of change. Extract key contextual information from the yaw angle data, including the absolute deviation of the current yaw angle relative to the prevailing wind direction and the recent yaw angle fluctuation variance; The numerical feature vector is concatenated with the key context information to obtain a concatenated vector. A timestamp and a wind turbine operating mode identifier are then added to the front of the concatenated vector to obtain the correction decision input vector.

[0010] In one embodiment, the step of performing feasibility correction on the preliminary yaw correction parameter set in conjunction with the current operating conditions of the wind turbine to generate yaw correction parameters includes: Obtain real-time operating data of the wind turbine; The preliminary yaw correction parameter set and the real-time operating condition data are input into the safety assessment model to obtain the predicted mechanical load change and predicted power fluctuation. If the predicted mechanical load change and / or the predicted power fluctuation does not exceed the safe operation threshold, then the preliminary yaw correction parameter set will be directly used as the yaw correction parameter. If the predicted mechanical load change and / or the predicted power fluctuation exceeds the safe operating threshold, the correction amplitude and rate in the preliminary yaw correction parameters are adjusted according to the preset optimization strategy to obtain the adjusted preliminary yaw correction parameters. Then, the process returns to the step of inputting the preliminary yaw correction parameter set and the real-time operating condition data into the safety assessment model to obtain the predicted mechanical load change and predicted power fluctuation, until the safety constraints are met, and the adjusted preliminary yaw correction parameters are used as the yaw correction parameters.

[0011] In one embodiment, the step of controlling the wind turbine according to the yaw correction parameters to correct the yaw error includes: The yaw correction parameters are converted into yaw motor control commands that can be recognized by the wind turbine main controller; The wind turbine is controlled in stages based on the yaw motor control commands, and the dynamic response data of the wind turbine is received. The correction result is evaluated in real time based on the dynamic response data. When the correction result does not meet the termination condition, cyclic fine-tuning is performed based on the yaw motor control command.

[0012] In one embodiment, the step of evaluating the correction result in real time based on the dynamic response data includes: The average power output within a specific time window is determined based on the dynamic response data, and the average power output within the specific time window is compared with the average power before control to obtain the power correction result. Based on the dynamic response data, the vibration data changes of the key components of the wind turbine before and after control are determined, and vibration correction results are obtained. A correction result is generated based on the power correction result and the vibration correction result.

[0013] In one embodiment, the method further includes: Record the input conditions, parameters used, and correction results for each correction, and establish a yaw correction case knowledge base based on the input conditions, parameters used, and correction results; Based on a preset statistical analysis period, statistical analysis is performed on the cases in the yaw correction case knowledge base to identify the target correction strategies under different wind direction and wind speed combinations. When the environmental conditions match the wind direction and wind speed corresponding to the target correction strategy, the correction parameters included in the target correction strategy are recommended.

[0014] Furthermore, to achieve the above objectives, this application also proposes a wind turbine yaw error self-learning correction system, which includes: The data acquisition module is used to collect wind speed, wind direction, and yaw angle data in real time during wind turbine operation; The error determination module is used to obtain the actual yaw error of the wind turbine based on the wind speed data, the wind direction data, and the yaw angle data. The feature extraction module is used to determine the error feature set based on historical yaw error, and to obtain the error evolution feature according to the matching length between the actual yaw error and the error feature set; The parameter correction module is used to generate yaw correction parameters by self-learning based on the error evolution characteristics and the yaw angle data. The wind turbine control module is used to control the wind turbine according to the yaw correction parameters in order to correct the yaw error.

[0015] In addition, to achieve the above objectives, this application also proposes a wind turbine yaw error self-learning correction device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the wind turbine yaw error self-learning correction method described above.

[0016] In addition, to achieve the above objectives, the present invention also proposes a storage medium, which is a computer-readable storage medium, and stores a computer program on the storage medium. When the computer program is executed by a processor, it implements the steps of the wind turbine yaw error self-learning correction method described above.

[0017] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the wind turbine yaw error self-learning correction method described above.

[0018] This application provides a self-learning correction method for wind turbine yaw error. It collects wind speed, wind direction, and yaw angle data in real time during wind turbine operation. Based on these data, the actual yaw error of the wind turbine is obtained. An error feature set is determined based on historical yaw errors. The error evolution feature is obtained based on the matching length between the actual yaw error and the error feature set. Based on the error evolution feature and yaw angle data, yaw correction parameters are generated through self-learning. The wind turbine is then controlled according to these parameters to correct the yaw error. This method solves the technical problem of delayed and inaccurate error correction caused by the lack of dynamic self-learning capability in existing wind turbine yaw systems. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating an embodiment of the wind turbine yaw error self-learning correction method of this application. Figure 2 This is a wind speed-power mapping curve of an embodiment of the wind turbine yaw error self-learning correction method of this application; Figure 3 This is a schematic diagram of the wind turbine response and correction effect according to an embodiment of the wind turbine yaw error self-learning correction method of this application; Figure 4 This is a schematic diagram of the module structure of the wind turbine yaw error self-learning correction system according to an embodiment of this application; Figure 5 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the wind turbine yaw error self-learning correction method in the embodiments of this application.

[0022] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0023] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0024] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0025] The main solution of this application embodiment is: real-time acquisition of wind speed data, wind direction data, and yaw angle data during wind turbine operation; The actual yaw error of the wind turbine is obtained based on the wind speed data, the wind direction data, and the yaw angle data. Based on historical yaw errors, an error feature set is determined, and the error evolution characteristics are obtained according to the matching length between the actual yaw error and the error feature set. Based on the error evolution characteristics and the yaw angle data, yaw correction parameters are generated through self-learning. The wind turbine is controlled according to the yaw correction parameters to correct the yaw error.

[0026] Currently, traditional wind turbine yaw control relies heavily on fixed parameters or simple feedback adjustments, making it difficult to adapt to complex and changing wind conditions and environmental disturbances. During long-term operation, the yaw system is susceptible to mechanical wear, sensor drift, and other factors, leading to deviations between the actual yaw angle and the theoretical optimal value. This results in decreased wind energy capture efficiency and accelerated component fatigue. Existing technologies typically depend on periodic manual calibration or static error compensation, lacking the ability to adaptively learn from the dynamic evolution of errors and thus failing to achieve real-time accurate correction. Therefore, there is an urgent need for a method that can autonomously optimize the yaw strategy based on historical data and real-time operating conditions to improve the economy and reliability of wind turbine operation.

[0027] This application provides a solution that collects wind speed, wind direction, and yaw angle data in real time during wind turbine operation. Based on these data, the actual yaw error of the wind turbine is obtained. An error feature set is determined based on historical yaw errors. The error evolution feature is obtained based on the matching length between the actual yaw error and the error feature set. Based on the error evolution feature and yaw angle data, yaw correction parameters are generated through self-learning. The wind turbine is then controlled according to these parameters to correct the yaw error. This approach solves the technical problem of delayed and inaccurate error correction in existing wind turbine yaw systems due to the lack of dynamic self-learning capabilities.

[0028] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions, such as a wind turbine yaw error self-learning correction device. This embodiment does not specifically limit it in this way. The following uses a wind turbine yaw error self-learning correction device as an example to describe this embodiment and the following embodiments.

[0029] All actions involving the acquisition of signals, information, or data in this application are carried out in accordance with the relevant data protection laws and policies of the country where the application is located, and with the authorization of the owner of the relevant device.

[0030] This application provides a self-learning correction method for wind turbine yaw error, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the wind turbine yaw error self-learning correction method of this application.

[0031] In this embodiment, the wind turbine yaw error self-learning correction method includes steps S10~S50: Step S10: Collect wind speed data, wind direction data, and yaw angle data in real time during wind turbine operation; It should be noted that wind speed data refers to the airflow velocity value, reflecting the magnitude of wind energy resources, measured by an anemometer installed on the wind turbine nacelle. Wind direction data refers to the angular information reflecting the horizontal direction of airflow, measured by a wind vane, usually with true north as the 0-degree reference. Yaw angle data refers to the angle between the wind turbine nacelle axis and a reference direction (usually true north), measured by angle sensors such as encoders, reflecting the current windward direction of the wind turbine.

[0032] Understandably, wind speed, wind direction, and yaw angle data during wind turbine operation are typically collected in real-time by sensor arrays (such as ultrasonic anemometers and encoders) mounted on the turbine. These raw signals undergo preprocessing such as filtering and noise reduction by a data acquisition system, and are then converted into standard format digital signals. This data is transmitted via fieldbus or industrial Ethernet to the turbine's central controller (such as a PLC or dedicated controller), providing high-precision and timely data input for subsequent error calculation and intelligent correction.

[0033] In data acquisition systems, raw signals from anemometers, wind vanes, and angle sensors are typically mixed with high-frequency electronic noise, transient pulse interference (such as lightning strikes or equipment start-ups and shutdowns), and low-frequency fluctuations caused by mechanical vibrations. Preprocessing first involves digital filtering to eliminate these interferences. For example, a low-pass filter is used to remove high-frequency noise above the wind turbine's response frequency, while median filtering or moving average algorithms are employed to smooth random pulses and short-term fluctuations, extracting effective signals that accurately reflect wind conditions and the system's status. Subsequently, the filtered analog signals are converted to digital signals by an analog-to-digital converter. This process discretizes the continuous signal at a rate much higher than the signal's frequency of change, such as 10 times per second, mapping continuous voltage values ​​to the nearest discrete digital level and converting the quantized values ​​into binary code. Finally, this binary data is encapsulated into standard data frames conforming to specific industrial communication protocols (such as CANopen and Modbus TCP), containing information such as timestamps, sensor IDs, data values, and check bits, facilitating reliable transmission and subsequent processing within the control system network.

[0034] Step S20: Based on the wind speed data, the wind direction data, and the yaw angle data, obtain the actual yaw error of the wind turbine; It should be noted that the actual yaw error refers to the angle difference between the actual orientation of the wind turbine nacelle (i.e., the yaw angle) and the current wind direction (i.e., the wind direction data). It quantifies the degree of inaccuracy of the wind turbine's "wind alignment".

[0035] Understandably, the controller uses the synchronously acquired wind direction data (the direction from which the wind is coming) as the target angle and the yaw angle data (the current direction the cabin is pointing) as the actual angle. The two are then vectoredly subtracted, and the absolute value of the result is the instantaneous actual yaw error angle. To eliminate the calculation jitter caused by rapid wind direction fluctuations, the average wind direction is usually calculated based on data over a period of time (such as 10 minutes), or the wind direction data is processed using methods such as first-order hysteresis filtering before being subtracted from the yaw angle to obtain a more reliable yaw error value that better reflects the steady-state deviation.

[0036] In one feasible implementation, the step of obtaining the actual yaw error of the wind turbine based on the wind speed data, the wind direction data, and the yaw angle data includes: Based on the wind turbine power data and the wind speed data, a wind speed-power mapping relationship is constructed; Based on the wind direction data, the dominant wind direction range is determined, and the target power output point corresponding to the dominant wind direction range is selected from the wind speed-power mapping relationship; The theoretical yaw angle corresponding to the target power output point is compared with the yaw angle data to obtain the yaw angle deviation. The yaw angle deviation is standardized to obtain the actual yaw error.

[0037] It should be noted that the wind speed-power mapping relationship refers to the curve showing the correspondence between the output power of a wind turbine and the wind speed under specific conditions. (Refer to...) Figure 2 , Figure 2 This is the wind speed-power mapping curve. This curve is the basic performance curve obtained through testing when the wind turbine leaves the factory, reflecting the theoretical maximum efficiency of the wind turbine in converting wind energy into electrical energy. The dominant wind direction range refers to the range of wind direction angles that occur most frequently or have the greatest impact on energy capture within a specific time period (such as the past 10 minutes). (For example, a ±15° fan-shaped area centered on the average wind direction) is used to filter out unstable or secondary wind directions and focus on the main, sustainable wind energy sources.

[0038] The target power output point refers to the theoretical maximum power value that the wind turbine should be able to achieve in the prevailing wind direction range under the current real-time wind speed, based on the constructed wind speed-power mapping relationship. It represents the ideal power generation target under the current wind resources.

[0039] The theoretical yaw angle is the optimal direction that the wind turbine nacelle should theoretically align with to achieve the target power output point. In most cases, this angle is equal to the center wind direction of the prevailing wind direction range, that is, completely facing the oncoming wind direction, so that the wind turbine can capture the maximum wind energy.

[0040] Yaw angle deviation is the raw difference between the calculated theoretical yaw angle and the actual measured yaw angle data.

[0041] In practice, yaw error can be inferred from the actual power generation performance of the wind turbine. This is a performance-based calibration method that is indirect but closer to maximizing power generation. Technically, the controller first constructs or retrieves a baseline wind speed-power curve based on historical operating data (power and wind speed). Then, the system analyzes wind direction data in real time, and calculates the standard deviation or frequency distribution of wind direction through a sliding time window to determine a stable dominant wind direction range. Next, the system will query the current wind speed. Below, historical data points falling within the prevailing wind direction range are used, and the point with the highest maximum power or the highest average power is identified as the target power output point. Theoretically, the wind turbine should be directly facing the wind direction to generate this optimal power; therefore, the theoretical yaw angle is... The average value of the prevailing wind direction can be directly taken. Then the original yaw angle deviation was calculated. Finally, to more fairly assess the importance of the error, it is standardized, for example, by dividing it by its sensitivity coefficient to power impact (which can be estimated from the curve slope) or a standard angle threshold, to obtain the final actual yaw error. The calculation formula is:

[0042] K is a standardization factor, which can effectively filter out the interference of instantaneous wind direction fluctuations, directly correlate with power generation efficiency, and thus provide a more reliable error signal.

[0043] Step S30: Determine the error feature set based on historical yaw error, and obtain the error evolution feature according to the matching length between the actual yaw error and the error feature set; It should be noted that historical yaw error refers to the sequence of actual yaw error data continuously recorded by the wind turbine over a period of time, such as the past few hours or days. This sequence encapsulates the operating status and error variation patterns of the yaw system under various conditions. The error feature set is a template library of error sequences that characterizes typical behavioral patterns of the yaw system. For example, it can include short-term error patterns with specific forms, such as "persistent positive deviation," "periodic oscillation," "trend-increasing," and "random fluctuation." Matching length refers to the maximum number of data points or the duration for which the currently generated real-time yaw error sequence can maintain continuous consistency with a certain feature pattern in the error feature set. The longer the matching length, the higher the degree of agreement between the current error evolution trend and that feature. Error evolution features are used to describe the degree of similarity and persistence between the current yaw error behavior and known historical features. It is usually measured or derived from the matching length and reflects the pattern strength and development stage of the dynamic changes in error.

[0044] Understandably, the system maintains a set of error features containing various typical error patterns as a comparison benchmark. During real-time operation, the controller concatenates the latest actual yaw error data with historical data into a short-term sequence and compares it against each pattern in the feature set using a sliding window, calculating similarity (e.g., using correlation coefficient or Euclidean distance). The matching length is the length of the sequence segment whose similarity consistently exceeds a preset threshold. For example, if the current error sequence matches the "increasing trend" pattern for 10 consecutive data points (e.g., a sequence lasting 10 minutes), its error evolution characteristic is judged as "a continuous increasing trend," and the matching length of 10 minutes is used as the strength indicator of this characteristic. This transforms instantaneous error values ​​into a more dynamic and predictive understanding of system behavior, providing crucial information for subsequent decision-making.

[0045] In one feasible implementation, the step of determining the error feature set based on historical yaw error and obtaining the error evolution characteristics according to the matching length between the actual yaw error and the error feature set includes: Historical yaw error samples in different wind direction and wind speed ranges are extracted from historical operational data to form an error feature set. Determine the sequence of actual yaw errors and each feature sequence in the error feature set, and calculate the matching degree between the sequence of actual yaw errors and each feature sequence in the error feature set; The sequence of actual yaw errors with matching degrees exceeding a preset threshold is selected as the target sequence, and the matching length is determined based on the number of consecutive data points matched between the target sequence and each feature sequence in the error feature set. Based on the matching length and the corresponding historical error evolution trend, the error evolution characteristics are obtained.

[0046] In practice, the massive historical operational data is first preprocessed, grouping it according to wind direction and wind speed intervals. Within each specific operating condition group, recurring and representative error variation patterns, i.e., feature sequences, are identified using algorithms such as K-means clustering or feature extraction techniques. This allows for the construction of a set of error features labeled with operating conditions. Then, the actual yaw error sequence within a sliding window is maintained in real time. Compare this sequence with all feature sequences corresponding to the current operating condition in the error feature set. Perform similarity calculation. Match score. Algorithms such as dynamic time warping or Pearson correlation coefficient are typically used to calculate the similarity in shape, rather than just the numerical difference.

[0047]

[0048] Set a matching threshold Find all that satisfy the condition. The target sequence consists of segments of data points. The matching length L is directly the number of data points contained in the target sequence. For example, if the subsequence consisting of the most recent 8 data points matches the feature sequence... High-level matching, then the matching length The final error evolution feature E is a comprehensive conclusion that combines the matching length L and the matched feature sequence. The corresponding historical error evolution trend. It can be formally expressed as:

[0049] in, It is the type of the matched pattern (by...) (Determined), L is the match length (representing the strength and persistence of the current pattern). It is the confidence level that the pattern will continue to develop, calculated based on L and historical statistical information. This feature can accurately predict specific failures or performance degradation that may occur in the yaw system.

[0050] Step S40: Based on the error evolution characteristics and the yaw angle data, generate yaw correction parameters through self-learning.

[0051] It should be noted that yaw correction parameters refer to system control parameters used to compensate for or eliminate identified yaw errors, such as proportional-integral-derivative (PID) parameters of the yaw controller, yaw advance angle, yaw dead zone threshold, etc.

[0052] Understandably, the control system uses the error evolution characteristics as its core input to determine the current error pattern (such as systematic deviation, response hysteresis, or oscillation) and severity (reflected by indicators such as match length). Then, it combines real-time yaw angle data to evaluate the control effect under the current parameters. Based on this information, the system invokes built-in adaptive algorithms (such as reinforcement learning, fuzzy logic, or model-based predictive control) to dynamically adjust the yaw correction parameters. For example, if the characteristics indicate a persistent positive deviation, the system may learn and generate a negative yaw advance angle for compensation; if the characteristics are oscillations, it will adjust the derivative gain of the PID controller to suppress overshoot. The entire process, through continuous trial and error and effect evaluation, allows the yaw correction parameters to automatically approach their optimal values, achieving intelligent and precise calibration of the yaw system.

[0053] In one feasible implementation, the step of generating yaw correction parameters through self-learning based on the error evolution characteristics and the yaw angle data includes: The error evolution characteristics are fused with the yaw angle data to obtain the correction decision input vector; The correction decision input vector is input into the parameter generation model, and a preliminary yaw correction parameter set is output. Based on the current operating conditions of the wind turbine, the preliminary yaw correction parameter set is modified to generate yaw correction parameters.

[0054] Understandably, the system first fuses the abstract error evolution characteristics with continuous yaw angle data to construct a unified correction decision input vector X:

[0055] in, Represented as pattern type encoding, with a normalized match length of The current yaw angle is and instantaneous yaw rate This vector X provides the model with the complete contextual information needed for decision-making.

[0056] Input vector X into the pre-trained parameter generation model By learning from a large amount of historical optimal control data, a nonlinear mapping relationship from system state to control parameters is built-in, directly outputting the preliminary yaw correction parameter set. .

[0057]

[0058] The system defines a constraint function to verify whether the parameter set P is within the allowable range. For the initial yaw correction parameter set... The final yaw correction parameters are generated by correcting the parameters through a projection or optimization algorithm. .

[0059]

[0060] Where Ω represents the set of all parameters that satisfy the operating condition constraints. This formula means: among all feasible parameters, find the value that best approximates the original suggested value of the model. parameter group This ensures both the safety and feasibility of the parameters while respecting the model's intelligent decision-making to the greatest extent possible. Ultimately, It is then sent to the yaw controller for execution.

[0061] In one feasible implementation, the step of fusing the error evolution characteristics with the yaw angle data to obtain the correction decision input vector includes: The error evolution characteristics are quantized and encoded, transforming them from a trend description into a numerical feature vector that includes the direction and rate of change. Extract key contextual information from the yaw angle data, including the absolute deviation of the current yaw angle relative to the prevailing wind direction and the recent yaw angle fluctuation variance; The numerical feature vector is concatenated with the key context information to obtain a concatenated vector. A timestamp and a wind turbine operating mode identifier are then added to the front of the concatenated vector to obtain the correction decision input vector.

[0062] Understandably, the system first quantizes and encodes the descriptive error evolution characteristics. For example, a trend like "continuously increasing" is converted into a vector containing specific numerical values, such as [1, 0.85], where the first number encodes the direction of change (1 represents positive, -1 represents negative), and the second number represents the normalized rate of change calculated based on the matching length. This conversion transforms the ambiguous trend language into a mathematical signal that the model can directly process.

[0063] Next, the system extracts key contextual information from the continuous yaw angle data stream. This includes the absolute deviation of the current yaw angle relative to the prevailing wind direction, reflecting the instantaneous absolute value of the error; and the recent yaw angle fluctuation variance, which quantifies the severity of cabin oscillations over a short period and helps distinguish between stable deviations and dynamic oscillations. These two metrics provide the model with a crucial snapshot of the system's current dynamic behavior.

[0064] Then, the numerical feature vectors and key contextual information generated in the above steps are concatenated to form a more comprehensive state description. To ensure the accuracy and traceability of the decision, the system also adds a timestamp (to identify the specific time the data was generated, associating it with external conditions such as wind speed and turbulence) and a wind turbine operating mode identifier (such as normal power generation, start-up, shutdown, or fault status, because the correction strategy may be completely different under different modes) to the beginning of this concatenated vector. The final composite vector, which contains all dimensions of time series, operating conditions, error dynamics, and system state, is the correction decision input vector, providing the parameter generation model with all the contextual information needed to make informed decisions.

[0065] In one feasible implementation, the step of performing feasibility correction on the preliminary yaw correction parameter set in conjunction with the current operating conditions of the wind turbine to generate yaw correction parameters includes: Obtain real-time operating data of the wind turbine; The preliminary yaw correction parameter set and the real-time operating condition data are input into the safety assessment model to obtain the predicted mechanical load change and predicted power fluctuation. If the predicted mechanical load change and / or the predicted power fluctuation does not exceed the safe operation threshold, then the preliminary yaw correction parameter set will be directly used as the yaw correction parameter. If the predicted mechanical load change and / or the predicted power fluctuation exceeds the safe operating threshold, the correction amplitude and rate in the preliminary yaw correction parameters are adjusted according to the preset optimization strategy to obtain the adjusted preliminary yaw correction parameters. Then, the process returns to the step of inputting the preliminary yaw correction parameter set and the real-time operating condition data into the safety assessment model to obtain the predicted mechanical load change and predicted power fluctuation, until the safety constraints are met, and the adjusted preliminary yaw correction parameters are used as the yaw correction parameters.

[0066] Understandably, the system will first acquire real-time operating data of the wind turbine, including but not limited to current wind speed, wind direction, rotor speed, generator power, stress and temperature of key components, etc.

[0067] Next, the system inputs a preliminary yaw correction parameter set, such as a new set of control gains, along with the aforementioned real-time operating data into a safety assessment model. This model is typically a digital twin model based on the wind turbine's physical characteristics or a predictive model trained on a large amount of data. Its purpose is to quickly simulate and predict how the wind turbine's key safety indicators will change in the near future if this new set of parameters is applied immediately. Specifically, it outputs predicted changes in mechanical loads (such as the expected increase in tower base bending moment and blade root load) and predicted power fluctuations (such as whether the amplitude of output power fluctuations will intensify).

[0068] The system then compares the predicted results with a preset safe operating threshold, a hard limit set based on the wind turbine's design limits and grid requirements. A decision branch exists here: Safe Path: If the prediction results show that both mechanical load and power fluctuations are within safe thresholds, it indicates that the new parameters are safe and will not pose a risk to the wind turbine structure or grid stability. The system will then directly adopt the preliminary yaw correction parameter set as the final yaw correction parameters and implement it.

[0069] Optimization Path: If any prediction exceeds a safety threshold, an optimization loop is triggered. The system softens the initial parameters according to a preset optimization strategy (e.g., scaling the correction magnitude proportionally, reducing the parameter adjustment rate), generating a more conservative set of adjusted initial yaw correction parameters. Subsequently, the system feeds this new set of parameters back into the safety assessment model for a new round of prediction and verification. This "prediction-assessment-adjustment" loop continues until the mechanical load and power fluctuations predicted by the generated parameters fully meet the safety constraints. Finally, the parameters that passed the last iteration and are now considered safe are output as the yaw correction parameters.

[0070] Step S50: Control the wind turbine according to the yaw correction parameters to correct the yaw error.

[0071] Understandably, the calculated optimal and safe yaw correction parameters are applied to the final execution phase of the wind turbine's main control system. The control system loads these parameters, such as the new PID gain or yaw angle offset, into the real-time control loop. Subsequently, the system continuously compares the prevailing wind direction measured by the wind vane with the actual yaw angle fed back by the encoder, calculates the yaw error in real time, and generates control signals based on the new correction parameters. (Refer to...) Figure 3 , Figure 3 This diagram illustrates the wind turbine's response and correction effect. The signal drives the yaw motor to slowly and precisely adjust the nacelle's orientation, thereby dynamically reducing or eliminating yaw error. Ultimately, this ensures the wind turbine disk is continuously and optimally aligned with the wind direction, improving power generation efficiency and stabilizing mechanical loads.

[0072] In one feasible implementation, the step of controlling the wind turbine according to the yaw correction parameters to correct the yaw error includes: The yaw correction parameters are converted into yaw motor control commands that can be recognized by the wind turbine main controller; The wind turbine is controlled in stages based on the yaw motor control commands, and the dynamic response data of the wind turbine is received. The correction result is evaluated in real time based on the dynamic response data. When the correction result does not meet the termination condition, cyclic fine-tuning is performed based on the yaw motor control command.

[0073] Understandably, the system first translates the upstream calculated, higher-level yaw correction parameters (such as the target angle or control gain) into lower-level yaw motor control commands that the wind turbine's main controller can directly understand and execute. These commands include specific pulse signals, torque setpoints, or speed commands. This is the crucial bridge connecting decision-making and execution. Subsequently, control is not implemented all at once, but rather through a phased strategy. For example, it first rapidly approaches the target angle, then performs low-speed, precise positioning to avoid impacting the wind turbine structure. During this process, the system continuously receives dynamic response data from sensors, such as the actual rotation angle and speed of the nacelle, as well as the motor's current and torque. Finally, the system evaluates the correction results based on this real-time data, determining whether preset termination conditions have been met (such as the error entering the dead zone or the system stabilizing). If these conditions are not met, it indicates overshoot or insufficient response. The system will immediately perform small-scale cyclical fine-tuning of the control commands based on the current state, forming a closed-loop feedback until the yaw error is stably and accurately eliminated.

[0074] In one feasible implementation, the step of evaluating the correction result in real time based on the dynamic response data includes: The average power output within a specific time window is determined based on the dynamic response data, and the average power output within the specific time window is compared with the average power before control to obtain the power correction result. Based on the dynamic response data, the vibration data changes of the key components of the wind turbine before and after control are determined, and vibration correction results are obtained. A correction result is generated based on the power correction result and the vibration correction result.

[0075] Understandably, the system first extracts key evidence of power generation performance from dynamic response data: it calculates the average power output of the wind turbine within a pre-set specific time window (e.g., a few minutes after completing the yaw maneuver and stabilizing). This value is then compared to the average power output over the same period before the correction parameters were applied. This power correction result directly quantifies the power generation benefit brought about by the yaw correction and is a core indicator for evaluating its effectiveness.

[0076] Simultaneously, the system evaluates the impact of the corrective actions on the wind turbine's mechanical structure. It analyzes vibration sensor data from key components (such as the main shaft, gearbox, and tower) and compares the changes in vibration amplitude or spectral characteristics before and after control. This vibration correction result is crucial in determining whether the correction process, while increasing power, has introduced harmful mechanical loads or oscillations, ensuring the safety of the optimization.

[0077] Ultimately, the system integrates the power correction and vibration correction results to generate a comprehensive correction result judgment. For example, even if the power is significantly increased, if the vibration also increases abnormally, the overall result may be judged as unsatisfactory, thus triggering subsequent fine-tuning or alarms. This evaluation mechanism ensures that yaw correction always strikes a balance between pursuing power generation efficiency and ensuring equipment lifespan.

[0078] In one feasible implementation, the method further includes: Record the input conditions, parameters used, and correction results for each correction, and establish a yaw correction case knowledge base based on the input conditions, parameters used, and correction results; Based on a preset statistical analysis period, statistical analysis is performed on the cases in the yaw correction case knowledge base to identify the target correction strategies under different wind direction and wind speed combinations. When the environmental conditions match the wind direction and wind speed corresponding to the target correction strategy, the correction parameters included in the target correction strategy are recommended.

[0079] Understandably, the system first possesses comprehensive data logging capabilities. It saves each instance of yaw correction as a complete "case," including the input conditions (such as the specific wind speed, wind direction, and turbulence intensity), the parameters used (such as the actual PID gain or yaw angle setting), and the final correction result (such as the power increase percentage and vibration change value). These historical cases collectively form a continuously growing yaw correction case knowledge base, essentially acting as an "experience log" for the wind turbine.

[0080] Secondly, the system conducts regular big data analysis. Following a pre-set statistical analysis cycle (e.g., weekly or monthly), the system uses data mining techniques to perform retrospective analysis of the entire knowledge base. Its core objective is to identify patterns: pinpointing which correction strategies produce the best overall effect under specific wind direction and speed combinations. These strategies, proven to be efficient, are then refined into target correction strategies.

[0081] Ultimately, the system applies the analysis results to real-time control, enabling proactive parameter recommendations. When the system detects that the current real-time environmental conditions (wind direction, wind speed) highly match the historically advantageous conditions corresponding to a certain target correction strategy, it proactively recommends the optimized parameters included in that strategy, which have been validated in practice. This skips the conventional calculation process and directly applies the historically optimal solution that is more likely to succeed. This is essentially a closed-loop optimization process, enabling the wind turbine to "remember" and "reuse" past successful experiences, continuously improving overall performance.

[0082] This embodiment provides a self-learning correction method for wind turbine yaw error. It collects wind speed, wind direction, and yaw angle data in real time during wind turbine operation. Based on these data, the actual yaw error of the wind turbine is obtained. An error feature set is determined based on historical yaw errors. The error evolution feature is obtained based on the matching length between the actual yaw error and the error feature set. Based on the error evolution feature and yaw angle data, yaw correction parameters are generated through self-learning. The wind turbine is then controlled according to these parameters to correct the yaw error. This method solves the technical problem of delayed and inaccurate error correction caused by the lack of dynamic self-learning capability in existing wind turbine yaw systems.

[0083] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the wind turbine yaw error self-learning correction method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0084] This application also provides a wind turbine yaw error self-learning correction system; please refer to [reference needed]. Figure 4 The wind turbine yaw error self-learning correction system includes: Data acquisition module 10 is used to collect wind speed data, wind direction data and yaw angle data in real time when the wind turbine is running; The error determination module 20 is used to obtain the actual yaw error of the wind turbine based on the wind speed data, the wind direction data and the yaw angle data. The feature extraction module 30 is used to determine an error feature set based on historical yaw error, and to obtain error evolution features based on the matching length between the actual yaw error and the error feature set. Parameter correction module 40 is used to generate yaw correction parameters by self-learning based on the error evolution characteristics and the yaw angle data; The wind turbine control module 50 is used to control the wind turbine according to the yaw correction parameters in order to correct the yaw error.

[0085] In one feasible implementation, the error determination module 20 is further configured to construct a wind speed-power mapping relationship based on the wind turbine power data and the wind speed data; Based on the wind direction data, the dominant wind direction range is determined, and the target power output point corresponding to the dominant wind direction range is selected from the wind speed-power mapping relationship; The theoretical yaw angle corresponding to the target power output point is compared with the yaw angle data to obtain the yaw angle deviation. The yaw angle deviation is standardized to obtain the actual yaw error.

[0086] In one feasible implementation, the feature extraction module 30 is further used to extract historical yaw error samples from historical operating data in different wind direction and wind speed ranges to form an error feature set. Determine the sequence of actual yaw errors and each feature sequence in the error feature set, and calculate the matching degree between the sequence of actual yaw errors and each feature sequence in the error feature set; The sequence of actual yaw errors with matching degrees exceeding a preset threshold is selected as the target sequence, and the matching length is determined based on the number of consecutive data points matched between the target sequence and each feature sequence in the error feature set. Based on the matching length and the corresponding historical error evolution trend, the error evolution characteristics are obtained.

[0087] In one feasible implementation, the parameter correction module 40 is further configured to fuse the error evolution characteristics with the yaw angle data to obtain a correction decision input vector; The correction decision input vector is input into the parameter generation model, and a preliminary yaw correction parameter set is output. Based on the current operating conditions of the wind turbine, the preliminary yaw correction parameter set is modified to generate yaw correction parameters.

[0088] In one feasible implementation, the parameter correction module 40 is further configured to quantize and encode the error evolution characteristics, converting the error evolution characteristics from a trend description into a numerical feature vector containing the direction and rate of change. Extract key contextual information from the yaw angle data, including the absolute deviation of the current yaw angle relative to the prevailing wind direction and the recent yaw angle fluctuation variance; The numerical feature vector is concatenated with the key context information to obtain a concatenated vector. A timestamp and a wind turbine operating mode identifier are then added to the front of the concatenated vector to obtain the correction decision input vector.

[0089] In one feasible implementation, the parameter correction module 40 is also used to acquire real-time operating condition data of the wind turbine. The preliminary yaw correction parameter set and the real-time operating condition data are input into the safety assessment model to obtain the predicted mechanical load change and predicted power fluctuation. If the predicted mechanical load change and / or the predicted power fluctuation does not exceed the safe operation threshold, then the preliminary yaw correction parameter set will be directly used as the yaw correction parameter. If the predicted mechanical load change and / or the predicted power fluctuation exceeds the safe operating threshold, the correction amplitude and rate in the preliminary yaw correction parameters are adjusted according to the preset optimization strategy to obtain the adjusted preliminary yaw correction parameters. Then, the process returns to the step of inputting the preliminary yaw correction parameter set and the real-time operating condition data into the safety assessment model to obtain the predicted mechanical load change and predicted power fluctuation, until the safety constraints are met, and the adjusted preliminary yaw correction parameters are used as the yaw correction parameters.

[0090] In one feasible implementation, the wind turbine control module 50 is further configured to convert the yaw correction parameters into yaw motor control commands recognizable by the wind turbine main controller. The wind turbine is controlled in stages based on the yaw motor control commands, and the dynamic response data of the wind turbine is received. The correction result is evaluated in real time based on the dynamic response data. When the correction result does not meet the termination condition, cyclic fine-tuning is performed based on the yaw motor control command.

[0091] In one feasible implementation, the wind turbine control module 50 is further configured to determine the average power output within a specific time window based on the dynamic response data, and compare the average power output within the specific time window with the average power before control to obtain a power correction result; Based on the dynamic response data, the vibration data changes of the key components of the wind turbine before and after control are determined, and vibration correction results are obtained. A correction result is generated based on the power correction result and the vibration correction result.

[0092] In one feasible implementation, the learning iteration module 60 is further configured to record the input conditions, parameters used, and correction results for each correction, and to establish a yaw correction case knowledge base based on the input conditions, parameters used, and correction results. Based on a preset statistical analysis period, statistical analysis is performed on the cases in the yaw correction case knowledge base to identify the target correction strategies under different wind direction and wind speed combinations. When the environmental conditions match the wind direction and wind speed corresponding to the target correction strategy, the correction parameters included in the target correction strategy are recommended.

[0093] The wind turbine yaw error self-learning correction system provided in this application employs the wind turbine yaw error self-learning correction method in the above embodiments, which can solve the technical problems of lag and inaccuracy in error correction caused by the lack of dynamic self-learning capability in wind turbine yaw systems. Compared with the prior art, the beneficial effects of the wind turbine yaw error self-learning correction system provided in this application are the same as those of the wind turbine yaw error self-learning correction method provided in the above embodiments, and other technical features of the wind turbine yaw error self-learning correction system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0094] This application provides a wind turbine yaw error self-learning correction device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the wind turbine yaw error self-learning correction method in the above embodiment 1.

[0095] The following is for reference. Figure 5 The diagram illustrates a structural schematic suitable for implementing a wind turbine yaw error self-learning correction device according to embodiments of this application. The wind turbine yaw error self-learning correction device in embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 5 The wind turbine yaw error self-learning correction device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0096] like Figure 5As shown, the wind turbine yaw error self-learning correction device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in ROM (Read Only Memory) 1002 or the program loaded from storage device 1003 into RAM (Random Access Memory) 1004. RAM 1004 also stores various programs and data required for the operation of the wind turbine yaw error self-learning correction device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, LCDs (Liquid Crystal Displays), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the wind turbine yaw error self-learning correction device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a wind turbine yaw error self-learning correction device with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0097] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0098] The wind turbine yaw error self-learning correction device provided in this application adopts the wind turbine yaw error self-learning correction method in the above embodiments, which can solve the technical problem of wind turbine yaw error self-learning correction. Compared with the prior art, the beneficial effects of the wind turbine yaw error self-learning correction device provided in this application are the same as the beneficial effects of the wind turbine yaw error self-learning correction method provided in the above embodiments, and other technical features in the wind turbine yaw error self-learning correction device are the same as the features disclosed in the method of the previous embodiment, and will not be repeated here.

[0099] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0100] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0101] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the wind turbine yaw error self-learning correction method in the above embodiments.

[0102] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory or Flash Memory), optical fibers, CD-ROM (CD-Read Only Memory), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0103] The aforementioned computer-readable storage medium may be included in the wind turbine yaw error self-learning correction device; or it may exist independently and not be assembled into the wind turbine yaw error self-learning correction device.

[0104] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the wind turbine yaw error self-learning correction device, cause the wind turbine yaw error self-learning correction device to: collect wind speed data, wind direction data, and yaw angle data in real time during wind turbine operation; The actual yaw error of the wind turbine is obtained based on the wind speed data, the wind direction data, and the yaw angle data. Based on historical yaw errors, an error feature set is determined, and the error evolution characteristics are obtained according to the matching length between the actual yaw error and the error feature set. Based on the error evolution characteristics and the yaw angle data, yaw correction parameters are generated through self-learning. The wind turbine is controlled according to the yaw correction parameters to correct the yaw error.

[0105] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including LAN (Local Area Network) or WAN (Wide Area Network)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0106] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0107] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0108] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described wind turbine yaw error self-learning correction method, thereby solving the technical problem of wind turbine yaw error self-learning correction. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the wind turbine yaw error self-learning correction method provided in the above embodiments, and will not be repeated here.

[0109] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the wind turbine yaw error self-learning correction method described above.

[0110] The computer program product provided in this application can solve the technical problem of self-learning correction of wind turbine yaw error. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the wind turbine yaw error self-learning correction method provided in the above embodiments, and will not be repeated here.

[0111] The above are only some embodiments of this application and do not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A self-learning correction method for wind turbine yaw error, characterized in that, The wind turbine yaw error self-learning correction method includes: Real-time acquisition of wind speed, wind direction, and yaw angle data during wind turbine operation; The actual yaw error of the wind turbine is obtained based on the wind speed data, the wind direction data, and the yaw angle data. Based on historical yaw errors, an error feature set is determined, and the error evolution characteristics are obtained according to the matching length between the actual yaw error and the error feature set. Based on the error evolution characteristics and the yaw angle data, yaw correction parameters are generated through self-learning. The wind turbine is controlled according to the yaw correction parameters to correct the yaw error; The step of determining the error feature set based on historical yaw error and obtaining the error evolution characteristics according to the matching length between the actual yaw error and the error feature set includes: Historical yaw error samples in different wind direction and wind speed ranges are extracted from historical operational data to form an error feature set. Determine the sequence of actual yaw errors and each feature sequence in the error feature set, and calculate the matching degree between the sequence of actual yaw errors and each feature sequence in the error feature set; The sequence of actual yaw errors with matching degrees exceeding a preset threshold is selected as the target sequence, and the matching length is determined based on the number of consecutive data points matched between the target sequence and each feature sequence in the error feature set. Based on the matching length and the corresponding historical error evolution trend, the error evolution characteristics are obtained.

2. The method as described in claim 1, characterized in that, The step of obtaining the actual yaw error of the wind turbine based on the wind speed data, the wind direction data, and the yaw angle data includes: Based on the wind turbine power data and the wind speed data, a wind speed-power mapping relationship is constructed; Based on the wind direction data, the dominant wind direction range is determined, and the target power output point corresponding to the dominant wind direction range is selected from the wind speed-power mapping relationship; The theoretical yaw angle corresponding to the target power output point is compared with the yaw angle data to obtain the yaw angle deviation. The yaw angle deviation is standardized to obtain the actual yaw error.

3. The method as described in claim 1, characterized in that, The step of generating yaw correction parameters by self-learning based on the error evolution characteristics and the yaw angle data includes: The error evolution characteristics are fused with the yaw angle data to obtain the correction decision input vector; The correction decision input vector is input into the parameter generation model, and a preliminary yaw correction parameter set is output. Based on the current operating conditions of the wind turbine, the preliminary yaw correction parameter set is modified to generate yaw correction parameters.

4. The method as described in claim 3, characterized in that, The step of fusing the error evolution characteristics with the yaw angle data to obtain the correction decision input vector includes: The error evolution characteristics are quantized and encoded, transforming them from a trend description into a numerical feature vector that includes the direction and rate of change. Extract key contextual information from the yaw angle data, including the absolute deviation of the current yaw angle relative to the prevailing wind direction and the recent yaw angle fluctuation variance; The numerical feature vector is concatenated with the key context information to obtain a concatenated vector. A timestamp and a wind turbine operating mode identifier are then added to the front of the concatenated vector to obtain the correction decision input vector.

5. The method as described in claim 3, characterized in that, The step of making feasibility corrections to the initial yaw correction parameter set and generating yaw correction parameters by combining the current operating conditions of the wind turbine includes: Obtain real-time operating data of the wind turbine; The preliminary yaw correction parameter set and the real-time operating condition data are input into the safety assessment model to obtain the predicted mechanical load change and predicted power fluctuation. If the predicted mechanical load change and / or the predicted power fluctuation does not exceed the safe operation threshold, then the preliminary yaw correction parameter set will be directly used as the yaw correction parameter. If the predicted mechanical load change and / or the predicted power fluctuation exceeds the safe operating threshold, the correction amplitude and rate in the preliminary yaw correction parameters are adjusted according to the preset optimization strategy to obtain the adjusted preliminary yaw correction parameters. Then, the process returns to the step of inputting the preliminary yaw correction parameter set and the real-time operating condition data into the safety assessment model to obtain the predicted mechanical load change and predicted power fluctuation, until the safety constraints are met, and the adjusted preliminary yaw correction parameters are used as the yaw correction parameters.

6. The method as described in claim 1, characterized in that, The step of controlling the wind turbine according to the yaw correction parameters to correct the yaw error includes: The yaw correction parameters are converted into yaw motor control commands that can be recognized by the wind turbine main controller; The wind turbine is controlled in stages based on the yaw motor control commands, and the dynamic response data of the wind turbine is received. The correction result is evaluated in real time based on the dynamic response data. When the correction result does not meet the termination condition, cyclic fine-tuning is performed based on the yaw motor control command.

7. The method as described in claim 6, characterized in that, The step of evaluating the correction result in real time based on the dynamic response data includes: The average power output within a specific time window is determined based on the dynamic response data, and the average power output within the specific time window is compared with the average power before control to obtain the power correction result. Based on the dynamic response data, the vibration data changes of the key components of the wind turbine before and after control are determined, and vibration correction results are obtained. A correction result is generated based on the power correction result and the vibration correction result.

8. The method according to any one of claims 1 to 7, characterized in that, The method further includes: Record the input conditions, parameters used, and correction results for each correction, and establish a yaw correction case knowledge base based on the input conditions, parameters used, and correction results; Based on a preset statistical analysis period, statistical analysis is performed on the cases in the yaw correction case knowledge base to identify the target correction strategies under different wind direction and wind speed combinations. When the environmental conditions match the wind direction and wind speed corresponding to the target correction strategy, the correction parameters included in the target correction strategy are recommended.

9. A wind turbine yaw error self-learning correction system, characterized in that, The wind turbine yaw error self-learning correction system includes: The data acquisition module is used to collect wind speed, wind direction, and yaw angle data in real time during wind turbine operation; The error determination module is used to obtain the actual yaw error of the wind turbine based on the wind speed data, the wind direction data, and the yaw angle data. The feature extraction module is used to determine the error feature set based on historical yaw error, and to obtain the error evolution feature according to the matching length between the actual yaw error and the error feature set; The parameter correction module is used to generate yaw correction parameters by self-learning based on the error evolution characteristics and the yaw angle data. The wind turbine control module is used to control the wind turbine according to the yaw correction parameters in order to correct the yaw error; The feature extraction module is further configured to extract historical yaw error samples from historical operational data in different wind direction and wind speed ranges to form an error feature set; determine the sequence of actual yaw errors and each feature sequence in the error feature set, calculate the matching degree between the sequence of actual yaw errors and each feature sequence in the error feature set; select the sequence of actual yaw errors with a matching degree exceeding a preset threshold as the target sequence, and determine the matching length based on the number of consecutive data points matching between the target sequence and each feature sequence in the error feature set; and obtain the error evolution characteristics based on the matching length and the corresponding historical error evolution trend.

Images