A wind turbine pitch control method and system based on wind speed correction

Abstract

This invention discloses a wind turbine pitch control method and system based on wind speed correction, relating to the field of wind turbine pitch control. The method includes: predicting a multi-moment wind speed sequence and its confidence interval within a future set time window; optimizing this sequence with a dual objective of minimizing pitch angle change and maximizing the expected total output power; and calculating a pitch angle adjustment sequence. When real-time monitoring detects that the actual wind speed deviates from the predicted confidence interval, the pitch angle adjustment sequence is immediately replanned starting from the current actual wind speed, and the deviation between the actual and expected output power is evaluated, dynamically adjusting the weight allocation of the optimization objectives. This invention establishes control decisions based on predictions over continuous time periods and effectively reduces mechanical wear of the pitch system through real-time replanning and weight adjustment.

Description

Technical Field

[0001] This invention belongs to the field of wind turbine pitch control technology, and relates to a wind turbine pitch control method and system based on wind speed correction. Background Technology

[0002] Wind power generation is an important form of renewable energy, and its power generation efficiency and operational stability are highly dependent on the capture and control of wind energy resources. Pitch control changes the aerodynamic characteristics of the wind turbine by adjusting the blade pitch angle, thereby maintaining the stability of the unit's speed and output power when wind speed changes, and ensuring equipment safety.

[0003] Currently, existing technologies have proposed methods for wind turbine pitch control. For example, the invention patent with publication number CN117967509A proposes an intelligent pitch control method, device, and storage medium for wind turbines. Based on historical wind speed and direction data, it obtains the predicted wind speed and direction values ​​for the next time point through a neural network predictor, and generates the pitch angle control value accordingly to solve the control lag problem, thereby reducing grid fluctuations and ensuring the output efficiency of wind turbines.

[0004] However, although the existing solutions mentioned above have certain effects on pitch control of wind turbine generators, they still have the following shortcomings: First, in wind power generation scenarios, wind speed exhibits continuous and variable fluctuations over time. Short-term abrupt changes and long-term trends have different impacts on pitch angle control. Existing technologies only predict wind characteristics for the next time point and do not fully consider the overall trend and fluctuation range of wind speed changes over the next continuous time period. This may result in control commands responding only to instantaneous wind speeds, lacking global optimization of wind speed temporal changes, and affecting the unit's operational stability and power output performance in the continuous time domain.

[0005] Secondly, the pitch angle adjustment process is physically limited by the actuator. Frequent and large-amplitude pitch changes will aggravate mechanical wear and affect the equipment life. Existing technologies are mostly based on single-point wind speed prediction, which leads to a step change in pitch angle adjustment. This causes frequent pitch angle oscillations or overshoots in actual control, increasing mechanical load and affecting equipment life. Summary of the Invention

[0006] In view of this, in order to solve the problems mentioned in the background art, the present invention provides a wind turbine pitch control method and system based on wind speed correction.

[0007] The objective of this invention can be achieved through the following technical solutions: In a first aspect, this invention provides a wind turbine pitch control method based on wind speed correction, comprising: S1, performing time-series prediction based on historical wind speed data to obtain a predicted wind speed sequence and corresponding confidence intervals for multiple consecutive moments within a future set time window.

[0008] S2. Based on the predicted wind speed sequence, with the goal of minimizing the change in pitch angle between adjacent moments and maximizing the total expected output power, the pitch angle at each moment within the future set time window is optimized and the pitch angle adjustment sequence is determined.

[0009] S3. Send the first pitch angle value in the pitch angle adjustment sequence as the target command for the current control cycle to the pitch actuator to drive the blades to rotate to the corresponding angle.

[0010] S4. Monitor the actual wind speed at the wind turbine in real time. If the actual wind speed deviates from the confidence interval of the predicted wind speed sequence, immediately start from the current actual wind speed and repeat the above steps to generate and execute a new pitch angle adjustment sequence.

[0011] S5. During continuous operation, evaluate the deviation between the actual output power and the sum of the expected output power, and adjust and optimize the weight allocation of the target calculation based on the magnitude of the deviation.

[0012] Secondly, the present invention provides a wind turbine pitch control system based on wind speed correction, including the following modules: a wind speed prediction module, used to predict the wind speed sequence and confidence interval at multiple moments within a future set time window based on historical wind speed data.

[0013] The pitch angle optimization module is used to predict wind speed sequences with the goal of minimizing the pitch angle change between adjacent moments and maximizing the total expected output power. It optimizes the calculation of the pitch angle at each moment and determines the adjustment sequence.

[0014] The blade adjustment module is used to drive the blades to rotate according to the first value of the blade pitch angle adjustment sequence.

[0015] The pitch angle replanning module is used to monitor the actual wind speed of the wind turbine in real time. If it deviates from the confidence interval, a new adjustment sequence is generated and executed starting from the current actual wind speed.

[0016] The weight allocation module is used to evaluate the deviation between the actual and expected output power, and adjust and optimize the target weight allocation based on the deviation.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention performs multi-time wind speed sequence prediction and its confidence interval for a future set time window, and optimizes the sequence based on the dual objectives of minimizing the pitch angle change and maximizing the expected total output power, calculates the pitch angle adjustment sequence, so that the pitch angle control command is globally planned based on the wind trend over a future period of time, rather than responding only to the next moment, which solves the problem of insufficient control foresight caused by the prior art only predicting the next time point, thereby improving the unit's operational stability and power output performance under dynamic wind conditions.

[0018] (2) When the actual wind speed deviates from the predicted confidence interval during real-time monitoring, the present invention immediately re-plans the pitch angle adjustment sequence based on the current actual wind speed, evaluates the deviation between the actual output power and the expected output power, and dynamically adjusts the weight allocation of the optimization target. This effectively avoids the step change of the pitch angle command caused by the sudden change of the predicted wind speed at a single point, achieves smooth adjustment of the pitch angle, reduces the mechanical wear of the pitch system, and improves the service life of the equipment. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a diagram illustrating the implementation steps of the method of the present invention.

[0021] Figure 2 This is a flowchart of the wind speed prediction logic of the present invention.

[0022] Figure 3 This is a schematic diagram showing the connections of the various modules in the system of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figure 1 As shown, the first aspect of the present invention provides a wind turbine pitch control method based on wind speed correction, including: S100, performing time-series prediction based on historical wind speed data to obtain a predicted wind speed sequence for multiple consecutive moments within a future set time window.

[0025] The specific future setting time window can be determined through the following steps: First, the window length must be greater than the dynamic response time of the wind turbine pitch control system, typically with a lower limit of no less than 30 seconds, to ensure that the optimization has forward-looking guidance; second, the window length must be less than the effective predictable time scale of the wind speed trend, typically with an upper limit of no more than 300 seconds, to avoid optimization based on predictions with excessively low confidence. Considering both wind turbine characteristics and wind field statistical features, the preferred length of the future setting time window is a fixed value between 30 and 300 seconds, such as 60 seconds or 120 seconds. Using a time window within this range ensures that the global optimization of the pitch angle adjustment sequence can be completed within a period when the wind speed prediction information is still reliable.

[0026] See Figure 2 As shown, the steps to obtain the predicted wind speed sequence for multiple consecutive moments within a future set time window are as follows: calculate the linear trend slope and standard deviation of the historical wind speed sequence, which are used as the trend intensity feature and fluctuation intensity feature, respectively.

[0027] Specifically, a linear fit is performed on a historical wind speed sequence, such as the historical wind speed sequence of the past 5 minutes. The trend intensity feature is calculated to reflect the trend and rate of wind speed increase or decrease; the fluctuation intensity feature is calculated to reflect the magnitude of random fluctuations in wind speed around the trend line. These two features respectively describe the deterministic and random components of wind speed changes.

[0028] If historical wind speed data is insufficient when the system starts up, the rated wind speed can be used as the initial prediction sequence, or the system can wait for sufficient data collection before entering the optimization calculation mode.

[0029] By comparing the absolute value of the trend intensity characteristic with the set resolution constant, it is determined whether the current wind speed is stable, rising, or falling.

[0030] The set resolution constant is used to distinguish between the actual wind speed change trend and random fluctuations caused by measurement noise and natural turbulence, ensuring that the state judgment only responds to changes exceeding the background noise level. Its determination method includes: first, calculating the fluctuation range of the wind speed change rate under stable weather conditions based on statistical analysis of long-term historical wind speed data of the target wind field; then, taking the standard deviation of this fluctuation range as the initial value of the set resolution constant. In practical applications, the value of the set resolution constant can be fine-tuned according to the meteorological characteristics of the wind turbine location and the measurement noise level of the wind speed sensor, typically between 0.05 m / s² and 0.2 m / s².

[0031] The determination of whether the wind speed is stable, rising, or falling can be based on the most suitable prediction method for different states. The rules are as follows: when the absolute value of the trend intensity feature is less than or equal to the set resolution constant, the current wind speed is determined to be in a stable state; this indicates that the wind speed within the historical window has not shown a directional trend of change, and its changes are mainly dominated by random fluctuations.

[0032] When the absolute value of the trend intensity feature is greater than the set resolution constant and the value of the trend intensity feature is positive, it is determined that the current wind speed is in an upward state; indicating that the wind speed within the historical window not only changes significantly, but also shows a clear increasing trend.

[0033] When the absolute value of the trend intensity feature is greater than the set resolution constant and the value of the trend intensity feature is negative, it is determined that the current wind speed is in a decreasing state; indicating that the wind speed within the historical window not only changes significantly, but also shows a clear decreasing trend.

[0034] The future time window is divided into a near-term segment and a long-term segment. For the near-term segment, if the current state is stable, a constant value is selected as the predicted wind speed value.

[0035] In one embodiment of the present invention, a fixed time length ratio is used for division. Generally, a future time window of 0.3 to 0.5 is selected as the near-term segment, and the remaining time window is the long-term segment.

[0036] For the recent period, if the historical wind speed data is determined to be in a stable state, the moving average of the wind speed within the recent historical window is calculated, and the moving average is used as the constant value for prediction. Essentially, the current state is used as the predictor.

[0037] If the wind speed is rising or falling, the predicted wind speed value is obtained by linear extrapolation using the trend strength characteristic as the slope.

[0038] Specifically, the predicted wind speed is calculated using the following linear extrapolation formula: Where k represents the trend intensity characteristic, v0 represents the current wind speed value, and t i Let t0 be any future discrete prediction time within the recent period, i be the predicted value within the recent period (i = 1, 2, ..., N), N be the prediction step within the recent period, and t0 be the current time. This formula shows that the predicted wind speed value starts from the current wind speed value, increasing at a constant rate over time during an upward trend and decreasing at a constant rate over time during a downward trend. When a sustained upward or downward trend in wind speed is detected, this trend often has inertia and will continue for some time in the recent period, assuming no sudden weather system influence.

[0039] For the long-term segment, the forecast value at the end of the short-term segment is used as the starting point. If the current state is stable, the forecast wind speed value for the long-term segment remains at the starting point value.

[0040] If the wind speed is rising or falling, the extrapolation slope decays to zero according to a negative exponential law, generating the predicted wind speed value for the long term.

[0041] Specifically, the predicted wind speed for the long term is determined by the following formula: , where v n The near-term end-of-range wind speed forecast is used as the starting point. λ is a negative attenuation coefficient used to control the rate of slope attenuation. Its value can be determined based on the statistical characteristics of trend persistence in historical data; the larger λ is, the faster the trend attenuation. k represents the trend strength characteristic, and t... i Let t0 be any future discrete prediction time within the recent segment, i be the prediction value within the recent segment, i = 1, 2, ..., N, N be the prediction step number within the recent segment, and t0 be the current time.

[0042] This represents the maximum or minimum total change in wind speed that is expected to reach before the trend completely decays, limiting long-term forecasts to a physically reasonable finite range and avoiding linear extrapolation to infinity.

[0043] It is a decay function between 0 and 1, indicating that the influence of a trend decays exponentially over time. The further away from the starting point, the smaller the impact of the current trend on the prediction.

[0044] It is a saturation curve function that monotonically increases from 0 to 1, describing the curve from its starting point to its asymptotic value. The progress made so that the wind speed forecast values ​​approached the equilibrium point in a non-linear manner, which was faster at first and then slower, which is conducive to generating smooth pitch commands.

[0045] From the starting point to t i The total predicted change in wind speed up to that point.

[0046] Apply non-negativity constraints and safety extreme value constraints to the preliminary predicted wind speed sequence, and then perform smoothing filtering on the constrained sequence to obtain the final predicted wind speed sequence.

[0047] To ensure that all predicted wind speed values ​​are non-negative and do not exceed the wind turbine's safety limit, a low-pass filter is applied to the constrained prediction sequence to guarantee that the prediction output conforms to physical principles, eliminate abrupt changes in predicted values ​​caused by data noise, smooth the prediction curve, and avoid oscillations in the results due to jumps in predicted values. The wind turbine's safety limit is determined according to the wind turbine's production manual.

[0048] S101. Based on historical wind speed data, perform time-series prediction to obtain confidence intervals for multiple consecutive moments within a future set time window.

[0049] Specifically, if the state is determined to be stable, an equal-width confidence interval is used, which is the larger of the fluctuation intensity characteristic and the fluctuation standard deviation of historical statistical wind speed data.

[0050] When the wind speed sequence is determined to be in a steady state, it is considered to be an approximately stationary random process. The distribution of the prediction error theoretically does not shift significantly with time. Therefore, using equal-width confidence intervals and taking the larger of the two values ​​ensures that the constructed confidence interval can cover the true value of future wind speed with a high probability.

[0051] If the trend is upward or downward, a confidence interval with a narrow front and a wide back is used. The interval width at each time point is linearly increased by adjusting the trend strength feature as the adjustment coefficient to obtain the interval width at each time point.

[0052] When wind speed is trending upwards or downwards, it is extrapolated from historical trends, and the prediction error accumulates and amplifies as the prediction time period lengthens. Therefore, the width of the confidence interval should increase over time to reflect this objective law of error growth.

[0053] A stronger trend indicates a faster rate of wind speed change, a higher risk of deviation of the prediction model from the initial conditions per unit time, and a correspondingly faster rate of error accumulation. This linear correlation means that the dynamic expansion of the confidence interval is directly proportional to the severity of the trend.

[0054] Using the predicted wind speed at each moment as the center, the initial boundary is constructed by the half-width of the interval at the corresponding moment. The lower boundary value is the larger of zero and the lower limit of the initial boundary, and the upper boundary value is the smaller of the wind speed cut off by the wind turbine and the upper limit of the initial boundary.

[0055] As a measure of atmospheric flow speed, wind speed cannot be negative in physics. Therefore, by taking the larger of zero and the initial lower boundary as the final lower boundary, the lower edge of the interval is forcibly raised to zero or above, ensuring the physical reality of the cognitive range.

[0056] Wind turbines are designed with an inherent upper limit for safe operating wind speed, known as the cut-out wind speed. When the wind speed exceeds this cut-out wind speed, the wind turbine must execute a shutdown protection procedure to prevent structural damage. Therefore, by taking the smaller value between the cut-out wind speed and the initial upper boundary as the final upper boundary, the upper edge of the interval is forcibly limited to the cut-out wind speed or below, ensuring that all predicted scenarios are within the normal operating range of the wind turbine.

[0057] S200: Based on the predicted wind speed sequence, with the goal of minimizing the change in pitch angle between adjacent moments and maximizing the total expected output power, the pitch angle is optimized and calculated for each moment within a future set time window to determine the pitch angle adjustment sequence.

[0058] The steps for determining the pitch angle adjustment sequence are as follows: the pitch angle at each moment within the future set time window is set as the variable to be determined; the allowable physical angle range of the wind turbine pitch angle is used as the boundary constraint for the value of each variable to be determined; and the maximum allowable angle change rate of the pitch actuator is used as the boundary constraint for the change between adjacent variables to be determined.

[0059] With minimizing the sum of pitch angle changes at all adjacent moments as the sole optimization objective, and under the premise of satisfying the value boundary constraints and change boundary constraints, the first-level optimization calculation is performed, and the minimum total change obtained from this optimization calculation is recorded.

[0060] The value boundary constraint means that each variable must be within its maximum and minimum values, and the maximum and minimum values ​​are determined by the physical structural limits of the blade; the change boundary constraint means that the change in pitch angle at any adjacent time is less than or equal to the maximum angle change rate that the pitch actuator can achieve in a single control cycle.

[0061] Pitching is a heavy machinery motion, and frequent angle changes directly lead to accelerated mechanical wear of the actuator, generate additional energy loss, and may induce structural vibration. Therefore, from the perspective of equipment lifespan, priority should be given to selecting the smallest total change in motion with the smoothest movement.

[0062] All pitch angle sequences that satisfy the constraints and achieve the minimum total change are collected into a first-level optimization solution set.

[0063] Within the first-level optimization solution set, the second-level optimization calculation is performed with the goal of maximizing the sum of expected output power at each time step.

[0064] From the first-level optimization solution set, select the pitch angle sequence that maximizes the total expected output power and determine it as the final pitch angle adjustment sequence.

[0065] By separating equipment protection objectives from power generation revenue objectives at the source, and through priority ranking and sequential optimization, the irreconcilable trade-offs in single-weighted objective optimization are completely avoided, ensuring the smoothest equipment operation. At the same time, under the stability constraint, the power generation is maximized, and the two objectives achieve optimal synergy.

[0066] S300: The first pitch angle value in the pitch angle adjustment sequence is sent as the target command for the current control cycle to the pitch actuator, driving the blades to rotate to the corresponding angle.

[0067] The steps for driving the propeller blades to rotate to the corresponding angle are as follows: extract the first pitch angle value from the determined pitch angle adjustment sequence, and use it as the target pitch angle command for the current control cycle.

[0068] Extracting and executing the first pitch angle value in the sequence demonstrates that the control system always makes current control decisions based on the latest optimization results. However, wind conditions are time-varying and uncertain, and the actual wind speed in the future may deviate from the prediction. Therefore, in each control cycle, the control strategy re-solves the finite-time domain optimization problem based on the latest wind speed, pitch angle, and prediction information to obtain a new optimal plan sequence, but only the control command corresponding to the first time point in the nearest future in this plan sequence is implemented.

[0069] The target pitch angle command is compared with the current actual pitch angle fed back by the pitch actuator, and the angle deviation is calculated.

[0070] Based on the angle deviation, the pitch controller generates a corresponding drive signal and sends it to the power unit of the pitch actuator.

[0071] The power unit drives the blades to rotate according to the drive signal until the actual pitch angle reaches the target pitch angle command, and then sends back a confirmation signal.

[0072] S400: Real-time monitoring of the actual wind speed at the wind turbine.

[0073] Specifically, the real-time wind speed measurement value obtained by the wind speed sensor at the wind turbine and the real-time power value generated by the wind turbine are acquired simultaneously.

[0074] The real-time power value of the wind turbine is obtained through the power measurement unit of the wind turbine power generation system.

[0075] To ensure the synchronous acquisition of real-time wind speed and real-time power values, a common clock source is used to generate a unified sampling trigger pulse, driving all channels to sample and perform analog-to-digital conversion at the same time. This results in real-time wind speed and real-time power values ​​with strictly aligned timestamps, and the sampled data is temporarily stored in a buffer register.

[0076] The power value of the wind turbine at zero pitch angle, measured in real time with wind speed, is used as the theoretical upper limit of power.

[0077] The power characteristic curve of a wind turbine describes the theoretical upper limit of its efficiency in converting wind energy into mechanical energy and then into electrical energy at a specific wind speed and pitch angle. When the pitch angle is zero degrees, the blade's angle of attack is at its optimal design state, at which point the wind energy capture coefficient reaches its theoretical maximum value. Therefore, for any given real-time wind speed measurement, the power value corresponding to it on the zero-pitch angle power characteristic curve represents the absolute maximum possible electrical power that the wind turbine can output under ideal operating conditions at that wind speed.

[0078] It should be added that the implementation of this method depends on the power characteristic curve of the wind turbine, which is provided by the wind turbine manufacturer or obtained through field testing and calibration.

[0079] The power value of the wind turbine at the current actual pitch angle, measured in real time, is used as the lower limit of the theoretical power.

[0080] In practical applications, the power characteristic curve can be calibrated periodically based on the actual operating data of the wind turbine. For slight deviations in power caused by changes in air density, mild turbulence, etc., a reasonable threshold buffer zone, such as ±10% of the theoretical power value, can be set to avoid false triggering.

[0081] When a wind turbine is in normal power generation mode, its actual output power is mainly governed by two factors: one is the real-time wind energy determined by the wind speed, and the other is the energy conversion state of the turbine determined by the actual pitch angle. The family of power characteristic curves of a wind turbine depicts the mapping relationship between wind speed and theoretical output power at different pitch angles.

[0082] For any given real-time wind speed measurement and current actual pitch angle, the power value obtained by querying the power characteristic curve of the corresponding pitch angle represents the minimum possible electrical power that should theoretically be generated at that wind speed under the current aerodynamic configuration of the wind turbine.

[0083] The power value shown on the power characteristic curve of the wind turbine at the current actual pitch angle, based on real-time wind speed measurement, does not account for all mechanical and electrical losses. Therefore, in actual operation, due to various losses, the power is usually slightly lower than this value, but this will not result in a significant deviation.

[0084] If the real-time power value is greater than the lower limit of the theoretical power but less than the upper limit of the theoretical power, the real-time wind speed measurement value is marked as reliable; otherwise, it is marked as pending verification. This effectively identifies single or multiple faults such as wind speed sensor reading drift or jamming, abnormal power measurement, or incorrect pitch angle feedback.

[0085] The real-time power value is greater than the lower limit of the theoretical power but less than the upper limit of the theoretical power, indicating that the measured power is consistent with the theoretical power range expected based on the current wind speed and pitch angle. Therefore, the wind speed measurement reliably reflects the current actual wind conditions.

[0086] Conversely, if the measured power deviates from the theoretical range expected by the current measurement parameters, this inconsistency indicates that at least one measurement link is abnormal, faulty, or severely interfered with.

[0087] S401. If the actual wind speed deviates from the confidence interval of the predicted wind speed sequence, the above steps are immediately repeated starting from the current actual wind speed to generate and execute a new pitch angle adjustment sequence.

[0088] The steps for generating and executing a new pitch angle adjustment sequence are as follows: obtain the actual wind speed measurement value at the wind turbine at the current moment and its confidence mark, as well as the prediction wind speed confidence interval corresponding to the current moment.

[0089] If the actual wind speed value is marked as reliable, and its value is less than the lower boundary value of the confidence interval at the current time, or greater than the upper boundary value, then it is determined that a deviation has occurred, triggering replanning.

[0090] A confidence interval is a range of uncertainty added to the predicted wind speed value. Its upper and lower boundaries define the reasonable range of fluctuations in future wind speed under the current prediction model. When the actual wind speed value, which has been verified as credible, still falls outside this range, it indicates that the actual wind conditions have deviated from the most probable evolution path described by the prediction model and its expected range of uncertainty.

[0091] This deviation indicates that the preconditions of the pitch angle optimization sequence based on the original prediction are no longer met. Continuing to execute it will not be able to guarantee the optimization target of control, and may even cause safety hazards. Therefore, it is necessary to trigger replanning immediately.

[0092] To avoid replanning oscillations caused by instantaneous measurement noise or minor fluctuations, a replanning trigger condition is set: within M consecutive control cycles, the reliable actual wind speed value deviates from the confidence interval at the current time, where M can be 2 to 3. Replanning can only be triggered after the condition is met.

[0093] Once a replanning is triggered, the ongoing pitch angle adjustment sequence is immediately paused, and the trigger time is marked as a reliable actual wind speed value, serving as the starting point for new historical wind speed data.

[0094] Triggered replanning indicates that the original predictions based on historical data have deviated significantly from reality. At this point, continuing to use old historical data that includes outdated or invalid trends will lead to a systematic bias in the starting point of the new predictions.

[0095] Marking the trigger time as a reliable actual wind speed value as the starting point for new historical data serves to eliminate accumulated errors and cut off erroneous trend information in old data that leads to prediction deviations.

[0096] Using the updated data starting point as a reference, a new predicted wind speed sequence and its corresponding confidence interval are regenerated. The new predicted wind speed sequence and its corresponding confidence interval are determined according to steps S100 and S101.

[0097] Based on the new predicted wind speed sequence, a new pitch angle adjustment sequence is regenerated, and the blades are driven to move according to the new sequence. The new pitch angle adjustment sequence is determined according to step S200.

[0098] S500: During continuous operation, assess the deviation between the actual output power and the sum of the expected output power.

[0099] The steps for evaluating the deviation between the actual output power and the expected output power are as follows: obtain the actual output power and the expected output power at each time point, and calculate the power deviation sequence.

[0100] Time-frequency analysis was performed on the power deviation sequence to decompose it into frequency band components representing the variation characteristics at different time scales.

[0101] Specifically, the power deviation sequence is first subjected to DC removal processing, that is, its mean is calculated and subtracted from each data point to obtain a deviation sequence with zero mean, thereby eliminating the influence of constant offset on subsequent frequency domain analysis.

[0102] Then, discrete wavelet transform is used as a time-frequency analysis tool. Key parameters are set as follows: the wavelet basis function is selected from wavelets with tight support and appropriate regularity, such as the Daubechies 4th order wavelet; the number of decomposition levels, J, is determined based on the time scale of interest and signal sampling frequency of the wind turbine control system. For example, J=4 is set.

[0103] Next, the preprocessed zero-mean deviation sequence is input into the discrete wavelet transform algorithm. The signal is iteratively decomposed by cascaded high-pass and low-pass filter banks and combined with a decimation operation.

[0104] Finally, after J-level decomposition, the original signal is resolved into J detail coefficient sequences, where the detail coefficient sequence of the j-th level is D. j Carrying the original signal at scale 2 j The fluctuation information on the scale, D1 represents the fastest fluctuations at the finest scale; one approximate coefficient sequence A J Carrying the original signal at scale 2 j Information on gradual changes and above.

[0105] The above J+1 coefficient sequences together constitute the multi-scale frequency band component of the power deviation.

[0106] The origins and physical meanings of different frequency components in the power deviation signal are different: the high-frequency components are mainly caused by rapid disturbances such as wind speed turbulence that are not effectively suppressed by the controller or cause overshoot, reflecting the dynamic response bandwidth and short-term tracking performance of the pitch system; the low-frequency components mainly originate from the systematic tracking error of the prediction model on the long-term trend of wind speed, reflecting the accuracy of feedforward prediction.

[0107] By using discrete wavelet transform for multi-scale decomposition, the overall deviation of temporal aliasing can be effectively separated and traced according to its physical generation time scale, thereby transforming comprehensive performance evaluation into an assessment of control problems with different root causes.

[0108] The total energy of the power deviation sequence is calculated, and the proportion of energy of each frequency band component to the total energy is calculated to obtain the multi-scale distribution characteristics of the deviation energy.

[0109] By comparing the multi-scale distribution characteristics with those of the same historical period, if the proportion of high-frequency band energy increases, it is determined that the short-term fluctuation response is poor; if the proportion of low-frequency band energy increases, it is determined that the long-term trend tracking is poor.

[0110] The multi-scale energy distribution characteristics of power deviation involve decomposing the total time-domain deviation according to its dynamic rate of change. High-frequency energy concentration indicates the rapidly changing portion of the deviation signal, primarily corresponding to the control system's response to and suppression of short-term and rapid disturbances. An increase in the proportion of this energy indicates that the actual power output cannot closely follow these rapid changes. The root cause lies in the insufficient dynamic response speed of the pitch actuator, inappropriate short-term tracking gain of the control loop, or insufficient damping, resulting in poor short-term fluctuation response.

[0111] Conversely, the low-frequency energy concentration reflects the slowly changing trend portion of the deviation signal, and its physical root cause mainly corresponds to the accuracy of the prediction-based feedforward control in tracking the long-term evolution trend of wind speed. If the proportion of this energy increases, it indicates a deviation between the actual power output and the expected value based on long-term prediction. The root cause lies in the persistent bias of the wind speed prediction model, or the optimizer being too conservative in tracking the long-term trend, i.e., poor long-term trend tracking.

[0112] S501. Adjust and optimize the weight allocation of the target calculation based on the magnitude of the deviation.

[0113] The weight allocation has adjustment range limits and change rate limits. After each adjustment, the new weight will remain unchanged for at least 5 minutes to observe the control effect and avoid frequent adjustments.

[0114] The specific implementation steps of the present invention are as follows: if it is determined that the short-term fluctuation response is poor, the weight of minimizing the change in pitch angle between adjacent moments is increased, while the weight of maximizing the total expected output power is reduced.

[0115] When the short-term fluctuation response is deemed poor, it indicates insufficient ability to suppress high-frequency disturbances. Essentially, the pitch angle changes are too aggressive or frequent, causing the actual power output to fail to smoothly track the rapidly changing wind speed.

[0116] By increasing the weight of minimizing the pitch angle change between adjacent moments, it is equivalent to adding equivalent damping to the control loop, thereby suppressing high-frequency disturbance response and smoothing power output. At the same time, reducing the weight of maximizing the sum of expected output power reduces instantaneous power tracking, and at the cost of acceptable short-term power loss, gains a greater margin for smoothness of action and avoids frequent actions that aggravate wear.

[0117] If the long-term trend tracking is deemed poor, the weight of maximizing the total expected output power is increased, while the weight of minimizing the pitch angle change between adjacent time points is decreased.

[0118] When a long-term trend tracking error is identified as poor, the power maximization weight is increased to incentivize more forward-looking pitch angle adjustments to capture trend performance; at the same time, the change minimization weight is reduced to provide a constraint margin for necessary forward-looking actions, thus jointly correcting systematic tracking bias.

[0119] Compared to the same period in previous years, the greater the increase in the proportion of energy in a specific frequency band during the current assessment period, the higher the level of weight change in the corresponding adjustment direction.

[0120] The increase in the energy percentage indicates the severity of the current performance defects. The more severe the defects, the more significant the mismatch between the original control strategy and the current actual wind conditions. Therefore, a more substantial adjustment to the weight allocation of the optimization objectives is needed to generate a sufficiently strong corrective effect.

[0121] It should be added that when evaluating the multi-scale distribution characteristics of power deviation, if the energy proportions of both the high-frequency and low-frequency bands increase simultaneously, it indicates that the system is simultaneously facing problems of insufficient short-term fluctuation suppression and inaccurate long-term trend tracking. In this case, a composite adjustment strategy can be adopted: first, to ensure equipment safety and lifespan, prioritize increasing the weight of minimizing pitch angle changes to suppress high-frequency movements; under this constraint, further increase the weight of maximizing the total expected output power, but the adjustment range should be lower than that when only poor long-term trend tracking occurs.

[0122] refer to Figure 3 As shown, a second aspect of the present invention provides a wind turbine pitch control system based on wind speed correction, including a wind speed prediction module, a pitch angle optimization module, a blade adjustment module, a pitch angle replanning module, and a weight allocation module. All modules are connected in the order described above.

[0123] Specifically, the wind speed prediction module is used to predict wind speed sequences and confidence intervals at multiple moments within a set future time window based on historical wind speed data.

[0124] The pitch angle optimization module is used to predict wind speed sequences with the goal of minimizing the pitch angle change between adjacent moments and maximizing the total expected output power. It optimizes the calculation of the pitch angle at each moment and determines the adjustment sequence.

[0125] The blade adjustment module is used to drive the blades to rotate according to the first value of the blade pitch angle adjustment sequence.

[0126] The pitch angle replanning module is used to monitor the actual wind speed of the wind turbine in real time. If it deviates from the confidence interval, a new adjustment sequence is generated and executed starting from the current actual wind speed.

[0127] The weight allocation module is used to evaluate the deviation between the actual and expected output power, and adjust and optimize the target weight allocation based on the deviation.

[0128] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0129] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0130] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0131] The above description is merely a specific embodiment 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.

[0132] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A wind turbine pitch control method based on wind speed correction, characterized in that: include: Based on historical wind speed data, time-series predictions are made to obtain a sequence of predicted wind speeds and corresponding confidence intervals for multiple consecutive moments within a set future time window. Based on the predicted wind speed sequence, with the goal of minimizing the change in pitch angle between adjacent moments and maximizing the total expected output power, the pitch angle at each moment within a future set time window is optimized and the pitch angle adjustment sequence is determined. The first pitch angle value in the pitch angle adjustment sequence is sent to the pitch actuator as the target command for the current control cycle, driving the blades to rotate to the corresponding angle. The actual wind speed at the wind turbine is monitored in real time. If the actual wind speed deviates from the confidence interval of the predicted wind speed sequence, the above steps are immediately repeated starting from the current actual wind speed to generate and execute a new pitch angle adjustment sequence. During continuous operation, the deviation between the actual output power and the sum of the expected output power is evaluated, and the weight allocation of the optimization calculation target is adjusted according to the magnitude of the deviation. The steps to obtain the predicted wind speed sequence are as follows: calculate the linear trend slope and standard deviation of the historical wind speed sequence, which are used as the trend intensity feature and fluctuation intensity feature, respectively; By comparing the absolute value of the trend intensity feature with a set resolution constant, the current wind speed is determined to be stable, rising, or falling. The future set time window is divided into a near-term segment and a long-term segment. For the near-term segment, if the current state is stable, a constant value is selected as the predicted wind speed value; if the state is rising or falling, the trend intensity feature is used as the slope for linear extrapolation to obtain the predicted wind speed value. For the long-term segment, the predicted value at the end of the near-term segment is used as the starting point. If the current state is stable, the predicted wind speed value for the long-term segment remains at the starting point; if the state is rising or falling, the extrapolation slope decays to zero according to a negative exponential law to generate the predicted wind speed value for the long-term segment. Wind speed non-negativity constraints and safety extreme value constraints are applied to the preliminary predicted wind speed sequence, and the constrained sequence is smoothed and filtered to obtain the final predicted wind speed sequence.

2. The wind turbine pitch control method based on wind speed correction according to claim 1, characterized in that: The corresponding confidence interval is obtained through the following steps: If the state is determined to be stable, an equal-width confidence interval is used, which is the larger of the fluctuation intensity characteristic and the fluctuation standard deviation of the historical wind speed data for the same period. If the trend is upward or downward, a confidence interval with a narrow front and a wide back is used. The interval width at each time point is linearly increased with the trend strength feature as the adjustment coefficient to obtain the interval width at each time point. Using the predicted wind speed at each moment as the center, the initial boundary is constructed by the half-width of the interval at the corresponding moment. The lower boundary value is the larger of zero and the lower limit of the initial boundary, and the upper boundary value is the smaller of the wind speed cut off by the wind turbine and the upper limit of the initial boundary.

3. The wind turbine pitch control method based on wind speed correction according to claim 1, characterized in that: The determination of the pitch angle adjustment sequence includes: The pitch angle at each moment within the future set time window is set as the variable to be determined. The allowable physical angle range of the wind turbine pitch angle is used as the boundary constraint for the value of each variable to be determined. The maximum allowable angle change rate of the pitch actuator is used as the boundary constraint for the change between adjacent variables to be determined. With the goal of minimizing the sum of pitch angle changes at all adjacent time points, the first-level optimization calculation is performed under the premise of satisfying the value boundary constraints and the change boundary constraints, and the minimum total change obtained in this optimization calculation is recorded. All pitch angle sequences that satisfy the constraints and achieve the minimum total change are compiled into a first-level optimization solution set; Within the first-level optimization solution set, the second-level optimization calculation is performed with the goal of maximizing the sum of expected output power at each time step. From the first-level optimization solution set, select the pitch angle sequence that maximizes the total expected output power and determine it as the final pitch angle adjustment sequence.

4. The wind turbine pitch control method based on wind speed correction according to claim 1, characterized in that: The process of rotating the propeller blade to the corresponding angle includes: From the determined pitch angle adjustment sequence, extract the first pitch angle value as the target pitch angle command for the current control cycle; The target pitch angle command is compared with the current actual pitch angle fed back by the pitch actuator, and the angle deviation is calculated. Based on the angle deviation, the pitch controller generates a corresponding drive signal and sends it to the power unit of the pitch actuator. The power unit drives the blades to rotate according to the drive signal until the actual pitch angle reaches the target pitch angle command, and then sends back a confirmation signal.

5. The wind turbine pitch control method based on wind speed correction according to claim 1, characterized in that: The real-time monitoring of the actual wind speed at the wind turbine includes: Simultaneously acquire the real-time wind speed measurement value obtained by the wind speed sensor at the wind turbine, as well as the real-time electrical power value generated by the wind turbine; The power value of the wind turbine at zero pitch angle is taken as the theoretical upper limit of the power characteristic curve of the wind turbine and the real-time wind speed measurement value. The power value of the wind turbine at the current actual pitch angle is taken as the lower limit of the theoretical power. If the real-time power value is greater than the lower limit of the theoretical power but less than the upper limit of the theoretical power, the real-time wind speed measurement value is marked as reliable; otherwise, it is marked as pending verification.

6. The wind turbine pitch control method based on wind speed correction according to claim 5, characterized in that: The generation and execution of the new pitch angle adjustment sequence includes: Obtain the actual wind speed measurement value at the wind turbine at the current moment and its confidence mark, as well as the corresponding prediction wind speed confidence interval at the current moment; If the actual wind speed value is marked as reliable, and its value is less than the lower boundary value of the confidence interval at the current time, or greater than the upper boundary value, it is determined that a deviation has occurred and a replanning is triggered. Once replanning is triggered, the ongoing pitch angle adjustment sequence is immediately paused, and the trigger time is marked as a reliable actual wind speed value, serving as the starting point for new historical wind speed data. Based on the updated data starting point, a new predicted wind speed sequence and corresponding confidence interval are regenerated; Based on the new predicted wind speed sequence, a new pitch angle adjustment sequence is generated, and the blades are driven to move according to the new sequence.

7. The wind turbine pitch control method based on wind speed correction according to claim 1, characterized in that: The deviation between the assessed actual output power and the sum of the expected output power includes: Obtain the actual output power and expected output power at each time point, and calculate the power deviation sequence; Time-frequency analysis was performed on the power deviation sequence to decompose it into frequency band components representing the variation characteristics at different time scales; The total energy of the power deviation sequence is calculated, and the proportion of energy of each frequency band component to the total energy is calculated to obtain the multi-scale distribution characteristics of the deviation energy. By comparing the multi-scale distribution characteristics with those of the same historical period, if the proportion of high-frequency band energy increases, it is determined that the short-term fluctuation response is poor; if the proportion of low-frequency band energy increases, it is determined that the long-term trend tracking is poor.

8. The wind turbine pitch control method based on wind speed correction according to claim 7, characterized in that: The process of adjusting and optimizing the weight allocation of the target calculation based on the magnitude of the deviation includes: If the short-term fluctuation response is deemed poor, the weight of minimizing the pitch angle change between adjacent time points is increased, while the weight of maximizing the total expected output power is decreased. If the long-term trend tracking is deemed poor, the weight of maximizing the total expected output power is increased, while the weight of minimizing the change in pitch angle between adjacent time points is decreased. Compared to the same period in previous years, the greater the increase in the proportion of energy in a specific frequency band during the current assessment period, the higher the level of weight change in the corresponding adjustment direction.

9. A wind turbine pitch control system based on wind speed correction, used to execute the steps of a wind turbine pitch control method based on wind speed correction as described in any one of claims 1-8, characterized in that: include: The wind speed prediction module is used to predict wind speed sequences and confidence intervals at multiple moments within a set future time window based on historical wind speed data. The pitch angle optimization module is used to predict wind speed sequences with the goal of minimizing the pitch angle change between adjacent time moments and maximizing the total expected output power. It optimizes the calculation of the pitch angle at each time moment and determines the adjustment sequence. The blade adjustment module is used to drive the blades to rotate according to the first value of the blade pitch angle adjustment sequence; The pitch angle replanning module is used to monitor the actual wind speed of the wind turbine in real time. If it deviates from the confidence interval, a new adjustment sequence is generated and executed starting from the current actual wind speed. The weight allocation module is used to evaluate the deviation between the actual and expected output power, and adjust and optimize the target weight allocation based on the deviation.

Images