A wind turbine cluster yaw error correction method based on wake model

Abstract

The application relates to a yaw error correction method for a wind turbine group based on a wake model, which is characterized in that a downstream wind turbine is used as a sensor to realize group collaborative optimization by inversing the upstream error through a wake model; a correction wake model containing a yaw angle function is constructed to quantify the influence of an asymmetric wake on the downstream; the error is dynamically inversely calculated based on SCADA data, and stable online correction is realized by combining a gain coefficient k to avoid the hysteresis of single machine optimization; the operation data of the downstream wind turbine are used as the sensor to inversely deduce and correct the yaw error of the upstream wind turbine, so that the wake field of the whole group is optimized while the performance of the upstream wind turbine is improved, and finally the overall power generation of the wind farm is improved; the limitation of single machine correction is broken, the flow field and power generation performance of the whole group are optimized by correcting the error of the upstream wind turbine from the overall efficiency of the wind farm, the cost is low, the precision is high, the anti-interference capability is strong, the self-adaptability is strong, and the power generation is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of wind power generation technology, specifically to a yaw control method for wind turbine units, and particularly to a method for correcting the yaw error of upstream wind turbines in a wind turbine cluster based on a wind turbine wake interaction model. Background Technology

[0002] The yaw system is a key component of a horizontal axis wind turbine. Its function is to adjust the nacelle position according to the wind direction, ensuring the rotor faces the wind head-on to maximize wind energy capture. Yaw error refers to the angle between the rotor plane and the wind direction. Even the most advanced yaw systems have unavoidable measurement and execution errors, causing the actual yaw angle to deviate from the ideal angle. Even a yaw error of a few degrees can lead to a significant decrease in turbine power capture (typically following a cosine square relationship).

[0003] Currently, mainstream yaw error correction methods are mainly based on single-aircraft behavior, for example: 1. Direct measurement method based on wind vane: A wind vane is installed on the top of the nacelle to directly measure the wind direction and control the yaw. However, the wake and turbulence behind the nacelle can severely interfere with the accuracy of the wind vane measurement, resulting in "wind vane error".

[0004] 2. LiDAR Measurement Method: This method uses airborne or ground-based lidar to directly measure the direction of the incoming wind. While highly accurate, this method requires expensive equipment, incurring high maintenance costs, and is difficult to deploy on a large scale across all wind turbines.

[0005] 3. Data-driven optimization method: This method involves continuously fine-tuning the yaw angle to find the angle that maximizes the unit's power output. This method has a slow response time and is ineffective in turbulent conditions and with rapidly changing wind speeds. Frequent adjustments also increase wear and tear on the yaw system.

[0006] For wind turbine clusters in a wind farm, the yaw error of the upstream turbines not only affects their own power generation, but also severely interferes with the inflow conditions and power output of the downstream turbines due to the asymmetric wake they generate. Current technologies only focus on the error correction of the individual turbines, completely ignoring the "secondary damage" caused to the downstream turbines by the yaw error of the upstream turbines through the wake, resulting in the entire cluster's power generation efficiency failing to reach its optimal level.

[0007] Therefore, there is an urgent need in this field for a new method that can take into account the wake interaction between wind turbines from the perspective of the entire wind farm system, so as to correct yaw error more accurately and efficiently. Summary of the Invention

[0008] In view of the above situation and to overcome the shortcomings of the existing technology, the purpose of this invention is to provide a method for correcting the yaw error of wind turbine groups based on the wake model, which can effectively solve the problem of accurate and efficient correction of yaw error.

[0009] The technical solution solved by this invention is: A method for correcting yaw error in wind turbine clusters based on a wake model includes the following steps: Step S1: Identify the target correction fan and its associated downstream fans. In the wind farm, select a target upstream wind turbine that needs yaw error correction, denoted as turbine T. Based on the wind farm layout and the current prevailing wind direction, take upstream turbine T as the origin and draw a ray along the prevailing wind direction. In the fan-shaped area downstream of this ray, identify a wind turbine that is significantly affected by the wake of the target upstream wind turbine T, that is, the wind turbine that is closest to the upstream wind turbine in this sector, denoted as downstream wind turbine D. Step S2: Data Acquisition and Preprocessing Simultaneously collect the operating data of the upstream wind turbine T and the downstream wind turbine D within the same time period, including at least: The yaw angle γ_T, power P_T, and wind speed V_T of the upstream wind turbine T; The power P_D and wind speed V_D of the downstream wind turbine D; Step S3: Construct a wake model based on yaw angle A mathematical model is established to describe the relationship between the yaw angle of the upstream wind turbine and the wake velocity distribution at the location of the downstream wind turbine. This model is a modified wake model, and its expression is as follows: Where: V wake V_T is the predicted wake velocity at the hub height of downstream wind turbine D; C_T is the wind velocity at the hub height of upstream wind turbine T; d is the rotor diameter; x, y, z are the position coordinates of downstream wind turbine D relative to upstream wind turbine T; V w The predicted wind speed in the three-dimensional wake does not take into account the effect of yaw; k wake γ is the wake expansion coefficient; f(γ,d, x, y, z, I) is a function of the yaw angle γ, used to describe the wake deflection and velocity deficit caused by yaw. When the yaw angle γ=0, the model degenerates into a standard centrosymmetric wake model. Step S4: Invert the yaw error of the upstream wind turbine based on the measured data of the downstream wind turbine. 1. Forward calculation: Using the current yaw angle γ_T of the upstream wind turbine T as the initial value, input it into the wake model established in step S3 to calculate the predicted wind speed V at the hub height of the downstream wind turbine D. wake_cal ; 2. Construct the loss function: Combine the measured wind speed V_D of the downstream wind turbine D with the model-predicted wind speed V_D. wake_cal For comparison, the mean squared error is used to construct the function J(γ): 3. Optimization solution: Use optimization algorithms (such as gradient descent, Newton's method or genetic algorithm) to find the yaw angle γ_opt that minimizes the loss function J(γ). This γ_opt is the theoretically optimal yaw angle that best matches the measured data of the downstream wind turbine. 4. Calculate the yaw error: the yaw error of the upstream wind turbine T. That is: ; Step S5: Yaw Error Correction and Verification 1. The calculated yaw error Δγ is used as a correction value and sent to the main controller of the upstream wind turbine T; 2. The controller corrects the yaw angle setpoint of the wind turbine T: Here, k is a gain coefficient between 0 and 1, used to prevent overcorrection and ensure system stability.

[0010] 3. After correction, return to step S2 and continuously collect new operating data to verify the correction effect. If the power of downstream wind turbine D increases and / or the wake effect weakens, and the power of upstream wind turbine T itself does not decrease significantly, then the correction is considered effective; otherwise, the wake expansion coefficient k is determined by combining the real-time turbulence level with the correction. wake Make adjustments; This invention innovatively utilizes downstream wind turbines as "sensors" to invert upstream errors through a wake model, achieving coordinated optimization of the wind farm cluster. It constructs a corrected wake model incorporating a yaw angle function to quantify the impact of asymmetric wakes on the downstream. Based on SCADA data, it dynamically inverts errors and achieves stable online correction using a gain coefficient k, avoiding the lag of single-unit optimization. By using the operating data of downstream wind turbines as "sensors," it reverse-engineers and corrects the yaw errors of upstream wind turbines, thereby improving the performance of upstream turbines while optimizing the wake field of the entire wind farm cluster, ultimately increasing the overall power generation of the wind farm. Compared with existing technologies, this invention has the following beneficial technical effects: 1. Systematic optimization: This invention breaks through the limitations of single-unit correction. Starting from the overall efficiency of the wind farm, it optimizes the flow field and power generation performance of the entire wind farm by correcting the errors of the upstream wind turbines.

[0011] 2. Indirect measurement, low cost: Utilizing existing SCADA data from downstream wind turbines as an indirect measurement method eliminates the need for expensive additional sensors (such as lidar), greatly reducing implementation costs.

[0012] 3. High accuracy and strong anti-interference capability: The power and wind speed data of downstream wind turbines are less affected by the wake duct, and can more accurately reflect the incoming flow. The yaw error obtained by inverting the wake duct model can effectively avoid the "weather vane error" problem.

[0013] 4. Strong adaptability: This method can run continuously and adapt to changes in environmental conditions such as wind direction and wind speed, so as to achieve online and dynamic correction of yaw error.

[0014] 5. Significantly Increased Power Generation: By simultaneously optimizing the upstream wind turbine's own windward angle and the downstream wind turbine's inflow conditions, the overall power generation of the wind turbine cluster can be effectively increased. To address the above issues and overcome the shortcomings of existing technologies, the purpose of this invention is to provide a customized correction method for the entire nacelle transfer function. Attached Figure Description

[0015] Figure 1 This describes the closed-loop control process of the entire yaw error correction method of the present invention.

[0016] Figure 2 This is a layout diagram of the wind turbines in the wind farm.

[0017] Figure 3 This is a block diagram illustrating the principle of retrieving upstream wind turbine yaw error based on downstream wind turbine data. Detailed Implementation

[0018] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0019] like Figure 1-3 As shown, this invention discloses a method for correcting the yaw error of a wind turbine cluster based on a wake model, comprising the following steps: Step S1: Identify the target correction fan and its associated downstream fans. In the wind farm, select a target upstream wind turbine that needs yaw error correction, denoted as turbine T. Based on the wind farm layout and the current prevailing wind direction, take upstream turbine T as the origin and draw a ray along the prevailing wind direction. In the fan-shaped area downstream of this ray, identify a wind turbine that is significantly affected by the wake of the target upstream wind turbine T, that is, the wind turbine that is closest to the upstream wind turbine in this sector, denoted as downstream wind turbine D. The fan-shaped area is defined as a fan-shaped area within ±15° of the upstream wind turbine T as the origin and the ray along the prevailing wind direction as the reference.

[0020] Step S2: Data Acquisition and Preprocessing Simultaneously collect the operating data of the upstream wind turbine T and the downstream wind turbine D within the same time period, including at least: The yaw angle γ_T, power P_T, and wind speed V_T of the upstream wind turbine T; The power P_D and wind speed V_D of the downstream wind turbine D; The collected data is filtered, and invalid and abnormal data points are removed. Step S3: Construct a wake model based on yaw angle A mathematical model is established to describe the relationship between the yaw angle of the upstream wind turbine and the wake velocity distribution at the location of the downstream wind turbine. This model is a modified wake model, and its expression is as follows: Where: V wake V_T is the predicted wake velocity at the hub height of downstream wind turbine D; C_T is the wind velocity at the hub height of upstream wind turbine T; d is the rotor diameter; x, y, z are the position coordinates of downstream wind turbine D relative to upstream wind turbine T; V w The predicted wind speed in the three-dimensional wake does not take into account the effect of yaw; k wake is the wake expansion coefficient, which is taken as 0.75 in the wake model; f(γ, d, x, y, z, I) is a function of the yaw angle γ, used to describe the wake deflection and velocity deficit caused by yaw. When the yaw angle γ=0, the model degenerates into a standard centrosymmetric wake model. The parameter c is determined by the actual operating conditions of the wind turbine, and is taken as 2.58 based on the characteristics of Gaussian distribution. Step S4: Invert the yaw error of the upstream wind turbine based on the measured data of the downstream wind turbine. 1. Forward calculation: Using the current yaw angle γ_T of the upstream wind turbine T as the initial value, input it into the wake model established in step S3 to calculate the predicted wind speed V at the hub height of the downstream wind turbine D. wake_cal ; 2. Construct the loss function: Combine the measured wind speed V_D of the downstream wind turbine D with the model-predicted wind speed V_D. wake_cal For comparison, the mean squared error is used to construct the function J(γ): 3. Optimization solution: Use optimization algorithms (such as gradient descent, Newton's method or genetic algorithm) to find the yaw angle γ_opt that minimizes the loss function J(γ). This γ_opt is the theoretically optimal yaw angle that best matches the measured data of the downstream wind turbine. 4. Calculate the yaw error: the yaw error of the upstream wind turbine T. That is: ; Step S5: Yaw Error Correction and Verification 1. The calculated yaw error Δγ is used as a correction value and sent to the main controller of the upstream wind turbine T; 2. The controller corrects the yaw angle setpoint of the wind turbine T: Where k is a gain coefficient between 0 and 1, used to prevent overcorrection and ensure system stability. k is a gain coefficient between 0 and 1, preferably 0.8.

[0021] 3. After correction, return to step S2 and continuously collect new operating data to verify the correction effect. If the power of downstream wind turbine D increases and / or the wake effect weakens, and the power of upstream wind turbine T itself does not decrease significantly, then the correction is considered effective; otherwise, the wake expansion coefficient k is determined by combining the real-time turbulence level with the correction. wake Adjustments are made, including the wake expansion coefficient k. wake The value increases with the increase of turbulence level. The empirical value is 0.75, and the range is 0.5-1. After further optimization of the wake model, it is recalibrated.

[0022] To verify the accuracy of the established yaw error correction method for wind turbine clusters, it was tested at an 8MW distributed wind farm in a plain area. The wind farm has four turbines arranged in two rows, with the prevailing annual wind direction being westerly (270°). To ensure the accuracy of the test results, the test wind direction was selected with turbine No. 1 in the first row as the target upstream turbine T1, and turbine No. 2 to its east as the associated downstream turbine T2.

[0023] 1. Collect 10-minute average γ_T1, P_T1, V_T1, P_T2, V_T2 data from the SCAT2A system, with a time period from 17:00 on March 1, 2025 to 15:00 on April 1, 2025.

[0024] 2. In data processing, it is assumed that the current yaw angle γ_T1 of T1 is 8°. Substituting γ=8° into the model, the predicted wind speed V of the downstream T2 is calculated. wake_cal =7 m / s, while the measured wind speed V_T2 = 7.45 m / s. Substituting this into the loss function J(γ), and solving it through an optimization algorithm, it was found that J(γ) is minimized when γ_opt = 0.5°, and at this time the predicted wind speed V wake_cal The closest value is 7.45 m / s. Therefore, the calculated yaw error of T1 is Δγ = 0.5° - 8° = -7.5°. This indicates that T1 is veered 7.5 degrees to the left of the current wind direction.

[0025] 3. Send Δγ = -7.5° to the controller of wind turbine T1. Taking a gain coefficient k = 0.5, the new yaw setting is: γs et_new = γ set_old + 0.8 * (-7.5°). This means that the fan T will deflect 6 degrees to the right.

[0026] 4. After correction, the results were verified by observing the data: the power P_T1 of the fan T1 increased by 1.5% due to the correction of the wind direction; at the same time, due to the reduction of its wake deflection, the blocking effect on the downstream fan T2 was weakened, and the inflow velocity V_T2 and power P_T2 of T2 also increased by 0.9% accordingly.

Claims

1. A method for correcting yaw error in wind turbine clusters based on a wake model, characterized in that, Includes the following steps: Step S1: Identify the target correction fan and its associated downstream fans. In the wind farm, select a target upstream wind turbine that needs yaw error correction, denoted as turbine T. Based on the wind farm layout and the current prevailing wind direction, take upstream turbine T as the origin and draw a ray along the prevailing wind direction. In the fan-shaped area downstream of this ray, identify a wind turbine that is significantly affected by the wake of the target upstream wind turbine T, that is, the wind turbine that is closest to the upstream wind turbine in the fan-shaped area downstream of this ray, denoted as downstream wind turbine D. Step S2: Data Acquisition and Preprocessing Simultaneously collect the operating data of the upstream wind turbine T and the downstream wind turbine D within the same time period, including at least: The yaw angle γ_T, power P_T, and wind speed V_T of the upstream wind turbine T; The power P_D and wind speed V_D of the downstream wind turbine D; Step S3: Construct a wake model based on yaw angle A mathematical model is established to describe the relationship between the yaw angle of the upstream wind turbine and the wake velocity distribution at the location of the downstream wind turbine. This model is a modified wake model, and its expression is obtained by simultaneously applying the following four formulas: Where: V wake V_T is the predicted wake velocity at the hub height of downstream wind turbine D; a is the thrust coefficient of upstream wind turbine T; d is the rotor diameter; x, y, z are the position coordinates of downstream wind turbine D relative to upstream wind turbine T; V w The predicted wind speed in the three-dimensional wake does not take into account the effect of yaw; k wake γ is the wake expansion coefficient; f(γ, d, x,y, z, I) is a function of the yaw angle γ, used to describe the wake deflection and velocity deficit caused by yaw. When the yaw angle γ=0, the model degenerates into a standard centrosymmetric wake model. Step S4: Invert the yaw error of the upstream wind turbine based on the measured data of the downstream wind turbine. Forward calculation: The current yaw angle γ_T of the upstream wind turbine T is used as the initial value and input into the wake model established in step S3 to calculate the predicted wind speed V at the hub height of the downstream wind turbine D. wake_cal ; Constructing a loss function: The measured wind speed V_D of the downstream wind turbine D is compared with the model-predicted wind speed V_D. wake_cal For comparison, the loss function J(γ) is constructed using the mean squared error: Optimization solution: Using an optimization algorithm, find the yaw angle γ_opt that minimizes the loss function J(γ). This γ_opt is the theoretically optimal yaw angle that best matches the measured data of the downstream wind turbine. Calculate the yaw error: the yaw error of the upstream wind turbine T That is: Δγ = γ_opt - γ_T; Step S5: Yaw Error Correction and Verification The calculated yaw error Δγ is sent as a correction value to the main controller of the upstream wind turbine T; The controller corrects the yaw angle setpoint of the wind turbine T: Where k is a gain coefficient between 0 and 1, used to prevent overcompensation and ensure system stability; After correction, return to step S2 and continuously collect new operating data to verify the correction effect. If the power of downstream wind turbine D increases and the power of upstream wind turbine T does not decrease significantly, the correction is deemed effective; otherwise, the wake expansion coefficient k is evaluated in conjunction with the real-time turbulence level. wake Adjustments will be made.

2. The method for correcting yaw error of wind turbine clusters based on wake model according to claim 1, characterized in that, The fan-shaped area in step S1 is a fan-shaped area within ±15° with the upstream wind turbine T as the origin and the ray along the prevailing wind direction as the reference.

3. The method for correcting yaw error of wind turbine clusters based on wake model according to claim 1, characterized in that, In the wake model of step S3, the wake expansion coefficient k wake The value is 0.

75.

4. The method for correcting yaw error of wind turbine clusters based on wake model according to claim 1, characterized in that, In the wake model of step S3, parameter c is determined by the actual operating conditions of the wind turbine, and is set to 2.58 based on the characteristics of Gaussian distribution.

5. The method for correcting yaw error of wind turbine clusters based on wake model according to claim 1, characterized in that, In step S5, the formula for the controller to correct the yaw angle setting of the wind turbine T has a gain coefficient k of 0.8.

Images