System and method for calibration and control of wind turbines

Abstract

Systems and methods for performing calibration and calculating consensus inputs in the context of wind farm control (e.g., consensus control of wind turbines) are disclosed herein. A method for calibrating a wind turbine includes receiving internal wind turbine data from a wind turbine, receiving external wind turbine data for the wind turbine from a source external to the wind turbine, comparing the internal wind turbine data to the external wind turbine data, and determining a north offset for the wind turbine based on the comparison of the internal wind turbine data to the external wind turbine data.

Description

Attorney Docket No. P316557.W0.01_520049-21SYSTEM AND METHOD FOR CALIBRATION AND CONTROL OF WIND TURBINESCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U.S. Provisional Patent Application No. 63 / 676,060, filed 26 July 2024, U.S. Provisional Patent Application No. 63 / 676,092, filed 26 July 2024, and U.S. Provisional Patent Application No. 63 / 676,111, filed 26 July 2024, the disclosures of which are incorporated by reference herein in their entirety.CONTRACTUAL ORIGIN

[0002] The United States Government has rights in this invention under Contract No. DE-AC36-08GO28308 between the United States Department of Energy and Alliance for Sustainable Energy, LLC, the Manager and Operator of the National Renewable Energy Laboratory.FIELD

[0003] The described embodiments relate generally to wind farms including a plurality of wind turbines, and more particularly, to systems and methods for calibrating offsets for wind turbines and performing wind farm level control of wind turbines in a wind farm.BACKGROUND

[0004] Wind turbines in a wind farm typically operate individually, controlling their own yaw direction and other operating parameters to maximize their own performance. The wind turbines do not take into account information from nearby turbines or effects of their operation on the nearby turbines. Wind turbine yaw controllers use nacelle-based wind measurements to determine local wind direction or an offset of the nacelle relative to the local wind direction and align the wind turbine based on these determinations. Yaw controllers observe wind direction continuously while in a fixed nacelle position. When a -1- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 persistent or large enough difference between the nacelle position and the wind direction is detected, the yaw controller moves the nacelle to a new location that is maintained until another correction is commanded. Wind turbines can react slowly to changes in wind direction, can be susceptible to errors in the detected wind direction, and can undesirably reduce power generation from nearby turbines due to their wakes.SUMMARY

[0005] One aspect of the present disclosure relates to a method for calibrating a wind turbine, the method including receiving internal wind turbine data from a wind turbine, receiving external wind turbine data for the wind turbine from a source external to the wind turbine, comparing the internal wind turbine data to the external wind turbine data, and determining a north offset for the wind turbine based on the comparison of the internal wind turbine data to the external wind turbine data.

[0006] In some examples, the method can further include updating the internal wind turbine data based on the north offset, generating wind turbine wake data based on the updated internal wind turbine data, and updating the north offset based on the wind turbine wake data. In some examples, the internal wind turbine data can include at least one of turbine yaw angle data or turbine wind direction data.

[0007] In some examples, the method can further include adjusting the north offset for the wind turbine in response to a difference between the internal wind turbine data and the external wind turbine data being greater than a threshold value. In some examples, the threshold value can be determined based on the quality or quantity of the internal wind turbine data and the external wind turbine data.

[0008] In some examples, the method can further include determining a consensus wind estimate for a plurality of wind turbines based on internal wind turbine data from the plurality of wind turbines. The wind turbine can be included in the plurality of wind turbines. The external wind turbine data can include the consensus wind estimate. In some examples, the method can further include excluding the wind turbine from the determination of the consensus wind estimate in response to a difference between the internal wind turbine data and the external wind turbine data being greater than a threshold value.-2- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0009] In some examples, the external wind turbine data can include reanalysis wind direction data. In some examples, the external wind turbine data can include a wind turbine yaw angle determined based on satellite imagery of the wind turbine.

[0010] In some examples, the method can further include determining an aggregate wind direction for a plurality of wind turbines based on internal wind turbine data from the plurality of wind turbines. The wind turbine can be included in the plurality of wind turbines. The external wind turbine data can include the aggregate wind direction.

[0011] Another aspect of the present disclosure relates to a method for managing northing offsets for wind turbines, the method including determining a consensus wind estimate for a plurality of wind turbines, receiving wind direction data from a first wind turbine of the plurality of wind turbines, detecting a bias between the wind direction data and the consensus wind estimate, and adjusting a northing offset for the first wind turbine based on the bias.

[0012] In some examples, the northing offset can be adjusted when the bias is greater than a threshold value. In some examples, the method can further include removing the first wind turbine from the plurality of wind turbines for the determination of the consensus wind estimate in response to the bias being greater than a threshold value.

[0013] In some examples, the method can further include receiving yaw data from the first wind turbine, detecting a bias between the yaw data and the consensus wind estimate, and adjusting the northing offset for the first wind turbine based further on the bias between the yaw data and the consensus wind estimate. In some examples, the determination of the consensus wind estimate can be based at least partly on reanalysis weather data.

[0014] Yet another aspect of the present disclosure relates to a non-transitory computer-readable medium storing code for determining northing offsets for wind turbines, the code including instructions executable by a processor to group a plurality of wind turbines into a plurality of clusters, compare wind turbine data for each of the wind turbines to a consensus wind estimate for the plurality of wind turbines, and apply a northing offset to each of the wind turbines in each of the clusters to minimize a difference between the wind turbine data for each of the wind turbines and the consensus wind estimate.

[0015] In some examples, the plurality of wind turbines can be grouped into the plurality of clusters based on geographic proximity. In some examples, the plurality of wind turbines-3- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 can be grouped into the plurality of clusters based on an average wind direction at each of the respective wind turbines.

[0016] In some examples, the code can further include instructions executable by the processor to re-group the plurality of wind turbines into the plurality of clusters based on a difference between a first wind turbine of the plurality of wind turbines and the consensus wind estimate being greater than a threshold value.

[0017] In some examples, the code can further include instructions executable by the processor to determine a wake profile for each of the wind turbines, and adjust the northing offset for each of the clusters based on the determined wake profiles.

[0018] One aspect of the present disclosure relates to a non-transitory computer-readable medium storing code for determining wind veer for wind turbines, the code including instructions executable by a processor to receive yaw direction data for a plurality of wind turbines for a period of time, determine wind farm aggregate yaw direction data based on the yaw direction data for the plurality of wind turbines, receive wind direction data for a first wind turbine of the plurality of wind turbines for the period of time, and determine a wind direction offset for the first wind turbine based on a difference between the wind farm aggregate yaw direction data and the wind direction data for the first wind turbine.

[0019] In some examples, the code can further include instructions executable by the processor to dynamically adjust a yaw direction for the first wind turbine based on real-time wind farm aggregate yaw direction data offset by the wind direction offset. In some examples, the code can further include instructions executable by the processor to determine consensus wind direction data based on weather data. The wind direction offset can be determined based on the consensus wind direction data.

[0020] In some examples, the code can further include instructions executable by the processor to calculate consensus wind direction data based on wind direction data received from each of the wind turbines of the plurality of wind turbines. A respective wind direction offset can be applied to the wind direction data received from each of the wind turbines of the plurality of wind turbines. In some examples, the wind direction data for the first wind turbine can be received from a wind sensor or a yaw controller of the first wind turbine.

[0021] In some examples, the wind direction offset can be determined after the period of time. An initial wind direction offset of 0 can be used before the period of time has elapsed.-4- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 In some examples, different wind direction offsets can be determined for different consensus wind directions in the consensus wind direction data. In some examples, different wind direction offsets can be determined depending on an estimated atmospheric stability.

[0022] In some examples, the plurality of wind turbines can further include a second wind turbine. The wind direction offset can be determined based further on a predicted wake generated by the second wind turbine.

[0023] Another aspect of the present disclosure relates to a method including defining a cluster including a plurality of wind turbines, detecting a condition in a first wind turbine of the plurality of wind turbines, and in response to detecting the condition, re-defining the cluster.

[0024] In some examples, the cluster is re-defined by removing the first wind turbine from the cluster and adding a second wind turbine to the cluster. In some examples, the detected condition can include at least one of a sensor in the first wind turbine going offline or a volatility of a data stream received from the sensor in the first wind turbine reaching a threshold value. In some examples, each cluster of a plurality of clusters can be re-defined in response to detecting the condition. The plurality of clusters can include the cluster.

[0025] In some examples, the cluster can be defined by grouping wind turbines that are geographically proximal to one another. The cluster can be re-defined by replacing the first wind turbine with a second wind turbine geographically proximal to the first wind turbine.

[0026] In some examples, the cluster can be defined by grouping wind turbines that have wind direction offsets closest to one another. The cluster can be re-defined by replacing the first wind turbine with a second wind turbine having a wind direction offset closest to the first wind turbine.

[0027] Yet another aspect of the present disclosure relates to a system for providing yaw control for wind turbines, the system including a wind farm including a plurality of wind turbines, a first wind turbine of the plurality of wind turbines including a wind direction sensor and a yaw controller, and a consensus controller coupled to the wind turbines of the plurality of wind turbines, the consensus controller configured to calculate a consensus wind direction for the wind farm. The yaw controller can be configured to determine a wind direction offset for the first wind turbine based on a first condition detected by comparing data from the wind direction sensor to the consensus wind direction. The consensus controller -5- 4935-3462-485511Attorney Docket No. P316557.W0.01_520049-21 can be configured to define one or more groupings of the wind turbines of the plurality of wind turbines based on a second condition detected by comparing data from the wind direction sensor to the consensus wind direction.

[0028] In some examples, the first condition can include a persistent difference between the data from the wind direction sensor and the consensus wind direction. In some examples, the second condition can include an offline state in the wind direction sensor or a varying difference between the data from the wind direction sensor and the consensus wind direction.

[0029] In some examples, the wind direction offset can be determined and the groupings of the wind turbines can be defined based on geographic positions of the wind turbines in the wind farm. In some examples, the wind direction offset can be determined and the groupings of the wind turbines can be defined based on the consensus wind direction or an estimated atmospheric stability.

[0030] One aspect of the present disclosure relates to a wind turbine including a wind sensor configured to output wind data, a yaw controller, and a non-transitory computer-readable medium storing code for correcting the wind data. The wind data can include at least one of wind direction data or wind speed data. The code can include instructions executable by a processor to command the yaw controller to misalign a rotor of the wind turbine relative to the wind data, receive the wind data, apply a correction factor to the wind data to account for the effect of the misalignment of the rotor, and output corrected wind data.

[0031] In some examples, the correction factor can be in a range of 0.7 to 1. In some examples, the correction factor can vary based on at least one of the wind direction data or the wind speed data. In some examples, the correction factor can vary based on a magnitude of the misalignment of the rotor relative to the wind data. In some examples, the correction factor can vary based on an estimated atmospheric stability.

[0032] In some examples, the wind turbine can be a first wind turbine and the code can further include instructions executable by the processor for determining the correction factor based on a difference between the wind data and wind data received from a second wind turbine. In some examples, the code can further include instructions executable by the processor for determining the correction factor through machine learning based on differences between historical yaw misalignment commands and historical wind data from the wind sensor.-6- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0033] Another aspect of the present disclosure relates to a system for providing yaw control for a wind turbine, the system including a first wind turbine including a wind direction sensor configured to output a wind direction signal and a yaw controller, and a consensus controller coupled to a plurality of wind turbines. The plurality of wind turbines can include the first wind turbine. The consensus controller can be configured to determine a consensus wind direction for the plurality of wind turbines based at least in part on the wind direction signal and output a consensus wind direction signal. The yaw controller can be configured to yaw the first wind turbine based on the consensus wind direction signal.

[0034] In some examples, the system can further include a filter configured to filter the consensus wind direction signal and output a filtered consensus wind direction signal to the yaw controller. In some examples, the yaw controller can be configured to yaw the first wind turbine based on the consensus wind direction signal without filtering the consensus wind direction signal. In some examples, the consensus controller can be configured to output the consensus wind direction signal with a lower variability than the wind direction sensor. In some examples, a gain of the yaw controller can be selected based on energy production by the first wind turbine and yaw activity of the first wind turbine.

[0035] Yet another aspect of the present disclosure relates to a method including receiving wind direction data from a wind direction sensor on a wind turbine, adjusting the wind direction data to produce corrected wind direction data, receiving the corrected wind direction data in a yaw controller of the wind turbine, and adjusting a yaw of the wind turbine through the yaw controller based on the corrected wind direction data.

[0036] In some examples, adjusting the wind direction data can include applying a correction factor to the wind direction data to produce the corrected wind direction data. In some examples, the wind turbine is a first wind turbine. The method can further include yawing the first wind turbine to misalign a rotor of the first wind turbine with a current wind direction and determining the correction factor based on a difference between the wind direction data and wind direction data received from a second wind turbine proximal the first wind turbine. In some examples, the method can further include determining the correction factor based on current atmospheric conditions at the wind turbine.

[0037] In some examples, adjusting the wind direction data can include determining a consensus wind direction based on the wind direction data and wind direction data received -7- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 from wind direction sensors from one or more additional wind turbines. In some examples, producing the corrected wind direction data can include applying a wind direction offset specific to the wind turbine to the consensus wind direction.

[0038] In some examples, adjusting the wind direction data can include comparing the wind direction data to weather data and producing the corrected wind direction data based on the comparison. In some examples, the corrected wind direction data can be filtered before being received by the yaw controller.BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The disclosure will be readily understood by the following detailed description in conjunction with the accompanying drawings, wherein like reference numerals designate like structural elements, and in which:

[0040] FIG. 1 A and IB show perspective views of a wind turbine.

[0041] FIG. 2 shows a schematic view of collective control for a plurality of wind turbines.

[0042] FIG. 3 shows a schematic view of a wind turbine north offset.

[0043] FIG. 4 shows a schematic view of a wind turbine north offset calibration.

[0044] FIG. 5 shows a schematic view of a wind farm with a wind wake.

[0045] FIG. 6 shows a satellite image of a wind turbine and illustrates a method for determining north offsets.

[0046] FIG. 7 shows a flow chart of a method for determining north offsets.

[0047] FIG. 8 shows a schematic view of a wind turbine with a detected wind direction offset from a consensus wind direction.

[0048] FIGS. 9 A and 9B show schematic views of wind farms with clusters of wind turbines.

[0049] FIG. 10 shows a flow chart of a method for determining a misalignment correction factor.

[0050] FIG. 11 shows a flow chart of a method for yawing a wind turbine based on consensus control.-8- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0051] FIG. 12 shows a block diagram of a controller that supports techniques for performing calibration and consensus control for wind turbines.DETAILED DESCRIPTION

[0052] Reference will now be made in detail to representative embodiments illustrated in the accompanying drawings. It should be understood that the following descriptions are not intended to limit the embodiments to one preferred embodiment. To the contrary, it is intended to cover alternatives, modifications, and equivalents as can be included within the spirit and scope of the described embodiments as defined by the appended claims.

[0053] The following disclosure relates to systems and methods for calibrating north offsets of wind turbines and performing wind farm level control of wind turbines in a wind farm. In a typical wind farm, each wind turbine operates individually and a yaw controller for each wind turbine sets the yaw for that wind turbine to match a wind direction detected at that wind turbine. This maximizes energy production for each of the respective wind turbines. However, errors in wind direction sensors of the wind turbines can reduce energy production from wind turbines with faulty sensors. Moreover, wind turbines that operate individually can have detrimental effects on neighboring wind turbines, which can reduce overall energy production from a wind farm. Some approaches to yaw control of wind turbines can be subject to limitations of measuring the wind direction with sensors that are typically located toward a downwind end of a nacelle of each respective wind turbine, where the sensors are subject to wake effects from a rotor of the wind turbine. These unwanted effects are measured by the direction sensors and require signal processing to remove them, which may not be completely effective and can lead to suboptimal yaw control.

[0054] Two approaches can be used to increase energy production from a wind farm. First, wind turbines can share information in order to make better decisions. This can provide the wind turbines with more accurate information regarding the correct wind direction and can reduce the impact of short term, turbulent wind direction readings from sensors of the wind turbines, faulty sensors on the wind turbines, and the like. Second, wind turbines can operate in a coordinated manner. Operating wind turbines in a coordinated manner can be referred to as consensus control. For example, some wind turbines can operate in a way that reduces energy production at those turbines, while increasing overall energy production from the -9- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 wind farm. For example, a first wind turbine can be misaligned with the current wind direction in order to increase energy production from a neighboring second wind turbine. Energy production at the second wind turbine can increase by a greater amount than the energy production decrease at the first wind turbine, thereby increasing energy production from a wind farm including both with turbines. Sacrificing energy production at certain wind turbines to increase energy production from the wind farm can be referred to as wake steering. These and other approaches can be used to increase energy production from a wind farm.

[0055] In order for wind turbines to accurately share information and operate in a coordinated manner, the wind turbines yaw directions should be calibrated to the same reference system. As an example, the wind turbines of a wind farm can each be calibrated to true north. A yaw controller of a wind turbine can track the wind turbine’s yaw with respect to a reference point, such as true north. When a wind turbine is installed, the yaw controller can read an arbitrary direction as north, and a north offset can be applied in order to ensure that the yaw controller is correctly calibrated to true north. A typical process for calibrating the north offset in a wind turbine includes determining the north offset based on a visual alignment of the wind turbine (e.g., a visual alignment of a nacelle of the wind turbine) with a reference point, such as a mountain, another wind turbine, or the like. However, this visual method can have low accuracy. Moreover, the north offset can be reset during service or maintenance events, during other operational events (e.g., a loss of power), or the like.

[0056] The present disclosure includes improved methods and systems for calibrating north offsets in wind turbines. The north offset for a wind turbine can be calibrated by tracking internal sensor data from the wind turbine, comparing this internal sensor data to data from an external source, and calibrating the north offset for the wind turbine such that the internal sensor data is aligned to the external source data. The internal sensor data can be obtained from a yaw controller of the wind turbine and various wind direction sensors, such as wind vanes or anemometers (e.g., mechanical, ultrasonic, sonic, LIDAR, and other anemometers), of the wind turbine. The external source data can be obtained from weather services (e.g., reanalysis weather data sources), satellite imagery, a consensus controller of a wind farm including the wind turbine, and the like. The north offset can then be determined by fitting the internal sensor data to the external source data. This calibration can be refined by determining wind wakes generated by the wind turbines of the wind farm and improving -10- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 the fit of the internal sensor data to the external source data based on these wind wakes. By calibrating north offsets for wind turbines based on internal and external data, the north offsets can be calibrated with improved accuracy. The north offsets can be calibrated in realtime and on-line so that any errors in the north offset, such as accidental resets or the like, are automatically detected and corrected.

[0057] The present disclosure further includes improved methods and systems for sharing data between wind turbines. Wind turbines in a wind farm can each detect different wind directions, and each of these wind turbines can be correct. Wind farms can be located in complex terrain that causes wind veer (e.g., variations in local wind directions) throughout the wind farm. In order to correct for wind veer, a wind direction offset can be determined for each wind turbine in the wind farm. The wind direction offset can be determined by comparing internal wind turbine sensor data from a long period of time to a consensus wind direction (e.g. obtained from an external source) over the same period and driving the difference between the two to 0 using a wind direction offset. The wind direction offset can be applied to wind direction data from each wind turbine by a wind farm-level yaw controller, which can be used to generate a consensus wind direction signal for the wind farm with improved accuracy. This wind farm-level wind direction signal can then be sent to yaw controllers of each wind turbine. The yaw controller of each wind turbine can then use the wind direction offset and the consensus wind direction as an input to determine which direction to face the wind turbine, improving energy production by the wind turbine. The wind veer across the wind farm can vary dynamically based on a number of factors. For example, the wind veer across the wind farm can change based on a predominant wind direction (e.g., with winds originating from the north or the south), based on atmospheric conditions (e.g., caused by changing seasons, weather patterns, or the like), based on wind turbine operation parameters (e.g., rotor RPM and the like), and the like. As such, dynamic values of the wind direction offset can be determined based on the current conditions of the wind farm.

[0058] Wind turbines in a wind farm that are geographically close to each other can be grouped into clusters, which can be used to provide control to groups of wind turbines that are smaller in number than the wind farm. Faulty sensors in wind turbines of a cluster can lead to poor data for the cluster and can negatively impact energy production by the cluster. Thus, when sensors from a wind turbine are determined to be faulty or inoperable (e.g., by -11- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 comparing signals from the sensors to signals from sensors of other wind turbines in the cluster), the wind turbine can be removed from the cluster. Depending on the size of the cluster, the wind turbine can be replaced by another wind turbine that is geographically proximal to the cluster or that has a wind direction offset close to wind direction offsets of the wind turbines in the cluster. In some examples, the clusters for the wind farm can be redefined when wind turbines are removed from clusters. This can ensure that wind turbines with faulty or inoperable sensors are not included in data used to control other wind turbines and increase energy production from the wind farm.

[0059] Wind turbines in a wind farm can be misaligned with a wind direction in order to increase energy production by neighboring wind turbines. When a wind turbine is misaligned with the wind, the sensors on the wind turbine can read the wind direction incorrectly, as the wind turbines and included sensors are designed to operate while the wind turbine is aligned with the wind. As such, a correction factor can be applied to data generated by the sensors to correct the data produced by the sensors. The correction factor can be determined by comparing data from the sensors to data from sensors of neighboring wind turbines, both when the turbines are aligned with the wind and misaligned with the wind. The correction factor can also be determined by comparing a misalignment commanded by a yaw controller of a misaligned wind turbine with the actual misalignment resulting from the command of the misaligned wind turbine. The data produced by the sensors of the misaligned wind turbine can be adjusted by the correction factor such that accurate data is provided by the misaligned wind turbine. This can improve energy production by a wind farm that uses data from the misaligned wind turbine.

[0060] By operating wind turbines according to the above methods and systems, the yaw controllers of the wind turbines can receive improved data that is filtered before the data is received by the yaw controller. This allows for the yaw controller to act on the data without filtering the data. Moreover, the yaw controller can have an increased gain, and can act on incoming signals more quickly, which results in the wind turbines reacting more quickly to changes in wind direction. This improves energy production from wind farms operating according to the systems and methods of the present disclosure. Increasing the gain of the yaw controller can result in greater yaw activity, which can be measured as yaw movements per hour, and which can increase wear and tear on wind turbines. As such, the gain can be-12- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 limited by a maximum yaw activity, or yaw activity can be balanced with energy production in order to balance energy production with wind turbine longevity.

[0061] The systems and methods of the present disclosure can be used in combination with any consensus control algorithm or other wind farm or energy generation control scheme. For example, the systems and methods of the present disclosure can be used in combination with the consensus control algorithm described in U.S. Patent No. 11,725,625, the disclosure of which is incorporated herein by reference in its entirety.

[0062] These and other embodiments are discussed below with reference to FIGS. 1 through 12. However, those skilled in the art will readily appreciate that the detailed description given herein with respect to these Figures is for explanatory purposes only and should not be construed as limiting. Furthermore, as used herein, a system, a method, an article, a component, a feature, or a sub-feature including at least one of a first option, a second option, or a third option should be understood as referring to a system, a method, an article, a component, a feature, or a sub-feature that can include one of each listed option (e.g., only one of the first option, only one of the second option, or only one of the third option), multiple of a single listed option (e.g., two or more of the first option), two options simultaneously (e.g., one of the first option and one of the second option), or combination thereof (e.g., two of the first option and one of the second option).

[0063] FIGS. 1 A and IB illustrate perspective views of a wind turbine 100. The wind turbine 100 includes a tower 102. A nacelle 104 is mounted on the tower 102. Rotor blades 106 are mounted to a rotor hub 108, which is connected to a main flange that turns a main rotor shaft. The main rotor shaft is connected to wind turbine power generation and control components housed within the nacelle 104. Power is generated through the main rotor shaft driving an electrical generator. Although three rotor blades 106 are illustrated in FIGS. 1 A and IB, the wind turbine 100 can include any number of rotor blades 106.

[0064] FIG. IB illustrates control and sensing components housed within or on the nacelle 104. A yaw system 110 can be included between the tower 102 and the nacelle 104. The yaw system 110 can be used to rotate the nacelle 104 around a longitudinal axis of the tower 102. The yaw system 110 can be used to align (or misalign) the main rotor shaft with the wind. The yaw system 110 can include a yaw drive, a yaw brake, and the like, which can be controlled by a yaw controller.-13- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0065] Various sensors can be mounted on the nacelle 104 in order to detect the wind direction at the wind turbine 100. In the example of FIG. IB, the sensors include an anemometer 112 and a wind vane 114. The anemometer 112 can be a mechanical anemometer, an ultrasonic anemometer (e.g., a spinner ultrasonic anemometer), a sonic anemometer, an anemometer mounted on the spinner, a nacelle-mounted LIDAR, or the like. Any number of anemometers 112, wind vanes 114, and additional sensors (e.g., sonic sensors or the like) that detect wind direction and / or speed can be included on the wind turbine 100. The yaw controller, the anemometer 112, and the wind vane 114 can generate wind direction data that reflects the local wind direction at the wind turbine 100. The wind turbine 100 of FIGS. 1 A and IB is provided for illustrative purposes only, and the present disclosure is not limited to any particular type of wind turbine configuration.

[0066] FIG. 2 illustrates a schematic view of a wind turbine control system 200 for a wind farm. The wind turbine control system 200 includes a control unit 202 (also referred to as a consensus controller) that is used to control wind turbines 204, 206, 208, 210, and 212. The wind turbines 204-212 can be the same as or similar to the wind turbine 100, discussed above with respect to FIGS. 1 A and IB. Each of the wind turbines 204-212 can include sensors that can be used to detect a local wind direction at each respective wind turbine 204-212 and a yaw direction of each respective wind turbine 204-212. The sensors can include yaw controllers, wind vanes, anemometers, and the like. Data produced by the sensors of each respective wind turbine 204-212 and used by the respective wind turbine 204-212 can be referred to as internal turbine data, internal data, internal wind data, internal direction data, or the like. Data from the sensors can be provided from the wind turbines 204-212 to the control unit 202.

[0067] In order for the wind turbines 204-212 to accurately share information with the control unit 202, each of the wind turbines 204-212 can be calibrated based on the same reference point. This ensures that the data provided to the control unit 202 from the sensors of each of the wind turbines 204-212 is consistent. As will be discussed in greater detail below, the yaw controllers for each of the wind turbines 204-212 can be calibrated based on true north as the reference point. Accurately calibrating each of the wind turbines 204-212 in a wind farm ensures that data provided by the sensors of the wind turbines 204-212 is consistent and allows for the control unit 202 to perform accurate wind farm level yaw control.-14- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0068] The control unit 202 can determine a consensus wind direction based on the data provided by the sensors of the wind turbines 204-212. The control unit 202 can use a consensus algorithm in order to determine a consensus wind direction based on data provided from the wind turbines 204-212. In some examples, the consensus wind direction can be determined based on an aggregate wind direction, such as a median, mode, or mean wind direction of the wind turbines 204-212. In order to account for the fact that 0° and 360° are identical angles (e.g., to deal with 360° to 0° wrapping), the aggregate wind direction can be calculated using directional or circular statistics. For example, the median can be calculated as a circular median, the mode can be calculated as a circular mode, and the mean can be calculated as a circular mean. By aggregating individual wind turbine measurements of the wind direction from each of the wind turbines 204-212, the consensus algorithm can produce a more reliable and predictive estimate of the wind direction. This can be used to improve alignment of the wind turbines 204-212 to incoming wind, provide higher quality information to the yaw controllers of the wind turbines 204-212, and be used to identify faulty sensors of the wind turbines 204-212. As will be discussed below, the consensus wind direction can be used to accurately calibrate the wind turbines 204-212 to true north. Further, the control unit 202 can detect and filter out short-lived changes in the wind direction detected by individual wind turbines 204-212, which reduces yaw activity from chasing short-lived changes in wind direction.

[0069] The control unit 202 can send yaw commands to the wind turbines 204-212 based on the consensus wind direction. The yaw controllers of the wind turbines 204-212 can yaw the wind turbines 204-212 based on the yaw commands from the control unit 202. By accurately calibrating each of the wind turbines 204-212 to true north or another reference point, the yaw controllers of the wind turbines 204-212 accurately yaw the wind turbines 204-212 in response to the yaw commands. Data produced by the control unit 202 received by the wind turbines 204-212 can be referred to as external data, external wind data, external direction data, or the like.

[0070] The control unit 202 can determine the consensus wind direction based on data from external sources in addition to or in place of data from the wind turbines 204-212. Data from external sources can include reanalysis weather data, data from weather stations, data from weather or wind sensors, and any other relevant data obtained from sources external to the wind turbines 204-212. Determining the consensus wind direction based on external -15- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 sources can further improve the accuracy of the consensus wind direction and yaw controls sent from the control unit 202 to the wind turbines 204-212. A single control unit 202 can be provided for a wind farm, or control units 202 can be provided for subsets of the wind turbines in the wind farm (e.g., for clusters of wind turbines).

[0071] FIG. 3 illustrates a wind turbine 300 having a perceived north 302 that is offset from true north 304. The perceived north 302 can be offset from true north by a north offset 306. In order to align the perceived north 302 of the wind turbine 300 to true north 304, the north offset 306 can be determined and applied to the perceived north 302. More specifically, data produced by (e.g., based on the perceived north 302) and received by (e.g., commands from a consensus controller) the wind turbine 300 can be offset by the north offset 306 in order to align the wind turbine 300 to true north 304. This ensures that data produced by the wind turbine 300 is accurate, and the wind turbine 300 is accurately able to yaw the wind turbine 300 based on consensus control signals provided to the wind turbine 300. FIGS. 4 through 7 describe methods for determining the north offset 306, which can be used to offset data produced by and received by the wind turbine 300 in order to align the wind turbine 300 with true north 304.

[0072] FIG. 4 illustrates a method of calibrating a north offset 402 (also referred to as a northing offset) for a wind turbine 400. The method illustrated in FIG. 4 is based on a comparison of a perceived wind direction 404 detected by the wind turbine 400 to a consensus wind direction 406 provided by a consensus controller (e.g., a consensus controller of a wind farm, a cluster of wind turbines, or another subset of wind turbines in which the wind turbine 400 is included). The wind turbine 400 can be the same as or similar to the wind turbines 100, 204-212, 300, discussed above with respect to FIGS. 1 A through 3, and the consensus controller can be the same as or similar to the control unit 202, discussed above with respect to FIG. 2. The perceived wind direction 404 can be detected by sensors of the wind turbine 400, such as a yaw controller and wind direction sensors (e.g., wind vanes, anemometers, and the like). The wind direction sensors can also detect wind direction and can be broadly referred to as wind sensors. The consensus wind direction 406 can be determined by the consensus controller based on data provided from wind turbines and / or external weather data (e.g., data from reanalysis weather services, weather stations, weather sensors, or the like).-16- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0073] The north offset 402 can be determined by comparing the perceived wind direction 404 detected by the wind turbine 400 to the consensus wind direction 406. The north offset 402 can be determined based on the assumption that the sensors of the wind turbine 400 and the sensors used to determine the consensus wind direction 406 are generally accurate over a long period of time. The perceived wind direction 404 detected by the wind turbine 400 can be offset by the north offset 402 in order to produce an offset wind direction 408, which can be aligned with the consensus wind direction 406. The north offset 402 can be adjusted in order to align the offset wind direction 408 with the consensus wind direction 406. In some examples, the north offset 402 is adjusted when a difference (also referred to as a bias) between the offset wind direction 408 and the consensus wind direction 406 is greater than a threshold value. For example, the north offset 402 can be adjusted or updated when a difference between the offset wind direction 408 and the consensus wind direction 406 is greater than about 5 degrees, greater than about 2 degrees, greater than about 1.5 degrees, greater than about 1 degree, or the like. The threshold value that is used to determine whether to adjust the north offset 402 can be determined based on a quantity or quality of internal and external wind turbine data that is used to determine the north offset 402. For example, the threshold value may be lower when the north offset 402 is determined based on a larger quantity of data, such as data obtained from a greater number of sensors, a greater variety of sensors, over a greater time period, or the like. Similarly, the threshold value may be lower when the north offset 402 is determined based on higher quality of data, such as data that is consistent (e.g., internally, with respect to external sources, or the like). This ensures that the north offset 402 is not erroneously altered based on noise from sensors, local wind fluctuations, or the like, and improves the accuracy of the north offset 402.

[0074] In some examples, the north offset 402 can be adjusted or updated periodically, such as on a seasonal, monthly, weekly, daily, or other basis. In some examples, the north offset 402 can be adjusted or updated following a maintenance event, an operational event, or another event that is likely to reset or otherwise alter the north offset 402. For example, the north offset 402 can be adjusted or updated following a loss of power to the wind turbine 400, following a maintenance event performed on the wind turbine 400, during or after the installation of the wind turbine 400, at a predetermined time after the installation of the wind turbine 400, or the like.-17- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0075] The north offset 402 can be determined based on data that is obtained over a long period of time and is highly accurate. For example, the north offset 402 can be determined based on data received over the course of at least 1 month, at least 3 months, at least 6 months, at least a year, or the like. An initial north offset 402 can be determined when the wind turbine 400 is installed, and the north offset 402 can be updated and improved over time as more data becomes available. As a result, the north offset 402 can have a high level of accuracy, such as within about 2 degrees, within about 1 degree, or the like. Further, by calculating the north offset 402 based on the perceived wind direction 404 at the wind turbine 400, any misalignments within in the wind turbine 400 can be accounted for. By accurately determining the north offset 402, the wind turbine 400 can be accurately aligned with true north. The wind turbine 400 can produce accurate offset wind direction 408 data and can accurately respond to consensus commands provided by a consensus controller. When each of the wind turbines that provide data to the consensus controller are accurately northed, the accuracy of the consensus wind direction determined by the consensus controller is improved. As such, the wind turbine 400 can be accurately aligned to the current wind direction, and energy production by the wind turbine 400 can be increased.

[0076] In some cases, the wind turbine 400 can have a static yaw misalignment. This can be exemplified as an offset of a wind direction sensor of the wind turbine 400 relative to a nacelle of the wind turbine 400 changing over time. As an example, wind veer over the nacelle can change over time due to blades of the wind turbine 400 aging, which can result in the wind direction sensor detecting a different wind direction relative to the nacelle. This static yaw misalignment can be constantly checked during operation of the wind turbine 400 (e.g., by comparing local wind direction at the wind turbine 400 with the wind direction detected by the wind direction sensor of the wind turbine 400) and can be corrected prior to performing consensus control. The static yaw misalignment can be corrected before wind direction data is sent from the wind turbine 400 and before the wind turbine 400 is yawed in response to a consensus control signal.

[0077] FIG. 5 illustrates a wind farm 500 including a plurality of wind turbines 502. FIG. 5 illustrates additional details for determining the north offset 402 for the wind turbine 400. More specifically, FIG. 5 illustrates a wind wake 506 produced by a wind turbine 504 neighboring the wind turbine 400. As illustrated in FIG. 5, the wind wake 506 can extend from the wind turbine 504 to the wind turbine 400 and the wind wake 506 can influence the -18- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 local wind direction (e.g. reflected in the perceived wind direction 404) at the wind turbine 400 as well as the wind speed at the wind turbine 400. The north offset 402 can be adjusted based on predicted wind wakes produced by neighboring wind turbines 502 in order to account for the influence of the wind wakes. The predicted wind wakes can be stored as wake profiles or wind wake profiles in a controller that determines the predicted wind wakes. The wind turbines 502, 504, 400 can be the same as or similar to the wind turbines 100, 204-212, 300, 400, discussed above with respect to FIGS. 1 A through 4.

[0078] The wind wakes that will be produced by each of the wind turbines 502 can be predicted based on a consensus wind direction 508, a consensus wind speed, and geometry. The wind wakes produced by the wind turbines 502 can be generally Gaussian, can be aligned with the wind direction, and an angle of the vertex of each of the wind wakes can be dependent on the wind speed. The predicted wind wakes can be used to improve the estimation of the north offset 402 by accounting for any differences in the perceived wind direction 404 caused by wind wakes from neighboring wind turbines 502 (e.g, the wind wake 506 produced by the wind turbine 504). During analysis of wind wakes, it can be assumed that a wake nadir or wake center occurs at a downwind wind turbine 400 when wind direction is the same as a bearing between the upwind turbine 504 and the downwind turbine 400. Data can be analyzed to determine the perceived wind direction where the wake center occurs and compare this to the bearing between the upwind turbine 504 and the downwind turbine 400. If the perceived wind direction where the wake center occurs is different from the bearing between the upwind turbine 504 and the downwind turbine 400, this implies that a correction is needed. This analysis can be run for every combination of the wind turbines 502 in the wind farm 500. Thus, the offset wind direction 408 produced by the wind turbine 400 based on the perceived wind direction 404 and the north offset 402 can be adjusted based on the predicted wind wakes (e.g, the wind wake 506 for the wind turbine 400). This further improves the accuracy of the north offset 402 that is determined for each of the wind turbines 502 and the accuracy of the offset wind direction 408 data produced by each of the wind turbines 502. The consensus wind direction 508 and the consensus wind speed can be determined by a consensus controller, which can be the same as or similar to the control unit 202 and the consensus controller discussed above with respect to FIGS. 2 through 4. The consensus wind direction 508 and the consensus wind speed can be determined by the consensus controller based on data provided from each of the wind turbines 502 and / or -19- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 external weather data (e.g., data from reanalysis weather services, weather stations, weather sensors, or the like). Thus, the accuracy of the consensus wind direction 508 can also be improved by predicting and accounting for wind wakes in the determination of north offsets for each of the wind turbines 502.

[0079] FIG. 6 illustrates a satellite image 600 of a wind turbine 602. The wind turbine 602 can be the same as or similar to the wind turbines 100, 204-212, 300, 400, 502, 504, discussed above with respect to FIGS. 1 A through 5. In FIG. 6, a north offset is determined for the wind turbine 602 based on the satellite image 600. The satellite image 600 can include a time stamp for when the satellite image 600 was taken, can be georeferenced such that the satellite image 600 is aligned with true north 604, and can include metadata that indicates an orientation of the satellite relative to earth when the satellite image 600 was taken.

[0080] Based on the satellite image 600, an orientation of a nacelle 606 of the wind turbine 602 relative to true north 604 can be determined. The orientation of a nacelle 606 relative to true north 604 can be determined based on reference features on the wind turbine 602 and geometry. The reference features can include a tower 608 of the wind turbine 602, a reference feature 610 included on a top surface of the nacelle 606, edges 612 of the nacelle 606, rotor blades 614 of the wind turbine 602, a shadow 616 of the wind turbine 602, aspects thereof, and the like. Aerial or satellite images can also be used to determine the exact latitude and longitude for the wind turbine 602, which can be used to determine bearing between wind turbines for wake center analysis and can be used to improve other data used in consensus yaw control.

[0081] The orientation of the nacelle 606 relative to true north 604 can be determined based on any of the reference features, based on interrelationships between the reference features, or the like. An orientation of the nacelle 606 or other features of the wind turbine 602 relative to the tower 608 can be examined. The reference feature 610 can be an arrow, a line, or another feature painted or otherwise affixed on the top surface of the nacelle 606. The reference feature 610 can be used to illustrate the orientation of the nacelle 606 or a main rotor of the wind turbine 602. Lines can be fit to the edges 612 in order to determine the orientation of the nacelle 606. Features of the rotor blades 614, such as orientations, relative lengths, and the like can be examined. In some examples, the rotor blades 614 can be examined in particular orientations, such as when one of the rotor blades 614 is in a vertical-20- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 or horizontal orientation. The rotor blades 614 can be examined to determine the orientation of the nacelle 606. A reference circle 618 can be drawn around tips of the rotor blades 614, and the orientation of the nacelle 606 can be determined based on a shape or orientation of the reference circle 618. Positions of features in the shadow 616 can be compared to positions of features in the wind turbine 602. In some examples, the reference features of the wind turbine 602 can be examined by machine learning or artificial intelligence, and the orientation of the nacelle 606 relative to true north 604 can be detected automatically.

[0082] Once the orientation of the nacelle 606 relative to true north 604 is determined, the orientation of the nacelle 606 can be compared to an offset yaw direction of the wind turbine 602 at the time the satellite image 600 was taken. The offset yaw direction of the wind turbine 602 can be determined by the wind turbine 602 by offsetting a perceived direction detected by sensors of the wind turbine 602 by a north offset. The north offset can be adjusted based on the orientation of the nacelle 606 determined from the satellite image 600 such that the offset yaw direction is aligned with the orientation of the nacelle 606. Thus, the north offset can be adjusted in order to accurately calibrate the wind turbine 602 and align the wind turbine 602 with true north. As discussed above in other examples, the north offset can be adjusted or updated in response to a difference between the offset yaw direction and the orientation of the nacelle 606 being greater than a threshold value, in response to an event, periodically, or the like.

[0083] FIG. 7 illustrates a method 700 for calibrating a north offset in a wind turbine. The method 700 can be the same as or similar to the methods discussed above with respect to FIGS. 4 through 6. In block 702, internal turbine data is received from a wind turbine. The internal turbine data can include yaw position data, wind direction data, or the like for the wind turbine. The yaw position data can be provided by a yaw controller of the wind turbine. The wind direction data can be provided by wind sensors of the wind turbine, such as a wind vane, an anemometer (e.g., a mechanical, ultrasonic, sonic, LIDAR, or other anemometer), or the like. The internal turbine data can include offset yaw direction data and / or offset wind direction data. The offset yaw direction data can be determined by offsetting perceived yaw direction data from the yaw controller of the wind turbine by the north offset. The offset wind direction data can be determined by offsetting perceived wind direction data from the wind sensors of the wind turbine by the north offset.-21- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0084] In block 704, external turbine data is received. The external turbine data can be received from a consensus controller or the like. The external turbine data can include a consensus wind direction, an orientation of the wind turbine relative to north, or the like. The consensus wind direction can be determined by the consensus controller based on internal turbine data received from a plurality of wind turbines. The consensus wind direction can be determined by the consensus controller based on external weather data, such as external wind direction data received from reanalysis weather services or databases, weather stations, weather sensors, or the like. The orientation of the wind turbine relative to north can be determined from satellite imagery or the like, as described above with respect to FIG. 6.

[0085] In block 706, the north offset is calibrated for the wind turbine. The north offset can be calibrated by aligning the internal turbine data with the external turbine data. Specifically, the north offset that is used to offset the perceived yaw direction data and / or the perceived wind direction data can be adjusted such that the offset yaw direction data and / or the offset wind direction data included in the internal turbine data are aligned with the external turbine data.

[0086] The internal turbine data and the external turbine data can be received and collected over long periods of time, which results in the internal turbine data and the external turbine data having high levels of accuracy. By calibrating the north offset based on the internal turbine data and the external turbine data, the north offset can be calibrated with a high level of accuracy. Improving the accuracy of the north offset for each of the wind turbines used to determine the consensus wind direction improves the accuracy of the consensus wind direction. Based on these improved accuracies, the wind turbines are able to more accurately align with the wind direction, and energy production by the wind turbines is increased.

[0087] The north offset can be calibrated according to any appropriate schedule. For example, the north offset can be calibrated in response to the internal turbine data being misaligned with the external turbine data by more than a threshold amount, in response to events, according to a periodic schedule, or the like. The north offset can be updated or adjusted when the difference between the internal turbine data and the external turbine data is greater than about 5 degrees, greater than about 5 degrees, greater than about 5 degrees, greater than about 5 degrees, or the like. The north offset can be updated or adjusted when a wind turbine is installed, after a prescribed period following installation of the wind turbine,-22- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 in response to maintenance being performed on the wind turbine, in response to a power outage at the wind turbine, or the like. The north offset can be updated or adjusted seasonally, monthly, weekly, daily, or according to any other schedule.

[0088] In block 708, the north offset can be adjusted. As described in reference to FIG. 5, wind wakes produced by wind turbines neighboring a wind turbine can impact the wind direction and wind speed at the wind turbine. As such, wind wakes for neighboring wind turbines can be predicted based on the external turbine data, and these predicted wind wakes can be used to adjust the north offset of the wind turbine. More specifically, turbine data can be analyzed in order to determine when a wake center occurs at a downwind turbine. This can be compared to a bearing between an upwind turbine and the downwind turbine. If the wake center occurs with a perceived wind direction different from the bearing between the upwind turbine and the downwind turbine, a correction can be applied to the north offset. This improves the accuracy and precision of the north offset.

[0089] Predicted wind wakes for the wind turbines of a wind farm can be predicted based on the external turbine data, relative positions of the wind turbines, and geometry. The predicted wind wakes for wind turbines neighboring the wind turbine for which the north calibration is performed can then be used in the calibration of the north offset. This accounts for changes to the wind speed and wind direction the wind turbine experiences due to the wind wakes produced by the neighboring wind turbines. Block 708 can be optional and can be omitted in some examples.

[0090] FIG. 8 illustrates a method of calibrating a wind direction correction factor 802 for a wind turbine 800. The method illustrated in FIG. 8 is based on a comparison of an offset wind direction 804 detected by the wind turbine 800 to a consensus wind direction 806 provided by a consensus controller (e.g., a consensus controller of a wind farm, a cluster of wind turbines, or another subset of wind turbines in which the wind turbine 800 is included). The wind turbine 800 can be the same as or similar to the wind turbines, discussed above with respect to FIGS. 1 A through 6, and the consensus controller can be the same as or similar to the control units or consensus controllers, discussed above with respect to FIGS. 2 through 7. The offset wind direction 804 can be determined by offsetting a perceived wind direction detected by sensors of the wind turbine 800 by a north offset, as discussed above. The consensus wind direction 806 can be determined by the consensus controller based on-23- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 data provided from wind turbines and / or external weather data (e.g., data from reanalysis weather services, weather stations, weather sensors, or the like).

[0091] As illustrated in FIG. 8, the offset wind direction 804 can be misaligned with the consensus wind direction 806, even though the north offset can be correctly calibrated. This misalignment can be caused by local variations in the wind direction throughout a wind farm. For example, wind farms can be located in large, geographically diverse areas. Features of the wind farm, such as the layout of the wind turbines, mountains, valleys, bodies of water, other terrain and geographic features, and the like can cause the wind direction to vary across the wind farm. This variation can be referred to as wind veer. This wind veer can cause the offset wind direction 804 to be different from the consensus wind direction 806, even when both wind directions are accurate.

[0092] The wind turbine 800 can be yawed based on a consensus command that reflects the consensus wind direction 806. If the wind turbine 800 is yawed directly based on this consensus command, the wind turbine 800 will be misaligned with the local wind direction. As such, the wind direction correction factor 802 can be applied to the consensus wind direction 806 of the consensus command in order to align the wind turbine 800 with the local wind direction (e.g., the offset wind direction 804). Each of the wind turbines that report offset wind direction data to the consensus controller can report the offset wind direction data without applying a wind direction correction factor. As such, wind veer can be taken into account at the respective wind turbines.

[0093] The wind direction correction factor 802 can be determined by comparing the offset wind direction 804 determined by the wind turbine 800 to the consensus wind direction 806. The wind direction correction factor 802 can be determined based on the assumption that the sensors of the wind turbine 800 and the sensors used to determine the consensus wind direction 806 are generally accurate over a long period of time. The offset wind direction 804 determined by the wind turbine 800 can be offset by the wind direction correction factor 802 in order to produce a corrected wind direction 808, which can be aligned with the consensus wind direction 806. The wind direction correction factor 802 can be adjusted in order to align the corrected wind direction 808 with the consensus wind direction 806.

[0094] In some examples, the wind direction correction factor 802 is adjusted when a difference between the corrected wind direction 808 and the consensus wind direction 806 is -24- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 greater than a threshold value. For example, the wind direction correction factor 802 can be adjusted or updated when a difference between the corrected wind direction 808 and the consensus wind direction 806 is greater than about 5 degrees, greater than about 2 degrees, greater than about 1.5 degrees, greater than about 1 degree, or the like. This ensures that the wind direction correction factor 802 is not erroneously altered based on noise from sensors, local wind fluctuations, or the like, and improves the accuracy of the wind direction correction factor 802.

[0095] The wind direction correction factor 802 can be determined based on data that is obtained over a long period of time and is highly accurate. For example, the wind direction correction factor 802 can be determined based on data received over the course of at least 6 months, at least a year, or the like. In some examples, a wind direction correction factor 802 of 0 can be used before a prescribed period of time (e.g., 1 month, 3 months, 6 months, a year, or the like) has elapsed, and the wind direction correction factor 802 can be updated after the prescribed period of time has elapsed. The accuracy of the wind direction correction factor 802 can be dependent upon the accuracy of north offsets applied to the wind turbine 800 and other wind turbines used to produce the consensus wind direction 806, and the wind direction correction factor 802 can be calibrated after the north offsets are accurately calibrated. By accurately determining the wind direction correction factor 802, the wind turbine 800 can be accurately aligned with the local wind direction, even when there are variations in the wind direction throughout wind turbines used to determine the consensus wind direction 806. Further, the wind turbine 800 can be controlled through the consensus controller, which improves yaw control of the wind turbine 800, as discussed above.

[0096] Wind veer across the wind farm can vary dynamically based on a number of factors. For example, the wind veer across the wind farm can change based on a predominant wind direction (e.g., with winds originating from the north or the south), based on atmospheric conditions (e.g., caused by changing seasons, weather patterns, or the like), based on wind turbine operation parameters (e.g, rotor RPM and the like), and the like. As such, different wind direction correction factors 802 can be determined for and applied to the wind turbine 800 based on current conditions. In other words, a first wind direction correction factor 802 can be determined for and applied to the wind turbine 800 when the consensus wind direction 806 is from the north in the winter, a second wind direction correction factor 802 can be determined for and applied to the wind turbine 800 when the consensus wind -25- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 direction 806 is from the south in the summer, and so forth. An appropriate wind direction correction factor 802 can be applied to the consensus command in order to correctly offset the consensus wind direction 806 based on the current conditions at the wind turbine 800. The wind direction correction factor 802 can be contrasted with the north offsets discussed above in that the wind direction correction factor 802 varies dynamically based on current conditions, whereas the north offset is an internal offset in the yaw controller of a wind turbine that does not typically vary based on current conditions.

[0097] In some examples, different wind direction correction factors 802 can be applied to the wind turbine 800 based on an estimated or measured atmospheric stability at the wind turbine 800. Atmospheric stability can vary based on a variety of atmospheric stability factors, including time of day (e.g., the atmosphere can be more stable at night and less stable in the day), time of year (e.g., the atmosphere can be more stable in the winter and less stable in the summer), environmental conditions (e.g., temperature, wind speeds, other weather factors, and the like), solar position, and the like. The current atmospheric stability of the wind turbine 800 can be estimated based on the atmospheric stability factors and an appropriate wind direction correction factor 802 can be applied to the wind turbine 800 based on the estimated atmospheric stability.

[0098] FIGS. 9 A and 9B illustrate clusters 902a, 902b, 902c, 902d (collectively referred to as clusters 902) of wind turbines 904 that can be defined within a wind farm 900. In FIG. 9A, the clusters 902a, 902b of the wind turbines 904 are defined based on geographic proximity of the wind turbines 904 to one another. A consensus controller 906a, 906b can provide consensus control for the wind turbines 904 in the respective clusters 902a, 902b, and can determine a consensus wind direction for the respective clusters 902a, 902b. The wind turbines 904 of the clusters 902a, 902b in the example of FIG. 9 A may have relatively small wind direction correction factors, as the wind turbines 904 near each other experience similar wind conditions. In FIG. 9B, the clusters 902c, 902d are defined based on similarity of wind conditions (e.g. wind directions 908 and wind speeds) at the respective wind turbines 904. In other words, the clusters 902c, 902d can be defined based on average wind directions that occur at each of the wind turbines 904. A consensus controller 906c, 906d can provide consensus control for the wind turbines 904 in the respective clusters 902c, 902d, and can determine a consensus wind direction for the respective clusters 902c, 902d. The wind turbines 904 of the clusters 902c, 902d in the example of FIG. 9B may have relatively small -26- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 wind direction correction factors, as the wind turbines 904 within each of the clusters 902c, 902d experience similar wind conditions.

[0099] In the examples of FIGS. 9A and 9B, the consensus controllers 906a, 906b, 906c, 906d (collectively referred to as consensus controllers 906) can be provided for each of the clusters 902 and consensus control can be performed on the basis of the clusters 902 rather than the wind farm 900. This can help to account for variations in the wind direction across the wind farm 900. For example, the wind direction can vary differently across the wind farm 900, but similarly within each of the clusters 902. Performing consensus control on the basis of the clusters 902 helps to account for this variation. The number of wind turbines 904 included in each of the clusters 902 can be selected to be large enough to achieve benefits from consensus control. The number of wind turbines 904 included in each of the clusters 902 can be selected to be small enough to avoid significant communication delays between the wind turbines 904 and the consensus controllers 906 and limit variations in local wind directions within the clusters 902.

[0100] Sensors of the wind turbines 904 can become faulty or inoperable over time. For example, the sensors can stop producing data, produce data that is wrong or has excessive variation or volatility, or the like. Faulty or inoperable sensors can be detected by the sensors going offline, data from the sensors flitting between values, data from the sensors drifting over time, or the like. The sensors of a wind turbine can be determined to be faulty or inoperable when a threshold is reached, such as the sensors not reporting data for a period of time, the data from the sensors having a volatility greater than a prescribed value, or the like. In these cases, wind turbines with faulty or inoperable sensors can be removed from the clusters 902. This allows for the consensus controllers 906 to exclude data from the faulty or inoperable sensors (e.g., from wind direction sensors and yaw direction sensors) such that any bad data from the faulty or inoperable sensors does not impact the determination of the consensus wind direction and consensus controls provided by the consensus controllers 906.

[0101] The clusters 902 can be redefined when a wind turbine 904 is removed from a cluster 902, or in order to maintain a minimum number of the wind turbines 904 in each of the clusters 902. FIGS. 9 A and 9B illustrate two methods for redefining the clusters 902. In the example of FIG. 9A, when a wind turbine 904 is excluded from the cluster 902b, a wind turbine 910 from the cluster 902a that is geographically proximal to the cluster 902b is-27- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 moved from the cluster 902a to the cluster 902b e.g., the wind turbine 910 is added to the cluster 902b). In the example of FIG. 9B, when a wind turbine 904 is excluded from the cluster 902c, a wind turbine 912 from the cluster 902d that has a wind direction 914 closest to the wind directions 908a of the wind turbines 902c is moved from the cluster 902d to the cluster 902c (e.g., the wind turbine 912 is added to the cluster 902c). In some examples, in response to a wind turbine 904 being removed from a cluster 902, maintenance can be scheduled for the wind turbine 904 with the faulty or inoperable sensor.

[0102] FIG. 10 illustrates a method 1000 of correcting wind direction data produced by a wind turbine when the wind turbine is yawed out of the wind. In block 1102, a wind turbine is yawed out of the wind. Overall energy production of a cluster or a wind farm can be increased by yawing selected wind turbines out of the wind and this process can be referred to as wake steering. Wake steering can be used in addition to or independently from consensus control. In other words, a selected wind turbine can be purposefully misaligned relative to the wind direction at the selected wind turbine. This can result in reduced energy production at the selected wind turbine, increased energy production at nearby wind turbines, and can be done to achieve a net gain in the energy production of the cluster or the wind farm.

[0103] In block 1004, a correction factor is determined. The wind direction sensors of wind turbines can be designed to detect the wind direction when the wind turbine is aligned with the wind. As a wind turbine is yawed out of the wind or becomes more misaligned relative to the wind direction at the wind turbine, the wind direction detected by the wind direction sensors of the wind turbine can become increasingly incorrect. However, the wind direction detected by the wind direction sensors of a yawed wind turbine can be consistently incorrect relative to the actual wind direction at the yawed wind turbine. As such, a correction factor can be applied to the wind direction detected by the wind direction sensors of the yawed wind turbine in order to accurately detect the actual wind direction at the yawed wind turbine while the yawed wind turbine is yawed out of the wind. The wind direction sensors can also detect wind speed and can be broadly referred to as wind sensors. The correction factor can be based on the wind direction and / or the wind speed detected by the wind direction sensors.-28- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0104] The correction factor can be referred to as a misalignment correction factor, as the correction factor is used to correct a wind direction detected by wind direction sensors when a wind turbine is misaligned with the wind. The correction factor can be a ratio or percentage that is applied to wind direction data generated by wind direction sensors of a yawed wind turbine. The correction factor can generally be in a range of about 0.7 to about 1 (e.g., in a range of about 0% to about 30%), in a range from about 0.6 to about 1 (e.g., in a range of about 0% to about 40%), in a range from about 0.7 to about .8 (e.g., in a range of about 20% to about 30%), in a range from about 0.8 to about .9 (e.g., in a range of about 10% to about 20%), in a range from about 0.9 to about 1 (e.g., in a range of about 0% to about 10%), or the like. The correction factor can be determined by comparing wind direction data generated by wind direction sensors of a misaligned or yawed wind turbine to neighboring wind turbines. In some examples, the correction factor can be determined by comparing wind direction data generated by wind direction sensors of a yawed wind turbine to wind turbines that have the same or similar wind direction correction factors. The correction factor can be determined by comparing a yaw movement of the yawed wind turbine commanded by a yaw controller of the yawed wind turbine (in other words, a commanded yaw misalignment) to an actual yaw misalignment of the yawed wind turbine. The actual yaw misalignment of the yawed wind turbine can be detected by wind direction sensors and yaw direction sensors of the yawed wind turbine and neighboring wind turbines. For example, a 20 degree yaw movement can be commanded by the yaw controller of the yawed wind turbine and a 25 degree difference in detected wind direction can be detected by the wind direction sensors of the yawed wind turbine. In this case, a correction factor of 0.8 can be determined by dividing the commanded yaw movement by the detected wind direction change.

[0105] The correction factor can vary dynamically based on a number of factors. For example, the appropriate correction factor can change based on atmospheric conditions, geographic features around the wind turbine, the distance of the yaw movement commanded by the yaw controller, and the like. Moreover, different correction factors may be appropriate for different wind turbines (e.g., specific wind turbines, models of wind turbines, or the like). As such, different correction factors can be determined for and applied to wind turbines based on the current conditions at the respective wind turbines. Further the correction factor can be improved over time by continuously monitoring wind turbine yawing. Machine learning and-29- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 artificial intelligence can be used to predict correction factors that will be appropriate in varying situations.

[0106] In block 1006, wind direction data is adjusted. Once an appropriate correction factor for a yawed wind turbine is determined, the correction factor can be applied to wind direction data produced by wind direction sensors of the yawed wind turbine. This ensures that the wind direction sensors can accurately detect the wind direction at a wind turbine, even when the wind turbine is yawed out of the wind or misaligned with the wind direction at the wind turbine.

[0107] FIG. 11 illustrates a method 1100 of controlling the yaw of a wind turbine based on signals received from a consensus controller. In block 1102, internal turbine data is generated by sensors of a wind turbine. The internal turbine data can include yaw direction data for the wind turbine, wind direction data for the wind turbine, and the like. The internal turbine data generated by sensors of a plurality of wind turbines can be provided to the consensus controller, which can provide yaw commands for the plurality of wind turbines (e.g., for wind turbines in a wind farm, a cluster, or the like). As discussed above, the internal turbine data can be offset by a north offset to ensure that each of the wind turbines send internal turbine data to the consensus controller that is calibrated to the same reference point.

[0108] In block 1104, the wind turbine receives consensus yaw commands from the consensus controller. The consensus controller can determine consensus yaw commands based on a consensus wind direction. The consensus wind direction can be determined from the internal turbine data received from the plurality of wind turbines. The consensus wind direction can be determined based on external wind data (e.g., wind data provided by reanalysis weather services or databases, weather stations, weather sensors, or the like). In determining the consensus wind direction, the consensus controller can compare and filter signals from any of the data sources used to determine the consensus wind direction. As such, the signals provided by the consensus controller (e.g., the consensus yaw commands) can have a lower variability relative to the signals provided by the sensors of the wind turbines.

[0109] In block 1106, the wind turbine is yawed based on the consensus yaw commands. The consensus yaw command sent to the yaw controller of the wind turbine can account for the wind direction correction factor, such that the yaw controller directly yaws the wind turbine based on the consensus yaw commands. In some examples, the yaw controller of the -30- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 wind turbine can offset the consensus yaw command based on a wind direction correction factor. Because the consensus controller filters signals from the data sources used to determine the consensus wind direction, the consensus yaw commands can accurately represent the current wind direction with minimal noise. As such, the yaw controller can yaw the wind turbine based on the consensus yaw command with high gain and can quickly respond to changes in the wind direction (e.g., non-turbulent changes in the wind direction). The yaw controller of the wind turbine can omit filters and can respond directly to the consensus yaw commands without filtering signals from the consensus controller. In contrast, traditional systems that control the yaw of wind turbines without consensus control include filters in the yaw controller, and this can slow reactions of the yaw controller to changes in the wind direction.

[0110] FIG. 12 shows a diagram of a system 1200 including a controller 1202 that supports techniques for calibrating wind turbines and performing consensus control on wind turbines in accordance with aspects of the present disclosure. The controller 1202 can be referred to as a consensus controller or the like. The controller 1202 may be an example of or include the components of a computer, a computing device, a computing system, or another electronic device. The controller 1202 may be an example of a portable electronic device, a computer, a laptop computer, a tablet computer, a smartphone, a cellular phone, a wearable device, an internet-connected device, a server, a database, or the like. In some examples, the controller 1202 may be configured for bi-directional wireless communication with other systems or devices using a base station or access point. For example, the controller 1202 may be configured for bi-directional communication with any of the wind turbines discussed above with respect to FIGS. 1A through 11. The controller 1202 may be configured to perform or facilitate any of the methods discussed above with respect to FIGS. 1A through 11.[OHl] The controller 1202 may include a processor 1204, a memory 1206, software 1208, a network transceiver 1210, and an I / O controller 1212. These components may be in electronic communication with one another via one or more buses (e.g., a bus 1214). The controller 1202 may communicate wirelessly with one or more other devices or computing systems over a network using the network transceiver 1210. For example, the controller 1202 may communicate wirelessly with one or more wind turbines in a wind farm, a cluster of wind turbines, or the like.-31- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0112] The processor 1204 may include an intelligent hardware device (e.g., a general-purpose processor, a DSP, a central processing unit (CPU), a microcontroller, an ASIC, an FPGA, a programmable logic device, a discrete gate or transistor logic component, a discrete hardware component, or any combination thereof). The processor 1204 may be configured to execute computer-readable instructions stored in the memory 1206 to perform various functions (e.g., functions or tasks supporting techniques for calibrating wind turbines and performing consensus control on wind turbines).

[0113] The memory 1206 may include random access memory (RAM) and read only memory (ROM). The memory 1206 may store computer-readable, computer-executable software 1208 including instructions that, when executed, cause the processor 1204 to perform various functions described herein. In some cases, the memory 1206 may contain, among other things, a BIOS which may control basic hardware or software operation such as the interaction with peripheral components or devices.

[0114] The software 1208 may include code to implement aspects of the present disclosure, including code to support techniques for calibrating wind turbines and performing consensus control on wind turbines. The software 1208 may be stored in a non-transitory computer-readable medium such as system memory or other memory (e.g., the memory 1206). In some cases, the software 1208 may not be directly executable by the processor 1204, but may cause a computer (e.g., when compiled and executed) to perform functions described herein.

[0115] The transceiver 1210 may communicate bi-directionally, via one or more antennas, wired, or wireless links. For example, the transceiver 1210 may represent a wireless transceiver and may communicate bi-directionally with another wireless transceiver. The transceiver 1210 may also include a modem to modulate the packets and provide the modulated packets to the antennas for transmission, and to demodulate packets received from the antennas.

[0116] The I / O controller 1212 may manage input and output signals for the controller 1202. The I / O controller 1212 may also manage peripherals not integrated into the controller 1202. In some examples, the VO controller 1212 may represent a physical connection or port to an external peripheral. In some examples, the I / O controller 1212 may utilize an operating system such as iOS®, ANDROID®, MS-DOS®, MS-WINDOWS®, OS / 2®, UNIX®,-32- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 LINUX®, or another known operating system. In some examples, the I / O controller 1212 may represent or interact with a modem, a keyboard, a mouse, a touchscreen, or a similar device. In some examples, the I / O controller 1212 may be implemented as part of a processor. In some examples, a user may interact with the controller 1202 via the I / O controller 1212 or via hardware components controlled by the I / O controller 1212.

[0117] The controller 1202 may receive data from external devices, such as wind turbines, reanalysis weather services or datasets, satellites, weather stations, weather sensors, and the like, as discussed above with respect to FIGS. 1A through 11. The data may include internal turbine data, external turbine data, and the like. The controller 1202 may determine a consensus wind direction from the internal turbine data and the external turbine data and may send the consensus wind direction and consensus command signals to the wind turbines. Various calibration operations (including determination of the north offset, wind direction correction factors, misalignment correction factors, and the like) for the wind turbines can be performed by the controller 1202, by yaw controllers of the wind turbines, or by a combination thereof. The controller 1202 may determine clusters within a wind farm and may redefine the clusters based on data received from the wind turbines. In some examples, any or all of the calculations described as being performed by the controller 1202 can be performed by an external computing device (such as a computer, laptop, or other computing device), and updating settings can be provided to the controller 1202 through any wired or wireless (e.g., through cloud-based communications or the like) communication means. The settings for the controller 1202 can be updated periodically or automatically, and updated settings can be determined by the controller 1202, the external computing device, or the like. By performing consensus control and calibrating wind turbines using the controller 1202, the wind turbines can be accurately calibrated, receive accurate yaw commands based on current wind conditions, and respond more quickly and accurately to the current wind conditions at each respective wind turbine. As a result, energy production by the wind turbines controlled by the controller 1202 can be increased.

[0118] Although the present disclosure has been described in the context of performing consensus control of wind turbines, the disclosed systems and methods can be used to perform consensus control in any application that provides control based on sensors.-33- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21

[0119] Other examples and implementations are within the scope and spirit of the disclosure and appended claims. For example, features implementing functions may also be physically located at various positions, including being distributed such that portions of functions are implemented at different physical locations. Also, as used herein, including in the claims, “or” as used in a list of items prefaced by “at least one of indicates a disjunctive list such that, for example, a list of “at least one of A, B, or C” means A or B or C or AB or AC or BC or ABC (i.e., A and Band C). Further, the term “exemplary” does not mean that the described example is preferred or better than other examples.

[0120] The foregoing description, for purposes of explanation, uses specific nomenclature to provide a thorough understanding of the described embodiments. However, it will be apparent to one skilled in the art that the specific details are not required in order to practice the described embodiments. Thus, the foregoing descriptions of the specific embodiments described herein are presented for purposes of illustration and description. They are not targeted to be exhaustive or to limit the embodiments to the precise forms disclosed. It will be apparent to one of ordinary skill in the art that many modifications and variations are possible in view of the above teachings.-34- 4935-3462-4855U

Claims

Attorney Docket No. P316557.W0.01_520049-21CLAIMSWhat is claimed is:

1. A method for calibrating a wind turbine, the method comprising:receiving internal wind turbine data from a wind turbine;receiving external wind turbine data for the wind turbine from a source external to the wind turbine;comparing the internal wind turbine data to the external wind turbine data; and determining a north offset for the wind turbine based on the comparison of the internal wind turbine data to the external wind turbine data.

2. The method of claim 1, further comprising:updating the internal wind turbine data based on the north offset;generating wind turbine wake data based on the updated internal wind turbine data; andupdating the north offset based on the wind turbine wake data.

3. The method of claim 1, wherein the internal wind turbine data comprises at least one of turbine yaw angle data or turbine wind direction data.

4. The method of claim 1, further comprising adjusting the north offset for the wind turbine in response to a difference between the internal wind turbine data and the external wind turbine data being greater than a threshold value.

5. The method of claim 4, wherein the threshold value is determined based on a quality or quantity of the internal wind turbine data and the external wind turbine data.

6. The method of claim 1, further comprising determining a consensus wind estimate for a plurality of wind turbines based on internal wind turbine data from the plurality of wind -35- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 turbines, wherein:the wind turbine is included in the plurality of wind turbines; andthe external wind turbine data comprises the consensus wind estimate.

7. The method of claim 6, further comprising excluding the wind turbine from the determination of the consensus wind estimate in response to a difference between the internal wind turbine data and the external wind turbine data being greater than a threshold value.

8. The method of claim 1, wherein the external wind turbine data comprises reanalysis wind direction data.

9. The method of claim 1, wherein the external wind turbine data comprises a wind turbine yaw angle determined based on satellite imagery of the wind turbine.

10. The method of claim 1, further comprising determining an aggregate wind direction for a plurality of wind turbines based on internal wind turbine data from the plurality of wind turbines, wherein:the wind turbine is included in the plurality of wind turbines; andthe external wind turbine data comprises the aggregate wind direction.

11. A method for managing northing offsets for wind turbines, the method comprising:determining a consensus wind estimate for a plurality of wind turbines; receiving wind direction data from a first wind turbine of the plurality of wind turbines;detecting a bias between the wind direction data and the consensus wind estimate; and adjusting a northing offset for the first wind turbine based on the bias.

12. The method of claim 11, wherein the northing offset is adjusted when the bias is greater than a threshold value.-36- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-2113. The method of claim 11, further comprising removing the first wind turbine from the plurality of wind turbines for the determination of the consensus wind estimate in response to the bias being greater than a threshold value.

14. The method of claim 11, further comprising:receiving yaw data from the first wind turbine;detecting a bias between the yaw data and the consensus wind estimate; and adjusting the northing offset for the first wind turbine based further on the bias between the yaw data and the consensus wind estimate.

15. The method of claim 11, wherein the determination of the consensus wind estimate is based at least partly on reanalysis weather data.

16. A non-transitory computer-readable medium storing code for determining northing offsets for wind turbines, the code comprising instructions executable by a processor to: group a plurality of wind turbines into a plurality of clusters;compare wind turbine data for each of the wind turbines to a consensus wind estimate for the plurality of wind turbines; andapply a northing offset to each of the wind turbines in each of the clusters to minimize a difference between the wind turbine data for each of the wind turbines and the consensus wind estimate.

17. The non-transitory computer-readable medium of claim 16, wherein the plurality of wind turbines is grouped into the plurality of clusters based on geographic proximity.

18. The non-transitory computer-readable medium of claim 16, wherein the plurality of wind turbines is grouped into the plurality of clusters based on an average wind direction at each of the respective wind turbines.-37- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-2119. The non-transitory computer-readable medium of claim 16, wherein the code further comprises instructions executable by the processor to re-group the plurality of wind turbines into the plurality of clusters based on a difference between a first wind turbine of the plurality of wind turbines and the consensus wind estimate being greater than a threshold value.

20. The non-transitory computer-readable medium of claim 16, wherein the code further comprises instructions executable by the processor to:determine a wake profile for each of the wind turbines; andadjust the northing offset for each of the clusters based on the determined wake profiles.

21. A non-transitory computer-readable medium storing code for determining wind veer across a wind farm, the code comprising instructions executable by a processor to:receive yaw direction data for a plurality of wind turbines for a period of time; determine wind farm aggregate yaw direction data based on the yaw direction data for the plurality of wind turbines;receive wind direction data for a first wind turbine of the plurality of wind turbines for the period of time; anddetermine a wind direction offset for the first wind turbine based on a difference between the wind farm aggregate yaw direction data and the wind direction data for the first wind turbine.

22. The non-transitory computer-readable medium of claim 21, wherein the code further comprises instructions executable by the processor to dynamically adjust a yaw direction for the first wind turbine based on real-time wind farm aggregate yaw direction data offset by the wind direction offset.-38- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 23. The non-transitory computer-readable medium of claim 21, wherein the code further comprises instructions executable by the processor to determine consensus wind direction data based on weather data, wherein the wind direction offset is determined based on the consensus wind direction data.

24. The non-transitory computer-readable medium of claim 21, wherein:the code further comprises instructions executable by the processor to calculate consensus wind direction data based on wind direction data received from each of the wind turbines of the plurality of wind turbines; anda respective wind direction offset is applied to the wind direction data received from each of the wind turbines of the plurality of wind turbines.

25. The non-transitory computer-readable medium of claim 21, wherein the wind direction data for the first wind turbine is received from a wind sensor or a yaw controller of the first wind turbine.

26. The non-transitory computer-readable medium of claim 21, wherein the wind direction offset is determined after the period of time, and wherein an initial wind direction offset of 0 is used before the period of time has elapsed.

27. The non-transitory computer-readable medium of claim 21, wherein different wind direction offsets are determined for different consensus wind directions in the consensus wind direction data.

28. The non-transitory computer-readable medium of claim 21, wherein different wind direction offsets are determined depending on an estimated atmospheric stability.

29. The non-transitory computer-readable medium of claim 21, wherein:the plurality of wind turbines further comprises a second wind turbine; and-39- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 the wind direction offset is determined based further on a predicted wake generated by the second wind turbine.

30. A method compri sing :defining a cluster comprising a plurality of wind turbines;detecting a condition in a first wind turbine of the plurality of wind turbines; and in response to detecting the condition, re-defining the cluster.

31. The method of claim 30, wherein the cluster is re-defined by removing the first wind turbine from the cluster and adding a second wind turbine to the cluster.

32. The method of claim 30, wherein the detected condition comprises at least one of a sensor in the first wind turbine going offline or a volatility of a data stream received from the sensor in the first wind turbine reaching a threshold value.

33. The method of claim 30, wherein:each cluster of a plurality of clusters is re-defined in response to detecting the condition; andthe plurality of clusters comprises the cluster.

34. The method of claim 30, wherein:the cluster is defined by grouping wind turbines that are geographically proximal to one another; andthe cluster is re-defined by replacing the first wind turbine with a second wind turbine geographically proximal to the first wind turbine.

35. The method of claim 30, wherein:the cluster is defined by grouping wind turbines that have wind direction offsets closest to one another; and-40- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 the cluster is re-defined by replacing the first wind turbine with a second wind turbine having a wind direction offset closest to the first wind turbine.

36. A system for providing yaw control for wind turbines, the system comprising:a wind farm comprising a plurality of wind turbines, a first wind turbine of the plurality of wind turbines comprising:a wind direction sensor; anda yaw controller; anda consensus controller coupled to the wind turbines of the plurality of wind turbines, the consensus controller configured to calculate a consensus wind direction for the wind farm,wherein:the yaw controller is configured to determine a wind direction offset for the first wind turbine based on a first condition detected by comparing data from the wind direction sensor to the consensus wind direction; andthe consensus controller is configured to define one or more groupings of the wind turbines of the plurality of wind turbines based on a second condition detected by comparing data from the wind direction sensor to the consensus wind direction.

37. The system of claim 36, wherein the first condition comprises a persistent difference between the data from the wind direction sensor and the consensus wind direction.

38. The system of claim 36, wherein the second condition comprises an offline state in the wind direction sensor or a varying difference between the data from the wind direction sensor and the consensus wind direction.

39. The system of claim 36, wherein the wind direction offset is determined and the groupings of the wind turbines are defined based on geographic positions of the wind turbines in the wind farm.-41- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 40. The system of claim 36, wherein the wind direction offset is determined and the groupings of the wind turbines are defined based on the consensus wind direction or an estimated atmospheric stability.

41. A wind turbine comprising:a wind sensor configured to output wind data comprising at least one of wind direction data or wind speed data;a yaw controller; anda non-transitory computer-readable medium storing code for correcting the wind data, the code comprising instructions executable by a processor to:command the yaw controller to misalign a rotor of the wind turbine relative to the wind data;receive the wind data;apply a correction factor to the wind data to account for the effect of the misalignment of the rotor; andoutput corrected wind data.

42. The wind turbine of claim 41, wherein the correction factor is in a range of 0.7 to 1.

43. The wind turbine of claim 41, wherein the correction factor varies based on at least one of the wind direction data or the wind speed data.

44. The wind turbine of claim 41, wherein the correction factor varies based on a magnitude of the misalignment of the rotor relative to the wind data.

45. The wind turbine of claim 41, wherein the correction factor varies based on an estimated atmospheric stability.-42- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 46. The wind turbine of claim 41, wherein:the wind turbine is a first wind turbine; andthe code further comprises instructions executable by the processor for determining the correction factor based on a difference between the wind data and wind data received from a second wind turbine.

47. The wind turbine of claim 41, wherein the code further comprises instructions executable by the processor for determining the correction factor through machine learning based on differences between historical yaw misalignment commands and historical wind data from the wind sensor.

48. A system for providing yaw control for a wind turbine, the system comprising:a first wind turbine comprising:a wind direction sensor configured to output a wind direction signal; and a yaw controller; anda consensus controller coupled to a plurality of wind turbines, the plurality of wind turbines comprising the first wind turbine,wherein:the consensus controller is configured to determine a consensus wind direction for the plurality of wind turbines based at least in part on the wind direction signal and output a consensus wind direction signal; andthe yaw controller is configured to yaw the first wind turbine based on the consensus wind direction signal.

49. The system of claim 48, further comprising a filter configured to filter the consensus wind direction signal and output a filtered consensus wind direction signal to the yaw controller.-43- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-21 50. The system of claim 48, wherein the yaw controller is configured to yaw the first wind turbine based on the consensus wind direction signal without filtering the consensus wind direction signal.

51. The system of claim 48, wherein the consensus controller is configured to output the consensus wind direction signal with a lower variability than the wind direction sensor.

52. The system of claim 48, wherein a gain of the yaw controller is selected based on energy production by the first wind turbine and yaw activity of the first wind turbine.

53. A method compri sing :receiving wind direction data from a wind direction sensor on a wind turbine; adjusting the wind direction data to produce corrected wind direction data; receiving the corrected wind direction data in a yaw controller of the wind turbine; andadjusting a yaw of the wind turbine through the yaw controller based on the corrected wind direction data.

54. The method of claim 53, wherein adjusting the wind direction data comprises applying a correction factor to the wind direction data to produce the corrected wind direction data.

55. The method of claim 54, wherein the wind turbine is a first wind turbine, the method further comprising:yawing the first wind turbine to misalign a rotor of the first wind turbine with a current wind direction; anddetermining the correction factor based on a difference between the wind direction data and wind direction data received from a second wind turbine proximal the first wind turbine.-44- 4935-3462-4855UAttorney Docket No. P316557.W0.01_520049-2156. The method of claim 54, further comprising determining the correction factor based on current atmospheric conditions at the wind turbine.

57. The method of claim 53, wherein adjusting the wind direction data comprises determining a consensus wind direction based on the wind direction data and wind direction data received from wind direction sensors from one or more additional wind turbines.

58. The method of claim 57, wherein producing the corrected wind direction data comprises applying a wind direction offset specific to the wind turbine to the consensus wind direction.

59. The method of claim 53, wherein adjusting the wind direction data comprises comparing the wind direction data to weather data and producing the corrected wind direction data based on the comparison.

60. The method of claim 53, wherein the corrected wind direction data is filtered before being received by the yaw controller.-45- 4935-3462-4855U

Images