Controlling activation of individual pitch control of wind turbine rotor blades based on detected wind events

Abstract

A controller for a wind turbine having three rotor blades is provided for controlling activation of individual pitch control of the rotor blades. The controller is configured to receive a flapping load signal from the flapping load sensor of each of the three rotor blades, the flapping load signal indicating a flapping load on each of the respective rotor blades. The controller is configured to determine a statistical discrete parameter of the flapping load of each rotor blade based on the received flapping load signal, the statistical discrete parameter being indicative of a wind event in a wind farm in which the wind turbine operates. The controller is configured to control activation of the individual pitch control based on the respective statistical discrete parameters.

Description

Technical Field

[0001] The present invention relates to the activation (start-up) of individual pitch controls for controlling rotor blades of a wind turbine. In particular, the activation of the individual pitch controls is based on wind events detected in a wind farm of the wind turbine, such as turbulence and / or shear. Background Art

[0002] Wind turbines known in the art include a wind turbine tower supporting a nacelle and a rotor on which are mounted a plurality (usually three) of adjustable pitch rotor blades. Wind turbines are susceptible to vibrations, such as movement of the tower, nacelle or rotor blades. It is known that certain types of vibrations can be suppressed by active pitching of the rotor blades or by adjusting the generator torque. Control strategies for adjusting the blade pitch can be used to maximize the energy production of the wind turbine while minimizing the loads borne by the various components of the wind turbine.

[0003] The rotor blades may be adjusted as part of a collective pitch control routine, where each of the (three) blades is adjusted simultaneously in the same manner, where such collective pitch control may be used to control wind turbine speed, etc. The rotor blades may also be adjusted as part of an individual pitch control (IPC) routine, where each blade has its own individual pitch reference, possibly as an adjustment to a collective pitch reference from a collective pitch controller. Individual pitch control may be used to mitigate loads caused by rotating wind farms near the wind turbine as the rotor rotates.

[0004] Continuous or excessive activation of the individual pitch controllers may result in excessive wear of the pitch bearings. Therefore, the individual pitch control scheme may be combined with an activation strategy that allows certain critical load issues associated with wind turbine operation to be addressed or resolved without placing excessive demands on the blade bearings. For example, the individual pitch control may be activated to ensure that there is sufficient clearance between the tip of the rotor blade and the wind turbine tower as the rotor blade sweeps across the tower, and / or to reduce extreme resultant blade moments.

[0005] Further improvements in individual blade control are needed to balance reducing loads while ensuring that excessive demands are not placed on the blade bearings.

[0006] The present invention is proposed just against this background. Summary of the invention

[0007] According to one aspect of the invention, there is provided a controller for a wind turbine having three rotor blades, the controller for controlling activation of individual pitch controls of the rotor blades. The controller is configured to receive a flapping load signal from a flapping load sensor of each of the three rotor blades, the flapping load signal being indicative of a flapping load on each of the respective rotor blades. The controller is configured to determine a statistically discrete parameter of the flapping load of each of the rotor blades based on the received flapping load signal, the statistically discrete parameter being indicative of a wind event in a wind park in which the wind turbine is operating. The controller is configured to control activation of the individual pitch controls based on the respective statistically discrete parameter.

[0008] The wind event may include or may be turbulence in the wind field.

[0009] Each statistical discrete parameter may be a standard deviation of the flapping loads of the respective rotor blade.

[0010] The controller may be configured to activate individual pitch control if the statistical discrete parameter indicates that the turbulence of the wind farm is above a prescribed turbulence threshold.

[0011] The controller may be configured to determine which of the statistically discrete parameters has a highest value.The controller may be configured to activate individual pitch control if the highest value statistically discrete parameter is above a prescribed turbulence threshold.

[0012] The controller may be configured to deactivate individual pitch control if the highest valued statistical discrete parameter falls below a prescribed second turbulence threshold value which is less than a prescribed turbulence threshold value.

[0013] In some examples, before determining the statistical discrete parameter, the controller may be configured to remove the frequency component targeted by the individual pitch control from the received flap load signal to obtain a modified flap load signal. The statistical discrete parameter may be determined based on the modified flap load signal.

[0014] Individual pitch control may be used to target a 1P frequency component. In such an example, a controller for controlling activation of individual pitch control may be configured to: obtain a flap load vector in a rotor coordinate frame of the wind turbine based on a received flap load signal; apply an m-blade coordinate transformation to the flap load vector to obtain mutually orthogonal first and second components in a fixed coordinate frame of the wind turbine, wherein the m-blade coordinate transformation may transform the 1P frequency component in the rotor coordinate frame into an 0P frequency component in the fixed coordinate frame; remove the 0P frequency component from the mutually orthogonal first and second components to obtain 1P-removed first and second components; and apply an inverse m-blade coordinate transformation to the 1P-removed first and second components to obtain a modified flap load signal.

[0015] Individual pitch control may be used to target the 2P frequency component. Prior to applying the inverse m-blade coordinate transformation, the controller may be configured to apply a further transformation to the first and second components from which 1P has been removed - i.e., to the so-called "first and second components from which 1P has been removed" - to isolate and remove the frequency component corresponding to the 2P frequency component in the rotor coordinate frame. The controller may be configured to apply a further inverse transformation to obtain the first and second components from which 1P and 2P have been removed, the inverse m-blade coordinate transformation being applied to the first and second components from which "1P and 2P have been removed" to obtain a modified flapping load signal.

[0016] Prior to determining the statistically discrete parameter, the controller may be configured to remove low frequency components indicative of an imbalance in the wind turbine, optionally a rotor imbalance, from the received flap load signal.

[0017] The wind event may include or may be wind shear in the wind farm. In such an example, the controller may be configured to: determine an estimate of at least one of a tilting moment and a yaw moment relative to a rotor plane of the wind turbine, the determination being based on the received flapping load signal, wherein determining the statistical discrete parameter may include determining a respective statistical discrete parameter indicative of wind shear in the wind farm based on the determined tilting moment estimate and the yaw moment estimate or each of the determined tilting moment estimate and the yaw moment estimate; determining a rate of change parameter of the respective statistical discrete parameter or each respective statistical discrete parameter indicative of wind shear; and controlling activation of the individual pitch control based on the rate of change parameter or each rate of change parameter.

[0018] The tilting moment and / or the yaw moment may be determined based on the received flapping load signal by a lookup table, wherein the values ​​of the tilting moment and / or the yaw moment in the lookup table correspond to the tilting moment and / or the yaw moment without individual pitch control of the rotor blades.

[0019] The controller may be configured to activate individual pitch control if: the highest valued statistical discrete parameter indicative of turbulence is above a prescribed turbulence threshold; or at least one or each of the rate of change parameters is above a prescribed shear threshold.

[0020] According to one aspect of the present invention, there is provided a wind turbine comprising a controller as described above.

[0021] According to one aspect of the invention, a method for a wind turbine having three rotor blades is provided, the method for controlling activation of individual pitch controls of the rotor blades. The method comprises receiving sensor data indicative of a flapping load on each of the respective rotor blades from a flapping load sensor of each of the three rotor blades. The method comprises determining a statistically discrete parameter of the flapping load of each of the rotor blades based on the received sensor data, the statistically discrete parameter being indicative of a wind event in a wind park in which the wind turbine is operating. The method comprises controlling activation of the individual pitch controls based on the respective statistically discrete parameter.

[0022] According to another aspect of the present invention, a non-transitory computer-readable storage medium is provided, on which instructions are stored. When executed by one or more processors, the instructions cause the one or more processors to perform the above method. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Examples of the present invention will now be described with reference to the accompanying drawings, in which:

[0024] Figure 1 schematically illustrates a wind turbine according to an aspect of the invention;

[0025] Figure 2 Schematically illustrates an embodiment of the present invention. Figure 1 The master controller of the wind turbine;

[0026] Figure 3 Schematically illustrates Figure 2 A module or component of a pitch adjustment unit of a controller;

[0027] Figure 4 Schematically illustrates Figure 3 an activation module or controller of a pitch adjustment unit;

[0028] Figure 5 The diagram shows a first wind shear event in which the Figure 2 a graph of the results obtained by the controller; and,

[0029] Figure 6 Summary Figure 2 The controller performs the steps of the method. DETAILED DESCRIPTION

[0030] Figure 1 An example of a wind turbine 10 is illustrated in schematic form. The wind turbine 10 includes a tower 102, a nacelle 103 disposed at the apex or top of the tower 102, and a rotor 104 operably coupled to a generator housed within the nacelle 103. In addition to the generator, the nacelle 103 also houses other components required to convert wind energy into electrical energy and various components required to operate, control, and optimize the performance of the wind turbine 10. The rotor 104 of the wind turbine 10 includes a central hub 105 and three rotor blades 106 protruding outwardly from the central hub 105. In addition, the wind turbine 10 includes a control system or controller ( Figure 1 106 . The rotor blades 106 may be pitch-adjustable. The rotor blades 106 may be adjusted according to a collective pitch setting, wherein each blade is set to the same pitch value. In addition, the rotor blades 106 may be adjusted according to an individual pitch setting, wherein each blade 106 may be set to an individual pitch set point.

[0031] The wind turbine 10 may include blade load sensors placed at or near each blade root 109 in a manner such that the sensors detect loads in the blade 106. Blade load signals from such sensors may be used to determine how to adjust the pitch of each individual blade 106. Depending on the location and type of the sensors, loads may be detected in the flapwise (along-flapwise) direction 110 (in-plane / out-of-plane) or edgewise (along-edge) direction 108 (in-plane). For example, such sensors may be strain gauge sensors or optical Bragg sensors. When the sensors are placed on the rotating blades 106, such load signals for each adjustable rotor blade 106 are measured in the rotating reference frame of the rotor 104. Figure 1 A blade load sensor 112 is shown in each blade 106 of FIG. 1 . In various examples, more than one blade load sensor may be provided in each blade. The blade load sensor 112 of the illustrated example is arranged to measure the flapping load on the blade 106 .

[0032] Figure 2 An example of an overall controller 20 of a wind turbine 10 is schematically illustrated, which is implemented to determine individual pitch actuation signals capable of reducing blade loads experienced by rotor blades 106. In the illustrated embodiment, a speed controller (control module / block) 202 of the overall controller 20 minimizes an actual rotor speed ω and a reference rotor speed ω ref The speed error between ref) in order to output the requested power P (in the form of a power set point) and the collective pitch reference θ col The collective pitch reference determined by the speed controller 202 in view of the rotor speed may also take into account further sensor values. Figure 2 In , this is referred to as the measurement set ms which is input into the speed controller 202. The feedback speed controller 202 may be implemented by a PI (proportional integral), PID (proportional integral derivative) or similar control scheme. In one example, the speed controller 202 may alternatively be a model predictive controller arranged to determine a collective pitch reference and / or a power reference based on minimizing a cost function.

[0033] Figure 2 Further illustrated is a control block / module or controller 204 of the overall controller 20, which may be referred to as a pitch actuation unit (PAU). In the pitch actuation unit 204, the pitch correction signal (Δθ1, Δθ2, Δθ3) is determined based on one or more input signals. In some examples, the input signal includes a blade load signal from a load sensor 112 in the blade 106.

[0034] The PAU 204 determines a pitch correction signal or pitch reference offset value (Δθ1, Δθ2, Δθ3) for each rotor blade 106. These offsets are added to the collective pitch reference to provide a result or aggregate pitch correction signal (θ 1, Δθ2, Δθ3) that can be applied to the pitch actuators of the rotor blades 106 individually. A ,θ B ,θ C ).

[0035] exist Figure 2 In the example shown, a collective pitch reference of the pitch-adjustable rotor blades 106 is determined based on the rotor speed, and a resulting pitch correction signal is applied to the pitch-adjustable rotor blades 106. The resulting pitch correction signal is applied individually to the pitch-adjustable rotor blades 106, and for each individual blade 106, the resulting pitch correction signal is based on the signal of the collective pitch reference and the corresponding individual pitch correction signal. In one example, the individual pitch correction signals are applied in a cyclic manner. Therefore, based on the pitch correction signal of each rotor blade 106, a pitch actuation signal of each pitch-adjustable rotor blade 106 is determined.

[0036] The controller 20 may be in the form of any suitable computing device, such as one or more functional units or modules implemented on one or more computer processors. Such functional units may be provided by suitable software running on any suitable computing substrate using conventional or custom processors and memory. One or more functional units may use a common computing substrate (e.g., they may run on the same server) or a separate substrate, or one or two functional units themselves may be distributed among multiple computing devices. Computer memory may store instructions for executing the method performed by the controller, and the processor may execute the stored instructions to perform the method.

[0037] The present invention relates to an improved activation strategy for individual or cyclic pitch control of wind turbine rotor blades. The improved activation strategy aims to optimize the balance between "component load reduction that can be achieved by implementing individual pitch control (IPC)" and "not placing excessive demands on wind turbine blade bearings due to activation of individual pitch control". In particular, the present invention is used to detect wind events in the vicinity of a wind turbine that are associated with an increased likelihood of high component loads, and to activate component load reduction by individual pitch control when such wind events are detected.

[0038] In some examples, the present invention advantageously takes into account the level of turbulence (as a form of wind events) in the wind farm in which the wind turbine is operating, as at least part of the activation strategy of the individual pitch control. Highly turbulent wind farms are relatively uncommon, which means that they do not contribute (significantly) to net energy production. Therefore, there is little economic incentive to design wind turbine rotors that can withstand the high wind turbine component loads associated with energy capture in such wind farms. Therefore, the present invention advantageously provides for the activation of individual pitch control - to relieve loads on certain components - which individual pitch control exhibits a bias towards relatively high turbulence wind farms. In particular, this is achieved by taking into account a statistically discrete parameter (such as a standard deviation) of the blade flapping loads on the wind turbine blades, which is indicative of wind turbulence in the vicinity of the wind turbine. It may be that only when certain turbulence levels are detected, activating the individual pitch controller to reduce the 1P and optionally the 2P frequency components may be an appropriate response: for higher levels of turbulence, it may be necessary to use different control functions (such as using a variable thrust limiter) to correct the wind turbine operation to quickly reduce power.

[0039] In some examples, the present invention advantageously considers wind shear (as a form of wind event) in a wind farm in which the wind turbine is operating as at least part of an activation strategy for individual pitch control. In particular, the statistical dispersion (e.g., standard deviation) of the tilt moment and yaw moment relative to the plane of the wind turbine rotor (i.e., the tilt moment and yaw moment on or experienced by the wind turbine rotor) can be indicative of vertical shear and horizontal shear, respectively. Since shear events that may result in high component loads may be relatively transient in nature, consideration of the rate of change of the corresponding statistically discrete parameters associated with the tilt moment and yaw moment can be used as part of an activation strategy for individual pitch control for load mitigation.

[0040] In the example described, Figure 2 The PAU 204 of FIG. 20 is used to determine a pitch adjustment value or pitch reference offset value Δθ1, Δθ2, Δθ3 for each rotor blade 106 to target certain frequencies or vibration modes to reduce component loads (as part of an individual pitch control routine). The PAU 204 is then further used to determine whether to activate such individual pitch control based on the detection of wind events such as turbulence or shear - that is, whether to apply the pitch reference offset value Δθ1, Δθ2, Δθ3 to the collective pitch reference θ output by the controller 20. col .

[0041] Figure 3 The functional or processing blocks / modules / units of the PAU 204 in one example according to the present invention are schematically illustrated. Specifically, the PAU 204 includes a separate pitch controller or module 31 for determining the pitch adjustment value of the rotor blades, for example to cancel certain frequency modes in the input signal 311.

[0042] The input signal 311 is sensor data from the blade load sensor 112. In the depicted example, the blade load sensor 112 is a flap load sensor that indicates the load on the corresponding blade 106 in the flap direction. As described above, in the depicted example, each blade 106 has a flap load sensor 112. The PAU 204 can receive data from each sensor 112.

[0043] In the described example, the individual pitch controller 31 targets or cancels out the so-called 1P and 2P frequency components in the input signal 311. The individual pitch controller 31 may actually be an independent 1P controller 312 and a 2P controller 313 for generating pitch adjustment values ​​for canceling out the 1P and 2P frequency components, respectively. The 1P frequency is the frequency of rotation of the wind turbine rotor, i.e., the frequency at which the rotor completes a full rotation. A 2P periodic disturbance in the rotor coordinate frame (rotating reference frame) is twice the rotor rotation frequency, but such a disturbance may appear at 3P in the fixed coordinate frame. Therefore, the 2P controller 313 may be used to cancel out vibration modes at 3P in the fixed frame, where 3P is the frequency at which the rotor blades in a three-bladed wind turbine pass the tower, i.e., three times for every rotation of the rotor.

[0044] In the IP controller 312, the sensor data 311 received from the flap load sensor 112 may be used to obtain a three-dimensional vector, wherein each value of the vector indicates a flap bending moment associated with a corresponding one of the three blades 112. The flap load signal is measured in a rotating reference frame or a rotor reference frame. The rotating reference frame may be located in a rotor rotation plane of the wind turbine 10 and may be centered on the rotor 104, for example.

[0045] The flapping load signal can be transformed by an m-blade (multi-blade) coordinate transformation. This transformation brings the three rotation signals along the first reference direction d and the second reference direction q into a fixed reference frame. The m-blade coordinate transformation can take the form of a Coleman transformation. The Coleman transformation can be defined as follows:

[0046]

[0047] Among them, M abc is the bending moment vector (in the rotor coordinate frame), which in the described example is in the flapping direction (but in a different example could be in the edgewise direction), ψ is the (1P) phase, M dq is a vector in a fixed coordinate frame.

[0048] When M abc When is a vector containing the out-of-plane blade root bending moment and the phase is set to the rotor azimuth angle, the directions d and q are the pitch and yaw directions; however, in general, the Coleman transform only transforms the rotation signal into a first component and a second component that are mutually orthogonal in a fixed reference frame.

[0049] Still referring to the 1P controller 312, the blade flapping bending moment M a 、M b 、M cPreprocessing or filtering may be performed before coordinate transformation to remove unwanted signal components. For example, frequency components associated with imbalances in blades 106 may be removed from input signal 311 by applying a high pass filter, particularly applying it equally to each of the three blade load signals.

[0050] The m-blade transformation can then be applied to the (possibly pre-filtered) flapping bending moment vector M abc , to obtain the first component M d and the second component M q The vector M dq , as described above. As a result of this transformation, the 1P component (i.e., the rotor rotation frequency component) in the input signal appears at 0P in the transformed signal. One or more further filters (e.g., a low-pass filter) may then be applied to the component in the fixed frame to remove cyclic loads other than the 1P cyclic load. Once the 1P component has been isolated in this manner, a control action may be determined to offset it or mitigate its effects. This may include applying a PI (proportional-integral) control module to the corresponding component to obtain a control component in the fixed frame. An inverse m-blade transformation may then be applied to the control component in the fixed frame to obtain a control component in the rotor frame, in particular in the form of a pitch correction signal which will be applied to the pitch angle setting of the corresponding rotor blade to offset the 1P component.

[0051] The 2P controller 313 may operate in a corresponding manner. As with the 1P controller 312, in the 2P controller 313, the sensor data 311 received from the flapping load sensor 112 may be used to obtain a three-dimensional flapping bending moment vector. The input signal may be pre-processed to remove frequency components associated with imbalance in the rotor 104. An m-blade transformation may be applied as in the 1P controller 312 to obtain the components in the fixed frame. As described above, this transformation causes the 1P component in the input signal to appear at 0P in the transformed signal. In addition, the 3P component in the input signal corresponds to the collective flapping load, which disappears in the transformed signal. In addition, the 2P and 4P components in the input signal (in the rotor frame) appear at 3P in the transformed signal (in the fixed frame); and the 5P and 7P components in the input signal (in the rotor frame) appear at 6P in the transformed signal (in the fixed frame).

[0052] In order to target the 2P component in the input signal (as required in the 2P controller 313), a further transformation may be applied to isolate this component in the transformed signal. Specifically, the transformation may be based on the phase of the 3P component. In one example, the following transformation is applied:

[0053]

[0054] where ψ 3P It is 3P phase, M d,3P 、M d,3P are the first and second 3P components.

[0055] In this reference frame, the 2P component in the original input signal (in the rotor frame) appears at 0P, while the 4P component in the original input signal now appears at 6P, and can therefore be removed, for example, by applying a low-pass filter. Similar to the 1P controller 312, once the 2P component is isolated in this way, a control action is determined to cancel it or mitigate its effects, in particular to obtain a control component in this reference frame. An inverse transform can then be applied to the control component (e.g., to return to the 1P phase and then back to the rotor frame) to obtain a pitch correction signal to be applied to the pitch angle setting of the corresponding rotor blade to cancel the 2P component.

[0056] Back to Figure 3 , the PAU 204 also includes an activation controller or module 32, which is used to determine whether the pitch adjustment value determined by the individual pitch controller 31 is to be included in the pitch reference setting output by the overall controller 20 or applied to the pitch reference setting as a control action on the relevant pitch actuator.

[0057] As described above, the activation controller 32 provides for biased activation of the individual pitch controllers 31 under certain wind event conditions. In the described example, the activation controller 32 considers wind events in the form of turbulence and wind shear. In a broad sense, the turbulence level of the wind farm in which the wind turbine 10 operates can be detected based on the flapping bending moment sensor data obtained from the blade load sensors 112. In particular, a statistical discrete parameter (e.g., a standard deviation) of the sensor data indicates the turbulence level of the wind farm. In addition, the wind shear level of the wind farm can be detected based on the estimated values ​​of the pitch moment and yaw moment of the rotor blades 106 (for vertical shear and horizontal shear, respectively). In particular, the rate of change of the statistical discrete parameters of the estimated values ​​(e.g., the rate of change of the standard deviation) indicates a transient shear event. If turbulence and wind shear are considered simultaneously in the activation controller 32, the corresponding detections can be combined to determine the total activation gain or parameter to be applied to the pitch correction signal determined by the individual pitch controller 31.

[0058] An important consideration in examples of the present invention relates to which portions of the blade load signal should be included in the turbulence detection. In particular, it may be desirable to have an activation routine that is not coupled to a control feature, as this may result in over-activation of pitch control with associated costs for blade bearings. In the described example, the overall controller 20 provides collective pitch control, as well as 1P and 2P periodic or individual pitch control. To prevent coupling with these control features, the determination or estimation of turbulence in the activation controller 32 may therefore be based on the components above 2P in the blade flapping load signal.

[0059] Figure 4 The modules of the activation controller 32 in one example are illustrated. The sensor data 311 received from the flap load sensor 112 provides the blade flap bending moment M for the corresponding rotor blade 106 a 、M b 、M c As previously described, before determining the turbulence level in the wind farm of the wind turbine 10 based on the received sensor data, the flap load signal is corrected to remove frequency components indicative of imbalance, which are the target of the individual pitch controllers 312, 313 of the overall controller 20 of the wind turbine 10. Therefore, in the activation controller 32, the received blade flapping signal is first processed in the control block 321 to remove relatively low frequency components that may indicate an imbalance in the wind turbine (e.g., rotor imbalance). This can be performed, for example, by applying a high pass filter to each respective signal.

[0060] In order to remove the 1P frequency component in the input signal 311 targeted by the 1P controller 312, a control block 322 is applied in the activation controller 32 to apply an m-blade transformation to the flap load vector formed by the (pre-processed) blade flap bending moment. Specifically, a Coleman transformation can be applied as described above to transform the input signal from the rotor rotating frame to the fixed frame, which places the 1P term in the input signal at 0P in the fixed frame. This component can then be removed by applying an appropriate high-pass filter in the control block 323.

[0061] It is desirable to retain the collective flapping load components 3P, 6P, 9P, etc. from the input signal for subsequent analysis. The output of the Coleman transform control block 322 includes components in the d and q directions, namely M d and M q , which are input to the high pass filter control block 323. Component M d and M q Only the asymmetric periodic load in the input signal is included. That is, the component M d and M qThe components associated with 1P, 2P, 4P, 5P, 7P, 8P, etc. in the rotating frame are retained. The independent channel or output 336 of the Coleman transform control block 322 then includes the symmetric or collective load component in the input signal, that is, the component associated with 3P, 6P, 9P, etc. in the rotating frame. This symmetric component corresponds only to the bending moment vector M defined above. abc Note that the 0P component in the rotation frame of the input signal does not appear in the output of the Coleman transform control block 322 because it is filtered out in the high pass filter control block 321. Also note that although the component M d and M q Input to control block 323, but the collective or symmetric component 326 bypasses this block.

[0062] In order to remove the 2P frequency component in the input signal 311 targeted by the 2P controller 313, the control block 324 applies a further transformation to the component in the fixed frame obtained from the 1P removal control block 324. Specifically, the further transformation can be the above-mentioned 3P phase-based transformation. As described above, this means that the 2P component in the original input signal (in the rotor frame) appears at 0P. Then, applying an appropriate filter (such as a low-pass filter) at the control block 325 can be used to remove other frequency components in the component (from the original input signal), in particular 4P, 5P, 7P, 8P, etc. Therefore, the 2P component remains in the signal component, which is then inversely transformed (relative to the 3P phase transformation) at the control block 326.

[0063] By subtracting the components obtained from control block 326 from the components obtained from control block 323 at subtraction module 327 (i.e., before applying the 3P phase transform), the 2P component in the original signal is removed, while leaving the 4P, 5P, 7P, 8P, etc. components. This is input into the inverse m-blade transformation control block 328 along with the collective components 3P, 6P, 9P, etc. 336 obtained from control block 322 to return the processed signal from the fixed frame to the rotor frame. That is, in the example described, the processed signal includes all frequency components above 2P, i.e., 3P, 4P, 5P, 6P, 7P, etc.

[0064] The processed signal components (i.e., processed to remove low frequency, 1P and 2P components) are now used to determine the standard deviation (or other statistical discrete parameters in different examples) of each signal (i.e., the signal of each blade 106) at the control block 329. The standard deviation can be determined in any suitable manner. The standard deviation can be based on data sampled over a defined time period. The standard deviation can be updated at each sampling instant of the controller (e.g., each instant that sensor data is received from the sensor).

[0065] In the described example, the maximum of the three calculated standard deviation values ​​is then determined or identified at control block 330 for input into a repeater function block 331. That is, in the described example, only the blade load signal with the highest standard deviation is used to determine the activation gain of the individual pitch controls. The highest standard deviation value indicates the highest turbulence level detected in the wind farm of wind turbine 10.

[0066] In some examples, only the detected turbulence level is used to determine the activation gain output by the activation controller 32. In such examples, the highest standard deviation signal is passed to the repeater function block 331 to be compared with a threshold. If the highest standard deviation value is greater than the threshold, the repeater will be set to 1 and the activation gain output by the activation controller 32 will be 1, which means that individual pitch control is activated (because the detected turbulence is high enough).

[0067] It will be appreciated that other indicators based on the standard deviation of the blade load signal may be used as additional or alternative indicators to describe turbulent wind events, such as the rate of change of the standard deviation, ie the time derivative of the standard deviation.

[0068] In the depicted example, the activation controller 32 also considers wind shear events when determining the activation gain output by the activation controller 32. Specifically, the activation controller 32 is provided with estimates of the pitch moment and the yaw moment of the rotor blades 106. The pitch moment and the yaw moment indicate blade moments that are out of a plane defined by the rotor 104 in the pitch direction and the yaw direction, respectively.

[0069] The tilt moment and yaw moment estimates (blade root bending moment) can be based on blade load sensor data. In particular, the tilt moment and yaw moment estimates can be performed by using a flapping load signal and a lookup table (such as a BEM (blade element momentum) lookup table). In particular, the actual tilt moment and yaw moment can be obtained from the measured flapping load signal; however, these actual tilt moments and yaw moments will include the effects of the individual pitch control (of the individual pitch control module 31). For control purposes, it is desirable to obtain estimates of the tilt moment and yaw moment observed in the absence of individual pitch control action. It is for this reason that the measured flapping load signal is combined with the BEM table to obtain the tilt moment and yaw moment estimates. The measured tilt moment and yaw moment can be derived from the measured flapping loads by appropriately applying the Coleman transform. The measured flapping loads can be corrected so that they represent the blade out-of-plane bending moment, and then the Coleman transform is applied to the out-of-plane bending moment. Then, to remove the effect of individual pitch control, the BEM lookup table can first be linearized to account for the effect of individual pitch control associated with relatively small rotor blade pitch angles, and then combined with the measured 1P and 2P periodic pitch signals (in the described example), where the pitch signals are determined by subtracting the collective pitch angle from the measured pitch angle on each blade. This gives an estimate of the effect of the individual pitch controllers on the tilt and yaw moments; therefore, adding these estimates to the tilt and yaw moments, it is possible to estimate what the tilt and yaw moments would be without individual pitch control. In this way, the individual pitch controllers can be activated based solely on the ambient wind conditions near the wind turbine.

[0070] The standard deviation of the respective tilt moment and yaw moment is calculated at control block 332, for example in a manner corresponding to control block 329. While turbulence may be considered more persistent in nature, and therefore the standard deviation itself is of interest, wind shear events may be considered more transient in nature. Therefore, in the described example, the rate of change of the standard deviation of the tilt moment and yaw moment estimates is of interest. Therefore, the time derivatives of the respective standard deviation values ​​from control block 332 are determined at control block 333 and these derivatives are passed to the repeater function block 331. It will be understood that in different examples, only one of the tilt moment and yaw moment may be used when determining the activation gain. It will also be understood that the standard deviation of the tilt moment and / or yaw moment estimates may be considered in addition or alternatively (rather than or in addition to the derivative).

[0071] In the repeater function control block 331, the repeater function is applied to each of the three received signals to obtain a binary output for each signal. Specifically, if the input value is greater than the corresponding threshold, the repeater function outputs 1, otherwise it outputs 0. These binary outputs are then passed to the control block 334, which identifies the maximum value of the three binary outputs as the activation gain. In other words, in the described example, if any of the three values ​​input to the repeater function is greater than their corresponding thresholds (which indicate the minimum level of turbulence or horizontal / vertical wind shear), the output activation gain 335 is 1, that is, the individual pitch control is activated. In different examples, two or all three binary outputs may need to be 1 to set the output activation gain 335 to 1.

[0072] In further different examples, there may be activation gains other than 0 or 1. Specifically, an activation gain between 0 and 1 may be determined based on a value input into block 331 relative to one or more corresponding thresholds.

[0073] Return to reference Figure 3 , it can be seen that the activation gain 335 output by the activation controller 32 is combined with the pitch correction signals output by the 1P and 2P controllers 312, 313 in the multiplier 315. Therefore, if the activation gain 335 is 1, the individual pitch control is activated so that the determined pitch correction signal is applied to the pitch reference output by the controller 20 (see Figure 2 ). On the other hand, if the activation gain 335 is 0, the individual pitch control is deactivated, ie the pitch correction signal is set to 0.

[0074] Figure 5 A graph showing the results obtained with individual pitch control activation as described above in the case of a horizontal wind shear event is shown. Figure 5 In the graph, the graphs with reference denoted by "a" at the end indicate the results when the activation scheme is not used, while the graphs with reference denoted by "b" at the end indicate the results when the activation scheme is used.

[0075] Figure 5 (a) to Figure 5 (f) shows the time-varying curves of blade pitch (unit: degree) 51, generator speed (rpm) 52, power (kW) 53, flapping moment (kNm) 54, blade tip deflection when passing the tower (m) 55 and main bearing yaw moment (kNm) 56. When the activation scheme is not used, the wind event is detected significantly later than when the activation scheme is used. This means that if Figure 5 As shown, when the described control scheme is used, the flapping loads experienced are reduced, resulting in reduced blade root moments, while achieving excellent power and generator speed performance.

[0076] Figure 6 Summarizing the steps of the method 60 performed by the PAU 204 including the activation controller 32 (as part of the overall controller 20) for controlling activation of individual pitch control of the rotor blades 106 of the wind turbine 10. In step 601, the controller 32 receives a flapping load signal from the flapping load sensor 112 of each of the three rotor blades 106, the flapping load signal indicating the flapping load on each of the corresponding rotor blades 106.

[0077] In step 602 of method 60, controller 32 determines a statistical discrete parameter of the flap load of each rotor blade based on the received flap load signal, the statistical discrete parameter being indicative of a wind event in the wind farm in which the wind turbine is operating. In the described example, the standard deviation is used as the statistical discrete parameter; however, in different examples, a statistical variance or other suitable parameter may be used.

[0078] In the described example, the activation controller 32 considers each of turbulent wind events, horizontal wind shear events, and vertical wind shear events. However, it will be understood that in different example embodiments of the activation controller 32, any combination of one or more of these types of wind events may be considered.

[0079] To account for turbulence, a standard deviation of the flapping load signal for each respective rotor blade 106 may be determined. In a different example, instead, a standard deviation of only one or two of the rotor blades 106 may be determined. Prior to determining the standard deviation, the received blade flapping load signal may be processed to remove certain frequency components, and then the standard deviation of the processed / corrected signal is determined. Generally, the blade flapping load signal may be processed to remove frequency components targeted by the individual pitch control routine implemented by the wind turbine. In the described example, the wind turbine 10 includes an individual pitch controller 31 for 1P and 2P frequency components, so in the described example, the activation controller 32 processes the blade load signal to remove the 1P and 2P frequency components prior to determining the standard deviation. That is, the standard deviation is determined based on frequency components above the frequency components targeted by the individual pitch controller (i.e., components above 2P in the described example). In examples where the individual pitch control is targeted only for the 1P component, the standard deviation may be determined based on components above 1P (i.e., including 2P). Low frequency components indicative of imbalance may also be removed from the received blade load signal prior to determining the standard deviation.

[0080] To account for vertical and horizontal shear, the flapping load signal may be used to estimate the pitch and yaw moments experienced by rotor 104, as described above. In the described example, standard deviations of the respective pitch and yaw moments may be determined, and then respective derivatives of the determined standard deviations may be determined.

[0081] In step 603 of method 60, controller 32 controls activation of individual pitch controls of wind turbine 10 based on the determined respective statistically discrete parameters. Determining activation based on the determined statistically discrete parameters may include the parameter itself (e.g., standard deviation) and / or a derivative of the parameter with respect to time (which may include first, second, third, etc. derivatives), such as the rate of change of the standard deviation, as in the described example.

[0082] In the described example, the standard deviation of the corresponding (corrected) blade load signal is determined and the maximum standard deviation value is identified. This maximum value is processed in a repeater function together with the derivatives of the standard deviations for the estimated tilt moment and yaw moment, respectively, to determine whether they are greater than the corresponding thresholds. If the input is below the corresponding threshold, the output of the three signals is 0, indicating that the intensity or severity of the particular wind event is below a certain level; if the input is above the corresponding threshold, the output is 1, indicating that the intensity or severity of the particular wind event is above a certain level. These binary values ​​can then be passed to a functional unit that identifies the maximum value, which is then used as a gain to be applied to the signal from the individual pitch controller 31. Therefore, in the described example, if any binary value is 1, the gain is set to 1, so that the individual pitch control is activated, and if each of the binary values ​​is 0, the gain is set to 0, so that the individual pitch control is deactivated.

[0083] Hysteresis can be introduced for the activation and deactivation thresholds to prevent repeated activation and deactivation of the individual pitch control. For example, the threshold for deactivating the individual pitch control can be less than the threshold for first activating it in the repeater function. Specifically, in the above example where three signals are input to the repeater function unit 331, when one or more of the outputs of the repeater function unit 331 is 1 (so that the individual pitch control is activated), each of the three outputs of the repeater function unit 331 needs to return to 0 before the individual pitch control can be activated. That is, if the turbulence level is deemed to have dropped below its deactivation threshold (which may be less than its activation threshold), the individual pitch control will remain activated if the wind shear level remains above its deactivation threshold.

[0084] In this regard, it is important to note that each of the three signals input to the repeater function unit 331 has its own activation and deactivation thresholds. That is, the activation and deactivation thresholds associated with the turbulence level may be different from the activation and deactivation levels associated with the rate of change of the tilt moment level, which in turn may be different from the activation and deactivation levels associated with the rate of change of the yaw moment level. In some examples, the rate of change of the tilt moment and the yaw moment may have the same activation and deactivation thresholds.

[0085] It is also important to note that the respective activation / deactivation thresholds for turbulence and wind shear levels may vary dynamically for different operating points of the wind turbine. For example, the activation (and deactivation) thresholds for turbulence may scale with wind speed. This is because the turbulence intensity is proportional to the product of the mean wind speed and the standard deviation of the wind speed (or other statistically discrete parameter). The higher the wind speed, the higher the standard deviation exhibited in the flapping load signal. Since examples of the present invention are intended to activate in the event of more extreme wind events (such as extreme turbulence), the activation threshold for turbulence level may increase with increasing wind speed, such that at high wind speeds "normal" or "standard" turbulence levels do not trigger activation of individual pitch control, but rather the activation is reserved for relatively extreme turbulence levels.

[0086] In a similar manner, the activation (and deactivation) thresholds for (rate of change of) wind shear may be scaled with the blade pitch angle. This is because wind shear events (which may be related to tower clearance as the blades rotate) may only be of interest near the elbow or knee of the power curve (i.e. just below the full load operating region of the wind turbine). Thus, for example, lower activation thresholds (higher sensitivity) for pitch and / or yaw shear may be set when the operating point of the wind turbine is near the elbow or knee of the power curve, whereas higher activation thresholds (lower sensitivity) may be set when the wind turbine is operating deeper into the full load region.

[0087] Many modifications may be made to the examples described without departing from the scope of the appended claims.

[0088] In the examples described, an activation strategy is described that takes into account both detected wind turbulence and wind shear; however, it will be appreciated that in different examples, an activation strategy that takes into account only one of them may be provided. It will also be appreciated that the activation strategy that takes into account one or both of the detected turbulence and shear may be combined with an activation strategy that takes into account different factors (such as other wind events or certain fatigue levels) to provide an overall activation strategy for individual pitch control.

[0089] In the described examples, activation gains (based on certain wind events) as part of the activation strategy are applied to the output of one or more individual pitch controllers (1P, 2P controllers, etc.). It will be appreciated that different activation gains may be applied within individual pitch controllers. For example, activation gains may be applied to pitch correction settings determined in a 1P or 2P controller, depending on the level of component excitation generated in one or more specific wind turbine components by the 1P or 2P frequency components in the input signal, respectively, so that individual pitch controls are activated only when needed.

[0090] In the above example, the input signal is transformed to a fixed frame to remove the 1P and 2P frequency components using appropriate high / low pass filters. In a different example, the 1P and / or 2P components may be removed in the rotor frame (i.e., without transforming to a fixed frame), for example, by using appropriate notch filters at 1P and / or 2P.

[0091] In the above example, the blade flapping load measurements / signals are processed to remove low frequency components and then the standard deviation of these processed signals is determined to estimate the turbulence level in the wind farm. Specifically, in the above example, the 1P and 2P frequency components are removed as part of this process (because separate pitch controllers for the 1P and 2P components are provided). In a different example, rather than removing the 2P frequency component component in the blade flapping load signal before determining the standard deviation, a term for compensating for the 2P periodic load is applied to the measured blade flapping signal (which may have been processed to remove the 1P component) and then the standard deviation is determined based on the obtained signal including the compensation term. The advantage of this approach is that more of the received input signal is retained from which the turbulence level can be determined and different types of turbulence conditions can be more easily identified. The compensation can be performed in any suitable manner. For example, this can be performed in a manner corresponding to the compensation for wind shear determination described above. Specifically, the measured blade load signal can be combined with a difference term (load increment) associated with the 2P periodic pitch activity, which is a term indicating the influence of the 2P separate pitch control. This difference term or load increment may be determined by isolating the 2P frequency component in the measured blade load signal and using a lookup table (e.g., a BEM lookup table, similar to that described above for wind shear) to estimate what this portion of the signal would be without individual pitch control. In a further example, rather than removing the 1P component prior to determining the standard deviation, a term to compensate for the 1P component difference caused by 1P individual pitch control may be applied to the measured blade load signal.

Claims

1. A controller for a wind turbine having three rotor blades, the controller for controlling activation of individual pitch controls of the rotor blades, the controller being configured to: receiving a flap load signal from a flap load sensor of each of the three rotor blades, the flap load signal indicating a flap load on each of the respective rotor blades; determining a statistically discrete parameter of a flap load of each rotor blade based on the received flap load signal, the statistically discrete parameter being indicative of a wind event in a wind park in which the wind turbine is operating; and, The activation of the individual pitch controls is controlled based on the respective statistically discrete parameters.

2. The controller according to claim 1, wherein: The wind event includes turbulence in the wind field.

3. The controller according to claim 2, wherein: Each statistical discrete parameter is the standard deviation of the flapping load of the corresponding rotor blade.

4. The controller according to claim 2 or claim 3, wherein: The controller is configured to activate the individual pitch control if the statistical discrete parameter indicates that turbulence of the wind farm is above a prescribed turbulence threshold.

5. The controller according to claim 4, wherein: The controller is configured to determine which of the statistically discrete parameters has a highest value, and wherein the controller is configured to activate the individual pitch control if the highest valued statistically discrete parameter is above the prescribed turbulence threshold.

6. The controller according to claim 5, wherein: The controller is configured to deactivate the individual pitch control if the highest valued statistical discrete parameter falls below a prescribed second turbulence threshold value that is less than the prescribed turbulence threshold value.

7. The controller according to any one of claims 2 to 6, wherein: Before determining the statistical discrete parameter, the controller is configured to remove a frequency component targeted by the individual pitch control from the received flapping load signal to obtain a corrected flapping load signal, and the statistical discrete parameter is determined based on the corrected flapping load signal.

8. The controller according to claim 7, wherein: The individual pitch control is for a 1P frequency component, and wherein a controller for controlling activation of the individual pitch control is configured to: obtaining a flap load vector in a rotor coordinate frame of the wind turbine based on the received flap load signal; applying an m-blade coordinate transformation to the flap load vector to obtain mutually orthogonal first and second components in a fixed coordinate frame of the wind turbine, wherein the m-blade coordinate transformation transforms a 1P frequency component in the rotor coordinate frame to an 0P frequency component in the fixed coordinate frame; removing the 0P frequency component from the mutually orthogonal first and second components to obtain the first and second components from which 1P is removed; and, An inverse m-blade coordinate transformation is applied to the first and second components from which IP is removed to obtain the modified flapping load signal.

9. The controller according to claim 8, wherein: The individual pitch control is for targeting a 2P frequency component, and wherein, prior to applying the inverse m-blade coordinate transformation, the controller is configured to apply a further transformation to the first and second components from which 1P has been removed to isolate and remove the frequency component corresponding to the 2P frequency component in the rotor coordinate frame, and to apply a further inverse transformation to obtain the first and second components from which 1P and 2P have been removed, the inverse m-blade coordinate transformation being applied to the first and second components from which 1P and 2P have been removed to obtain the modified flapping load signal.

10. The controller according to any one of claims 2 to 9, wherein: Prior to determining the statistical discrete parameter, the controller is configured to remove low frequency components indicative of an imbalance in the wind turbine from the received flap load signal.

11. A controller according to any preceding claim, wherein: The wind event includes wind shear in the wind farm, and the controller is configured to: determining an estimate of at least one of a tilting moment and a yaw moment relative to a rotor plane of the wind turbine, the determination being based on the received flapping load signal, wherein determining the statistically discrete parameter comprises determining a respective statistically discrete parameter indicative of wind shear in the wind farm based on the or each of the determined tilting moment and yaw moment estimates; determining a respective statistically discrete parameter or a rate of change parameter of each respective statistically discrete parameter indicative of wind shear; and, Activation of the individual pitch controls is controlled based on the or each rate of change parameter.

12. The controller according to claim 11, wherein: The tilting moment and / or the yawing moment is determined based on the received flapping load signal by means of a lookup table, wherein the values ​​of the tilting moment and / or the yawing moment in the lookup table correspond to the tilt and / or moment without individual pitch control of the rotor blades.

13. A controller according to claim 11 or claim 12 when dependent on claim 5, wherein: The controller is configured to activate the individual pitch control in the following situations: the highest value of the statistical discrete parameter indicative of turbulence is above said prescribed turbulence threshold; or, At least one or each of the rate of change parameters is above a specified shear threshold.

14. A wind turbine comprising a controller according to any preceding claim.

15. A method for a wind turbine having three rotor blades, the method for controlling activation of individual pitch controls of the rotor blades, the method comprising: receiving sensor data from a flap load sensor of each of the three rotor blades, the sensor data indicative of a flap load on each of the respective rotor blades; determining a statistically discrete parameter of a flapping load of each rotor blade based on the received sensor data, the statistically discrete parameter being indicative of a wind event in a wind park in which the wind turbine is operating; as well as, The activation of the individual pitch controls is controlled based on the respective statistically discrete parameters.

Images