METHOD FOR DETECTING THE INCORRECT POSITION OF A ROTOR BLADE ON A WIND POWER PLANT

DK4253753T3Active Publication Date: 2026-11-09WOBBEN PROPERTIES GMBH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
DK2022166371T
Authority / Receiving Office
DK · DK
Patent Type
Patents
Current Assignee / Owner
Filing Date
2022-04-01
Publication Date
2026-11-09
Estimated Expiration
2042-04-01

AI Technical Summary

Technical Problem

Wind turbines face challenges in detecting and correcting aerodynamic imbalances caused by blade misalignment, which can lead to uneven loads and increased costs due to complex and costly balancing procedures.

Method used

A method using a state observer and a Kalman filter to detect blade misalignment by monitoring the azimuth movement of the nacelle and adjusting the blade angles based on the detected misalignment, without requiring additional sensors, by utilizing existing variables from the azimuth adjustment device.

Benefits of technology

This approach allows for efficient and cost-effective detection and correction of blade misalignment, reducing operational costs and maintaining optimal turbine performance by automatically adjusting blade angles to balance aerodynamic loads.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to a method for detecting at least one blade misalignment of a rotor blade of a wind turbine rotor with multiple rotor blades whose blade angles are adjustable. The blade misalignment describes a deviation of the blade angle of a detected rotor blade from a reference blade angle. The wind turbine comprises a nacelle with the rotor and an azimuth adjustment device for adjusting the nacelle in an azimuth orientation. To adjust the azimuth orientation, the azimuth angle is followed by the azimuth adjustment device by means of a predefinable target azimuth angle. The blade misalignment is detected as a function of an azimuth movement of the nacelle. The invention aims to provide a solution for detecting aerodynamic imbalances with minimal cost or additional expense.At least an alternative to previously known solutions should be proposed.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] The present invention relates to a method for detecting at least one blade misalignment of a rotor blade of a wind turbine rotor. The present invention also relates to a corresponding wind turbine.

[0002] Wind turbines are well-known; they generate electricity from wind. This is achieved using an aerodynamic rotor with rotor blades, which are driven by the wind. Modern wind turbines feature rotor blades with adjustable blade angles. This allows for an optimal blade angle to be found for every wind condition, and, in particular, the rotor blades can be turned completely or partially out of the wind at high wind speeds to prevent excessive stress.

[0003] If the rotor blades are unevenly adjusted, meaning they have different blade angles, an uneven load can develop on the rotor. This can also be described as aerodynamic imbalance. One blade, for example, will then have a higher or lower load compared to the other blades, and this higher or lower load rotates as the rotor turns. This results in an asymmetrical, rotating load acting on the rotor. In particular, this can lead to a situation where the thrust load from the wind no longer acts purely axially on the rotor axis and is therefore no longer rotationally symmetrical.

[0004] Imbalances, which can involve mass and aerodynamic imbalances, generate loads on wind turbines. An aerodynamic imbalance occurs when a blade has a permanent misalignment at the blade angle, also known as pitch misalignment, causing the aerodynamic forces on that blade to differ from those on the other blades.

[0005] Minor imbalances can occur and must be tolerated by the wind turbine, which is also referred to simply as the turbine. Specified imbalance limits may be defined for this purpose; these limits must be adhered to and not exceeded. It is possible that a wind turbine may exceed a specified imbalance limit and then require adjustments, particularly balancing. Such adjustments or balancing can incur significant costs.

[0006] The basic procedure is as follows: once an imbalance has been quantified (its magnitude determined) and located (its position determined), it can be corrected to balance the wind turbine. In the case of a pitch misalignment, which is primarily caused by an aerodynamic imbalance, the blade angle (also called pitch angle) of the affected rotor blade can be corrected by an offset. For simplicity and synonymously, a rotor blade is also referred to simply as a blade.

[0007] An aerodynamic imbalance, i.e., a pitch misalignment of one or more blades, generates different bending moments in one or more blades, resulting in periodic changes in rotational speed and periodic loads on the tower. To detect such aerodynamic loads, these bending moments, periodic changes in rotational speed, or periodic loads could be evaluated. However, tower vibrations and rotational speed changes are also caused and influenced by factors such as the system control or a mass imbalance. Therefore, such detection can be difficult and complex, and consequently, costly.

[0008] The present invention is therefore based on the objective of addressing at least one of the aforementioned problems. In particular, a solution is to be proposed for capturing aerodynamic imbalances with the lowest possible cost or minimal additional costs. At the very least, an alternative to previously known solutions is to be proposed.

[0009] According to the invention, a method according to claim 1 is proposed. The method relates to detecting at least one blade misalignment of a rotor blade of a wind turbine rotor with multiple rotor blades whose blade angles are adjustable. The blade misalignment describes a deviation of the detected blade angle of the rotor blade from a reference blade angle (or vice versa). Ideally, the reference blade angle is the actual blade angle. The actual blade angle is thus the sum of the detected blade angle and the blade misalignment. The detected blade angle is the blade angle output by a corresponding encoder or other sensor. The aim is to quantitatively detect the blade misalignment in order to calculate the actual blade angle or to correct the detected blade angle.

[0010] Blade misalignment can also be referred to as misangle or blade misangle. Ideally, such a reference blade angle is the actual blade angle, but it has been recognized that in some cases, the deviation from the actual blade angle is not the crucial factor. Instead, the deviation between the blade angles of the rotor blades themselves is more important, as this can create an aerodynamic imbalance of the rotor. For example, if all three rotor blades are misaligned by 0.5° with the same sign, no aerodynamic imbalance will occur. The reference angle can then deviate from the actual blade angle by this example of 0.5°. The reference blade angle can then be an average blade angle of all rotor blades, especially when a target angle is specified for all rotor blades, also known as the collective blade angle.Regardless of this, it can usually be assumed that the reference blade angle corresponds to the actual blade angle.

[0011] Furthermore, the wind turbine is designed to have a nacelle containing the rotor and an azimuth adjustment mechanism. This azimuth adjustment mechanism is designed to move the nacelle to an azimuth angle. In particular, the azimuth adjustment mechanism is designed to align the rotor with the wind. This is usually achieved by adjusting the nacelle, which supports the rotor, accordingly. At least one azimuth adjustment motor, also known as an azimuth motor, is provided for this adjustment. The orientation of the nacelle, and thus of the rotor, can therefore be described by the azimuth angle. The azimuth angle thus quantifies the azimuth orientation.

[0012] To adjust the azimuth alignment, the azimuth angle is tracked by the azimuth adjustment mechanism to follow a predefined target azimuth angle. This target azimuth angle can be set, for example, based on a detected wind direction. In the simplest case, the target azimuth angle corresponds to the wind direction or—depending on the definition—is offset from it by 180°. However, other criteria can also be taken into account. In particular, a target azimuth angle typically follows a changing wind direction with a certain degree of inertia. This results in the target azimuth angle being a relatively stable value that changes only slowly.

[0013] If the target azimuth angle changes, the azimuth alignment is adjusted accordingly. Such adjustment is also possible if the target azimuth angle remains unchanged or its change can be neglected, but the azimuth alignment is altered by external influences. In particular, an uneven wind field acting on the rotor can lead to a change in the azimuth alignment.

[0014] Such a change in azimuth alignment is also counteracted by the azimuth adjustment device, which returns the azimuth alignment to this value.

[0015] It is further proposed that the blade misalignment be detected as a function of the nacelle's azimuth movement. This is based on the understanding that aerodynamic imbalance can cause the nacelle's azimuth orientation to change. For example, aerodynamic imbalance can cause one rotor blade to absorb more thrust from the wind than the other blades due to misalignment. If this blade is on one side during the rotor's rotation, e.g., in the 9 o'clock position, greater pressure is exerted on that side of the rotor, which can lead to an azimuth movement of the nacelle towards that side. The rotor then continues to rotate, and when the same blade is then in the 3 o'clock position, i.e., on the other side, the rotor experiences a higher aerodynamic load, and thus higher pressure, on that side.This can lead to a change in the azimuth orientation of the gondola towards this second side.

[0016] Naturally, the rotor rotates continuously, so the rotor blade continuously moves from the 9 o'clock position to the 3 o'clock position. Accordingly, in the example given, this increased load continuously decreases on one side and then continuously increases on the other side until the rotor blade has rotated from the 9 o'clock to the 3 o'clock position. The continued rotation from the 3 o'clock to the 9 o'clock position then leads to a continuous decrease in the load on this second side and an increase in the load on the first side.

[0017] It was recognized that the aerodynamic imbalance of the rotor, and thus a blade misalignment of a rotor blade, could be derived from this azimuth movement of the nacelle.

[0018] According to one aspect, it is proposed that blade misalignment as a function of an azimuth movement of the nacelle be detected by recording the adjustment activity of the azimuth adjustment device. Specifically, an azimuth adjustment device includes at least one azimuth motor, which effects this azimuth adjustment, thus acting as an actuator for the azimuth adjustment. It is proposed that the adjustment activity of the at least one azimuth motor of the azimuth adjustment device be recorded.

[0019] The azimuth adjustment device, in particular the azimuth motor, adjusts the azimuth orientation of the gondola, which is therefore an adjustment activity of the azimuth adjustment device or at least one azimuth motor. This adjustment activity is coordinated by a system control unit. In particular, setpoint values ​​can be specified, and actual values ​​are often recorded as well. Such actual values ​​can be, for example, motor torques, motor speeds, or motor currents of the azimuth motor. However, other parameters of the azimuth adjustment device are also possible, including an actual value and / or a setpoint value for the azimuth orientation. Electric motors are particularly suitable as azimuth motors, but hydraulic drives can also be used.

[0020] In any case, various parameters of the azimuth adjustment device and / or the azimuth motor are specified and / or recorded by the system control, and thus the adjustment activity of the azimuth setting or the azimuth motor can be well monitored and evaluated.

[0021] It was particularly noted that the tracking of the azimuth alignment according to the target azimuth angle is achieved by a control system that detects the actual azimuth angle and, based on this, controls the azimuth adjustment mechanism, especially the azimuth motor. From the control signal and / or the motor's behavior, conclusions can thus be drawn about the influence of an aerodynamic imbalance on the nacelle's azimuth alignment.

[0022] It was also recognized here that not only can an azimuth movement be detected, but that in particular a control variable of a controller also allows conclusions to be drawn about the forces acting.

[0023] It was also observed that when using a good or fast controller in the azimuth adjustment mechanism, only comparatively small azimuth movements occur, as they are compensated for by the controller. In this case, observing only an azimuth movement can be inaccurate, and using a control variable of the controller, in particular, achieves higher accuracy. Therefore, it is proposed to record the adjustment activity of the azimuth adjustment mechanism, or at least the adjustment activity of one azimuth motor of the azimuth adjustment mechanism.

[0024] According to one aspect, it is proposed that the azimuth adjustment device be designed to counteract changes in the nacelle's azimuth orientation caused by wind by controlling the azimuth angle. In particular, it is proposed that an azimuth adjustment device without a holding brake be used.

[0025] It was particularly noted that the position control of the azimuth adjustment mechanism counteracts, or rather, counteracts, the effects of the wind. Such position control eliminates the need for a holding brake, which could otherwise be used to keep the gondola in a fixed azimuth orientation as long as the wind direction remains constant. Accordingly, it is proposed not to include a holding brake. The position control then takes over the function of the holding brake. Thus, if the wind slightly alters the gondola's azimuth orientation, this is possible without a holding brake, but the position control of the azimuth adjustment mechanism counteracts it. This is precisely what generates the aforementioned adjustment activities, which counteract this change in the gondola's azimuth orientation caused by the wind and can be evaluated to detect aerodynamic imbalance.

[0026] In particular, the azimuth angle is continuously tracked. The actual azimuth angle is continuously measured and compared to the target azimuth angle, and the azimuth angle is adjusted accordingly. For simplicity, the actual azimuth angle is referred to here simply as the azimuth angle. This continuous tracking naturally requires implementation on a digital computer, especially when the actual values ​​are sampled at typical sampling rates.

[0027] This continuous tracking ensures that the described signals are constantly present, thus providing a continuous signal profile. It is particularly important that, due to the continuous tracking, the adjustment activity of the azimuth adjustment mechanism reflects the azimuth movements of the gondola.

[0028] One aspect proposes using a state monitor to detect blade misalignment. Such a state monitor would thus take into account the azimuth movement of the nacelle, in particular the adjustment activity of the azimuth adjustment mechanism or the azimuth motor.

[0029] Thus, a model can be defined that represents the behavior of the wind turbine. At the very least, a model can be provided that represents a relevant aspect of the wind turbine's behavior, namely the influence of blade misalignment. The model can operate in parallel with the actual behavior of the wind turbine, or in parallel with this specific aspect, and can be calibrated using at least one output variable. For this purpose, an output variable from the wind turbine, or the part under consideration, can be compared with the corresponding output variable of the model, allowing the model to be calibrated based on this comparison. The observed variable can then be read directly from the model.

[0030] According to one aspect, it is proposed that the state observer, in order to detect leaf misalignment, has at least a gondola azimuth moment and a leaf misalignment as state variables. The state variables are those quantities that are observed or estimated in the state observer; these terms are synonymous here.

[0031] It was recognized that the nacelle azimuth moment characterizes an important dynamic of the wind turbine in connection with aerodynamic imbalance. From the difference between the nacelle azimuth moment and a nacelle acceleration moment, which can be calculated from the rotational acceleration of the nacelle in the nacelle's direction of rotation, an azimuth driving torque can be calculated and compared with a measured azimuth driving torque in order to adjust the observer states.

[0032] Once adjusted, it can then be compared as the output variable of the state observer with the actual azimuth driving torque of the nacelle of the wind turbine.

[0033] The azimuth drive torque of the gondola, i.e., the torque with which the gondola is driven to adjust, can be determined from the adjustment activity of the azimuth adjustment device. Specifically, it can be determined from the motor current of one azimuth motor or from the combined motor currents of all azimuth motors in the azimuth adjustment device.

[0034] The blade misalignment of at least one rotor blade, which is therefore also a state variable of the state observer and thus a state variable of the state observer's model, can be directly read from the model as an observed quantity. This makes the blade misalignment available as a constant variable. The actual blade misalignment can then be corrected depending on the observed blade misalignment. Furthermore, it can be continuously monitored to determine whether a blade misalignment, or at least an excessive blade misalignment, occurs.

[0035] No additional sensors are required if the parameters already known from the azimuth adjustment device are used, which are already present in the plant control system.

[0036] In particular, it is proposed that the state observer includes the nacelle azimuth torque, an azimuth torque offset, and a blade misalignment of each pair of rotor blades as state variables. The nacelle azimuth torque points in one nacelle rotation direction and is composed of an azimuth torque component of an aerodynamic blade torque (M aero,A , M aero,B , M a-ero,c ) of each rotor blade, directed in the nacelle rotation direction, and the azimuth torque offset, such that the sum of the torque components of all rotor blades and the azimuth torque offset constitutes the nacelle azimuth torque.

[0037] The gondola's direction of rotation is therefore a direction of rotation around a tower axis, i.e., the direction of rotation around which the gondola rotates during azimuth adjustment.

[0038] For each rotor blade, an aerodynamic blade torque can be calculated based on its aerodynamic properties, wind speed, and blade angle. The aerodynamic blade torque depends on its actual blade angle. If this angle is known, it can be calculated using a method described and explained below.

[0039] The actual blade angle is composed of a given or assumed blade angle and a blade misalignment. Therefore, the assumed blade angle and the blade misalignment can be used instead of the actual blade angle to calculate the aerodynamic blade torque. Particularly for modeling the state observer, it can be useful to base the calculation on such a procedure and to treat this blade misalignment of at least one rotor blade as a state variable.

[0040] The aerodynamic blade torque essentially acts at the root of the respective rotor blade. When the rotor blade is horizontal, i.e., in a 3 o'clock or 9 o'clock position, its aerodynamic blade torque acts in its full magnitude as a torque in the direction of nacelle rotation, thus contributing to the maximum nacelle azimuth torque. When it is vertical, it does not act in the direction of nacelle rotation at all and therefore contributes nothing to the nacelle azimuth torque. This is reflected in an azimuth torque component, which can be calculated from the aerodynamic blade torque and the corresponding rotor angle.

[0041] The transitions between these positions are gradual, and the contribution to the nacelle azimuth torque depends accordingly on the rotor's rotational position, i.e., the rotor angle. This contribution to the nacelle azimuth torque is the azimuth torque component of the aerodynamic blade torque of the respective rotor blade, directed in the nacelle's direction of rotation.

[0042] The nacelle azimuth moment is thus composed of these azimuth torque components of all blades directed in the nacelle's direction of rotation, plus the azimuth moment offset. The azimuth moment offset therefore includes all other components of the nacelle torque that do not originate from the aerodynamic torques of the rotor blades. These include, for example, oblique airflow onto the rotor.

[0043] It was particularly recognized that the nacelle azimuth torque is the sum of the nacelle azimuth torque offset and the aerodynamic blade torques transformed in the nacelle rotation direction, i.e., the azimuth torque components. Ideally, these azimuth torque components add up to zero, and deviations from this ideal, where the sum is zero, result from blade misalignment of the rotor blades. The aerodynamic blade torques are calculated from the blade angle, blade misalignment, blade characteristics, and wind speed and transformed in the nacelle rotation direction. The offset, i.e., the azimuth torque offset, includes all other torques acting on the azimuth drive. This can also include crosswinds. A permanent offset from a sensor, such as a decoder possibly present on the azimuth motor or another torque sensor that may have an incorrect zero point, is also a possible cause.

[0044] When considering two blade misalignments, it is specifically assumed that the wind turbine, or rather its rotor, has exactly three rotor blades. It is also assumed that the sum of all blade misalignments is zero. This is based on the understanding that the main problem with blade misalignment is that it leads to undesirable differences in the angles between the rotor blades. If the sum of the blade misalignments is zero, this also means that their average value is zero.

[0045] In steady-state conditions, when all rotor blades are set to the same target angle, the reference blade angle can correspond to the mean of all measured blade angles. If all blade misalignments are corrected, all rotor blades will have the same blade angle, namely the reference blade angle. While this steady-state case is particularly important, as the problem of aerodynamic imbalance can especially arise under these conditions, the measurement of blade misalignment is not limited to this scenario, and this case served only as an illustration. The assumption that the sum of all blade misalignments is zero can also be applied in non-steady-state conditions; in particular, this assumption can generally underlie the model of the state observer.

[0046] If two leaf misalignments are known, the third misalignment can be determined if the sum of these three misalignments is zero. This allows the state observer to be ordered more efficiently compared to a variant that uses all three leaf misalignments as state variables. It also avoids the risk of linear dependencies between state variables.

[0047] The azimuth moment offset can also be considered as a state variable and thus essentially factored out. In particular, the azimuth moment offset is also considered dynamically. It is not strictly necessary to consider or know this azimuth moment offset as an input variable; it can suffice to include it in the structure of the state observer. It has been found that the azimuth moment offset is slower in its dynamic behavior than the component of the nacelle azimuth moment caused by the aerodynamic blade torques, i.e., the aerodynamic imbalance.

[0048] One aspect of this proposal suggests using a Kalman filter as a state observer to detect blade misalignment. It was recognized that quantities such as wind speed, rotor speed, and motor current are relevant parameters for the observer to detect and that these can be noisy. This can be accounted for by a Kalman filter, and therefore its use is proposed.

[0049] According to one aspect, it is proposed that the adjustment activity of the azimuth adjustment device is carried out by evaluating an azimuth drive torque, which denotes a drive torque with which the gondola is adjusted in its azimuth orientation. In particular, it is proposed that the azimuth drive torque be determined as a function of a motor current or motor torque of the at least one azimuth motor. Specifically, the azimuth drive torque is determined as the product of the motor current or motor torque and a proportionality factor.

[0050] It is also conceivable that several azimuth motors are used. In this case, it is proposed that the azimuth drive torque be determined as a function of several motor currents or several motor torques of the several azimuth motors, in particular that the azimuth drive torque be determined as the product of a sum of the several motor currents or a sum of the several motor torques and a proportionality factor.

[0051] To adjust the gondola's azimuth alignment—in fact, to adjust the gondola at all, since azimuth adjustment is the only adjustment typically provided for a gondola—at least one azimuth motor is used. This motor provides torque, which, in a typical azimuth motor, can be proportional to the motor current.

[0052] Neglecting friction losses, the azimuth drive torque is proportional to the motor torque. Also neglected is the dynamics, particularly the stiffness and inertia, of the azimuth adjustment mechanism's gearbox. If there were only one azimuth motor, the proportionality factor would correspond to a gear ratio between the azimuth motor and the nacelle. However, multiple identical azimuth motors are usually present, so the number of azimuth motors is also factored into the proportionality factor. Ideally, with multiple motors, the azimuth drive torque corresponds to the sum of the converted motor torques. The sum of the motor currents of all azimuth motors can be used as the considered motor current. Alternatively, for simplification, a single motor current can be recorded and extrapolated to a total current based on the number of all motors.

[0053] It was particularly recognized here that the motor current or motor torque of at least one azimuth motor provides a very good basis for determining the azimuth drive torque. This makes it possible to accurately determine the azimuth drive torque.

[0054] When using a permanent magnet DC motor, the motor current is proportional to the motor torque, and in this case the motor current and thus the azimuth drive torque can be easily determined.

[0055] The azimuth drive torque thus recorded provides good information about the dynamic behavior of the adjustment activity and therefore allows good conclusions to be drawn about the nacelle azimuth torque and thus about the aerodynamic imbalance, or about the blade misalignments that cause it.

[0056] According to one aspect, it is proposed that the azimuth driving torque is used as the output of the state observer, so that the state observer outputs an estimated azimuth driving torque and a difference between detected and estimated azimuth driving torque is fed back to the state observer as observation error for adjusting observer states.

[0057] It was particularly noted that the azimuth drive torque is very characteristic of aerodynamic imbalance, or rather, that aerodynamic imbalance is well reflected in the azimuth drive torque. Furthermore, it was recognized that the azimuth drive torque, as described above, can be accurately measured and thus provides a good comparative parameter. In particular, it was recognized that the azimuth drive torque is also easily measurable insofar as a directly proportional quantity, namely the aforementioned motor current, can be directly measured. The motor torque is correspondingly also easily measurable.

[0058] Instead of directly using the azimuth drive torque, a representative quantity can be used, referred to here as the azimuth drive quantity. This can be both measured and estimated in the state observer. The motor torque or current of the azimuth motor is particularly suitable as the azimuth drive quantity. The motor torque can be proportional to the azimuth drive torque, with one proportionality factor resulting from the gear ratio of a transmission between the azimuth motor and the nacelle. An additional proportionality factor can be added for the motor current, indicating a relationship between motor torque and motor current.

[0059] According to one aspect, it is proposed that the condition observer for detecting leaf misalignment is based on a model description in which The aerodynamic blade torque M aero for each rotor blade is described by the equation: M aero = 1 3 ∗ 1 2 ∗ π ∗ ρ air ∗ c s λ , α + γ ∗ l bl 3 ∗ v Wind + ω Yaw ∗ l bl ∗ sin ϕ 2 myth: ρ air = Luftdichte λ = ω Rot ∗ l bl v Wind + ω Yaw ∗ l bl ∗ sin φ = Schnelllaufzahl ω Rot = Rotorgeschwindigkeit bzw . Rotordrehzahl in rad / s l bl = Rotorblattlänge α = Blattwinkel γ = Blattfehlstellung des Rotorblattes c s λ α = Aerodynamische Eigenschaften des Rotorblattes ω Yaw ∗ l bl ∗ sin ϕ = Scheinbare Windgeschwindigkeit durch eine Azimutbewegung am Rotorblatt wherein each blade torque M aero describes a torque acting on the rotor through the rotor blade, namely in a direction of rotation transverse to the axis of rotation of the rotor, and wherein it is provided in particular that the sum of the blade misalignments of all rotor blades is zero and the blade misalignment of one of the rotor blades is calculated from the blade misalignments of the other rotor blades, wherein the model description is also based on calculations in which each of the blade torques is transformed into an azimuth torque component directed in the nacelle direction of rotation by means of an angle transformation as a function of a rotor angle related to the respective blade, which can also be referred to as the rotation angle of the rotor, by means of an angular transformation, and the nacelle azimuth torque is calculated as the sum of the azimuth torque components and the azimuth torque offset,Depending on a change in the azimuth rotational speed and the moment of inertia of the gondola, a gondola acceleration torque is calculated; depending on a difference between the gondola azimuth torque and the gondola acceleration torque, an azimuth driving torque is determined, wherein the azimuth driving torque, or a representative, in particular proportional, quantity thereof, forms an output variable of the state observer's model, and in particular two of the blade misalignments, the azimuth torque offset and the azimuth torque, form system states of the model.

[0060] This describes the model underlying the state observer. According to this model description, an aerodynamic blade torque is determined for each of the three rotor blades, namely from wind speed, rotor speed, and blade angle. Furthermore, the aerodynamic blade torque also depends on the blade misalignment of the respective blade. The calculation is also based on properties of the rotor blade, which result particularly from its airfoil, and such properties can be summarized in a coefficient. The coefficient Cs is provided for this purpose.

[0061] The coefficient CS can be described as the aerodynamic bending moment coefficient. It establishes a quantitative relationship between the wind acting on the blade and the resulting aerodynamic blade torque Maero. It depends on the aerodynamic properties of the rotor blade and the blade angle. It is known during the design and / or dimensioning of the rotor blade, but can also be determined through investigations of the rotor blade and / or simulations of the rotor blade.

[0062] In this way, the aerodynamic blade torque, which can be simplified but synonymously referred to as blade torque, can be calculated for each of the three rotor blades. The calculation for each rotor blade naturally uses the value assigned to that specific blade. Most values ​​are the same for all three rotor blades, assuming they are of the same design. Differences exist, or can exist, in the blade misalignment of each individual blade (γA, γB, γC). The blade angles of the rotor blades can also differ, for example, if individual blade control is used.

[0063] The calculation also incorporates a rotor angle ϕ specific to each blade, and this rotor angle ϕ is different for each of the three rotor blades, with each blade-specific rotor angle differing by 120°. Therefore, the rotor angle of one blade can be converted into the rotor angles for the other two blades.

[0064] Blade torque is therefore a torque that results from the wind acting on the rotor blade, essentially bending it around the blade root. This torque has an axis of rotation, even if it doesn't cause rotation around this axis, which is perpendicular to the rotor axis. For example, if the rotor blade is in the 12 o'clock position, this axis of rotation is horizontal and, as in any position, perpendicular to the rotor axis. If the rotor blade is in the 9 o'clock or 3 o'clock position, this axis of rotation, which is fictitious in this respect, is vertical. This axis of rotation thus changes its position with the rotation of the rotor.

[0065] Each blade torque is then converted into an azimuth torque component via an angle transformation. The azimuth torque component thus describes an azimuth torque, i.e., a torque about the vertical tower axis, caused solely by the respective rotor blade in a headwind. This azimuth torque component therefore depends on the rotor angle, i.e., the rotation angle of the rotor relative to the respective blade. At the 12 o'clock or 6 o'clock position of the respective rotor blade, the azimuth torque component is zero, whereas at the 3 o'clock or 9 o'clock position of the respective rotor blade, it corresponds to the aerodynamic blade torque. For the other rotor angles, values ​​in between are obtained.

[0066] A wind blowing directly from the front is defined here as one that flows horizontally in the direction of the rotor axis and contains no other wind components. It therefore flows in accordance with the azimuth orientation of the nacelle. Other wind components can occur and also act on the rotor blade, but they are not taken into account via the aerodynamic blade torque, but rather via an azimuth torque offset.

[0067] This results in three azimuth torque components, which are added together, and the azimuth torque offset is also added. This azimuth torque offset essentially corresponds to the other force effects that can arise from oblique inflows or other wind influences, which are not reflected in the (idealized) wind speed used in the calculation of the aerodynamic blade torques and which, incidentally, is assumed to be the same for all three rotor blades.

[0068] The resulting azimuth moment is called the nacelle azimuth moment. It is therefore the sum of all partial torques resulting from external forces acting on the nacelle, in particular forces acting on the rotor from the outside.

[0069] For the blade misalignment, it is assumed that its sum is zero, and thus one of the blade misalignments can be calculated from the other two, assuming a rotor with three rotor blades.

[0070] Furthermore, the change in azimuth rotational speed is considered. This can be achieved, in particular, by measuring the rotor speed of an azimuth motor and converting it, via a transmission gear, into the rotational speed of the azimuth movement, i.e., the movement of the nacelle around the vertical tower axis. The direct azimuth rotational speed is, of course, very small and can, for example, represent a rotation of just a few degrees, e.g., a few degrees per second or less. However, a more representative, and especially a proportional, quantity can be used, in particular the rotational speed of the azimuth motor, i.e., the motor speed.

[0071] This rotational speed, whether the motor speed or the azimuth speed, is differentiated with respect to time and, if necessary, multiplied by a proportionality factor, particularly a gear ratio between the azimuth motor and the gondola, to obtain an azimuth acceleration. This azimuth acceleration is then multiplied by the moment of inertia of the gondola, resulting in a gondola acceleration torque. The gondola acceleration torque thus describes a torque that would accelerate the gondola by the calculated azimuth acceleration if no external forces were present, i.e., if the gondola azimuth torque were zero.

[0072] In any case, the difference between the gondola azimuth torque and the gondola acceleration torque is calculated, resulting in an azimuth drive torque. This azimuth drive torque forms an output variable of this model description, i.e., this model of the state observer. In particular, a torque representative of the azimuth drive torque, specifically proportional to it, is output—namely, the motor torque of an azimuth motor. This motor torque then forms the output of the model and can be compared with a measured motor torque. The difference between these two can be used to adjust the state variables of the state observer or the state observer model. This is done in a manner known from the theory of the Kalman filter.

[0073] In particular, it is proposed that two of the blade misalignments, the azimuth moment offset and the nacelle azimuth moment, constitute system states of the model. These four states are adjusted when feeding back the difference between the estimated and measured output, i.e., the difference between the motor torque as the observer's output and the measured motor torque. An estimated quantity or estimated state can also be referred to synonymously as an observed quantity or observed state.

[0074] According to one aspect, it is proposed that the state observer is based on a model description that can be described by the following system of equations, where the rotor has three rotor blades A, B and C: M aero , A = f v w n R α γ A ; M aero , B = f v w n R α γ B ; M aero , C = f v w n R α γ C ; M yaw , A = f M aero , A , ϕ ; M yaw , B = f M aero , B , ϕ ; M yaw , C = f M aero , C , ϕ γ c = − γ A − γ B M yaw , Nac = M yaw , A + M yaw , B + M yaw , C + M yaw , off M acc = i ge . i dn DRV dt . J Nac M DRV = 1 i g . M yaw , Nac − M acc with M aero,A , M aero,B and M aero,C : aerodynamic blade torque M aero of rotor blade A, B or C; VW: Wind speed; n R : Rotor speed; α: Blade angle; ϕ : Rotor angle M yaw,A , M yaw,B and M yaw,C : Azimuth torque component of rotor blade A, B or C; γ A , γ B and γ C : Blade misalignment of rotor blade A, B or C; M yαw,off : Azimuth moment offset; M yaw,Nac : Gondola azimuth moment; M acc : Gondola acceleration moment; M acc = i ge .2 π dn DRV dt . J Nac ; and ge : Gear ratio between azimuth motor and gondola; n DRV : Speed ​​of azimuth motor; J Nac : Moment of inertia of gondola M DRV : Motor torque of azimuth motor or sum of all azimuth motor torques if several azimuth motors are used.

[0075] The model used in the state observer is thus described by the equations above.

[0076] The aerodynamic blade torque M aero of each rotor blade can be calculated using the equation specified above. The azimuth torque component of each rotor blade is calculated from the aerodynamic blade torque M aero by an angular transformation as a function of the rotor angle.

[0077] The azimuth moment offset appears only as a subcomponent in the total calculation of the gondola azimuth moment. It can be understood as a disturbance variable and constitutes a state variable of the observer, which is corrected by adjusting for the observer error.

[0078] The blade misalignments, at least two of them, can also be understood as disturbance variables. Each constitutes a state variable of the state observer and is adjusted by the feedback of the observer error, thus determining it in the first place. For these blade misalignments in the model, and the same applies to the azimuth moment offset, initial values ​​can be specified in the state observer during the calculation, for which the value zero is preferably chosen.

[0079] Regarding the state observer model, it is specifically proposed, both in this aspect and the others, that the rotor angle, wind speed, rotor speed, collective blade angle, and motor speed of an azimuth motor be used as input variables. Instead of the collective blade angle, an individual blade angle can also be used for each rotor blade.

[0080] The motor torque of the azimuth motor can be chosen as the output variable of the model and thus of the state observer.

[0081] According to the invention, a method for correcting at least one blade misalignment of a rotor blade of a wind turbine is proposed, based on at least one aspect of at least one of the methods for detecting a blade misalignment described above. This method is thus designed to correct at least one blade misalignment of a rotor blade of a wind turbine rotor with multiple rotor blades whose blade angles are adjustable. The blade misalignment describes a deviation of the blade angle of a detected blade angle of the rotor blade from a reference blade angle, as already explained above regarding the method for detecting a blade misalignment.

[0082] Similarly, this system also uses a wind turbine, which has a nacelle with a rotor and an azimuth adjustment device to move the nacelle into an azimuth orientation. Here too, the azimuth angle is adjusted by means of the azimuth adjustment device to follow a predefined target azimuth angle. The blade misalignment is detected as a function of the nacelle's azimuth movement. In particular, this is done using a method according to at least one of the aspects described above.

[0083] Furthermore, it is proposed that a correction angle be determined for each rotor blade depending on the respective detected blade misalignment, and that the blade angle be corrected by the correction angle.

[0084] It was particularly noted here that such a correction is also possible online using this method. This avoids the need to first detect a blade misalignment, for example in an offline procedure, and then perform a calibration to compensate for it. Instead, the blade misalignment can be detected as a function of the nacelle's azimuth movement and then converted into a correction value.

[0085] A particular advantage here is that the proposed method can run continuously and that leaf misalignments can be output specifically as observed states. The method works in such a way that knowledge of a previously performed correction is not required and therefore does not influence the method, especially not the proposed state observer. The data acquisition method thus does not need to be adapted to a completed correction and can run continuously during operation.

[0086] Blade misalignments can often be caused by manufacturing or installation errors and should only occur at the beginning of operation or after a component replacement. This means that if a blade misalignment occurs during operation, it is likely that the wind turbine is damaged. Such damage could be, for example, on the rotor blade itself, or it could be on an encoder for the corresponding blade angle or blade angle adjustment system. The proposed method can also detect such damage. While it can initially be corrected, it is preferable to issue a warning upon detection of such a blade misalignment during operation, so that the cause of the misalignment can be investigated.

[0087] Therefore, one aspect proposes that a detected leaf deviation be compared to a predefined deviation limit, and an error or warning message be generated if the detected leaf deviation exceeds the deviation limit. Such a deviation limit could, in particular, be in the range of 0.2° to 2°.

[0088] According to one aspect, it is proposed that the correction angle corresponds to the detected leaf misalignment, or that the correction angle is adjusted to match the detected leaf misalignment, in particular with a delay function which has a time constant of at least one hour.

[0089] It was particularly recognized that this enables online implementation and even fully automated implementation. The detected leaf misalignment can be directly converted into a correction value. The very slow tracking eliminates interactions with other control dynamics.

[0090] Due to the particularly preferred design with a long delay time during implementation, this can especially lead to all three rotor blades being adjusted evenly during the correction implementation, so that all blade misalignments are corrected to zero.

[0091] According to the invention, a wind energy system is also proposed which has a rotor with several rotor blades whose blade angles are adjustable, wherein the wind turbine has a nacelle with the rotor and an azimuth adjustment device to adjust the nacelle in an azimuth orientation having an azimuth angle, and to adjust the azimuth orientation the azimuth angle is followed by means of the azimuth adjustment device to a predefinable azimuth setpoint angle, wherein the wind turbine has a control device which is prepared to carry out a method according to one of the above aspects.

[0092] In particular, the control unit may contain a corresponding program that performs the procedure for detection according to at least one aspect described above and / or a procedure for correcting a leaf misalignment according to at least one aspect described above.

[0093] The invention is explained in more detail below by way of example with reference to the accompanying figures. Figure 1shows a wind turbine in a perspective view. Figure 2 shows a schematic front view of a wind turbine rotor. Figure 3 shows a gondola with a rotor schematically in a top view. Figure 4 A schematic representation of a proposed state observer is shown in a block diagram.

[0094] Figure 1 Figure 1 shows a wind turbine 100 with a tower 102 and a nacelle 104. A rotor 106 with three rotor blades 108 and a spinner 110 is mounted on the nacelle 104. During operation, the wind sets the rotor 106 into rotation, thereby driving a generator in the nacelle 104.

[0095] The wind turbine 100 has an electric generator 101, which is indicated in the nacelle 104. Electrical power can be generated by means of the generator 101. A feed-in unit 105, which can be configured as an inverter, is provided for feeding electrical power into an electrical supply network at a grid connection point PCC. A plant control unit 103 is provided for controlling the wind turbine 100 and also the feed-in unit 105.

[0096] Figure 2 Figure 200 shows a rotor 200 with 3 rotor blades 202 in a top view from the front, looking at a spinner 204. The rotor 200 is thus in Figure 2 Shown from the perspective of the wind. During operation, the wind causes the rotor 200 to rotate in the direction of rotation 206. The rotor 200, and thus each rotor blade 202, therefore has a rotor angle 208, which is also denoted by the Greek letter ϕ.

[0097] The rotor angle 208 or ϕ can be related to a rotor reference angle 212, which is in Figure 2 This is shown as an example of a dashed, vertical line. Accordingly, this vertical line would be the reference point, and for a rotor blade that is pointing vertically upwards (12 o'clock position), the rotor angle would therefore have a value of zero.

[0098] For example, in Figure 2 Only one rotor angle is shown as an example for one rotor blade. The rotor angles for the other two rotor blades are correspondingly 120° or 240° larger or smaller, respectively.

[0099] For example, in Figure 2 A blade angle direction 210 is also shown, but only as an example for a rotor blade, into which the rotor blade in question can be adjusted in its rotor blade angle α.

[0100] Figure 3Figure 1 schematically shows a nacelle 300 in a top view. It has a rotor 306 with first, second, and third rotor blades 301, 302, and 303. The second rotor blade 302 is in a 12 o'clock position, i.e., pointing vertically upwards. The first and third rotor blades 301 and 303 are in 8 o'clock and 4 o'clock positions, respectively.

[0101] The gondola 300 is rotatable about a vertical tower axis 308 in the gondola rotation direction 310, i.e., its azimuth orientation is adjustable. A gondola azimuth moment 312 also points in this gondola rotation direction 310.

[0102] To illustrate, in Figure 3A wind 314 acting on the nacelle 300, and thus on the rotor 306, is shown or indicated. The wind 314 acts on each of the three rotor blades 301 to 303, resulting in an aerodynamic blade torque 331 to 333 on each respective rotor blade 301 to 303. Each of these aerodynamic blade torques 331 to 333 results in an azimuth torque component directed in the nacelle rotation direction 310. This depends on the rotor angle ϕ, which is shown in Figure 2 which bears the reference number 208.

[0103] The extent and direction in which the respective aerodynamic blade torque 331 to 333 contributes to the nacelle azimuth moment 312 thus depends on this rotor angle ϕ. According to the definition of the rotor angle ϕ or 208 of the Figure 2The respective azimuth torque component is calculated from the respective aerodynamic blade torque 331 to 333, multiplied by sin(ϕ). The azimuth torque component of the first rotor blade 301 is thus opposite to the nacelle azimuth torque 312, whereas the azimuth torque component of the third rotor blade 303 points in the direction of the nacelle azimuth torque 312. With a constant wind, the azimuth torque components of the first and third rotor blades 301, 303 should cancel each other out in the position shown, and thus the azimuth torque components of all three rotor blades should cancel each other out, since the second rotor blade 302 has an azimuth torque component of zero in the vertical position.

[0104] However, if the individual rotor blades experience different forces from the wind and thus different aerodynamic blade torques because their blade angles α are different, these forces do not cancel each other out and lead to a noticeable contribution to the nacelle azimuth torque 312. This contribution changes as the rotor rotates, particularly cyclically and with a changing sign.

[0105] The gondola azimuth moment 312 also has a further component due to an azimuth moment offset, which is in Figure 3 however, it is not shown.

[0106] To illustrate, in Figure 3A profile is indicated for the second rotor blade 302, and a blade angle α is also shown within it for illustrative purposes. It is assumed here that an orientation of the rotor blade, in this case the second rotor blade 302, with its chord 316 pointing towards a rotor plane 318, corresponds to a blade angle of zero. The blade angle α shown here as an example thus has a blade angle α relative to this rotor plane 318 that, in the example or snapshot shown, has a value of approximately 30°. In partial load operation, the rotor blade angle α typically has a value of approximately 0° to 5°.

[0107] Such a blade angle α is set the same for all rotor blades and is called the collective blade angle. An individual component can optionally be superimposed. Instead of the collective blade angle, an individual blade angle can also be used for each rotor blade if individual blade pitch control is employed. In this case, an individual component can also be superimposed.

[0108] The leaf misalignment to be measured is therefore the deviation of a measured leaf angle from the actual leaf angle α, or from another reference leaf angle that is specified in Figure 3 The blade misalignment is not shown. It must be recorded individually for each rotor blade, but can be recorded in a single, combined process.

[0109] To detect the blade misalignment of each rotor blade, the use of a state observer is proposed, based on a model that is in Figure 4 is shown. Figure 4 This shows a model 400, which takes a rotor angle ϕ, a wind speed Vw, a rotor speed nR, and a collective blade angle α as input variables. It is assumed that there is no individual blade control, so the same setpoint for the blade angle exists for each rotor blade. This is set for each rotor blade, with the set blade angle then deviating from the setpoint by the blade misalignment. The motor speed nDRV of an azimuth motor is also taken as an input variable.

[0110] Furthermore, a blade misalignment γA of a first rotor blade A, a second blade misalignment γB of a second rotor blade B, and a third blade misalignment γC of a third rotor blade C are taken into account. The actual rotor blade angle then deviates from the target value, i.e., the collective blade angle α, by the amount of this blade misalignment for each respective rotor blade. The first two blade misalignments γA and γB are represented as input variables but are not inputted directly into Model 400. They are to be determined by the state observer using this Model 400 and each constitutes a system state of the observer.

[0111] The third leaf misalignment γC is calculated from the first two leaf misalignments γA and γB. It corresponds to the negative value of the sum of the first two leaf misalignments γA and γB. Subtraction block 402 is provided for this purpose. If the assumption that the sum of all three leaf misalignments is zero is not made, the third leaf misalignment γC can also be used as an additional input.

[0112] The model 400 calculates an aerodynamic blade torque M aero,A for each of the three rotor blades A, B and C in an aerodynamic block 411 to 413 , M aero,B and M a-ero,C , which the Figure 4 For the sake of clarity, they are only shown as MA, MB and Mc.

[0113] Each of these aerodynamic blade torques MA, MB and Mc is transformed in a corresponding first, second and third transformation block 421 to 423, taking further account of its rotor angle, into an azimuth torque component M yaw,A, M yaw,B and M yaw,C, which are further clarified in Figure 4 are designated as M y,A , M y,B and M y,C.

[0114] To take your respective rotor angle into account, the transformation blocks 421 to 423 receive the general rotor angle ϕ as an input and convert it accordingly into the individually applicable rotor angle, in which the rotor angle ϕ is increased or decreased by 120° or 240° if necessary.

[0115] The summation block 404 adds the three azimuth torque components M y,A , M y,B and M y,C to the nacelle azimuth torque M y , adding an azimuth torque offset M y,O . For simplicity, the nacelle azimuth torque M y is denoted by the symbol M yaw,Nac , and the azimuth torque offset M y,O is denoted by the symbol M yaw,Offset .

[0116] The azimuth moment offset M y,O is represented as an input variable, but is not entered into the model. Instead, it constitutes a system state of the model and thus of the state observer, which is to be determined by the state observer. The gondola azimuth moment M y is also a system state in this model and thus of the state observer.

[0117] The Model 400 also takes into account the motor speed n DRV as an input variable, which is converted into a nacelle speed n G via a gear ratio ig in the first amplification block 406. The gear ratio ig denotes the transmission ratio of an adjustable gear between the azimuth motor and the nacelle. The transmission ratio ig can also be denoted by the symbol i ge or i gear.

[0118] In derivation block 408, the gondola rotational speed nG is differentiated with respect to time, resulting in a gondola acceleration aG. For simplicity, a factor of 2π has been omitted from the representation. In the second amplification block 410, this gondola acceleration aG is calculated using the moment of inertia JG to determine the gondola acceleration moment MA.

[0119] The moment of inertia JG of the gondola can also be denoted by the symbol J Nac. The gondola acceleration moment MA can also be denoted by the symbol M acc.

[0120] In the differential block 414, the difference between the nacelle azimuth torque M y and the nacelle acceleration torque MA is calculated, resulting in an azimuth drive torque M AG. This can be multiplied in the third amplification block 416 by the inverse transmission ratio ig, yielding the motor torque M m, for which the symbol M Motor can also be used.

[0121] The azimuth drive torque M AG or the motor torque M m forms the output of the model and thus the output of the state observer. This estimated motor torque is compared with a measured motor torque and fed back into the model via a corresponding feedback mechanism to align the system states, i.e., the first blade error angle γ A, the second blade error angle γ B, the azimuth torque offset M y,O, and the nacelle azimuth torque MY.

[0122] According to the invention, the following was thus recognized and the following solutions were proposed.

[0123] In particular, automatic detection of an aerodynamic imbalance based on a Kalman filter and a position-controlled azimuth drive is proposed.

[0124] In wind turbines with azimuth holding brakes, aerodynamic imbalances result in tower torsion, which is difficult to measure. With the proposed use of position-controlled azimuth adjustment devices, such a holding brake is eliminated during operation, as the drive is position-controlled and thus counteracts the torque. This makes aerodynamic imbalances visible in the azimuth movement, which creates a control error, and in the drive torque of the motors, which is a control variable for the azimuth adjustment device.

[0125] This proposal presents a solution in which Kalman filters, which are essentially control system observers, perform online parameter estimation and control system state monitoring. An advantage of this solution is its lean structure and simple parameterization.

[0126] With a position-controlled azimuth adjustment device, the azimuth drive does not have a holding brake like previous azimuth systems, because it is position-controlled. This means the motors are regulated in such a way that the control error (difference between target position and actual position) is as small as possible by using the motor torque as the control variable. This means that the azimuth drive will yield slightly to a torque, e.g., from the wind, and then return to the target position.

[0127] The motor's torque is proportional to the current and can therefore be determined. Furthermore, each motor has a speed sensor, and due to the large gear ratio of >1:1000, the gondola's movement is also very well understood.

[0128] The aerodynamic imbalance results in a nacelle azimuth moment, which depends on the blade position (i.e., the rotor angle of the blade misalignment) and the wind speed. This moment is counteracted by the azimuth drive motors and the inertia of the nacelle.

[0129] The aerodynamic imbalance is continuously and online estimated using a Kalman filter based on the azimuth movements of the gondola. A key requirement is the measurement of the azimuth drive torque. Alternatively, the gondola azimuth torque could be measured, which can be determined from the azimuth rotation speed (or its change) and the torque of the azimuth motors on the position-controlled azimuth drive.

[0130] The Kalman filter has a total of 4 states: Gondola azimuth moment, blade misalignment, blade A, blade misalignment, blade B, azimuth moment offset.

[0131] The nacelle azimuth moment is the sum of the offset and the aerodynamic moments resulting from the pitch misalignment of blades A, B, and C, which can also be referred to as blade misalignment. The aerodynamic moments are calculated from the pitch angle (also known as blade angle), the pitch misalignment, the blade characteristics, and the wind speed. The offset includes all other moments acting on the azimuth drive, such as those caused by an oblique airflow.

[0132] The azimuth motor torque is used as the "measured value" in the context of the Kalman filter. This torque is compared to the sum of the nacelle azimuth torque and the acceleration torque (moment of inertia multiplied by azimuth acceleration). The Kalman filter continuously optimizes its internal states—that is, the states of the Kalman filter model—until the "measured value" from the Kalman filter matches the actual measured value as closely as possible.

[0133] As a result, the pitch errors in sheets A, B and C are obtained after a short time, assuming that the sum of all pitch errors corresponds to 0°.

[0134] Subsequently, the pitch misalignment can be automatically corrected during operation of the wind turbine, thus avoiding imbalance.

[0135] The proposed solution can achieve, or should ideally achieve, the following: 1. Optimization of plant operation: Pitch misalignments result in a reduction in wind turbine yield and increased noise emissions, depending on the degree of misalignment. 2. Compliance with certification requirements: Wind turbine certification is currently based on the assumption that the pitch angles deviate from each other by a maximum of + / -0.3°. 3. Load reduction / potential cost savings: Through the automated detection and elimination of aerodynamic imbalance, the assumed pitch error, i.e., the pitch misalignment of + / -0.3°, can be reduced in simulations, especially bladed load simulations, thereby achieving a load reduction. Lower aerodynamic imbalance reduces the load on the azimuth drive, potentially allowing for the use of fewer or less powerful motors.

[0136] The following alternative approaches can be avoided or improved: Optical methods: The pitch angle is checked optically → Disadvantage: no system operation; complex measuring technology required; manual procedure. Tower torsion monitoring: If the azimuth moment is unknown, it can also be determined from the tower torsion. However, this requires complex measuring technology and is therefore quite expensive.

[0137] One objective of the invention can be seen as the automatic detection of aerodynamic imbalance during the operation of the wind turbine without additional hardware.

[0138] This then allows the pitch misalignment to be corrected by adding an offset to the respective pitch angle.

[0139] The following advantages result: Cost savings: no additional costs for an azimuth adjustment system that uses position control. Risk minimization: aerodynamic imbalance is monitored and corrected in all wind turbines, not just those with noticeable imbalances. Simplified and shortened commissioning: determining aerodynamic imbalance by an external service provider is no longer necessary. Simple software maintenance and parameterization: many parameters describe the physics of the wind turbine and are known or only need to be specified in terms of their magnitude. No dependencies on the wind turbine's control system and no cross-sensitivity to wind shear, turbulence, or mass imbalance. After approximately 10 minutes of operation, the pitch misalignment of all blades is known.

[0140] The following should be added regarding the Kalman filter used.

[0141] The Kalman filter works by adjusting the system states of its model based on a comparison of the model's output and a corresponding measured value of the system. The adjustment is performed to minimize the difference between the model's output and the corresponding measured value. It is assumed that both the measured value and the observed system states may exhibit measurement noise.

[0142] The system description, and therefore also the model, uses a discrete-time description. This considers quantities at the current discrete time k and at the previous time k-1, i.e., at step k and the previous step k-1.

[0143] This describes the conditions xk to step k depending on the states of the previous step k-1 and depending on input variables UKThe current step k is calculated. This is represented by the following generalized relationship: − x k = f x k − 1 u k

[0144] The leaf misalignment and azimuth moment offset can be assumed to be constant over time. Then xk = xk-1 applies to these.

[0145] The states are represented by the vector xk represented in which all considered states are summarized. The input variables are represented by the vector UK As shown. In simple systems, there might only be one state and / or one input variable, but that is not the case here.

[0146] Furthermore, it is assumed that the states exhibit noise with covariance Q.

[0147] Measured values zk The output variables of the model are directly dependent on system states. xk and input variables UKcalculated, which is represented by the following generalized relationship: z k = h x k u k

[0148] The measured values ​​are represented by a vector. zk as shown. In the present case, or according to a proposed aspect, however, there is only one measured value or only one output quantity.

[0149] It is assumed that the measured quantities have a noise R.

[0150] In the previous equations, f ( x k- 1, UK ) and h ( xk,uk ) are referred to as transfer functions, namely as state transfer functions f ( x k- 1 ,uk ) and as output transfer function h ( xk,uk ). Their generalized representation suggests that the transfer functions may be nonlinear.

[0151] The functionality of the Kalman filter can be divided into two parts.

[0152] The first part can be considered a prediction. This involves linearizing the transfer functions: F k = ∂ f x u ∂ x | x k − 1 , u k und H k = ∂ h x u ∂ x | x k , u k

[0153] Linearization is therefore achieved by partial differentiation of the functions with respect to the respective states, specifically at the points x k-1 or xk and uk . This results in a linearized state transfer matrix. F k and an output transfer matrix H k .

[0154] The conditions xk are predicted by the relationship described above, or calculated for it, i.e., with the generalized relationship: x k = f x k − 1 u k

[0155] Furthermore, the covariance of the observed states is expressed as a covariance matrix. P k the estimated values ​​are predicted. The covariance matrix P k The estimated values ​​can also be synonymously referred to as the covariance matrix. P kThe observed states are referred to as the covariance matrix. For prediction purposes, it can be calculated from a previous covariance matrix. P k-1 , the linearized state transfer matrix F k , and the covariance matrix Q k of the process noise, which thus characterizes process noise that is superimposed on the current system states by input variables. The covariance matrix P k The estimated values ​​are calculated according to the equation: P k = F k ∗ P k − 1 ∗ F k T + Q k

[0156] The second part can be seen as the correction, namely of the conditions.

[0157] First, a residual covariance matrix is ​​calculated. S k calculated, namely from the current output transfer matrix HK, the covariance matrix PK the estimated values ​​and a measurement covariance matrix R k : S k = H k P k H k T + R k

[0158] Using the covariance matrix P k the estimated values ​​of the output transfer matrix HKand residual covariance matrix S k can the feedback gain matrix K k calculated, which can also be described as the optimal gain: K k = P k H k T S k − 1

[0159] Based on this, the states in the observer are determined xk adapted, which is also referred to as an update of these states xk This can be described as follows: A difference is calculated from the measured output quantity. zk and their calculation from the estimated conditions xk of the observer and the input variables UK of the observer, according to the output transfer function h ( xk,uk ) calculated using the following equation: x k = x k + K k z k − h x k u k

[0160] The difference between the measured and observed output quantity is thus multiplied by the optimal gain and added to the current states.

[0161] Finally, in preparation for the next step, the current covariance matrix is ​​updated. P k the estimated values, i.e. the observed system states, according to the following equation: P k = I − K k H k P k

[0162] The updated covariance matrix P k The estimated values ​​are then needed in the next step to predict, i.e., calculate, the new covariance matrix in the first part, namely the prediction, as shown above. The updated covariance matrix is ​​used for this purpose. P k The estimates from this step are used in the next step to replace the previous covariance matrix. P k-1 .

[0163] All these calculations are repeated step by step, so that the current step is then the previous step.

[0164] The Kalman filter is therefore a recursive method in which the feedback gain matrix K k, which can also be called the correction matrix or correction vector, is recalculated or adjusted in each step.

[0165] In summary, for each step k: xk :observed conditions UK :Input value zk:Output size F k State transfer matrix H k Output transfer matrix K k :Feedback amplification matrix P k Covariance matrix of the estimates Q k :Kovairance matrix of process noise R k : Covariance matrix of the output variable S k Residual covariance matrix l :Identity matrix (for each step)

[0166] For the in Figure 4 The following initial values ​​can be used in the described example case.

[0167] Pk can be a 4x4 matrix where only the main diagonal can be occupied for Po. The initial values, i.e., values ​​on the main diagonal, can be chosen according to a rough expectation, namely how well the initial values ​​x0 = 0 are. For example, the values ​​on the main diagonal can be chosen as P0(1,1) = (106)2; P0(2,2) = (106)2; P0(3,3) = 0.12; P0(4,4) = 0.12.

[0168] Precise initial values ​​are not important; the order of magnitude is more crucial than the exact value.

[0169] For simplification, Qk and Rk can be assumed to be constant. Qk characterizes how quickly the states can change independently of the model equations. Here, the main diagonal can be occupied by: Qo(1,1) = 0; Qo(2,2) = Q0(3,3) = (10-7)2; Q0(4,4) = (102)2

[0170] Rk describes the measurement noise and, for example, (104)2 can be assumed as the initial value.

[0171] According to one aspect, also with reference to Figure 4 : x k = M y M y , o γ A γ B T u k = ϕ v W n R α n DRV T z K = M m

[0172] The transfer functions result from the system description shown above, also from the relationships according to Figure 4 .

[0173] It should be noted that when using a Luenberger observer, no noise is taken into account, neither for states, input variables, nor output variables. Therefore, it is possible to compensate for observer error via a fixed correction vector. The recursive modification of the feedback gain matrix K k The necessary recursive calculations required for the Kalman filter are unnecessary for the Luenberger observer. The fixed correction vector can be calculated in advance.

[0174] System states can also be referred to simply and synonymously as states. They also include the states of the observer.

Claims

1. Method for detecting at least one blade misalignment of a rotor blade of a rotor of a wind turbine with several rotor blades adjustable in their blade angle, wherein - the blade misalignment describes a blade angle deviation of a detected blade angle of the rotor blade from a reference blade angle, - the wind turbine has a nacelle with the rotor and an azimuth adjustment device to adjust the nacelle in an azimuth orientation having an azimuth angle, and - for adjusting the azimuth orientation, the azimuth angle is followed by means of the azimuth adjustment device to a predefinable azimuth setpoint angle, and - the blade misalignment is detected as a function of an azimuth movement of the nacelle.

2. Method according to claim 1, characterized by the fact that- the blade misalignment as a function of an azimuth movement of the gondola is detected by - detecting an adjustment activity of the azimuth adjustment device, in particular by - detecting an adjustment activity of at least one azimuth motor of the azimuth adjustment device.

3. Method according to claim 1 or 2, characterized by the fact that - the azimuth adjustment device is designed in such a way that it counteracts a change in the azimuth orientation of the gondola due to wind influence by means of position control of the azimuth angle, in particular that - the azimuth adjustment device does not have a holding brake and / or - the tracking of the azimuth angle is continuous.

4. Method according to any of the foregoing claims, characterized by the fact that A condition monitor is used to detect leaf malposition.

5. Method according to any of the foregoing claims, characterized by the fact that- a state observer for detecting the blade misalignment has at least - a nacelle azimuth moment and - a blade misalignment as state variables, in particular that - the state observer has - the nacelle azimuth moment, - an azimuth moment offset and - a blade misalignment of each of two rotor blades as state variables, wherein the nacelle azimuth moment points in a nacelle rotation direction and is composed of - each an azimuth torque component directed in the nacelle rotation direction of an aerodynamic blade torque (M aero,A , M aero,B , M aero.C ) of each rotor blade, and - the azimuth moment offset, so that - a sum of the azimuth torque components of all rotor blades and the azimuth moment offset forms the nacelle azimuth moment.

6. Method according to any of the foregoing claims, characterized by the fact that A Kalman filter is used to detect leaf malposition, or as the state observer.

7. Method according to any of the foregoing claims, characterized by the fact that - the adjustment activity of the azimuth adjustment device is carried out by evaluating an azimuth drive torque, which denotes a drive torque with which the gondola is adjusted in its azimuth orientation, wherein in particular - the azimuth drive torque is determined as a function of one or more motor currents or one or more motor torques of the at least one azimuth motor, in particular that - the azimuth drive torque is determined as a product of the motor current, a sum of the several motor currents, the motor torque or a sum of the several motor torques and a proportionality factor.

8. Method according to any of the foregoing claims, characterized by the fact that- the azimuth drive torque is used as an output variable of the state observer, such that - the state observer outputs an estimated azimuth drive torque, or a representative, in particular proportional, azimuth drive quantity, and - a difference between the detected and estimated azimuth drive torque, or a difference between the respective representative, in particular proportional, detected and estimated azimuth drive quantity, is fed back to the state observer as an observation error for adjusting observer states.

9. Method according to any of the foregoing claims, characterized by the fact that The system used to detect blade misalignment is based on a model description for the state observer, in which an aerodynamic blade torque M is defined for each rotor blade. aero is described by the equation: M aero = 1 3 ∗ 1 2 ∗ π ∗ ρ air ∗ c s λ , α + γ ∗ l bl 3 ∗ v Wind + ω Yaw ∗ l bl ∗ sin ϕ 2 with: ρ air = Luftdichte λ = ω Rot ∗ l bl v Wind = Schnelllaufzahl ω Rot = Rotorgeschwindigkeit bzw . Rotordreh − zahl in rad / s l bl = Rotorblattlänge α = Blattwinkel γ = Blattfehlstellung des Rotorblattes c s λ α = Aerodynamische Eigenschaften des Rotorblattes oh Yaw * l bl * sin(ϕ) = Apparent wind speed due to an azimuthal movement of the rotor blade; with ϕ : Rotor angle where each blade torque M aeroa torque acting on the rotor through the rotor blade, namely in a direction of rotation perpendicular to the axis of rotation of the rotor, and wherein it is provided in particular that - the sum of the blade misalignments of all rotor blades is zero and the blade misalignment of one of the rotor blades is calculated from the blade misalignments of the other rotor blades, wherein the model description is also based on calculations in which - each of the blade torques is transformed into an azimuth torque component directed in the nacelle rotation direction by means of an angle transformation as a function of a rotor angle of the rotor referred to the respective blade, - a nacelle azimuth torque is calculated as the sum of the azimuth torque components and a nacelle azimuth torque.the azimuth moment offset is calculated, - a gondola acceleration moment is calculated as a function of a change in an azimuth rotational speed and a moment of inertia of the gondola, - an azimuth driving torque is determined as a function of a difference between the gondola azimuth moment and the gondola acceleration torque, wherein - the azimuth driving torque, or a representative, in particular proportional, quantity thereof, forms an output quantity of the state observer model, and in particular - two of the blade misalignments, the azimuth moment offset and the azimuth moment, form system states of the model.

10. Method according to any of the foregoing claims, characterized by the fact that The state observer is based on a model description that can be described by the following system of equations, where the rotor has three rotor blades A, B and C: M aero , A = f v w n R α γ A ; M aero , B = f v w n R α γ B ; M aero , C = f v w n R α γ C ; M yaw , A = f M aero , A , ϕ ; M yaw , B = f M aero , B , ϕ ; M yaw , C = f M aero , C , ϕ γ c = − γ A − γ B M yaw , Nac = M yaw , A + M yaw , B + M yaw , C + M yaw , off M acc = i g . i dn DRV dt . J Nac M DRV = 1 i g . M yaw , Nac − M acc with M aero,A , M aero,B andM aero,C : aerodynamic blade torque M aero of rotor blade A, B or C; v w : Wind speed; n R : Rotor speed; a: Blade angle; ϕ : Rotor angle M yaw,A , M yaw,B and M yaw,C : Azimuth torque component of rotor blade A, B or C; c A , c B and c C : Blade misalignment of rotor blade A, B or C; M yαw,off : Azimuth moment offset; M yaw,Nac : Gondola azimuth moment; M acc : Gondola acceleration moment; M acc = i ge ⋅ i dn DRV dt ⋅ J Nac ; i ge : Gear ratio between azimuth motor and gondola; n DRV : Speed ​​of the azimuth motor; J Nac : Moment of inertia of gondola M DRV : Motor torque azimuth motor.

11. Method for correcting at least one blade misalignment of a rotor blade of a rotor of a wind turbine with several rotor blades adjustable in their blade angle, wherein - the blade misalignment describes a blade angle deviation of a detected blade angle of the rotor blade from a reference blade angle, - the wind turbine has a nacelle with the rotor and an azimuth adjustment device for adjusting the nacelle in an azimuth orientation having an azimuth angle, and - for adjusting the azimuth orientation, the azimuth angle is followed by means of the azimuth adjustment device to a predefinable target azimuth angle, - the blade misalignment is detected as a function of an azimuth movement of the nacelle, in particular with a method according to one of the preceding claims,and - for each rotor blade - a correction angle is determined depending on the respective detected blade misalignment, and - the blade angle is corrected by the correction angle, and / or - a detected blade misalignment is compared with a predefinable deviation limit, and an error or warning message is generated if the detected blade misalignment is greater than the deviation limit.

12. Method according to claim 11, characterized by the fact that - the correction angle corresponds to the detected leaf misalignment, or - the correction angle of the detected leaf misalignment is adjusted, in particular with a delay function which in particular has a time constant of at least one hour.

13. Wind turbine with a rotor having several rotor blades adjustable in their blade angle, wherein - the wind turbine has a nacelle with the rotor and an azimuth adjustment device to adjust the nacelle in an azimuth orientation having an azimuth angle, and - to adjust the azimuth orientation, the azimuth angle is followed by means of the azimuth adjustment device to a predefinable azimuth setpoint angle, wherein - the wind turbine has a control device which is prepared to carry out a method according to one of the preceding claims.