METHOD AND DEVICE FOR DETERMINING A BENDING ANGLE OF A ROTOR BLADE OF A WIND TURBINE SYSTEM

Abstract

A method for determining a bending angle of a rotor blade of a wind turbine system includes step of reading in an acceleration signal which represents an acceleration of the rotor blade acting essentially perpendicularly with respect to a rotor plane. In addition, the method includes a step of determining the bending angle of the rotor blade of the wind turbine using the acceleration signal.

Description

The present invention relates to a method and a device for determining a bending angle of a rotor blade of a wind turbine system in accordance with the independent patent claims.

Wind energy systems are controlled via the adjustment of the rotor blades about their longitudinal axis, and by the generator torque. Controlled variable for the pitch control is the rotor rotational speed, and the manipulated variables are the pitch angles of the rotor blades. The collective pitch control CPC is used with conventional systems. Here, the three rotor blades are all adjusted with the same pitch angle. In the case of wind energy systems with a horizontal axis and at least two rotor blades, synchronous adjustment of the blade angles is used to control the rotational speed above the rated wind speed so that by changing the incidence angle the aerodynamic lift, and thus the drive torque, are reduced in such a way that the system can be operated in the region of the rated rotational speed. Given wind speeds above the switchoff speed, this blade adjustment mechanism is additionally used as a brake by setting the blades in a fashion nose into the wind so that the rotor can no longer supply any appreciable drive torques. Because of asymmetric aerodynamic loads, pitch and yaw torques on the nacelle are produced in the case of this collective blade adjustment. The asymmetric loads result, for example, from wind shears in a vertical direction (boundary layers), yaw angle errors, gusts and instances of turbulence, damming of the flow at the tower, etc. A known approach to reducing these asymmetric aerodynamic loads consists in adjusting the incidence angle of the blades individually (Individual Pitch Control=IPC). This control approach requires determination of the bending torques (in particular, flapping bending torques) that prevail at the rotor blade root. The bending torques then serve as controlled variable for the individual blade adjustment. Strain gauge sensors that are applied at the rotor blade root can be used to determine the bending torques. The problems in the case of the strain gauge sensors consist in the application and risk of breakage, and in the short service life.

Other methods such as are disclosed, for example, in WO 2008/041066 or DE 197 39 164 B4 determine the pitching and yawing torques by measuring the nacelle acceleration via gyrometers or by means of sensors that use distance measurements to measure the deformations, occurring during the loads, of system parts, and thereby determine the loads. The blade bending torques are very well suited as controlled variable from the point of view of the IPC. However, it has not so far been possible to find any measurement technique suitable for continuous use. Fiber Bragg sensors laminated into the blades to measure torques cannot be exchanged in the case of a defect, while strain gauge sensors bonded on have a far too short service life. Both methods additionally have the problem that the measurement is performed only locally on the blade. Local inhomogeneities in the laminate therefore lead to measuring errors, and so an inference concerning the global stress state in the blade root, and thus the torque acting there, is always affected by errors.

It is therefore the object of the present invention to provide a method and a device that enable an improved determination of the load of a rotor blade of a wind turbine system.

This object is achieved by the subject matter of the independent patent claims. Advantageous refinements follow from the subject matters of the subclaims, and from the following description.

The present invention provides a method for determining a bending angle of a rotor blade of a wind turbine system, the method having the following steps:

• • reading in at least one acceleration signal that represents an acceleration acting on the rotor blade, and
  • determining the bending angle of the rotor blade of the wind turbine system by using the acceleration signal.

Furthermore, the present invention produces a device for determining a bending angle of a rotor blade of a wind turbine system, the device having the following features:

• • an interface for reading in an acceleration signal that represents an acceleration acting on the rotor blade, and
  • a unit for determining the bending angle of the rotor blade of the wind turbine system by using the acceleration signal.

Also advantageous is a computer program product having program code that is stored on a machine readable carrier such as a semiconductor memory, a hard disk memory or an optical memory, and that is used to carry out the method according to one of the abovedescribed embodiments when the program is executed on a control unit or a device.

The invention is based on the finding that the bending of the rotor blade of the wind turbine system is related in a predetermined fashion to a bending torque of this rotor blade at the blade root. In order to determine the bending, use is made, in particular, of an acceleration or an acceleration signal that is measured in a fashion substantially perpendicular to the rotor plane. The acceleration in a longitudinal direction of the blade can also be used as additional acceleration signal. Denoted in this case as rotor plane is a virtual or actual plane in which the rotor blades rotate about the rotor axis of the wind turbine system. This means that the acceleration used in the present approach represents an acceleration in the direction of the rotor axis. Given knowledge of this predetermined relationship, it is possible in this case to use the acceleration of the rotor blade, or at least of a part of the rotor blade, to infer the bending torque present at the blade root of this rotor blade so that a conventional control unit can further be used to determine the blade pitch angle by using a modified control parameter. It is not mandatory in this case to infer the bending torque present at the blade root; rather, it is also possible to directly calculate the incidence angle to be set on the basis of the measured acceleration or acceleration read in. In this case, the acceleration determined is thus used to determine the blade deflection (that is to say a value beta), the IPC pitch angle or the incidence angle of the rotor blades being determined therefrom. The preset control can therefore use the bending torque (or the bending angle), and then determines the so-called pitch angle of the rotor blade. It is also possible in this case to use the term “incidence angle” for the term “pitch angle”. Thus, the incidence angle or the individual incidence angle for the rotor blade can be determined from the bending angle.

The present invention offers the advantage that conventional control units can continue to be used such that there is no need for a cost intensive redevelopment of a control unit for controlling the incidence angles of the rotor blades of the wind turbine system. At the same time, the sensor variables used can be provided by using sensors that are substantially more robust against aging phenomena and measuring errors. Because wind turbine systems are designed for a long running time and, in particular, it is very cost intensive to exchange rotor blades, the abovementioned advantage is gaining even further in importance. A simple and cost effective retrofitting is also possible with the approach presented here.

In accordance with a favorable embodiment of the present invention, a profile of the acceleration can be acquired in the read-in step, in the determination step a spectrum being determined from the acceleration profile, and the bending angle being determined by using the determined spectrum. Such an embodiment of the present invention offers the advantage that relatively small measuring errors can be compensated by using a spectrum that has been determined over a specific period and was subsequently transformed into the frequency domain. It is possible in this case to utilize the fact that the revolution of the rotor blade or the rotor blades periodically gives rise to physical influences at specific positions in the pitch circle of the rotor blade, specifically exactly when the rotor blade again reaches the specific position in the following revolution.

It is particularly advantageous when in the determination step the determined spectrum is compared with a provided spectrum, the bending angle being determined by using a result of comparison between the determined spectrum and the provided spectrum. Such an embodiment of the present invention offers the advantage of the possibility of a good and reliable determination of the provided spectrum. By way of example, it is possible in this case to determine a mean value from a multiplicity of recorded spectra, it then also being possible for such a provided spectrum to map certain variations of the environmental conditions.

In accordance with a particular embodiment of the present invention, in the read-in step a movement of the rotor blade in a fashion substantially perpendicular to the rotor blade can be actively effected. Such an embodiment of the present invention offers the advantage that spectra already to be expected for specific, frequently occurring scenarios can be measured or calculated and stored in a memory. For example, in this case a specific bending angle can be assigned to each spectrum stored in the memory. In practical use, it is possible in this way to determine the bending angle very easily numerically or by circuitry, since in essence a comparison of the determined spectrum with one or more spectra from the memory needs to be performed in order to obtain an already very accurate magnitude for the bending angle from the result of comparison when the determined spectrum corresponds approximately to a specific spectrum to which this bending angle is assigned.

In accordance with a further embodiment of the present invention, in the determination step the acceleration signal can be subjected to lowpass filtering and/or Kalman filtering. Such an embodiment of the present invention offers the advantage that filtering provides a smoothing of the measured value read in, which increases the stability of the control method for the incidence angle. In particular, high-frequency signal interference components are filtered out in this case, leaving the pure useful signal which carries the desired information, to be evaluated, relating to the gravitation and centrifugal force.

Furthermore, in a further embodiment of the invention it is also possible in the determination step for determining the bending angle to make use of an information item relating to bending stiffness or to an approximation of the bending stiffness, an information item relating to a distance of an acceleration sensor providing the acceleration signal from a rotor axis, an inclination angle of the rotor axis from the horizontal, and/or an acceleration of a tower head of the wind turbine system. Such an embodiment of the present invention offers the advantage that a very precise estimate of the bending torque occurring at the blade root is possible hereby so that a slight change in the parametrization of control units already in use is necessary. This is relevant, in particular, because the control units currently in use determine the control of the incidence angle for a rotor blade on the basis of an occurring bending torque, and so the control variable can be very easily exchanged.

In order to obtain a determination of the bending angle of the rotor blade that is as precise as possible, in the determination step it is possible to determine a time profile of the acceleration at a position of the rotor blade from the acceleration signal, and to determine the bending angle or the blade deflection of the rotor blade by using the determined profile. The time profile can in this case extend over a rotor blade revolution about the rotor axis. Such an embodiment of the present invention determines the bending of the rotor blade by virtue of the fact that the position relative to the gravitational acceleration, which acts periodically during the measurement of the acceleration of the rotor blade in a fashion now amplifying and now reducing the measured sensor signal, is determined.

In accordance with another embodiment of the present invention, in the read-in step it is possible to read in a further acceleration signal that is measured in the direction of the longitudinal axis of the rotor blade. In this case, in the determination step of the bending angle of the rotor blade of the wind turbine system is determined by using the further acceleration signal.

The invention is explained in more detail in exemplary fashion below with the aid of the attached drawings, in which:

FIG. 1 shows an illustration of a uniform definition of designations of the possible movements on a wind energy system;

FIG. 2 shows a block diagram of a control unit for the individual incidence angles of a rotor blade of a wind energy system, in the case of which an exemplary embodiment of the present invention can be used;

FIG. 3 shows a diagram of the principle of the relevant variables for a positioning of the sensor on a rotor blade;

FIG. 4 shows a diagram illustrating the relationship between a blade deflection and a root bending torque, plotted against time;

FIG. 5 shows an illustration of a sensor coordinate system on a rotor blade;

FIG. 6 shows an illustration showing the measurement principle and the processing of the sensor signal obtained;

FIG. 7 shows a diagram in which a sensor signal subjected to lowpass filtering and a reference signal are illustrated;

FIG. 8 shows a diagram illustrating the relationship between a blade deflection and a useful signal of the acceleration measurement of a sensor that is positioned at a distance of r=10 m from the rotor hub; and

FIG. 9 shows a flowchart in accordance with an exemplary embodiment of the present invention as method.

Identical or similarly acting elements can be provided in the following figures by identical or similar reference symbols. Furthermore, the figures of the drawings, their description and the claims include numerous features in combination. It is clear here to a person skilled in the art that these features can also be considered individually, or can be brought together to form further combinations that are not explicitly described here. Furthermore, in the following description the invention may be explained by using different measurements and dimensions, although the invention is not to be understood as confined to these measurements and dimensions. Furthermore, inventive method steps can be executed repeatedly and in a sequence other than that described. If an exemplary embodiment includes an “and/or” conjunction between a first feature/step and a second feature/step, this can be read to the effect that, in accordance with one embodiment, the exemplary embodiment has both the first feature/the first step and the second feature/the second step and, in accordance with a further embodiment, the exemplary embodiment has either only the first feature/the first step or only the second feature/the second step.

A particular aim of the invention is to provide a possibility of using a control method to minimize the yaw and pitch torques on the nacelle, which result from asymmetric aerodynamic loads. Manipulated variables are advantageously the individual incidence angles of the blades of the wind turbine system. An important aspect in this case is that, in accordance with the approach presented here, the controlled variables are determined via acceleration sensors on the rotor blades. To this end, there is installed in at least one rotor blade at least one acceleration sensor that can measure accelerations in the flapping direction (that is to say perpendicular to the rotor plane). This offers the advantage that it is possible hereby to use point sensors that can easily be applied in the blades, are easy to exchange, and do not acquire static errors such as stresses owing to temperature differences and the inhomogeneous blade material. In addition, in some circumstances the sensors are already present if condition monitoring of the blades is installed.

Recourse may be made to the illustration in accordance with FIG. 1 in the interest of standard definition of the following variables used regarding the possible movements of a wind turbine system. Here, a wind turbine system is understood as a system having a tower on which a nacelle is fastened. This nacelle contains a generator that is coupled to a rotor, the rotor having two rotor blades in the example illustrated in FIG. 1. In this case, the tower can execute a tower longitudinal bending 100 and a tower transverse bending 110 given an incident flow of wind and a transmission of forces of the rotor onto the nacelle and the tower. Further, the tower can execute a tower torsion 120 about its vertical axis. A movement of the tower about its vertical axis is also denoted as yawing 130 of the wind turbine system. Furthermore, it is also possible for forces to act on the tower or the wind turbine system, which leads to a rolling 140, that is to say a rolling movement about the rotor axis of the wind turbine system. If the effect of wind on the wind turbine system is to induce a movement that acts both perpendicular to the vertical axis of the tower and also to the rotor axis, the wind turbine system is said to pitch 150. The rotor blades can, on the one hand, execute a pivoting movement 160 or a flapping movement 170, or twist internally, this equally being denoted as torsion 180, referred now to the rotor blades. The pivoting movement 160 corresponds in this case to a desired movement of the rotor blades about the rotor axis, the flapping movement 170 denoting a movement, in particular of the tips of the rotor blades, out of the rotor plane, that is to say in the direction of extent of the rotor axis. Such a definition of movements of a wind turbine system follows the definition from the book of E. Hau, entitled “Wind turbine systems”, in which corresponding controlled variables are named for the yaw and pitch torque of the nacelle. The flapping movement leads to bending torques on the blade root, and is the cause of yaw and pitch torques of the nacelle.

An important aspect of the present invention can be seen in that it is possible to make use of acceleration signals from sensors on the blade, and to process these signals within a control method in order to reduce the yaw and pitch torques on the nacelle via the individual adjustment of the blade incidence angles.

A system in the case of which the present invention can be used in accordance with one exemplary embodiment is illustrated in a simplified fashion as a block diagram in FIG. 2. The system 200 for controlling the wind turbine system 210 comprises in this case a unit 220 for operational control, and a unit 230 for controlling the individual incidence angle 235 (β_IPC1,2,3) for each of the rotor blades of the wind turbine system 210. From a sensor of the wind turbine system 210, the unit 220 for operational control (also denoted as CPC; CPC=Collective Pitch Control) receives a signal for outputting, in particular a signal relating to the rotational speed ω of the rotor of the wind turbine system 210. The signal can now be used, on the one hand, by the unit 220 for operational control to determine a generator torque 240 that is to be set, and to make this available for controlling the wind turbine system 210 and, on the other hand, to determine for all rotor blades a common incidence angle 242 (β_CPC) for which the wind turbine system has an optimum output efficiency. From at least one sensor in or on a rotor blade of the wind turbine system 210, the control unit 230 (also denoted as IPC controller) for the individual incidence angle 235 receives a signal relating to an acceleration a₁ of this rotor blade at that position at which the sensor is arranged. In particular, the control unit 230 for the individual incidence angle 235 can receive signals relating to accelerations a_1,2,3 from a plurality of, for example from all rotor blades, and in this case it can provide for each rotor blade for which it receives a sensor signal a corresponding signal β_IPC1,2,3 for setting the individual incidence angle 235 of the relevant rotor blade. In this way, the signal relevant to the common incidence angle can be corrected for each individual rotor blade in order to take account of local wind inhomogeneities. Furthermore, the shear of the wind also leads to asymmetric loads. The signal relating to the common incidence angle 242 can then, for example, be combined additively with the different signals relating to the individual incidence angles 235 for the relevant rotor blades, thus yielding a control signal 250 for the individual relevant rotor blades of the wind turbine system 210. This adjustment of the incidence angle of the individual rotor blades of the wind turbine system 210 in accordance with the desired incidence angles is subsequently set by an actuator 255. Under the influence of varying wind conditions 260, the rotor blades are then deflected in a flapping direction with different degrees of intensity, this deflection or the acceleration occurring in this case being measured in turn by the appropriate sensors, and being fed via the sensor signals 265 to the operational control unit 220 and to the control unit 230 for the individual incidence angle. In this way, the control loop for controlling the individual incidence angles is closed.

A very simple modification of already existing control systems for the individual incidence angles of the rotor blades of a wind turbine system can be implemented by using acceleration signals that represent an acceleration of the individual rotor blades in the flapping direction. Specifically, conventional wind turbine systems mostly use the bending torques at the blade root of the rotor blades to set the individual incidence angles of the relevant rotor blades. However, since a simple relationship between a bending torque at the blade root of a rotor blade and an associated bending of the rotor blades in a flapping direction can mostly be detected, or is known, it is possible by means of a signal of a substantially more robust acceleration sensor to use for the control of the incidence angle of the rotor blade an adequately useful signal that represents the acceleration of the rotor blade or of a part of the rotor blade in a flapping direction by employing this signal to determine the bending angle of the rotor blade in accordance with the invention. Two variants can now be conceived in order to obtain and to process a signal relating to a bending angle of a rotor blade that can be processed well and is as free from interference as possible.

In a first variant, a natural frequency analysis of the determined accelerations or of the acceleration signals derived therefrom can be carried out. To this end, use is made of the natural vibrations of the (rotor) blade. The excitation during the operation of the system is performed by aerodynamically induced vibrations or via an additionally fitted shaker, that is to say a unit which actively sets the rotor blade vibrating. In this case, the acceleration sensors continuously acquire and store signals, and determine the amplitude spectrum of the natural vibrations after a specific measuring time (at most 1 s). This frequency spectrum is, for example, compared with desired spectra that are stored in the control/regulation device and are associated with specific loading states of the blade. Loading at the blade is reduced by adjusting the blade angle, and this is controlled by comparison with the desired spectra. The desired spectra are preliminarily determined by measurements on the blade without and with loadings, or else determined from calculations via natural frequency analysis. The advantage of this variant consists in that it is also possible to make use of the already available measurement technique for condition monitoring, which has already integrated the sensors, the acquisition of measured values, and the preparation and evaluation of the acceleration signals. Desired spectra already stored are also associated therewith. Desired spectra for loading cases that are stored in the control or regulation device should be added thereto for such an application scenario. Said spectra can be determined by measurements on blade test stands. It is probably easier to carry out reference measurements before mounting on the blade, and to carry out similar measurements on the rotor blade after mounting in the case of wind speeds below the startup speed. The deviations relating to these spectra in the case of loading are calculated via simulation starting from these spectra and the blade data, and stored as desired spectra.

A second variant for the use of the approach presented here is to be seen in the use of data from a direct acceleration measurement and evaluation thereof. In this case, the bending angle of the rotor blade is determined from the measured accelerations. The control aim is then to set the same bending angles at all rotor blades. The blade angles are once again manipulated variables. Because of aerodynamic effects such as turbulence and vortex shedding, vibrations of the blade are always excited, but they are of higher frequency than the vibrations to be removed by control in the region of the first natural frequency of blade and tower. Consequently, for control purposes the measured acceleration should be filtered by means of a lowpass filter. The lower half of the rotor blade is advantageous for the position of the first (acceleration) sensor, since the blade tip can be excited to vibrate strongly owing to the tapering and the transverse flows prevailing there, which also drive the tip vortex.

The following aspects may be adduced as advantages of the two abovedescribed variants. Firstly, it is possible to use known and possibly already present measuring devices and, if appropriate, data determined by the condition monitoring of the blades. Furthermore, there is no need for any application of strain gauges or the like, in the case of which it is not known in the prior art where and how they are to be exactly fitted. In addition, the temperature compensation for these sensors has not yet been satisfactorily resolved in technical terms. In addition, an acceleration sensor can easily be replaced in the case of a defect. This is impossible with strain sensors that have been laminated in. The signals supplied by strain sensors are possibly not indicative, since they acquire only the local strain. Again, when the abovedescribed approach is used there is no occurrence of errors owing to static loadings such as temperature stresses, excessive local stresses owing to the inhomogeneous material, ice coating (with simultaneous use of condition monitoring) etc., something which greatly increases the reliability of the control by making use of the variable of the bending angle which is calculated from the blade acceleration.

In other words, this means that the approach presented here constitutes a use as additional control function with the pitch drives of the applicant. On the basis of current market trends, future drives should be able to adjust the blades individually.

A further important aspect of the present invention consists in enabling, on the basis of an acceleration sensor (DCU), an improved IPC-suited measurement method in which the sensor has a longer service life, a simple exchangeability of the sensor system is ensured, and a variable equivalent to the global stress state in the blade root is acquired as far as possible.

In the case of the approach presented below, a substantial aspect resides in the use of a signal of an acceleration sensor that measures the acceleration of the rotor blade in the direction of the rotor axis. The acceleration sensor should be able to measure stationary acceleration. The measurement of the acceleration in the blade is known according to the prior art, and is used, inter alia, for condition monitoring. A twofold integration of this measured acceleration would provide the current blade deflection. However, this method has a drift that corrupts the calculated results over a longer time. This measured variable is therefore not suitable for IPC control.

In accordance with an exemplary embodiment, the invention presented here presents a measurement concept that enables a signal evaluation suitable for the IPC control. An online signal evaluation can be performed on the basis of sensor data of a mono-axial acceleration sensor. A possible use resides, for example, in the field of IPC control, or with experimental measurements on wind energy systems. The control of the blade angles of the rotor blades of a wind energy system requires the blade deflection in a flapping direction (that is to say perpendicular to the rotor plane given 0 degree pitch setting). In order to determine this variable, the blade deflection can be measured directly via strain gauge sensors on the rotor blade root. An alternative sensor concept for the measurement of the blade deflection is the use of acceleration sensors whose measurement equation is described by the so-called navigation equation (3), which reads as follows:

$\begin{array}{cc}\overrightarrow{a}=\frac{{}^2\overrightarrow{r}}{t^2}{|}_i+\overrightarrow{g}=\frac{{\overrightarrow{v}}_i}{t}{|}_i+\overrightarrow{g},& \left(3\right)\end{array}$

a corresponding to the measured acceleration, and g to the gravitational acceleration.

Given appropriate lowpass filtering of the local acceleration, the sensor signal can be used to estimate the projection of the gravitational vector, and thus the pitch angle of the sensor coordinate system. The orientation of the sensor can be used to infer the deflection of the rotor blade, and thus the corresponding flapping bending torque. The diagram from FIG. 4 shows a measured relationship for the deflection of the rotor blade and the corresponding flapping bending torque, in which time is represented on the abscissa, and the profile of the blade deflection (dashed line) and of the blade root bending torque (continuous line) is represented on the ordinate. It is to be seen in this case from FIG. 4 that the profiles for the measured blade deflection and the measured blade root bending torque correspond to one another such that in order to control the individual incidence angle of the rotor blade it is also possible to use the blade deflection, and thus also the acceleration that leads to the relevant blade deflection.

Assuming negligible torsion, it suffices to take account of the x-component of the sensor signal in order to determine the blade deflection. The x-component points in the direction of the normal vector on the blade surface, and lies in the bending flapping direction to the extent that no blade torsion is present. A sensor coordinate system 500 in the rotor blade, such as is shown in FIG. 5, is assumed to this end. In this case, the z-component is oriented in the direction of the rotor blade end, the x-component in a normal to the rotor plane, and the y-component in a pivoting direction of the rotor blade. Furthermore, a coordinate system 510 in the hub of the rotor, and a coordinate system 520 in the rotor shaft can be used for the conversion of the sensor acceleration values, as is further described in more detail below.

In order to determine the blade deflection, the first step to this end is to transform the coordinates of the tower into the rotor axis, the coordinates of the rotor axis into the rotor blade, and from the rotor blade into the bent rotor blade. The following transformation matrices can be used to this end:

$M_{\mathrm{tower}\mathrm{into}\mathrm{rotor}\mathrm{axis}}=\left(\begin{array}{ccc}\cos \Lambda & 0& -\sin \Lambda \\ 0& 1& 0\\ \sin \Lambda & 0& \cos \Lambda \end{array}\right)=M_{\mathrm{ST}},M_{\mathrm{rotor}\mathrm{axis}\mathrm{into}\mathrm{rotor}\mathrm{blade}}=\left(\begin{array}{ccc}1& 0& 0\\ 0& \cos \Omega & \sin \Omega \\ 0& \sin \Omega & \cos \Omega \end{array}\right)=M_{\mathrm{BS}},\mathrm{and}$ $M_{\mathrm{rotor}\mathrm{blade}\mathrm{into}\mathrm{deflected}\mathrm{rotor}\mathrm{blade}}=\left(\begin{array}{ccc}\cos \beta & 0& -\sin \sin \Omega \\ 0& 1& 0\\ \sin \beta & 0& \cos \beta \end{array}\right)=M_{B^{\prime }B},$

Λ representing the inclination angle of the rotor axis relative to a horizontal, Ω representing the rotor azimuth angle about the rotor axis, and β representing the torsion angle of the rotor blade at the location of the sensor from the rotor plane. In this case, a projection of the gravitational acceleration

${\underset{\_}{g}}_T={\left(\begin{array}{c}0\\ 0\\ -g\end{array}\right)}_T$

can be described in the following way:

${\underset{\_}{g}}_T=M_{B^{\prime }B}·M_{\mathrm{BS}}·M_{\mathrm{ST}}·{\underset{\_}{g}}_{B^{\prime }}=g·\left(\begin{array}{c}\cos \beta ·\sin \Lambda +\sin \beta ·\cos \Omega ·\cos \Lambda \\ -\sin \Omega ·\cos \Lambda \\ \sin \beta ·\sin \Lambda -\cos \beta ·\cos \Omega ·\cos \Lambda \end{array}\right).$

Furthermore, a measurement equation of the acceleration sensors can be specified as follows:

$\underset{\_}{a}={\left(\frac{}{t}\right)}^t\left[{\left(\frac{}{t}\right)}^f\underset{\_}{r}\right]+\underset{\_}{g},$

from which it follows that:

$a_{{\mathrm{Sensor}}^{\prime }}={\left(\begin{array}{ccc}{\omega }^2·\cos \beta ·\sin \beta ·r_s& +r_s·\stackrel{\_}{\beta }& +g·\left(\cos \beta ·\sin \Lambda +\sin \beta ·\cos \Omega ·\cos \Lambda \right)\\ 0& r_s·\left(2\omega ·\sin \beta ·\beta -\stackrel{.}{\omega }·\cos \beta \right)& -g·\left(\sin \Omega ·\cos \Lambda \right)\\ -{\omega }^2·{\left(\cos \beta \right)}^2·r_s& -r_s·{\stackrel{\_}{\beta }}^2& +g·\left(\sin \beta ·\sin \Lambda -\cos \beta ·\cos \Omega ·\cos \Lambda \right)\end{array}\right)}_{B^{\prime }}$

the first column of the above-specified matrix representing the centripetal acceleration, the second column of the above-specified formula the measured accelerations on the basis of the rotation of the sensor coordinate system, and the third column of the above specified formula the gravitational acceleration.

If the tower head acceleration is not to be neglected, it is necessary to expand the sensor equation by a_{tower head}, where

$a_{\mathrm{tower}\mathrm{head}}=\left(\begin{array}{c}+a_x\left(\cos \beta ·\cos \Lambda -\sin \beta ·\sin \Lambda ·\sin \Omega \right)+a_y·\sin \beta ·\sin \Omega \\ +a_x·\sin \Lambda ·\sin \Omega +a_y·\cos \Omega \\ +a_x\left(\sin \beta ·\cos \Lambda +\cos \beta ·\sin \Lambda ·\cos \Omega \right)-a_y·\cos \beta ·\sin \Omega \end{array}\right).$

Consequently, it therefore holds for the measured total acceleration a_sensor of the sensor that:

a _sensor =a _sensor′ +a _{tower head}.

The components based on the rotation of the sensor coordinate system can be filtered by a lowpass filter and thereby eliminated.

In accordance with the exemplary embodiment of the present invention presented here, it is possible to implement two measurement concepts or measurement methods. A mono-axially measuring acceleration sensor on the rotor blade is used for the first method. What is measured in this case is the acceleration that is, for example, directed normally onto the rotor blade surface. This acceleration is denoted as a_{x, sensor} and can be expressed as follows, neglecting the tower head acceleration:

a _x,Sensor=ω²·cos β·sin β·r_s+r_s·{umlaut over (β)}+g·(cos β·sin Λ+sin β·cos Ω·cos Λ),

the term g·(cos β·sin Λ+sin β·cos Ω·cos Λ) periodically repeated with the angle Ω. Furthermore, when β is small, and thus sin β tends to 0, the trajectory of the gravitational acceleration can be used uniquely to determine β.

Without blade deflection, it therefore holds that β=0. It follows herefrom that

(a_x)_{sensor,filtered}=+g·sin Λ=const.

The result with β≠0 is a rotational frequency component (sin β·cos Ω·cos Λ), it being possible to determine the bending angle from the following relationship:

$2·g·\left(\sin \beta ·\cos \Lambda \right)=A\Rightarrow \beta =\mathrm{arc}\sin \left(\frac{A}{2·g·\cos \Lambda }\right).$

The change in the projection trajectory of the gravitational acceleration is therefore determined by the deflection of the rotor blade (β≠0), and can be used to determine β. The first measurement method of the blade deflection on the basis of a mono-axially measuring acceleration sensor offers advantages with reference to a sensor that can be used cost effectively, and to a simpler evaluation of the sensor signals than in the case of the use of a plurality of sensor signals. However, it must be adduced as a disadvantage of this measurement method for the bending angle that a smaller useful signal is available, because only the signal amplitude can be used to determine β.

Variables as explained in more detail with reference to FIG. 3 can be used for the second method for determining the bending angle. Here, the acceleration of the rotor blade 300 in the direction of the rotor axis 310 is considered, the acceleration sensor being arranged at the distance r therefrom. The following accelerations are measured by the local rotation of the rotor blade at the location of the acceleration sensor by the angle β owing to the deflection of the rotor blade from the rotor plane (320):

a _x,Sensor=ω²·cos β·sin β·r_s+r_s·{umlaut over (β)}+g·(cos β·sin Λ+sin β·cos Ω·cos Λ)+a_x(cos β·cos Λ−sin β·sin Λ·sin Ω)+a_y·sin β·sin Ω

and

a _x,Sensor=−ω²·(cos β)²·r_s−r_s·{dot over (β)}²+g·(sin β·sin Λ−cos β·cos Ω·cos Λ)+a_x(sin β·cos Λ+cos β·sin Λ·cos Ω)−a_y·cos β·sin Ω.

The first term (that is to say the first product) of the two equations is constant in this case with reference to the angle β. The second term (that is to say the second product) is negligible when the acceleration sensor signal is subjected to lowpass filtering. The last term (that is to say the last product) is periodic with the angle Ω.

By way of example, a constant component of ω²·r_s=58 m/s² can be obtained given an angular velocity of ω=1.7 rad/s (which corresponds to a system rotational speed of 15 rpm) and a distance of the sensor r_s=20 m. The rotational frequency component is g=9.81 m/s² in this case. Consequently, the acceleration in the z-direction fluctuates, by way of example, from 68 m/s² to 48 m/s² within a revolution of the rotor blade above the rotor axis. All variables except β are now known in the filtered equation for a_z,sensor (that is to say the second term is filtered out). The equation can therefore be solved numerically for the desired torsion angle β.

${\omega }^2·r_s·{\cos }^2\beta \Rightarrow \beta =\mathrm{arc}\cos \sqrt{\frac{a_z}{{\omega }^2·r_s}}$

The above-specified equation for a_x,sensor specifies how the acceleration measured in the x-direction is composed of the known and unknown variables. If this acceleration is additionally measured, the accuracy of the determination of β can be increased. In particular, the use of a Kalman filter can lead to better results. In this case, a model of the rotor blade is simulated in the Kalman filter and the deflection is determined therefrom. The simulation is updated and/or corrected (for example, by means of a predictor, corrector method) in each time step with the aid of the two measurements (a_x, a_z). The blade deflection can be determined directly from the blade inclination with the aid of a model for the blade bending (that is to say the bending line). The blade root bending torque then also follows from the model for the blade bending. The bending stiffness EI so also requires to be known for this. Since the known IPC controllers use only the differences in the blade root bending torques of the blades for control purposes, there is no need for an absolutely accurate value, and an approximate value suffices for the bending stiffness EI. Such a previously mentioned measurement concept could also be established simply by the presence of an acceleration sensor in the rotor blade which is arranged for measuring the acceleration in a z-direction.

This the application of the above-described equations, it is then possible to use the acceleration signals to infer the bending angle of the rotor blade, which is then further used to control the incidence angle of the rotor blade. In particular, in this case the application of the second method has the advantage that because of a constant centrifugal force there is present over a rotor revolution a constant useful signal from which the g-projection can then be calculated or also used to determine β.

The projection component of the gravitational acceleration, which is to be ascribed purely to the flapping bending of the rotor blade, should be determined in order to be able to determine the blade deflection. To this end, it is possible to filter out the change in projection of the gravitational acceleration, which results from the rigid body movement of the system. Two degrees of freedom determine the rigid body movement: the rotation about the rotor axis and the blade angle adjustment about the pitch axis. In addition, the rotation of the azimuth bearing could also be considered, but this is neglected in this consideration.

The projection component responsible for the blade deflection is therefore yielded from:

{right arrow over (g)} _Bending ={right arrow over (g)} _Mess −{right arrow over (g)} _RBFilter   (4)

g_RBFilter corresponding to the projection vector of the gravitational acceleration, which is calculated on the basis of the rigid body movement. g_Mess is the gravitational acceleration component measured by the sensor. g_Bending is the appropriately filtered signal, which is to be ascribed purely to the elastic deformation of the rotor blade (that is to say corresponds to the bending in a flapping direction). Equation (4) can be used to filter out the projection component of the gravitational vector, which is not to be ascribed to the deflection. The calculation of the projection of the gravitational acceleration, which results from the rigid body movement, may be gathered from the following equation (5).

{right arrow over (g)} _RBFilter =T _{Blade — Hub} ·T _z(β)·T_Hub—Rotor·T_x(α)·{right arrow over (g)}_RotorCOS   (5)

where

• • T_Blade—Hub corresponds to a transformation matrix for a transformation into the blade segment COS,
  • T_z(β) represents a rotation by β (that is to say a pitch angle of the rotor blade) referred to the z-axis of the blade bearing COS,
  • T_Hub—Rotor corresponds to a transformation matrix for a transformation into the blade bearing COS, and
  • T_x(α) corresponds to a rotation by α (that is to say an azimuth angle of the rotor) referred to the x-axis of the rotor COS.

In this case, {right arrow over (g)}_RotorCOS denotes the gravitational vector expressed in the inertial rotor axis coordinate system. The rotor axis is upwardly inclined by approximately 5° by the so-called shaft angle.

The measurement principle in accordance with the first method is illustrated in the two partial figures of FIG. 6. In this case, the left-hand partial figure illustrates a measurement principle and an associated measurement signal, in the case of which the wind turbine system and/or the rotor blades are deflected. By contrast, the right-hand partial figure from FIG. 7 illustrates a measurement principle and an associated measurement signal in the case of application of this measurement principle, it being possible to infer the elastic deformation of the rotor blade from the variation in the amplitude.

The diagrams of FIGS. 7 and 8 represent the sensor signal (dashed line 700) at the blade tip (that is to say at a distance of r=36 m from the blade root) and the sensor signal subjected to lowpass filtering (continuous line 710) plotted against time, whereas FIG. 8 represents the profile of a sensor signal subjected to lowpass filtering. The reference signal corresponds in each case to the projection of the gravitational acceleration into the sensor coordinate system. It is to be seen that the lowpass filtering enables determination of the amplitude of the projection of the gravitational vector. The amplitude of the projection is relevant for the evaluation of the blade deflection.

Furthermore, there is a correlation between the amplitude of the blade deflection and the filtered sensor signal. Here, it is necessary to take account of the change in the g-projection, which may be ascribed to the deflection of the rotor blade. This means that a relatively large amplitude signal is also to be expected given a relatively large distance of the sensor from the blade root. This information can also be obtained from the amplitudes of the variable {right arrow over (g)}_Bending. Consequently, the blade root (flapping) bending torques can be determined directly with the aid of the blade acceleration sensors in the course of an appropriate calibration. It may be shown that a better evaluation is possible on the basis of the larger blade bending given a distance of r=20 meters of the acceleration sensor relative to the rotor hub. There is an analogous result given a distance of r=36 meters of the acceleration sensor relative to the rotor hub, the simulation results not being illustrated here.

However, there is still a problem in that the useful signal, that is to say the change in the inclination of the gravitational vector on the basis of the blade bending, is relatively small in relation to the interference signal the more closely the sensor is applied on the blade root. It is correspondingly more advantageous to measure further out on the blade, for example at a position of r=36 meters from the blade hub, because of the larger deflection by comparison with the measurement points located relatively close to the blade bearing at, for example, r=10 and r=20 meters. Furthermore, the measurement of the rotor rotational speed and of the pitch angle are important for the application of the measurement concept presented here. However, this is currently prior art for wind energy systems, and is used for conventional control methods.

Furthermore, the approach presented here enables an already well matured range of acceleration sensors (for example MM3, DCU) of the applicant to be used for the sensor signal evaluation described here, and said approach is capable of future use within a larger scope in the field of the control of wind energy systems.

In accordance with a further exemplary embodiment, the present invention comprises a method 900 for determining an incidence angle of a rotor blade of a wind turbine system as illustrated in the form of a flowchart in FIG. 9. The method 900 has a step of reading in 910 an acceleration signal that represents an acceleration of the rotor blade acting substantially perpendicular to a rotor plane of the wind turbine system. Furthermore, the method 900 comprises a step of determining 920 the incidence angle of the rotor blade of the wind turbine system by using the acceleration signal.

LIST OF REFERENCES

100 Tower longitudinal bending

110 Tower transverse bending

120 Tower torsion

130 Yawing

140 Rolling

150 Pitching

160 Pivoting movement

170 Flapping movement

180 Torsion

200 System for controlling the wind turbine system

210 Wind turbine system

220 Operational control unit

230 Control unit for the individual incidence angle

235 Individual incidence angles (β_IPC1,2,3)

240 Generator torque

242 Common incidence angle (β_CPC)

250 Control signal

255 Actuator

260 Local wind conditions

265 Sensor signals

300 Rotor blade

310 Rotor axis

320 Rotor plane

500 Coordinate system in the rotor blade

510 Coordinate system in the rotor hub

520 Coordinate system in the rotor shaft

700 Reference signal

710 Filtered sensor signal

900 Method for determining a bending angle of a rotor blade

910 Reading in an acceleration signal

920 Determining the bending angle of the rotor blade of the wind turbine system

Claims

1. A method for determining a bending angle of a rotor blade of a wind turbine system comprising: reading in at least one acceleration signal that represents a first acceleration acting on the rotor blade in a fashion substantially perpendicular to a rotor plane; anddetermining the bending angle of the rotor blade of the wind turbine system by using the at least one acceleration signal.

2. The method as claimed in claim 1, wherein: reading in the at least one acceleration signal includes acquiring an acceleration profile,determining the bending angle of the rotor blade includes determining a spectrum from the acceleration profile, anddetermining the bending angle includes using the determined spectrum.

3. The method as claimed in claim 2, wherein: determining the bending angle of the rotor blade includes comparing the determined spectrum with a provided spectrum, and using a result of the comparison.

4. The method as claimed in claim 2, wherein reading in the at least one acceleration signal includes actively exciting vibration of the rotor blade.

5. The method as claimed in claim 1, wherein determining the bending angle of the rotor blade includes subjecting the at least one acceleration signal to at least one of lowpass filtering and Kalman filtering.

6. The method as claimed in claim 1, wherein: determining the bending angle includes using an information item relating to at least one of: a distance of a first acceleration sensor providing the at least one acceleration signal from a rotor axis,an inclination angle of the rotor axis from horizontal, andan acceleration of a tower head of the wind turbine system, anddetermining the bending angle includes using an information item relating to rotational speed and rotary position of a rotor.

7. The method as claimed in claim 1, wherein determining the bending angle includes considering a time profile of the first acceleration at a position of the rotor blade.

8. The method as claimed in claim 6, further comprising reading in a further acceleration signal that represents a second acceleration, acting substantially in a longitudinal direction of the rotor blade, at a location of the first acceleration sensor, wherein determining a bending angle of the rotor blade of the wind turbine system includes using the further acceleration signal.

9. The method as claimed in claim 1, further comprising one of: determining individual incidence angles of the rotor blades based on the bending angle, anddetermining individual incidence angles of the rotor blades based on loading, determined from the bending angle, of the rotor blade or the rotor blades.

10. A device for determining a bending angle of a rotor blade of a wind turbine system, the device comprising: an interface configured to read in at least one acceleration signal that represents an acceleration acting on the rotor blade, anda unit configured to determine the bending angle of the rotor blade of the wind turbine system by using the acceleration signal.

11. A computer program product having program code for carrying out a method for determining the bending angle of a rotor blade of a wind turbine system, the computer program product comprising: a mechanism configured to read in at least one acceleration signal that represents a first acceleration acting on the rotor blade in a fashion substantially perpendicular to a rotor plane; anda mechanism configured to determine the bending angle of the rotor blade of the wind turbine system by using the at least one acceleration signal, wherein the computer program product is configured to determine the bending angle when the program is executed on one of a control unit and a device.