METHOD FOR PARAMETING A SENSOR ARRANGEMENT OF MULTIPLE LOAD SENSORS OF A ROTOR BLADE OF A WIND ENERGY PLANT

Description

[0001] The invention relates to a method for parameterizing a sensor arrangement consisting of several load sensors on a rotor blade of a wind turbine rotor. The invention also relates to a corresponding wind turbine in which such a method is implemented.

[0002] Wind turbines generate electrical power from wind by rotating their rotors. The rotor, and especially the rotor blades, are subjected to strong stresses due to the wind. These stresses include lift forces, which can vary depending on wind speed, wind direction, rotor orientation, and blade angle.

[0003] For safe operation, it is advisable to determine the loads in order to react promptly if the loads become too high. The wind turbine can then continue operating at reduced speed and power output, or be shut down if necessary. Furthermore, estimates of the expected service life can be derived from the recorded loads.

[0004] Especially with wind turbines featuring individual blade pitch control, where the rotor blades can be individually adjusted in their angle, it is advisable to determine the resulting loads. Such wind turbines allow the rotor blade to be individually adjusted during a rotor revolution, ensuring optimal alignment with the wind direction for maximum power output at any given time. This can result in individual loads being placed on the rotor blades.

[0005] Since the loads can act from different directions, it is common practice to measure the load on the rotor blades in at least two directions.

[0006] Known methods typically use four sensors for this purpose. Two sensors form a sensor pair, and the two sensors of each pair are mounted on opposite sides of the rotor blade. A disadvantage of this measurement method is that precise alignment is crucial when positioning the sensors. This complicates sensor installation and leads to measurement inaccuracies if the positioning is incorrect. Furthermore, using four sensors per rotor blade results in correspondingly high costs.

[0007] Methods for measuring loads on rotor blades using three sensors are also known, as for example in EP 3 317 513 B1.

[0008] The adjustment and calibration of the sensors is carried out individually in accordance with EP 3 317 513 B1. However, this has the disadvantage that each adjustment or calibration must be performed very precisely, requiring details such as the sensor position to be known very accurately or to be recorded precisely. The procedure is therefore prone to errors.

[0009] Document EP 2 431 607 A1 relates to an invention which calibrates a load based on data collected under broad observation conditions and calibrates the loads of a large number of wind turbine blades in a timely manner.

[0010] Document US 2019 / 010923 A1 describes a data processing and control extension system capable of detecting an overload of one or more wind turbine assemblies and providing information to a wind farm control system to reduce the power output of each overloaded turbine.

[0011] Document EP 3 141 746 A1 describes a method for calibrating a load measuring device for measuring a load on a wind turbine rotor blade based on strain data derived from the strain of the wind turbine rotor blades.

[0012] The object of the invention is to address one of the aforementioned problems. In particular, a solution is to be proposed according to which the most accurate and least error-prone possible setting is suggested for load sensors as a prerequisite for the use of the load sensors. At the very least, an alternative to known methods is to be proposed.

[0013] According to the invention, a method according to claim 1 is proposed. This method relates to parameterizing a sensor arrangement consisting of several load sensors of a rotor blade of a rotor of a wind turbine for detecting at least one load parameter.

[0014] The wind turbine therefore has a rotor with at least one rotor blade, in particular with three rotor blades.

[0015] Each rotor blade is equipped with a sensor array consisting of at least three load sensors. These are positioned primarily on the surface of the rotor blade in the area of ​​the blade root, where the rotor blade has a substantially circular circumference. Ideally, with three load sensors, these can be offset from each other by approximately 120° around the circumference to be able to detect loads in different directions.

[0016] Multiple load sensors can also be used. An even distribution of the load sensors, such as approximately 120° for three load sensors, is desirable, but minor deviations are acceptable.

[0017] Each load sensor detects a load-dependent physical quantity of the rotor blade and outputs a representative value as the measured sensor value. Specifically, a load sensor can detect a change in length as a physical quantity of the rotor blade and output, for example, a proportional electrical voltage amplitude as a sensor value. Alternatively, the sensor value can simply be a pre-processed number proportional to the physical quantity.

[0018] The voltage output as a sensor value, to stick with this example, is therefore representative of the recorded change in length. A bending moment of the rotor blade can be calculated or interpreted from this change in length. This is also possible here, but it is suggested to consider the entire sensor arrangement and also to take all the sensor signals from the arrangement into account together in order to calculate the bending moment.

[0019] To calculate at least one load parameter from the sensor parameters, at least one overall calculation formula is used, specifically one for each load parameter. The overall calculation formula establishes a relationship between the measured sensor parameters of all load sensors and the at least one load parameter of the rotor blade. The overall calculation formula includes several calculation parameters. Specifically, it is designed to calculate two load parameters, namely two bending moments, each in different directions.

[0020] The load magnitude is thus calculated according to the overall calculation formula. The individual recorded sensor values ​​are directly converted into the load magnitude based on the overall calculation formula. The overall calculation formula therefore takes an overall relationship into account.

[0021] Depending on the load magnitude to be determined, e.g., whether a bending moment in a direction of impact or a bending moment in a direction of rotation is to be calculated, a different overall calculation method or different calculation parameters can be used. It is also possible to combine several overall calculation methods; however, preferably a separate overall calculation method is used for each load magnitude.

[0022] The overall calculation method is based on the theory that, for each sensor individually, a corresponding load component, such as a bending moment in the sensor's measuring direction, can be calculated from the physical quantity (e.g., the change in length) by selecting appropriate parameters. These calculations for the individual sensors are, to put it simply, combined in the overall calculation method to also take other measuring directions into account.

[0023] In fact, the calculation parameters in the overall calculation procedure can no longer be assigned a relationship between the load magnitude (bending moment) and the recorded physical quantity (change in length at the sensor), at least not quantitatively. Instead, the overall calculation procedure and its calculation parameters are solely concerned with calculating the total load magnitude (bending moment in one direction) by taking into account the physical quantities (changes in length) of all load sensors. The individual calculation parameters of this overall calculation procedure are then no longer physically interpretable, which is why they are referred to as calculation parameters.

[0024] The calculation parameters therefore also depend on the number and arrangement of the load sensors. To illustrate this with an example, if a sensor is added to an existing sensor arrangement, the existing calculation parameters must be reduced, otherwise the added sensor would lead to a higher calculated load value (or the added sensor would have no effect at all).

[0025] To parameterize the sensor arrangement, the calculation parameters of the overall calculation procedure are now determined, taking into account the recorded sensor values ​​of all load sensors.

[0026] The load sensors are therefore not configured individually, but collectively. This parameterizes the overall calculation formula, which focuses solely on the relationship between the numerous input variables (the sensor values) and the output variable (the load value). Parameterization of individual sensors is avoided. Consequently, details of the individual sensors, such as their precise arrangement, no longer play a role, or at least a less significant one than when considering them individually.

[0027] In particular, it is proposed that the parameterization of the sensor arrangement, and at least of one overall calculation rule, be carried out in a single step, specifically in a single calculation step. This ensures that the calculation parameters of at least one overall calculation rule are determined together in one step.

[0028] The parameterization can be performed by recording sensor values ​​from all load sensors under different loads and determining the calculation parameters based on this data set. For each set of sensor values, a reference value for the load to be determined is calculated. The only unknowns are then the calculation parameters. If a sufficient number of data sets are recorded, each under different loads, the calculation parameters can be determined. This can be done by rearranging the overall calculation formula. A matrix notation is preferably used for this purpose.

[0029] The larger the dataset, the less influence statistical measurement deviations have. Parameterization is still possible even with just a few recorded load cases. However, at least as many load cases should be recorded for the determination as there are calculation parameters to be determined.

[0030] According to another aspect, no individual parameterization of individual load sensors is performed to determine the calculation parameters. In particular, no individual calibration or adjustment of individual load sensors is performed.

[0031] This is not about individual load sensors, and in particular not about any individual parameterization or adjustment of each sensor, as this can be done by the manufacturer of the individual sensors before use. The goal here is to find an overall relationship between a load parameter and the sensor parameters of all sensors considered. Even if each load sensor could be uniquely assigned a parameter in the overall calculation formula, these calculation parameters are not determined individually. Instead, the calculation parameters are determined collectively in order to parameterize the calculation formula as a whole and thus establish the overall relationship.

[0032] According to another aspect, the sensor parameters are successively acquired in several acquisition steps under varying operating conditions of the wind turbine. In each acquisition step, one sensor parameter is acquired for each load sensor. Each acquisition step thus acquires multiple sensor parameters, namely exactly as many as the sensor array contains load sensors. The sensor parameters acquired in a single acquisition step constitute a sensor parameter set.

[0033] Furthermore, in each acquisition step, at least one associated load parameter is determined as a reference load parameter for the sensor parameter set, specifically one per overall calculation rule. The reference load parameter is then calculated. The acquisition steps are repeated with a modified operating situation, specifically at least as often as the overall calculation rule has calculation parameters to be determined, and at least once more. The calculation parameters are then determined as a function of all sensor parameters acquired in the acquisition steps and the associated load parameters determined.

[0034] After a data acquisition step, at least one recorded load variable serves as a reference load variable, and several sensor variables are available—one per sensor, thus forming a set of sensor variables with a reference load variable. The at least one corresponding load variable recorded as a reference load variable corresponds to the at least one load variable that should actually be determined from the sensor variables if the parameterization were complete, specifically a blade bending moment. Without this parameterization, and therefore for the parameterization process, the blade bending moment, to continue with this example, must be determined by other means. For instance, the blade bending moment can be calculated from a known weight force that acts on the rotor blade and varies depending on its position. For each load variable that is to be determined by the sensor arrangement, a corresponding reference load variable is defined.

[0035] The overall calculation procedure for this single acquisition step can then be represented as a mathematical equation in which the reference load and the sensor parameters are known. If there were only one calculation parameter to be determined, the equation could be solved for it. However, several calculation parameters need to be determined, and therefore, to continue with this illustration, several equations must be formulated. For this purpose, the acquisition steps are repeated with modified operating conditions.

[0036] The operating conditions are varied in such a way that the resulting blade bending moment, to stick with this example, is always different, because it makes little sense to formulate the same equation twice. Mathematically speaking, the equations that result for each data entry or operating condition should be "linearly independent of each other".

[0037] If as many data acquisition steps are performed as there are calculation parameters in the overall calculation procedure, a system of equations with a corresponding number of equations could be set up and uniquely solved. However, if more data acquisition steps are performed, an overdetermined system of equations can be set up.

[0038] This is precisely the preferred approach, as it allows for the determination of a solution with the lowest statistical deviation instead of a unique one. This approach assumes that statistical measurement deviations may occur. Preferably, a large number of sensor parameter sets, each with a reference load, are recorded. It is particularly recommended to record at least twice, or even at least ten times, as many sensor parameter sets with reference loads as there are calculation parameters to be determined.

[0039] According to another aspect, the overall calculation procedure has a calculation equation, or the overall calculation procedure is the calculation equation itself. The calculation equation describes, using the calculation parameters, a relationship between the load quantity and the sensor parameters.

[0040] To determine the calculation parameters, a system of several equations is established. For each equation, a determined reference load is used. Furthermore, for each equation, sensor values ​​corresponding to the determined reference load are used.

[0041] The system of equations is solved for the calculation parameters. In particular, more calculation equations are set up than the total calculation procedure has calculation parameters, so that the system of equations is overdetermined. The system of equations is then solved in such a way as to obtain a statistically optimal solution.

[0042] The relationship between the load magnitude and the sensor parameters is thus defined by the calculation equation, in which the load magnitude depends on the sensor parameters and calculation parameters. To put it simply, if the determined sensor parameters are inserted into the calculation equation, which also contains the calculation parameters, the resulting load magnitude is obtained.

[0043] However, the calculation parameters must first be determined, for which a system of several calculation equations is set up. The calculation equations in this system differ only in the values ​​of the sensor parameters and the reference load parameter, which are substituted into the equations. Each calculation equation therefore has only the calculation parameters as unknowns.

[0044] To determine the calculation parameters, the system of equations is then solved for these parameters. In particular, the system of equations can be written in matrix notation, so that it is solved for a vector of calculation parameters. This assumes a linear system of equations.

[0045] To determine the calculation parameters, it is specifically proposed that the system of equations be overdetermined. Therefore, more calculation equations are formulated than there are calculation parameters. In particular, the calculation equations should differ based on the reference load quantity to avoid identical equations.

[0046] A statistically optimal solution is particularly one in which the calculation parameters are chosen such that the sum of the magnitudes or squares of the deviations between the reference load size and the load size, which in each case results from the associated sensor sizes with the calculation equation and the chosen calculation parameters, is minimal.

[0047] Such a statistically optimal solution, in which the sum of the squares of the deviations is minimal, can be achieved using the pseudoinverses by the methods described below.

[0048] According to another aspect, the system of equations from several calculation equations of the overall calculation procedure is set up in matrix form, with an output vector for the determined reference load quantities, a measurement matrix for the recorded sensor quantities, and a parameter vector for the calculation parameters to be determined. The system of equations is set up such that the output vector is equal to a product of the measurement matrix and the parameter vector.

[0049] The underlying idea here is that the load size L as the sum of the products k i · x i and the offset k which can result in 0, according to the formula: L = ∑ i = 1 n k i ⋅ x i + k 0 .

[0050] Here, ki are the parameters to be determined and xi are the respective sensor values ​​recorded by the same sensor. Further explanations can be found below. For simplicity, k 0 can also be assumed to be zero. In matrix form, the equation for both cases is: L = x → T ⋅ k → ,

[0051] To the parameters ki of the parameter vector k To determine the results, several such equations are formulated and summarized in matrix form as follows: L → = A ⋅ k → .

[0052] We now have a system of equations with several equations. To determine the parameters ki of the parameter vector k To determine the sensor parameters xi, a measurement can be performed for each equation. The respective load parameters L of the load parameter vector L However, they are not known. For simplicity, reference load values ​​can be used instead. L ref The reference load values ​​used represent loads calculated from the system states. Ideally, they correspond to the loads L determined from the sensor data, but in reality, they can exhibit slight deviations. In particular, the reference load values ​​do not capture various real-world effects, such as wind influences, nonlinearities, noise, and other effects. Therefore, they can differ from the loads to be determined. It is proposed that these deviations be neglected.

[0053] With this simplification or approximation, the load vector can be used instead. L the reference load vector the output vector L ref , can be used. Together with the parameter vector k , and the measurement matrix A can be approximated by the following system of equations in matrix form: L → ref = A ⋅ k → .

[0054] The measurement matrix shows A All recorded sensor parameters are displayed. The measurement matrix AThis means it has at least as many columns as there are load sensors and as many rows as there are measured values ​​per load sensor. For the sensor size set, from sensor size x 1 of the first load sensor, sensor size x 2 of the second load sensor, and sensor size x 3 of the third load sensor, i.e. for a system with exactly three load sensors, the measurement matrix is ​​therefore as follows: A = x 1 t = t 1 x 2 t = t 1 x 3 t = t 1 x 1 t = t 2 x 2 t = t 2 x 3 t = t 2 ⋯ ⋯ ⋯ x 1 t = t p x 2 t = t p x 3 t = t p , if at those times t 1 to t p was measured and therefore p Sensor size sets have been included.

[0055] For each sensor size set, the reference load size was also determined. L ref determined such that the output vector also has as many rows as there were measurements. It thus results as L → ref = L ref t = t 1 L ref t = t 2 ⋯ L ref t = t p .

[0056] The parameter vector is used to parameterize the sensor arrangement. k to determine, whereby kthe calculation parameters to be determined and therefore has at least as many lines as there are load sensors, i.e., with three load sensors k → = k 1 k 2 k 3 .

[0057] To achieve this, an approximate solution to the system of equations is found by forming a pseudoinverse of the measurement matrix. Therefore, the following holds: k → = A T ⋅ A − 1 ⋅ A T ⋅ L → ref .

[0058] This approach finds a solution where there is an error between L and L ref minimized.

[0059] By using pseudoinverses, the calculation parameters are determined in a single step, and all recorded sensor values ​​are considered simultaneously. Furthermore, forming pseudoinverses minimizes the sum of squared errors in the recorded load variable. The method thus becomes more robust the more measurements are recorded, i.e., the larger the load. pThe calculation remains computationally simple and can be solved quickly and easily. It is important to note that the order of the inverse matrix does not depend on the number of data acquisition steps, but only on the number of calculation parameters to be determined. The order can therefore be 3 or 4, even if 100 measurements were taken. The inverse matrix can thus be determined with minimal computational effort.

[0060] Alternatively or additionally, the parameter vector has an offset, and the measurement matrix has an additional column corresponding to the offset, which is filled with the value 1. Therefore, when the measurement matrix is ​​multiplied by the parameter vector, the offset is always multiplied by the value 1.

[0061] The parameter vector k results with the further offset k 0 therefore to k → = k 1 k 2 k 3 k 0 , and the measurement matrix also has another column filled with ones. A = x 1 t = t 1 x 2 t = t 1 x 3 t = t 1 1 x 1 t = t 2 x 2 t = t 2 x 3 t = t 2 1 ⋯ ⋯ ⋯ 1 x 1 t = t p x 2 t = t p x 3 t = t p 1 .

[0062] Alternatively or additionally, the measurement matrix contains columns with powers of the sensor quantities, in addition to the sensor quantities themselves. The parameter vector has corresponding calculation parameters.

[0063] This allows for the consideration of non-linear influences of sensor parameters alongside linear ones. The underlying calculation formula is a polynomial function. Depending on the degree of the polynomial, the measurement matrix is ​​extended by additional columns. For each degree of the polynomial and each load sensor, the measurement matrix then has one column, plus an optional column for the offset. The number of columns thus corresponds to the product of the degree of the polynomial and the number of load sensors, plus one optional additional column. The number of calculation parameters in the parameter vector corresponds to the number of columns in the measurement matrix.

[0064] According to another aspect, the calculation parameters include an offset. k 0. Each load sensor is assigned at least one calculation parameter. k i assigned. Furthermore, the overall calculation rule is derived by calculating a product of sensor size for each of the load sensors. x i of the respective load sensor i with at least one calculation parameter belonging to the load sensor k i The load quantity L is formed as the sum of the products. k i · x i and the offset k 0 according to the formula: L = ∑ i = 1 n k i ⋅ x i + k 0 .

[0065] The formula applies to n load sensors. For three load sensors, the formula for the overall calculation procedure is thus: L = k 1 ⋅ x 1 + k 2 ⋅ x 2 + k 3 ⋅ x 3 + k 0 , where the calculation parameter k 1 the first load sensor with sensor size x 1 is assigned. The same applies accordingly to the other calculation parameters.

[0066] The sensor size itself is proportional to the strain ε of the respective load sensor. However, determining this strain as an intermediate value is not necessary.

[0067] In matrix notation, the load quantity could also be described as L = x → T ⋅ k → , where the following applies x → = x 1 x 2 x 3 1 , und k → = k 1 k 2 k 3 k 0 .

[0068] According to another aspect, in order to successively record the sensor sizes under varying operating conditions, the operating conditions are chosen in such a way that different load sizes, in particular different bending moments, occur on the rotor blade.

[0069] The bending moments result from the rotor blade's own weight. They are varied by changing the rotor's rotational position. Alternatively or additionally, the blade angle can also be changed.

[0070] The goal is to generate as many different bending moments as possible on the rotor blade. With each measurement, i.e., with each recording of a set of sensor data, a different load is preferably applied to the rotor blade, in particular a different bending moment. Accordingly, the calculated reference load values ​​also differ from one another, as their calculation takes into account the altered operating conditions. The calculation parameters can be determined more precisely the greater the divergence of the measured values.

[0071] The load can also depend on environmental conditions, especially wind. Preferably, the sensor readings are taken in the weakest possible wind conditions, so that the load is approximately determined only by the weight of the rotor blade itself.

[0072] To vary the load, it is sufficient for the rotor blade to continue rotating, as the force of gravity acts on the blade differently depending on the rotor's position. Ideally, the measurement is taken when the wind is so weak that the rotor can just barely continue turning. Changing the blade angle can also be used to vary the applied load.

[0073] When calculating the reference load, it is important to note that the load sensors have their own coordinate system, which lies within the rotor blade and changes with the blade angle. A corresponding transformation depending on the blade angle must therefore be taken into account. In particular, the weight force can be transformed into the coordinate system of the load sensors to calculate the bending moment as a reference load.

[0074] According to another aspect, the sensor parameters are successively acquired in acquisition steps under varying operating conditions. These acquisition steps are predefined according to a sequence plan, ensuring that the varying operating conditions are configured according to this plan. Parameterization of the sensor array is performed after the sequence plan has been completed.

[0075] The operating sequence can therefore be determined in advance which operating conditions will be set. This generates a routine that ensures as many different operating conditions as possible. Preferably, the reference load values, especially the bending moments, are defined so that they are evenly distributed over a predetermined range, which is preferably chosen to be as large as possible, but still small enough to be reached by weight forces. After all operating conditions have been set according to the operating sequence, a sufficiently large dataset of sensor parameter sets, including reference load values, is determined to ensure proper parameterization of the sensor array.

[0076] According to another aspect, at least one operating setting is provided from the list to record the sensor parameters: the wind turbine is operated in a spin mode, a blade angle of the rotor blade is set, a power output is set, a torque of a generator connected to the rotor is set, and a rotational speed of the rotor is set.

[0077] To measure sensor values, at least one of the operating settings is varied. Each operating setting is maintained for a predetermined test duration. Preferably, each operating setting is maintained for several measurement steps.

[0078] Alternatively or additionally, at least one operating setting is maintained for such a long time that several operating situations are traversed, in each of which a recording step is carried out to record the sensor values.

[0079] An operating situation underlying a detection step is, in particular, a rotor position with a set blade angle. A set generator torque and / or a set power output may also be included. In spin mode, the rotor rotates very slowly, and thus, with fixed operating settings (i.e., spin mode and blade angle), the operating situation changes from one rotor position to the next.

[0080] This allows for the simulation of various operating conditions, particularly through full, half, or quarter rotor rotations, during which several sets of sensor parameters along with reference load values ​​are recorded. This process can then be repeated with a different blade angle. Furthermore, the rotor can be rotated during spin operation. Again, several operating conditions can be simulated through full, half, or quarter rotor rotations, and several sets of sensor parameters along with reference load values ​​can be recorded.

[0081] However, it is also possible to vary several operating settings from the list simultaneously.

[0082] During spin-start operation, the rotor of the wind turbine rotates, but the turbine does not feed any power into the grid. In particular, no power is generated. The rotor is therefore in a state where it can be assumed that the loads on the rotor blade are caused solely by the blade's own weight. Other loads can thus be neglected, leading to a more accurate result. This is especially true when the wind speed is low and aerodynamic loads can therefore be disregarded. Typically, the blade angle during spin-start operation is set to a value in the range of approximately 50° to 70° or more, and in particular, it is set to a value above 60°.

[0083] However, even when power is being generated, the sensor array can be parameterized. This allows for further adjustment of the rotor's power and / or torque. In this case, however, the load can no longer be considered solely due to the rotor's own weight, making it necessary to factor in the additional loads within the reference load parameter.

[0084] According to another aspect, the sensor parameters for configuring the sensor array are acquired at wind speeds within a first and a second wind speed range. The first wind speed range lies between the initial wind speed and a second wind speed greater than the initial wind speed. The second wind speed range lies between the first wind speed and a second wind speed greater than the first. Furthermore, acquisition at wind speeds within the first wind speed range is repeated if acquisition has previously occurred at wind speeds within the second wind speed range.

[0085] Preferably, the sensor arrangement is then parameterized when wind speeds are low. The wind speed only needs to be high enough to turn the rotor, i.e., above the starting wind speed. The starting wind speed is therefore the wind speed sufficient to turn the rotor of the wind turbine without generating power. This speed is sufficient to overcome bearing friction in the rotor.

[0086] If the wind speed exceeds the first wind speed, aerodynamic influences can no longer be neglected and must be considered in the calculation of the reference load. The first wind speed is therefore chosen so that aerodynamic loads can be neglected if the parameterization or the acquisition of measurements for this purpose takes place at wind speeds up to the first wind speed.

[0087] Should parameterization be necessary before such low wind speeds occur, the sensor array can also be parameterized above the first wind speed threshold. However, the parameterization is then repeated under better conditions, namely when the wind speed is within the first wind speed range, to improve the parameterization result.

[0088] According to another aspect, the first speed range is selected within the range of the starting wind speed at which the wind turbine begins operation. The starting wind speed is above the run-up wind speed. Specifically, the first speed range lies between 30% and 200% of the starting wind speed, preferably even between 50% and 150% of the starting wind speed, whereby the speed range is chosen in such a way as to avoid wind speeds below the run-up wind speed. The rotor thus rotates.

[0089] Alternatively or additionally, the first speed range is selected within the range starting from the approach wind speed. This first speed range lies specifically between 100% and 250% of the approach wind speed.

[0090] Alternatively or additionally, the first speed range is between 1.5 m / s and 8 m / s and particularly preferably between 2 m / s and 6 m / s.

[0091] The starting wind speed is the wind speed at which the wind turbine starts up, meaning it can generate and feed power into the grid. The starting wind speed is therefore higher than the run-up wind speed, at which the rotor begins to turn but no power is yet being generated. Both the starting wind speed and the run-up wind speed can vary for each wind turbine. Furthermore, the starting wind speed and the run-up wind speed can also depend on weather conditions, such as air density.

[0092] According to another aspect, the sensor arrangement is designed to detect at least two load variables. A separate overall calculation procedure with its own calculation parameters is provided for each load variable.

[0093] Once the sensor arrangement is parameterized, at least two load quantities can be calculated using the sensor arrangement. Preferably a bending moment in a pivoting direction and a bending moment in a flapping direction of the rotor blade.

[0094] According to another aspect, one of the load quantities is a blade bending moment of the rotor blade in a first direction, particularly in a flapping direction, and another is a blade bending moment of the rotor blade in a second direction, particularly in a pitching direction. Optionally, one of the load quantities is an axial force of the rotor blade. A separate overall calculation procedure with its own calculation parameters is used for each load quantity to calculate the respective load quantity from the sensor values. The calculation parameters of each overall calculation procedure are determined by simultaneously considering the acquired sensor values ​​from all load sensors, but preferably using the same acquired sensor values. To determine the calculation parameters, a measurement matrix of acquired sensor values ​​is recorded and used jointly for all overall calculation procedures. Thus, the same measurement matrix is ​​used for each load quantity.However, for each overall calculation rule, and thus for each load size, a separate initial vector from determined reference load sizes is used.

[0095] For example, if a bending moment is to be used as a load parameter M x in one direction x, For example, to determine the direction of rotation of the rotor blade, the formula for the overall calculation procedure could be derived for an example with three load sensors. M x = k x , 1 ⋅ x 1 + k x , 2 ⋅ x 2 + k x , 3 ⋅ x 3 + k x , 0 .

[0096] For the same set of sensor sizes, the bending moment could also be calculated using adapted calculation parameters. M y to determine a second direction y, e.g. a swivel direction of the rotor blade. M y = k y , 1 ⋅ x 1 + k y , 2 ⋅ x 2 + k y , 3 ⋅ x 3 + k y , 0 .

[0097] Optionally, the axial force acting on the rotor blade can also be calculated, with adapted calculation parameters. M z = k z , 1 ⋅ x 1 + k z , 2 ⋅ x 2 + k z , 3 ⋅ x 3 + k z , 0 .

[0098] The load quantities – bending moment in a first direction x, bending moment in a second direction y, and axial force in a third direction z – result from the same applied force, i.e., the same operating conditions. Therefore, the same data set of sensor parameters can be used to determine the respective calculation parameters, as they can thus be recorded under the same operating conditions. Only the reference load quantity needs to be calculated individually for each load quantity to be determined. This calculation can also be performed under the same operating conditions, with only the calculation of the reference load quantities differing.

[0099] According to another aspect, in each acquisition step, the reference load value for each load sensor is calculated as a function of a gravitational force acting on the rotor blade. For this purpose, the gravitational force is converted into a coordinate system of the rotor blade. This is done taking into account the blade angle of the rotor blade, the rotational position of the rotor, the tilt angle of the rotor relative to a horizontal, and / or the tilt angle of the rotor blade relative to a plane of rotation of the rotor.

[0100] The weight force can be calculated using the known mass and mass distribution of the rotor blade. Its effect on the load varies depending on the blade's position. To account for this, converting the weight force into the rotor blade's coordinate system is proposed. Rotation matrices can be used for this purpose. The rotor blade's coordinate system thus refers to the area where the load sensors are mounted on the rotor blade.

[0101] According to another aspect, in order to successively record the sensor parameters in varying operating situations in recording steps, the reference load parameter is additionally determined in each recording step depending on the aerodynamic forces and / or aerodynamic moments acting on the rotor blade.

[0102] Taking aerodynamic loads into account improves the results of the reference load calculations. The calculation parameters can thus be determined even more precisely. This is especially true when the wind speed is higher than the initial wind speed.

[0103] This also ensures that sensor parameters can be taken into account for operating situations where aerodynamic influences cannot be neglected.

[0104] According to another aspect, a non-linear relationship between the sensor sizes and the load size is taken into account by including an offset in the calculation parameters. k 0 include, each load sensor has a total of m calculation parameters k i,j are assigned, where m is greater than one, and the overall calculation rule for n load sensors for the load size L is as follows: L = ∑ i = 1 n ∑ j = 1 m k i , j ⋅ x i j + k 0 .

[0105] The load size is thus calculated using a polynomial that takes the higher powers of the sensor sizes. x i allowed.

[0106] As a first approximation, a second-degree polynomial can be chosen, for example, i.e. j = 2, from which the load magnitude is calculated as L = k 1 , 1 ⋅ x 1 + k 1 , 2 ⋅ x 1 2 + k 2 , 1 ⋅ x 2 + k 2 , 2 ⋅ x 2 2 + k 3 , 1 ⋅ x 3 + k 3 , 2 ⋅ x 3 2 + k 0 .

[0107] If nonlinearities are to be taken into account, this must also be considered when calculating the reference load values.

[0108] In a matrix notation, a measurement matrix A could contain the recorded sensor parameters in each column. x i with the corresponding power, and in one column always the value one, which corresponds to the power 0.

[0109] According to the invention, a wind turbine is also proposed, comprising a sensor arrangement of several load sensors on a rotor blade of the wind turbine rotor for detecting at least one load quantity representative of a load acting on the rotor blade, wherein the rotor blade has at least three load sensors. Each of the load sensors detects a load-dependent physical quantity of the rotor blade and outputs a quantity representative of this quantity as the detected sensor quantity. The wind turbine also includes a control unit, in particular a parameterization unit. The control unit is configured to parameterize the sensor arrangement according to one of the described aspects.

[0110] For this purpose, the load sensors, which are arranged on the rotor blade particularly in the area of ​​the blade root, are distributed around the circumference of the rotor blade. Preferably, three load sensors are attached, spaced approximately 120° apart around the circumference. However, this angle is not taken into account during parameterization. Therefore, the load sensors can also be attached at a different angle or distance from each other. It is also not necessary to know the angle. Thus, an inaccurate placement of the load sensors does not lead to errors in the parameterization. This is achieved by determining the calculation parameters of the overall calculation procedure while simultaneously considering the measured sensor values ​​of all load sensors. Therefore, no individual calibration of the individual sensors is performed.

[0111] In particular, the parameterization according to one of the described aspects is implemented on the control unit. Alternatively or additionally, a parameterization unit can also be provided on which the parameterization is implemented.

[0112] The control unit and / or parameterization unit thus sets various operating conditions and / or operating settings and performs several acquisition steps, in particular according to a schedule stored in the control unit and / or parameterization unit. Furthermore, sensor parameter sets are acquired, and the calculation parameters are calculated and stored. The control unit and / or parameterization unit can also be implemented in a process computer of the wind turbine. Figure 1 shows a wind turbine in perspective view. Figure 2 shows a rotor blade with a sensor arrangement. Figure 3 shows a sensor arrangement. Figure 4 shows a flowchart for a method according to the invention.

[0113] Fig. 1 Figure 1 shows a schematic representation of a wind turbine according to the invention. The wind turbine 100 has a tower 102 and a nacelle 104 on the tower 102. An aerodynamic rotor 106 with three rotor blades 108 and a spinner 110 is provided on the nacelle 104. During operation of the wind turbine, the aerodynamic rotor 106 is set into rotation by the wind and thus also rotates an electrodynamic rotor or rotor of a generator, which is directly or indirectly coupled to the aerodynamic rotor 106. The electric generator is arranged in the nacelle 104 and generates electrical energy. The blade angles of the rotor blades 108, which can also be referred to as pitch angles, can be changed by pitch motors at the rotor blade roots 109 of the respective rotor blades 108.

[0114] Three load sensors are arranged on each of the rotor blades 108.

[0115] Fig. 2 Figure 2 shows a rotor blade 208 with a sensor arrangement 210. The sensor arrangement 210 has three load sensors 211, 213, and 215. The sensor arrangement 210 is mounted in the area of ​​the blade root 220 of the rotor blade 208. The three load sensors 211, 213, and 215 are distributed around the circumference of the rotor blade. Ideally, the load sensors 211, 213, and 215 are offset from each other by an angle of 120° around the circumference of the blade. However, exact positioning is not critical.

[0116] Fig. 3 The figure schematically shows the sensor arrangement 310 with three load sensors 311, 313, 315, which are distributed around the circumference of the rotor blade in the area of ​​the blade root 320.

[0117] The three load sensors 311, 313, 315 together form the sensor arrangement 310, which is configured to detect a load quantity. The load quantity is to be measured in one of the spatial directions. x, y, z The coordinate system 320 is used to determine the position. The z-axis points into the figure and thus corresponds to a direction along the longitudinal axis of the sheet.

[0118] The load parameter to be measured could, for example, be a bending moment. M x of the rotor blade in a direction of movement x, a bending moment M y of the rotor blade in a pivoting direction y, or an axial force F z in the longitudinal direction z of the rotor blade. It is not necessary that the measuring directions of the load sensors 311, 313, 315 be along one of the coordinate axes. x, y, z of the 320 coordinate system.

[0119] If the rotor blade bends as a result of a load acting upon it, each of the load sensors 311, 313, 315 experiences a change in length or elongation. The physical quantity of the change in length is detected by the load sensors 311, 313, 315 using a representative sensor parameter.

[0120] For example, if a bending moment acts M x in a direction of impact x, Each of the load sensors 311, 313, 315 experiences a strain according to its position in the sensor arrangement 310 and detects a sensor quantity proportional to the strain.

[0121] To calculate the load size from the sensor sizes of the load sensors 311, 313, 315, the sensor arrangement 310 is parameterized.

[0122] A comprehensive calculation method is used to parameterize the sensor arrangement 310. The load magnitude can thus be calculated based on this method, which establishes a relationship between the measured sensor values ​​of the load sensors 311, 313, and 315 and the load magnitude. The load magnitude can therefore be determined directly from the measured sensor values. It is not necessary to first determine the physical quantity, namely the strain.

[0123] The overall calculation procedure includes several calculation parameters. Different calculation parameters are used depending on the load magnitude to be determined.

[0124] For the bending moment M x in the direction of impact x For example, the overall calculation rule M x = k x , 1 ⋅ x 1 + k x , 2 ⋅ x 2 + k x , 3 ⋅ x 3 + k x , 0 verwendet, with the calculation parameters k x, 1 , k x,2 , k x,3 , k x,0 , where k x,0 forms an offset. The first load sensor 311 measures the first sensor value. x 1, the second load sensor 313, the second sensor size x 2 and the third load sensor 311 the third sensor size x 3 .

[0125] For the same set of sensor sizes x 1 ,x 2 ,x 3 This can be done with the other calculation parameters. k y,1 , k y,2 , k y,3 , k y,0 also the bending moment M y Calculate in the direction of rotation y according to M y = k y , 1 ⋅ x 1 + k y , 2 ⋅ x 2 + k y , 3 ⋅ x 3 + k y , 0 .

[0126] Therefore, the calculation parameters are determined for parameterizing the sensor arrangement 310.

[0127] The first, second and third sensor size x 1 , x 2 and x The 3 values ​​are the same for both bending moments.

[0128] Fig. 4 shows a method 400 for parameterizing the sensor arrangement 310.

[0129] Parameterization is initiated in starting block 410. Parameterization is preferably initiated when there is sufficient wind to rotate the wind turbine rotor, but the wind speed is so low that only gravitational loads are relevant, while other loads can be neglected. The resulting bending moments thus result from the rotor blade's own weight.

[0130] It then becomes a time step t = t 1. A data acquisition step 420 is performed. In data acquisition step 420, a sensor size is determined for each load sensor 311, 313, 315 in a sensor size set determination step. x 1 ( t = t 1 ), x 2 ( t = t 1 ), x 3 ( t = t 1) detected. The sensor sizes x 1 ( t = t 1 ), x 2 ( t = t 1 ), x 3 ( t = t 1) All load sensors 311, 313, 315 together form the sensor size set for time step t = t 1 .

[0131] Furthermore, in the acquisition step 420, a reference load parameter determination step 423 is carried out. In this step, a reference load parameter is determined. L ref ( t = t 1) determined. The reference load size L ref ( t = t 1) corresponds to the load magnitude at the time step t = t 1 occurs as expected, and can be calculated via the weight force.

[0132] The reference load size L ref For example, a bending moment M x in the direction of impact x, a bending moment M y in the direction of rotation y or an axial force F z in the longitudinal direction. However, in the reference load determination step 423, several reference loads can also be determined.

[0133] In memory step 430, the reference load size is L ref ( t = t 1) in an output vector L ref saved. Likewise, the sensor size set, i.e., the sensor sizes, is also saved. x 1 ( t = t 1 ), x 2 ( t = t 1 ), x 3 ( t = t 1) in a measurement matrix A saved.

[0134] Once sufficient data has been stored, particularly after a predetermined number of repetitions, the process continues with parameterization step 440. Otherwise, another acquisition step 420 is performed in the next time step. With each new acquisition step, an operating condition of the wind turbine is varied. This change in the operating condition is intended to alter the reference load parameter to be determined. Specifically, this could mean that the rotor has rotated further. The rotation of the rotor causes the force of gravity to alter the extension of the rotor blade in the area of ​​the load sensors. This, in turn, changes the reference load parameter. L ref .

[0135] The operating situation can also be varied by, for example, changing the blade angle of the rotor blade.

[0136] Furthermore, the operating setting can also be changed. Changing the operating setting should also change the reference load value. However, the operating setting does not need to be varied after each acquisition step 420, but can be maintained for several acquisition steps 420. Thus, for example, continuing to rotate the rotor does not represent a change in the operating setting, but leads to a change in the operating situation. The operating setting can, however, be changed by varying the blade angle.

[0137] In acquisition step 420, the sensor parameter set and the reference load are repeatedly acquired under changing operating conditions. This occurs in sensor parameter set determination step 421 and in reference load parameter determination step 423. Operating settings can also be changed if necessary. The acquired parameters are stored in storage step 430 in the next row of the output vector or the measurement matrix.

[0138] After p acquisition steps (420) and storage steps (430), a measurement matrix is ​​thus available. A = x 1 t = t 1 x 2 t = t 1 x 3 t = t 1 1 x 1 t = t 2 x 2 t = t 2 x 3 t = t 2 1 ⋯ ⋯ ⋯ 1 x 1 t = t p x 2 t = t p x 3 t = t p 1 before, as well as an output vector L → ref = L ref t = t 1 L ref t = t 2 ⋯ L ref t = t p .

[0139] A flowchart is preferably used as a basis, so that in storage step 430 enough data has been saved precisely when the flowchart has finished. The flowchart can therefore determine the number of p The data collection steps are specified.

[0140] For example, a schedule can be set to perform a measurement step 420 after each 1° rotor rotation and to change the blade angle after two full rotations, measuring, for example, three different blade angles. The number p The number of data collection steps would in this case be 2*3*360=2160.

[0141] Once the process plan is complete, the sensor arrangement is parameterized in parameterization step 440 by determining the calculation parameters.

[0142] The calculation parameters are determined using the measurement matrix A, taking into account all sensor values ​​acquired during parameterization. The parameter vector is calculated in a single step. k , which has the calculation parameters, determines about k → = A T ⋅ A − 1 ⋅ A T ⋅ L → ref .

[0143] In the subsequent final step 450, the calculation parameters are saved and the parameterization is completed.

[0144] Is the sensor arrangement designed for a bending moment M x in the direction of impact x When parameterized, the calculation parameters are set accordingly. k x,1 , k x,2 , k x,3 , k x, 0 determined and saved.

[0145] Once the parameters have been configured, the sensor array can be used to calculate load values. For example, once the parameters for the bending moment have been configured... M x Once the calculation is carried out, it is based on the overall calculation rule. M x = k x , 1 ⋅ x 1 + k x , 2 ⋅ x 2 + k x , 3 ⋅ x 3 + k x , 0 .

[0146] To ensure that the calculation parameters are determined as accurately as possible, the procedure 400 for parameterizing the sensor arrangement 310 is repeated after a predetermined time, e.g. every three months.

[0147] According to the invention, the following was particularly recognized and the following is proposed.

[0148] A method for parameterizing a sensor arrangement of a wind turbine is proposed, wherein the sensor arrangement is located on a rotor blade of the wind turbine and consists of at least three load sensors.

[0149] The parameterization is preferably carried out in a two-stage process, consisting of a flow chart or parameterization routine and a parameterization step.

[0150] The goal of the parameterization routine is to record and store measured values ​​from the load sensors, along with a corresponding calculated reference bending moment, for different bending moments at the rotor blade. For this purpose, the wind turbine is operated in a setting where the bending moment acting on the rotor blade can be calculated with high accuracy.

[0151] It is particularly suggested that, for the parameterization routine, the wind turbine should be operated in a spin mode and varying blade angles should be specified for the rotor blade, so that the reference bending moment corresponds approximately to a natural moment.

[0152] To ensure that aerodynamic forces are negligible, the sensor array is preferably parameterized at low wind speeds. However, the wind must be strong enough to complete the parameterization within an acceptable timeframe. The specific wind speeds at which such parameterization is performed depend on the initial wind speed of the wind turbine at the respective location. If these conditions are not met, particularly if the wind is stronger than usual, but parameterization is necessary, it will still be performed and repeated later under favorable conditions.

[0153] Alternatively, the aerodynamic forces and moments could also be taken into account to enable precise parameterization of the sensor arrangement even in high winds.

[0154] After the parameterization routine is complete, the parameterization step is executed. This step calculates the calculation parameters. Once the parameterization routine is finished, the calculation parameters are determined from the recorded sensor values ​​and the reference bending moments, ensuring that the mean squared error between the reference bending moments and the bending moments calculated from the calculation parameters and sensor values ​​is minimized.

[0155] It is assumed that the bending moments are proportional to the elongation of the rotor blade. In reality, nonlinearity occurs above a certain elongation. This deviation from linearity can be inversely accounted for in the parameterization routine or parameterization step, if necessary.

Claims

1. A method for parameterizing a sensor arrangement comprising multiple load sensors of a rotor blade (208) of a rotor (106) of a wind power installation (100) for acquiring at least one load variable, which is representative of a load that acts on the rotor blade (208), wherein - the rotor blade (208) has at least three load sensors (211, 213, 215), - each of the load sensors (211, 213, 215) records a load-dependent physical variable of the rotor blade (208) and outputs a variable representative thereof as the acquired sensor variable, - for calculating the at least one load variable from the sensor variables, at least one overall calculation rule is used, forming a relationship between the acquired sensor variables of all the load sensors (211, 213, 215) and the at least one load variable of the rotor blade (208), and having multiple calculation parameters, and, - for parameterizing the sensor arrangement (210), the calculation parameters of the overall calculation rule are determined while at the same time taking into consideration acquired sensor variables of all the load sensors (211, 213, 215).

2. The method as claimed in claim 1, characterized in that - no individual parameterization of individual load sensors (211, 213, 215) takes place for determining the calculation parameters, in particular no individual calibration and / or adjustment of individual load sensors (211, 213, 215) takes place.

3. The method as claimed in claim 1 or 2, characterized in that - the sensor variables are successively acquired in multiple acquisition steps (420) in varying operational situations of the wind power installation (100), wherein - in each acquisition step (420) - a sensor variable is respectively acquired for each load sensor (211, 213, 215), so that multiple sensor variables are acquired in the acquisition step (420), forming a set of sensor variables, and - at least one associated load variable is ascertained, in particular is calculated, for the set of sensor variables as a reference load variable, - the acquisition steps (420) are respectively repeated with a changed operational situation, in particular at least as often as the overall calculation rule has calculation parameters to be determined, in particular at least once more often, and - the calculation parameters are determined in dependence on all the sensor variables acquired in the acquisition steps and ascertained associated load variables.

4. The method as claimed in one of the preceding claims, characterized in that - the overall calculation rule has or is a calculation equation, which describes by means of the calculation parameters a relationship between the load variable and the sensor variables, wherein, for determining the calculation parameters, - a system of equations comprising multiple calculation equations is set up, - for each calculation equation of the system of equations - an ascertained reference load variable is used for the load variable, and - acquired sensor variables associated with the ascertained reference load variable are used for the sensor variables, and - the system of equations is resolved on the basis of the calculation parameters, wherein in particular - more calculation equations are set up than the overall calculation rule has calculation parameters, so that the system of equations is overdetermined, and in particular the system of equations is resolved such that a statistically optimal solution is obtained.

5. The method as claimed in one of the preceding claims, characterized in that - a or the system of equations comprising multiple calculation equations of the overall calculation rule is set up in matrix form, with - an output vector for the ascertained reference load variables, - a measurement matrix for the acquired sensor variables and - a parameter vector for the calculation parameters to be determined such that - the output vector is equal to a product of the measurement matrix times the parameter vector, wherein in particular - an approximation solution for the system of equations is found by means of forming a pseudoinverse of the measurement matrix, and / or - the parameter vector has an or the offset, and the measurement matrix has an additional column corresponding to the offset, which is filled with the value 1, so that, in the multiplication of the measurement matrix by the parameter vector, the offset is respectively multiplied by the value 1.

6. The method as claimed in one of the preceding claims, characterized in that - the calculation parameters comprise an offset k0, - each load sensor (211, 213, 215) is assigned at least one calculation parameter ki, and - the overall calculation rule is obtained in that - for each of the load sensors (211, 213, 215), a product of the sensor variable xi of the respective load sensor i times the at least one calculation parameter ki associated with the load sensor is formed, and - the load variable L is obtained as the sum of the products ki · xi and the offset k0 on the basis of the formula: L = ∑ i = 1 n k i ⋅ x i + k 0 .

7. The method as claimed in one of the preceding claims, characterized in that - for the successive acquisition of the sensor variables in varying operational situations, - the operational situations are chosen such that different load variables, in particular different bending moments, occur at the rotor blade (208), and in particular - the bending moments result from a dead weight of the rotor blade (208), and - the bending moments are varied by the rotor (106) being changed in its rotational position, and / or a blade angle of the rotor blade being changed.

8. The method as claimed in one of the preceding claims, characterized in that - for the successive acquisition of the sensor variables in varying operational situations in acquisition steps (420), - a schedule is prescribed for the acquisition steps (420), so that - the varying operational situations are set according to the schedule, and - the parameterizing of the sensor arrangement is carried out after ending of the schedule.

9. The method as claimed in one of the preceding claims, characterized in that, for acquiring the sensor variables, at least one operational setting is provided from the list comprising - the wind power installation (100) is operated in an idling mode, - a blade angle of the rotor blade (208) is set, - a power output is set, - a torque of a generator connected to the rotor (106) is set, and - a rotational speed of the rotor (106) is set, wherein - for acquiring the sensor variables, at least one of the operational settings is varied, and in particular - each operational setting is retained for a predetermined test period, in particular for multiple acquisition steps (420) and / or - at least one operational setting is retained until multiple operational situations are in each case enacted, with an acquisition step (420) for acquiring the sensor variables being respectively carried out.

10. The method as claimed in one of the preceding claims, characterized in that - the acquisition of the sensor variables for parameterizing the sensor arrangement takes place at wind speeds in a first wind speed range and a second wind speed range, wherein - the first wind speed range lies between a start-up wind speed and a first wind speed, which is greater than the start-up wind speed, - the second wind speed range lies between the first wind speed and a second wind speed, which is greater than the first wind speed, and in particular - the acquisition is repeated at wind speeds in the first wind speed range if the acquisition previously took place at wind speeds in the second wind speed range.

11. The method as claimed in one of the preceding claims, characterized in that - the sensor arrangement is set up for acquiring at least two load variables, wherein - a dedicated overall calculation rule with dedicated calculation parameters is provided for each load variable, and in particular - one of the load variables is a blade bending moment of the rotor blade (108) in a first direction, in particular in a flapwise direction, and - one of the load variables is a blade bending moment of the rotor blade (108) in a second direction, in particular in an edgewise direction, and optionally - one of the load variables is an axial force of the rotor blade (108), and wherein, - for each load variable, - a dedicated overall calculation rule, with dedicated calculation parameters, is used, in order to calculate the respective load variable from the sensor variables, and - the calculation parameters of each overall calculation rule are respectively determined while at the same time taking acquired sensor variables of all the load sensors into consideration, - preferably using the same acquired sensor variables, wherein in particular - for determining the calculation parameters - a measurement matrix comprising acquired sensor variables for all the overall calculation rules together is recorded and used, but - a dedicated output vector comprising ascertained reference load variables is used for each overall calculation rule.

12. The method as claimed in one of the preceding claims, characterized in that in each acquisition step (420) a or the reference load variable is calculated in dependence on a gravitational force that is acting on the rotor blade for each load sensor (211, 213, 215), wherein in particular - the gravitational force is converted into a coordinate system of the rotor blade while taking into consideration a blade angle of the rotor blade (208), a rotational position of the rotor (106), an angle of inclination of the rotor (106) with respect to a horizontal, and / or an angle of inclination of the rotor blade (208) with respect to a rotational plane of the rotor (106).

13. The method as claimed in one of the preceding claims, characterized in that - for the successive acquisition of the sensor variables in varying operational situations in acquisition steps (420), the reference load variable is additionally ascertained in each acquisition step (420) in dependence on aerodynamic forces occurring and / or aerodynamic moments occurring, which act on the rotor blade (208).

14. The method as claimed in one of the preceding claims, characterized in that - a non-linear relationship between the sensor variables and the load variable is taken into consideration in that - the calculation parameters comprise a or the offset k0, - altogether m calculation parameters ki,j are assigned to each load sensor (211, 213, 215), where m is greater than one, and - the overall calculation rule in the case of n load sensors for the load variable L is obtained as L = ∑ i = 1 n ∑ j = 1 m k i , j ⋅ x i j + k 0 .

15. A wind power installation (100), with a sensor arrangement comprising multiple load sensors (211, 213, 215) of a rotor blade (208) of a rotor (106) of the wind power installation (100) for acquiring at least one load variable, which is representative of a load that acts on the rotor blade (208), wherein - the rotor blade (208) has at least three load sensors (211, 213, 215), - each of the load sensors (211, 213, 215) records a load-dependent physical variable of the rotor blade (208) and outputs a variable representative thereof as the acquired sensor variable, - the wind power installation (100) has a control device, in particular a parameterizing unit, and - the control device is set up to parameterize the sensor arrangement according to a method as claimed in one of claims 1 - 14.