Detecting blade weight anomaly of wind turbine

Abstract

A method of monitoring a blade of a rotor of a wind turbine. A series of bending values is obtained by measuring bending of the blade as the rotor rotates. The method comprises: identifying a local maximum bending value in the series of bending values; calculating an upper envelope value based on the local maximum bending value and at least one neighbouring bending value on each side of the local maximum bending value; identifying a local minimum bending value in the series of bending values; and calculating a lower envelope value based on the local minimum bending value and at least one neighbouring bending value on each side of the local minimum bending value. A delta value is calculated based on a difference between the upper and lower envelope values, and used to determine whether a blade weight anomaly exists and generate a corresponding output.

Description

[0001] DETECTING BLADE WEIGHT ANOMALY OF WIND TURBINE

[0002] FIELD OF THE INVENTION

[0003] The present invention relates to a method of monitoring a blade of a rotor of a wind turbine; a method of controlling a wind turbine; apparatus for monitoring a blade of a rotor of a wind turbine; and a wind turbine system.

[0004] BACKGROUND OF THE INVENTION

[0005] US2009246019 A1 discloses a method of detecting the formation of ice on the blades of a wind turbine. The wind turbine has at least one turbine blade mounted to a rotor and provided with at least a first strain sensor for measuring mechanical strain of the turbine blade. The method comprises detecting changes in an output signal of the strain sensor due to changes in the mass of the turbine blade caused by the formation of ice on the turbine blade. A relationship between the output value of the first strain sensor and the bending moment due to the mass of the turbine blade is determined by reference to the peak-to-peak amplitude of the output signal.

[0006] SUMMARY OF THE INVENTION

[0007] A first aspect of the invention provides a method of monitoring a blade of a rotor of a wind turbine, the method comprising: obtaining a series of bending values by measuring bending of the blade as the rotor rotates; identifying a local maximum bending value in the series of bending values; calculating an upper envelope value based on the local maximum bending value and at least one neighbouring bending value on each side of the local maximum bending value; identifying a local minimum bending value in the series of bending values; calculating a lower envelope value based on the local minimum bending value and at least one neighbouring bending value on each side of the local minimum bending value; calculating a delta value based on a difference between the upper and lower envelope values; and using the delta value to determine whether a blade weight anomaly exists and generate a corresponding output.

[0008] Optionally the series of bending values is obtained by: obtaining a series of data points; obtaining an azimuth value for each data point, wherein the azimuth value is indicative of an azimuth angle of the blade; defining a plurality of bins, each bin spanning a range of azimuth values; assigning each data point to a bin based on its azimuth value; and determining a bending value for each bin, wherein each bending value is representative of plural data points in the bin.

[0009] Optionally each bin spans an azimuth angle of more than 1 degree or more than 2 degrees.

[0010] Optionally the upper envelope value is based on the local maximum bending value and only a single neighbouring bending value on each side of the local maximum bending value; and the lower envelope value is based on the local minimum bending value and only a single neighbouring bending value on each side of the local minimum bending value.

[0011] Optionally the upper envelope value is based on the local maximum bending value and two or more neighbouring bending values on each side of the local maximum bending value; and the lower envelope value is based on the local minimum bending value and two or more neighbouring bending values on each side of the local minimum bending value.

[0012] Optionally the wind turbine is generating substantially no power as the series of bending values is obtained.

[0013] Optionally the rotor is rotating at a rate below 10 RPM or below 2 RPM as the series of bending values is obtained.

[0014] Optionally the delta value is compared with an output from a trained model, and the determination whether a blade weight anomaly exists is based on the comparison.

[0015] Optionally the trained model comprises a Kalman filter.

[0016] Optionally the blade comprises a sensor; and the bending of the blade is measured by the sensor.

[0017] Optionally the sensor comprises a strain sensor. Optionally the blade weight anomaly is an ice-related blade weight anomaly.

[0018] A further aspect of the invention provides a method of controlling a wind turbine, the method comprising monitoring a blade of a rotor of the wind turbine by a method according to the first aspect; and controlling the wind turbine based on the output.

[0019] A further aspect of the invention provides apparatus for monitoring a blade of a rotor of a wind turbine, wherein the apparatus is configured to: receive a series of bending values indicative of bending of the blade as the rotor rotates; identify a local maximum bending value in the series of bending values; calculate an upper envelope value based on the local maximum bending value and at least one neighbouring bending value on each side of the local maximum bending value; identify a local minimum bending value in the series of bending values; calculate a lower envelope value based on the local minimum bending value and at least one neighbouring bending value on each side of the local minimum bending value; calculate a delta value based on a difference between the upper and lower envelope values; and use the delta value to determine whether a blade weight anomaly exists and generate a corresponding output.

[0020] A further aspect of the invention provides a wind turbine system comprising: a rotor comprising a blade, the blade comprising a bending sensor configured to obtain bending values by measuring bending of the blade as the rotor rotates; and apparatus according to the preceding aspect configured to receive the bending values and monitor the blade accordingly.

[0021] BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Embodiments of the invention will now be described with reference to the accompanying drawings, in which:

[0023] Figure 1 is a front view of a wind turbine;

[0024] Figure 2 is a side view of the wind turbine;

[0025] Figure 3 shows a blade of the wind turbine

[0026] Figure 4 shows strain gauges in the root of the blade;

[0027] Figure 5 shows apparatus for monitoring the blades;

[0028] Figure 6 is a flow diagram showing a method of monitoring a blade of the wind turbine; Figure 7 shows an edge bending signal; Figure 8 shows a local maximum of the edge bending signal;

[0029] Figure 9 shows a local minimum of the edge bending signal; and

[0030] Figure 10 shows a series of cycles of the edge bending signal.

[0031] DETAILED DESCRIPTION OF EMBODIMENT(S)

[0032] Figures 1 and 2 show a wind turbine 1 including a tower 3 mounted on a foundation and a nacelle 4 disposed at the apex of the tower. The wind turbine 1 depicted here is an onshore wind turbine such that the foundation is embedded in the ground, but the wind turbine 1 could be an offshore installation in which case the foundation would be provided by a suitable marine platform.

[0033] A rotor is operatively coupled via a gearbox to a generator (not shown) housed inside the nacelle 4. The rotor includes a central hub and a plurality of rotor blades 2a-c, which project outwardly from the central hub. It will be noted that the wind turbine 1 is the common type of horizontal axis wind turbine (HAWT) such that the rotor is mounted at the nacelle to rotate about a substantially horizontal axis defined at the centre at the hub. The example shown has three blades, but it will be realised by the skilled person that other numbers of blades are possible.

[0034] When wind blows against the wind turbine 1, the blades generate a lift force which causes the rotor to rotate, which in turn causes the generator within the nacelle 4 to generate electrical energy.

[0035] All of the blades 2a-c are substantially identical, and an exemplary one of the blades 2a is shown in Figures 3 and 4. Each of the blades has a root end 6 proximal to the hub and a tip end 7 distal from the hub. A leading edge 5a and a trailing edge 5b extend between the root end 6 and the tip end 7, and each of the blades has a respective aerodynamic high pressure surface (i.e. the pressure surface) and an aerodynamic low pressure surface (i.e. the suction surface) extending between the leading and trailing edges of the blade.

[0036] Each blade has an edgewise direction 14 (between the leading edge 5a and the trailing edge 5b) and a flapwise direction 15 (at right angles to the edgewise direction). Ice accretion on the blades can impact both performance and safety of the wind turbine. A clear indication of the presence of ice is important to determine reasons for performance losses, changes to control strategy, or legal requirements. From a safety standpoint, ice presence can lead to ice throw which can pose risk if a turbine is placed near people, roads, or animals. Improved ice detection can also lead to bill of materials (BOM) savings, reducing the need for further material, making the wind turbine 1 lighter, cheaper, and easier to transport.

[0037] As shown in Figure 4, the blade includes a blade shell 10 defining a hollow interior space 11. Strain gauges 12a, 13a are adhered to the interior blade shell, or embedded within the blade shell, at the root end 6 of the blade.

[0038] Each strain gauge 12a, 13a measures linear extension of the blade root. An example of a suitable strain gauge is an optical strain gauge such as a fibre Bragg grating (FBG) as described in EP3317531 B1. Such an optical strain gauge comprises equally spaced reflection points in the core of the optical fibre that reflect difference wavelengths of light under different levels of strain. Such a sensor is well known to the skilled person.

[0039] A first one of the strain gauges is an edge strain gauge 12a which is positioned to measure bending of the blade in the edgewise direction 14. When the blade bends one way in the edgewise direction 14, the edge strain gauge 12a extends and generates a positive edge bending signal. When the blade bends the other way in the edgewise direction 14, the edge strain gauge 12a contracts and generates a negative edge bending signal. In this case there is only a single edge strain gauge 12a, but in other embodiments there may be two edge strain gauges positioned on opposite sides of the blade root 6.

[0040] A second one of the strain gauges is a flap strain gauge 13a which is positioned to measure bending of the blade in the flapwise direction 15. When the blade bends one way in the flapwise direction 15, the flap strain gauge 13a extends and generates a positive flap bending signal. When the blade bends the other way in the flapwise direction 15, the flap strain gauge 13a contracts and generates a negative flap bending signal. In this case there is only a single flap strain gauge 13a, but in other embodiments there may be two flap strain gauges positioned on opposite sides of the blade root 6. Figure 5 is a schematic diagram of apparatus 20 for monitoring the blades 2a-c of the rotor of the wind turbine 1. All elements of the apparatus 20 may be part of the wind turbine 1, or some elements may be located remotely from the wind turbine 1 (for instance at a central wind park controller).

[0041] The apparatus 20 is configured to perform the method shown by the flow diagram of Figure 6.

[0042] Each blade has a pair of strain gauges 12a, 13a; 12b, 13b; 12c, 13c which generate bending signals which are input into a computer-implemented processing system 21.

[0043] As the rotor rotates, each blade rotates through a series of azimuth angles. The azimuth angle may be defined in any way, but in this specification we define the azimuth angle as 0° when the blade is pointing down, 180° when the blade is pointing up, and 90° or 270° when the blade is pointing left or right.

[0044] An azimuth sensor 22 measures an azimuth angle of the rotor, from which an azimuth angle of each blade is determined by the processing system 21. The azimuth sensor 22 may comprise an encoder in the rotor hub, a magnetic ring or a gyroscope for example.

[0045] When the blade is vertical (i.e. at its highest point so it is pointing up, or at its lowest point so it is pointing down) the weight of the blade will not be bending the blade in the edgewise direction, so the magnitude of the edge bending signal is low. When the blade is horizontal (i.e. pointing left or right) the weight of the blade will be maximally bending the blade in the edgewise direction, so the edge bending signal is at a maximum or a minimum. Hence the edge bending signal follows a sinusoidal profile as shown in Figure 7, with a maximum at 90° and a minimum at 270°.

[0046] Note that the rotor axis may be tilted up slightly from the horizontal, as shown in Figure 2, so the blade 2a pointing up or down at 0° or 180° azimuth angle is not precisely vertical but is slightly tilted. For each blade 2a, 2b, 2c a series of edgewise bending data points is obtained from its respective edge strain gauge(s) 12a, 12b, 12c; and an azimuth value is obtained for each data point from the azimuth sensor.

[0047] Referring to Figure 6, edgewise bending data points 30 are binned at 31 based on their respective azimuth values 32. That is, a plurality of bins are defined, each bin spanning a range of azimuth values; and each data point is assigned to a bin based on its azimuth value. The range of each bin can vary, a typical range being 5 degrees. Typically each bin spans an azimuth angle of more than 1 degree or more than 2 degrees.

[0048] A bending value is then determined for each bin, where each bending value is representative of plural data points in the bin. By way of example, each bending value may be a sum or mean average of the data points in the bin.

[0049] Figure 8 gives an example of five bending values 50-54 derived from a signal 55 which passes through a local maximum at an azimuth angle of 90°.

[0050] The bending values are analysed at 33 and 34 to identify local maximum and local minimum bending values.

[0051] An upper envelope value 56 is calculated at 35, based on the latest local maximum bending value 52 and the neighbouring bending values 51 , 53 on each side of the local maximum bending value 52. By way of example, the upper envelope value 56 may be based on the sum or mean average of the local maximum bending value 52 and the neighbouring bending values 51, 53 on each side of the local maximum bending value 52. The upper envelope value 56 is indicated by a dashed line in Figure 8.

[0052] Figure 9 gives an example of five bending values 60-64 derived from the signal 55 which passes through a local minimum at an azimuth angle of 270°.

[0053] A lower envelope value 66 is calculated at 36, based on the latest local minimum bending value 62 and the neighbouring bending values 61 , 63 on each side of the local minimum bending value 62. By way of example, the lower envelope value 66 may be based on the sum or mean average of the local minimum bending value 62 and the neighbouring bending values 61, 63 on each side of the local minimum bending value 62. The lower envelope value 66 is indicated by a dashed line in Figure 9.

[0054] In this example the upper envelope value 56 is based on only a single neighbouring bending value 51, 53 on each side of the local maximum bending value 52. Similarly the lower envelope value 66 is based on only a single neighbouring bending value 61, 63 on each side of the local minimum bending value 62. In other examples the upper envelope value may be based on two or more neighbouring bending values (for example the values 50,51 ,53,54) on each side of the local maximum bending value 52; and the lower envelope value may be based on the local minimum bending value and two or more neighbouring bending values (for example the values 60,61,63,64) on each side of the local minimum bending value 62.

[0055] A delta value is then calculated at 37, based on a difference between the upper and lower envelope values; and the delta value is used to determine whether a blade weight anomaly exists and generate a corresponding output.

[0056] Figure 10 shows a series of cycles of the bending signal, showing an example of how the upper and lower envelope values 56, 66 may change from cycle to cycle. The delta value 57 is also indicated in Figure 10.

[0057] At 38 the delta value is compared with an output from a trained model, such as a Kalman filter 23a-c, and a determination whether a blade weight anomaly exists is based on the comparison.

[0058] The Kalman filter 23a-c for each blade is trained when a temperature sensor senses that the ambient temperature is above a threshold (for example 5 degC). Hence the output of each Kalman filter 23a-c provides an estimate of the delta value for its respective blade in the absence of ice on the blade.

[0059] The trained model determines its output based on input comprising wind speed, environmental temperature and blade loads, for example the bending values from the strain gauges.

[0060] A Kalman filter 23a-c is an example of a trained model for predicting the delta value 57, but other models are possible. The delta value 57 correlates with the weight of the blade, so if the delta value 57 is higher than the output from the Kalman filter, then this indicates that a blade weight anomaly exists, most likely due to accretion of ice or other material on the surface of the blade.

[0061] The processing system 21 determines whether a blade weight anomaly exists and generates a corresponding output 24 which is indicated in Figure 5. This output 24 may simply be an indication on a display device, or in a computer memory, that a blade weight anomaly does or does not exist. Alternatively, the output 24 may be fed to a wind turbine control system 25 which controls the wind turbine 1 based on the output 24. For example, referring to Figure 6, if the output 24 indicates a blade weight anomaly then the wind turbine control system 25 may initiate ice adaption measures 39 such as activating heaters on the blades 2a-c, pitching out the blades 2a-c to enter idle operation, suspending calibration procedures, or changing a control strategy of the wind turbine to extract more power from the rotor. Similarly, if the output 24 indicates that there is no blade weight anomaly, then such ice adaption measures may be stopped at 40 so the wind turbine returns to normal operation.

[0062] The detection method of Figure 6 may be performed when the wind turbine is in idle mode, or any similar mode in which the rotor rotates at a lower rate than nominal. For example the rotor may be rotating at a rate below 10 RPM, below 2 RPM or below 1 RPM. Optionally substantially no power is being generated by the generator, or a small amount of power may be being generated to charge batteries or to keep the wind turbine running.

[0063] The idle mode operation may have been triggered by a detection of ice during normal operation in which the generator is generating power. In this case, the rotor will remain in idle mode until a blade weight anomaly is no longer detected at 38, and the pitch of the blades is adjusted at 40 to return the rotor to normal operation.

[0064] The delta value 57 derived from upper and lower envelope values 56, 66 provides a useful measure for blade weight anomaly detection because it is more robust (for instance less sensitive to noise) than a simple peak-to-peak measure, as in US2009246019 A1. Binning based on azimuth angle provides further robustness. Checks may be done to ensure the quality of the delta value 57. Examples of checks include, but are not limited to, positive rotor speed for at least a minimum period of period time, valid status of signals, enough datapoints per azimuth position, etc.

[0065] Checks may also be done to ensure the quality of the outputs of the Kalman filter 23a-c, or other model. Examples of checks include, but are not limited to, residuals, model error, means / variance of residuals, etc.

[0066] The information on blade condition provide by the delta value 57 could enable advanced control strategies, that can extract more power from the rotor. Instead of traditional mixed profile assumptions, the control strategy could use the estimated blade condition to define the control strategy.

[0067] Typically the blade weight anomaly detected at 38 is an ice-related blade weight anomaly, i.e. an indication that ice has accreted on the blade. The method may also be used to indicate a blade weight anomaly caused by accretion of another material on the blade, or a low blade weight anomaly caused by degradation of the blade.

[0068] The embodiments above provide a method of monitoring a blade of the rotor of a wind turbine based on a series of bending values obtained by measuring bending of the blade as the rotor rotates. The bending values could be strain values from one or more edge strain gauges, from one or more flap strain gauges, or from a combination of both flap and edge strain gauges. Other embodiments of the invention may obtain the bending values from other sensors, for example a camera, lidar sensor or accelerometer. The bending values may be obtained directly from a specific sensor or from a fusion of different sensors. The bending values may be obtained by directly measuring bending of the blade, or by indirectly measuring bending of the blade using an engineering model. The bending values for a particular blade may also originate from a model based on the other two blades of the rotor.

[0069] Although the invention has been described above with reference to one or more preferred embodiments, it will be appreciated that various changes or modifications may be made without departing from the scope of the invention as defined in the appended claims

Claims

CLAIMS1. A method of monitoring a blade of a rotor of a wind turbine, the method comprising:obtaining a series of bending values by measuring bending of the blade as the rotor rotates; identifying a local maximum bending value in the series of bending values; calculating an upper envelope value based on the local maximum bending value and at least one neighbouring bending value on each side of the local maximum bending value; identifying a local minimum bending value in the series of bending values; calculating a lower envelope value based on the local minimum bending value and at least one neighbouring bending value on each side of the local minimum bending value; calculating a delta value based on a difference between the upper and lower envelope values; and using the delta value to determine whether a blade weight anomaly exists and generate a corresponding output.

2. A method according to claim 1 , wherein the series of bending values is obtained by:obtaining a series of data points; obtaining an azimuth value for each data point, wherein the azimuth value is indicative of an azimuth angle of the blade; defining a plurality of bins, each bin spanning a range of azimuth values; assigning each data point to a bin based on its azimuth value; and determining a bending value for each bin, wherein each bending value is representative of plural data points in the bin.

3. A method according to claim 2, wherein each bin spans an azimuth angle of more than 1 degree or more than 2 degrees.

4. A method according to any preceding claim, wherein the upper envelope value is based on the local maximum bending value and only a single neighbouring bending value on each side of the local maximum bending value; and the lower envelope value is based on the local minimum bending value and only a single neighbouring bending value on each side of the local minimum bending value.

5. A method according to any of claims 1 to 3, wherein the upper envelope value is based on the local maximum bending value and two or more neighbouring bending values on each side of the local maximum bending value; and the lower envelope value is based on the local minimum bending value and two or more neighbouring bending values on each side of the local minimum bending value.

6. A method according to any preceding claim, wherein the wind turbine is generating substantially no power as the series of bending values is obtained.

7. A method according to any preceding claim, wherein the rotor is rotating at a rate below 10 RPM as the series of bending values is obtained.

8. A method according to any preceding claim, wherein the blade weight anomaly is an ice-related blade weight anomaly.

9. A method according to any preceding claim, wherein the delta value is compared with an output from a trained model, and the determination whether a blade weight anomaly exists is based on the comparison.

10. A method according to claim 9, wherein the trained model comprises a Kalman filter.

11. A method according to any preceding claim, wherein the blade comprises a sensor;and the bending of the blade is measured by the sensor.

12. A method according to claim 11 , wherein the sensor comprises a strain sensor.

13. A method of controlling a wind turbine, the method comprising monitoring a blade of a rotor of the wind turbine by a method according to any preceding claim; and controlling the wind turbine based on the output.

14. Apparatus for monitoring a blade of a rotor of a wind turbine, wherein the apparatus is configured to: receive a series of bending values indicative of bending of the blade as the rotor rotates; identify a local maximum bending value in the series of bending values; calculate an upper envelope value based on the local maximum bending value and at least one neighbouring bending value on each side of the local maximum bending value; identify a local minimum bending value in the series of bending values; calculate a lower envelope value based on the local minimum bending value and at least one neighbouring bending value on each side of the local minimum bending value; calculate a delta value based on a difference between the upper and lower envelope values; and use the delta value to determine whether a blade weight anomaly exists and generate a corresponding output.

15. A wind turbine system comprising: a rotor comprising a blade; a bending sensor configured to obtain bending values by measuring bending of the blade as the rotor rotates; and apparatus according to claim 14 configured to receive the bending values and monitor the blade accordingly.

Images