Frequency content-based monitoring of wind turbine blade pitch systems

Abstract

A system, method, and computer program product for monitoring a wind turbine (10) are provided. The system receives a signal (426) indicating pitch forces being applied to blades (20) of the wind turbine (10). The signal (426) is sampled to generate a discrete time-domain signal (426) comprising a plurality of pitch force samples. The samples are selected for analysis using a sampling window (326) that excludes samples obtained under operating conditions determined to be detrimental to obtaining good data. The selected samples are processed to generate a spectral density (150) of the signal (426), and the frequency content of the spectral density (150) is analyzed to determine a condition of one or more components of the wind turbine (10). If the analysis indicates that a component of the wind turbine (10) requires attention, the system generates an alarm.

Description

Technical Field

[0001] The present invention relates generally to monitoring wind turbines and, in particular, to methods, systems, and computer program products for determining a condition of a wind turbine based on the spectral density of signals generated by a blade pitch drive system. Background Art

[0002] Wind turbines are a growing source of electricity primarily due to their low carbon footprint and concerns about the environmental impact of traditional power generation methods. However, in order to accelerate the replacement of fossil fuel-dependent plant equipment with wind turbines and still remain competitive with other forms of renewable energy, it is important to continue to reduce the cost of using wind turbines to generate electricity. One source of cost for wind turbines is maintaining the turbines. Maintenance must be frequent enough to prevent power outages due to component failures, but not unnecessarily frequent due to the cost of sending crews to perform inspections. Maintenance of wind turbines is particularly expensive due to a number of factors, such as the remote nature of many wind farms, the necessity to interrupt power production while maintenance is performed, and the fact that turbines are typically located atop tall towers and are therefore difficult to reach.

[0003] One way to monitor wind turbine operation is to place vibration sensors on or near components susceptible to wear. The vibrations picked up by the sensors are then analyzed to determine if there are vibration signatures of worn or failing components, thereby identifying the need for maintenance. However, these types of systems increase the cost and complexity of wind turbines due to the need for additional sensors, and the sensors themselves become potential points of failure. Furthermore, the sources of noise in wind turbines can be difficult to isolate, often leading to misdiagnosis of problems and unnecessary maintenance. In particular, operating conditions often influence vibration data, making it difficult to isolate noise generated at different locations within the wind turbine.

[0004] Therefore, there is a need for improved systems, methods, and computer program products that enable detection of vibrations in a wind turbine and diagnosis of component conditions based on the detected vibrations. Summary of the Invention

[0005] In an embodiment of the present invention, a system for monitoring the operation of a wind turbine including a rotor having blades is provided. The system includes one or more processors and a memory coupled to the one or more processors. The memory includes program code that, when executed by the one or more processors, causes the system to: receive a time-domain signal indicating a pitch force being applied to a blade; determine a first spectral density of the time-domain signal; determine a condition of the wind turbine based on the frequency content of the first spectral density; and generate an alarm signal in response to the condition indicating a problem with the wind turbine.

[0006] In one aspect of the invention, the time domain signal is indicative of the pressure of a fluid in a chamber of a hydraulic actuator of a pitch drive that controls the pitch of the blades.

[0007] In another aspect of the invention, the program code causes the system to determine the first spectral density by sampling the time domain signal to generate a discrete time domain signal; selecting a plurality of samples within a sampling window from the discrete time domain signal; and transforming the plurality of samples from the time domain to the frequency domain.

[0008] In another aspect of the invention, the rotor rotates in a rotor plane having a plurality of sectors, and the sampling window corresponds to a first period of time when the blade is in a selected sector of the plurality of sectors.

[0009] In another aspect of the invention, the blade passes through a horizontal position as the blade moves through the selected sector.

[0010] In another aspect of the present invention, the plurality of samples are selected such that each of the plurality of samples corresponds to a wind speed within a predetermined wind speed range or a rate of change of the wind speed within a predetermined rate of change of the wind speed range.

[0011] In another aspect of the invention, the samples are selected such that each sample of the plurality of samples corresponds to a pitch position within a predetermined pitch position range or a rate of change of the pitch position within a predetermined rate of change of the pitch position range.

[0012] In another aspect of the invention, the samples are selected such that each sample of the plurality of samples corresponds to a power output of the wind turbine within a predetermined power output range or a rate of change of the power output of the wind turbine within a predetermined rate of change of the power output range.

[0013] In another aspect of the invention, the rotor rotates in a rotor plane having the plurality of sectors, the sampling window is one of at least two sampling windows, the samples are selected such that each of the plurality of samples is within each of the at least two sampling windows, and the at least two sampling windows are selected from the following: a first sampling window corresponding to a first time period when the blade is in a selected sector of the plurality of sectors; a second sampling window corresponding to a wind speed within a predetermined wind speed range or a rate of change of the wind speed within a predetermined rate of change of the wind speed range; a third sampling window corresponding to a pitch position within a predetermined pitch position range or a rate of change of the pitch position within a predetermined rate of change of the pitch position range; and a fourth sampling window corresponding to a power output of the wind turbine within a predetermined power output range or a rate of change of the power output of the wind turbine within a predetermined rate of change of the power output range.

[0014] In another aspect of the invention, the samples are selected such that each sample of the plurality of samples corresponds to a load on the blade within a predetermined load range.

[0015] In another aspect of the invention, the program code causes the system to determine a condition of the wind turbine based on frequency content of the first spectral density by comparing the first spectral density to a second spectral density determined for the wind turbine during a second time period when the wind turbine is known to have been operating normally.

[0016] In another aspect of the invention, the program code causes the system to determine a condition of the wind turbine based on the first spectral density by: defining at least one frequency bin covering a portion of the first spectral density; determining one or more of a maximum amplitude, an average amplitude, and a minimum amplitude of the portion of the first spectral density covered by the at least one frequency bin; comparing the one or more of the maximum amplitude, the average amplitude, and the minimum amplitude of the portion of the first spectral density to corresponding alarm thresholds; and triggering an alarm if the one or more of the maximum amplitude, the average amplitude, and the minimum amplitude exceeds its corresponding alarm threshold.

[0017] In another aspect of the invention, the at least one frequency bin is one of a plurality of frequency bins, each of the plurality of frequency bins covering a different portion of the first spectral density, and the determining step, the comparing step, and the triggering step are performed for each frequency bin of the plurality of frequency bins.

[0018] In another aspect of the invention, the program code causes the system to determine a condition of the wind turbine based on the first spectral density by: defining at least one operating point having a frequency corresponding to a harmonic of the rotation of the rotor; determining one or more of a maximum amplitude, an average amplitude, and a minimum amplitude of the at least one operating point; comparing the one or more of the maximum amplitude, the average amplitude, and the minimum amplitude of the at least one operating point to corresponding alarm thresholds; and triggering an alarm if the one or more of the maximum amplitude, the average amplitude, and the minimum amplitude of the at least one operating point exceeds the alarm threshold.

[0019] In another aspect of the present invention, the at least one operating point is one operating point among a plurality of operating points, each of the plurality of operating points corresponding to a different harmonic of the rotation of the rotor, and the determining step, the comparing step, and the triggering step are performed for each of the plurality of operating points.

[0020] In another aspect of the present invention, the program code further causes the system to: in response to the first spectral density including a frequency component having a magnitude above a resonance threshold, enable a resonance control algorithm that suppresses resonance corresponding to the frequency component.

[0021] In another embodiment of the present invention, a method for monitoring operation of a wind turbine including a rotor having blades is provided. The method includes the steps of receiving a time domain signal indicative of a pitch force being applied to the blades; determining a first spectral density of the time domain signal; determining a condition of the wind turbine based on frequency content of the first spectral density; and generating an alarm signal in response to the condition indicating a problem with the wind turbine.

[0022] The above summary presents a simplified overview of some embodiments of the present invention to provide a basic understanding of certain aspects of the invention discussed herein. This summary is not intended to provide an extensive overview of the invention, nor is it intended to identify any key or important elements, or to delineate the scope of the invention. The sole purpose of this summary is simply to present some concepts in a simplified form as an introduction to the detailed description presented below. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate various embodiments of the invention and, together with the general description of the invention given above and the detailed description of the embodiments given below, serve to explain embodiments of the invention.

[0024] Figure 1 is a perspective view of an exemplary wind turbine including a nacelle and a rotor according to an embodiment of the present invention.

[0025] Figure 2 yes Figure 1A front view of a portion of a rotor showing the blades attached to the hub of the rotor by pitch bearings.

[0026] Figure 3 yes Figure 1 A perspective view of a portion of a wind turbine with a nacelle partially cut away to expose structure housed within the nacelle.

[0027] Figure 4 is a diagrammatic view of a control system including a pitch system that can be used to control Figures 1 to 3 The pitch of the blades in a wind turbine.

[0028] Figure 5 According to an exemplary embodiment of the present invention Figure 4 Schematic view of the pitch system.

[0029] Figure 6 According to another exemplary embodiment of the present invention Figure 4 Schematic view of the pitch system.

[0030] Figure 7 is a graph showing the use of sampling windows with different lengths to determine the indication Figure 1 Graph of the spectral density of signals of the pitch positions of individual blades of a wind turbine.

[0031] Figure 8 is a graph showing the use of different frequency scales to depict the indication from Figure 7 Graph of the spectral density of the signal of the pitch position of each blade.

[0032] Figure 9 is a graph showing the determination of the signal being applied to the Figure 1 FIG. 1 is a graph showing the spectral density of the pitch force signals of each blade of a wind turbine.

[0033] Figure 10 is a graph showing the use of different frequency scales to depict the indication from Figure 9 Graph of the spectral density of the pitch force signal being applied to each blade.

[0034] Figure 11 Is to show instructions Figure 9 Graph of the spectral density of a pitch force signal also shows frequency bins and operating points that can be used to analyze the spectral density of the pitch force signal.

[0035] Figure 12 It is shown that the embodiment of the present invention can be used to monitor Figure 1A diagrammatic view of the user interface of a wind turbine monitoring system.

[0036] Figure 13 is a diagrammatic view of a rotor plane comprising a plurality of sectors, Figure 1 The blades of the rotor rotate through these sectors.

[0037] Figure 14 It shows Figure 13 Graph showing the relationship between the blade azimuth angle in the rotor plane and the sampling window function.

[0038] Figure 15 It shows the wind speed and Figure 14 A graphical view of the relationship between the sampling window functions.

[0039] Figure 16 It shows Figure 1 The power output of a wind turbine is related to Figure 15 A graphical view of the relationship between the sampling window functions.

[0040] Figure 17 It shows Figure 14 The pitch position of the blades is Figure 16 A graphical view of the relationship between the sampling window functions.

[0041] Figure 18 is a graph showing the relationship between the pitch force signal and Figure 17 A graphical view of the relationship between the sampling window functions.

[0042] Figure 19 can be used to implement Figures 1 to 18 A diagrammatic view of one or more computers showing components or processes. DETAILED DESCRIPTION

[0043] It should be understood that the drawings are not necessarily drawn to scale and may present somewhat simplified representations of various features illustrating the basic principles of the invention. The specific design features of the operational sequences disclosed herein (e.g., including the specific dimensions, orientations, positions, and shapes of the various illustrated components) may be determined in part by the specific intended application and use environment. Certain features of the illustrated embodiments may be amplified or deformed relative to other features for ease of visualization and clear understanding. In particular, thin features may be thickened, for example, for clarity or illustration.

[0044] Embodiments of the present invention determine a condition of a wind turbine based on the amount of force applied to one or more blades of the wind turbine, thereby controlling the pitch of the blades. To this end, a time-domain signal representing the pitch force being applied to the blades of the wind turbine is received by a monitoring system. This signal can be directly related to the pitch force (e.g., a signal output by a strain gauge that measures the force being applied to the blade by a pitch drive) or can be indirectly related to the pitch force (e.g., a signal output by a pressure sensor that measures the pressure in the hydraulic actuator of the pitch drive). In either case, it has been determined that the frequency content of these types of signals can provide information about the operation of the wind turbine.

[0045] The spectral density of a time-domain signal can be obtained by converting the time-domain signal into a frequency-domain signal. A condition of a wind turbine or one or more components of a wind turbine can then be determined based on the spectral density. A sampling window can be applied to the signal in the time domain, which selects which portions of the signal are analyzed. Selectively filtering samples in the time domain based on various operating parameters has been determined to improve the ability of a monitoring system to detect certain conditions of a wind turbine, such as worn bearings or other components requiring maintenance.

[0046] Figure 1 An exemplary wind turbine 10 according to an embodiment of the present invention is shown. Wind turbine 10 includes a tower 12, a nacelle 14 disposed at the apex of tower 12, and a rotor 16 operably coupled to a generator in nacelle 14. In addition to generator 18, nacelle 14 typically houses various components required to convert wind energy into electrical energy and to operate and optimize the performance of wind turbine 10. Tower 12 supports the loads provided by nacelle 14, rotor 16, and other wind turbine components housed within nacelle 14. Tower 12 of wind turbine 10 elevates nacelle 14 and rotor 16 to a height above ground level that allows rotor 16 to spin freely and at which airflow with less turbulence and higher velocities is typically found.

[0047] The rotor 16 includes a hub 18 and one or more (e.g., three) blades 20 attached to the hub 18 at locations distributed around the circumference of the hub 18. The blades 20 project radially outward from the hub 18 and are configured to interact with the passing airflow to generate a rotational force that causes the hub 18 to spin about its longitudinal axis. This rotational energy is delivered to a generator housed within the nacelle 14 and converted into electrical energy. To optimize the performance of the wind turbine 10, the pitch of the blades 20 is adjusted by a pitch system in response to wind speed and other operating conditions.

[0048] Figure 21 is a front view of rotor 16 of wind turbine 10, illustrating pitch bearings 22 that operably couple blades 20 to hub 18. Each pitch bearing 22 has an axis of rotation that is generally aligned with the longitudinal axis of blade 20. Pitch bearings 22 are configured such that each blade 20 can rotate relative to hub 16 about the bearing's axis of rotation and transmit forces between hub 18 and blades 20. These forces include forces caused by gravity, centrifugal forces, wind loads, and loads provided by the generator.

[0049] The direction in which gravity acts on the pitch bearing 22 changes depending on the position of the blade 20, thereby generating a varying load 22 that tends to repeat with each rotation of the rotor 16. As the rotor 16 is rotating, the bearing 22 is also subjected to centrifugal forces that primarily generate axial tension in the pitch bearing 22. The forces generated by wind loads include those that rotate the rotor 16 and generally generate the greatest load on the pitch bearing 22. The pitch bearing 22 is configured to transfer these loads to the hub 18, which in turn transfers these loads to the rest of the wind turbine 10.

[0050] Changing the pitch of blades 20 generally changes the amount of lift and drag generated by blades 20 in response to wind, which changes the driving force provided by blades 20 to hub 18 and the lateral forces transmitted to tower 12. Thus, the pitch system can be used to help control wind turbine 10, optimize power production under varying wind conditions, and prevent damage from excessive wind.

[0051] Figure 3 A perspective view of the nacelle 14 is shown with the nacelle 14 partially cut away to expose the structure housed therein. The main shaft extending from the rotor 16 into the nacelle 14 can be held in place by a main bearing support 32, which supports the weight of the rotor 16 and transfers the load on the rotor 16 to the tower 12. The main shaft can be operably coupled to a gearbox 34, which transfers its rotation to a generator 36. As will be understood by one of ordinary skill in the art, the electricity generated by the generator 36 can be supplied to a grid (not shown) or an energy storage system (not shown) for subsequent release to the grid. In this way, the kinetic energy of the wind can be harnessed by the wind turbine 10 to generate electricity. The nacelle 14 can also house other equipment (not shown) used to operate the wind turbine 10, such as a hydraulic pump, a hydraulic accumulator, a cooling system, a controller, sensors, batteries, communication equipment, and the like.

[0052] The weight of the nacelle 14, including the components housed therein, can be carried by a load-bearing structure 38. The load-bearing structure 38 can include the outer shell of the nacelle 14 and one or more additional structural components (such as a frame or grid), as well as a gear bell that operably couples the load of the nacelle 14 to the tower 12 via a yaw bearing (not shown). The yaw bearing can be configured to allow the nacelle 14 to rotate into or out of the wind via the yaw system. The hub 18 can house at least a portion of a pitch system, including one or more pitch drives 40. Each pitch drive 40 can include one or more pitch actuators 42 (e.g., hydraulic cylinders, electric actuators, mechanical actuators, etc.) configured to provide a pitch force and operably coupled to corresponding blades 20 of the rotor 16 via linkages 44.

[0053] Figure 4 An exemplary control system 50 that can be used to control wind turbine 10 is illustrated. Control system 50 includes a wind turbine controller 52 in communication with a wind sensor 54, a pitch system 56, a yaw system 58, and a supervisory controller 60. Supervisory controller 60 can be configured to implement a system-wide control strategy for a group of wind turbines 10 (e.g., a wind farm) that optimizes the overall performance of wind turbines 10, for example, to maximize power production from the group of wind turbines and minimize overall maintenance. Yaw system 58 can be used by wind turbine controller 52 to control the direction in which nacelle 14 is pointed and can include one or more yaw controllers, drive systems, position sensors, etc. configured to implement a yaw command signal received from wind turbine controller 52. Pitch system 56 can be configured to collectively or individually adjust the pitch of blades 20 in response to the pitch command signal received from wind turbine controller 52. The mechanical force or “pitch force” required to pitch blades 20 may be provided by pitch drive 42 .

[0054] Figure 5 An exemplary pitch system 56 is illustrated and includes a pitch controller 70, a hydraulic actuator 72, and a hydraulic valve 74 (e.g., a proportional valve) coupling the hydraulic actuator 72 to a source of pressurized hydraulic fluid 76. The pitch controller 70 may be a stand-alone controller configured to control the pitch of one or more blades 20 of the wind turbine 10, or may be provided by another controller, such as the wind turbine controller 52.

[0055] Hydraulic actuator 72 includes a piston 78 located within a cylinder 80 that terminates at one end in a cylinder head 82 and at the other end in a cylinder head 84. Piston 78 is coupled to a piston rod 86 and divides the interior of cylinder 80 into a front chamber 88 (also referred to as the rod chamber) through which piston rod 86 passes, and a rear chamber 90 (also referred to as the bottom chamber) that terminates in cylinder head 82.

[0056] Piston rod 86 passes through a sealed opening in cylinder head 84 and includes a distal end operatively coupled to blade 20 via connecting rod 91. The connecting rod may include a ball joint, a pivot joint, or other joint that allows rotation about at least one axis between the distal end of piston rod 86 and blade 20. Movement of piston rod 86 may thereby cause blade 20 to rotate about the longitudinal axis of pitch bearing 22. Cylinder head 82 may be operatively coupled to hub 18 of rotor 16 (e.g., via another joint that allows rotation) such that movement of piston 78 causes angular displacement of blade 20 relative to hub 18. Pitch controller 70 thereby causes piston 78 to apply a pitch force to piston rod 86 in a forward direction (toward cylinder head 84) or in an aft direction (toward cylinder head 82) in response to actuation of hydraulic valve 74.

[0057] In response to receiving a signal from the pitch controller 70, the hydraulic valve 74 can selectively fluidly couple the output port of the fluid source 76 to one of the forward chamber 82 and the rearward chamber 90 of the hydraulic actuator 72, and can selectively fluidly couple the return port of the fluid source to the other of the forward chamber 82 and the rearward chamber of the hydraulic actuator 72. The pitch controller 70 can thereby control the flow of fluid between the fluid source 76 and the hydraulic actuator 72 via actuation of the hydraulic valve 74. The fluid source 76 can include one or more pumps, valves, accumulators, etc. configured to provide pressurized fluid. The fluid source 76 can be dedicated to the operation of a single hydraulic actuator 72, or can provide fluid to multiple hydraulic actuators 72 controlled by the pitch system 56.

[0058] The pitch system 56 may also include one or more sensors of a forward chamber pressure sensor 92, an aft chamber pressure sensor 94, and a position sensor 96. The forward chamber pressure sensor 92 is configured to sense the pressure of the fluid in the forward chamber 82 of the hydraulic actuator 72, or the fluid being supplied to the forward chamber. The aft chamber pressure sensor 94 is configured to sense the pressure in the aft chamber 86 of the hydraulic actuator 72, or the pressure being supplied to the aft chamber. Each pressure sensor 92, 94 may output a corresponding pressure signal 98, 100 indicative of the pressure sensed by the sensor. For example, each signal 98, 100 may have one or more characteristics (e.g., voltage, current, impedance, frequency, phase, etc.) that provide information indicative of the sensed pressure to the pitch controller 70.

[0059] Pitch position sensor 96 may be configured to provide a pitch position signal 102 indicative of the pitch position φ of blade 20 to pitch controller 70 in a manner similar to that described above with respect to pressure sensors 92, 94. For example, pitch position sensor 96 may be configured to measure the angular position of blade 20 using one or more sensors, such as optical sensors, magnetic sensors, or mechanical sensors, which may be used to generate a signal indicative of the pitch position φ of blade 20 to pitch controller 70.

[0060] Pitch controller 70 may use the pressure data received from pressure sensors 92, 94 to determine the pitch force being applied to blade 20 by hydraulic actuator 72. The pitch force may be determined, for example, using the following equation:

[0061] F P =P FC ×A FF -P RC ×A RF (Equation 1)

[0062] Among them, F P is the pitch force applied by the piston rod 86, P FC is the pressure in the antechamber 88, P RC is the pressure in the rear chamber 90, A FF is the effective area of ​​the piston 78 facing the front chamber 88, and A RF is the effective area of ​​the piston 78 facing the rear chamber 90. As can be seen from Equation 1, the pitch force F P A positive value of indicates that the piston rod 86 is pushing the blade 20, and the pitch force F P A negative value of indicates that the piston rod 86 is pulling the blade 20. Due to the presence of the piston rod 86, the effective area of ​​the piston 78 facing the rear chamber 90 is generally larger than the effective area of ​​the piston 90 facing the front chamber 88. Therefore, under certain operating conditions (e.g., when the piston 78 is not moving), the pressure of the fluid in the front chamber 88 may tend to be higher than the pressure of the fluid in the rear chamber 90. In some cases, a force sensor (not shown) may be used to directly measure the pitch force F P In this case, the pitch force F can be determined without pressure measurement P , or the force can be calculated based solely on the pressure measurement to check the operation of the force sensor.

[0063] Figure 6An exemplary pitch system 56 utilizing a mechanical actuator 110 according to an alternative embodiment of the present invention is illustrated that includes a motor 112 (e.g., an electric or hydraulic motor) operably coupled to a screw 114 that provides a pitch force to the blades 20. The screw 114 may include a cylindrical shaft 116 having a spiral ridge 118 and a threaded collar 120 including a bore having a spiral groove configured to engage the spiral ridge 118 of the cylindrical shaft 116. The screw 114 may be configured such that when the motor 112 rotates the cylindrical shaft 116, the collar 120 is urged longitudinally along the axis of the cylindrical shaft 116 in a direction dependent on the direction of rotation.

[0064] Collar 120 may be coupled to link 91 via connecting rod 122 so that the pitch force generated by screw 114 is operably coupled to blade 20, i.e., so that movement of collar 120 changes the pitch of blade 20. Pitch system 56 may also include a force sensor 124 (e.g., a strain gauge) that provides an indication to pitch controller 70 of the pitch force F being applied to blade 20 by mechanical actuator 110. P signal 126.

[0065] It should be understood that although pitch system 56 has been generally described with reference to pitch drive 40 including a single pitch actuator 42 per blade, the present invention is not limited thereto. Embodiments of the present invention may include pitch systems 56 having more than one pitch actuator per blade, as well as pitch actuators that rotate blades 20 relative to hub 18 along one or more axes of rotation.

[0066] One or more of the wind turbine controller 52, supervisory controller 60, pitch controller 70, or other suitable computing systems may be configured to execute a wind turbine monitoring process that collects and analyzes at least one time-domain signal indicative of a pitch force. The signal indicative of a pitch force may include a signal directly related to the pitch force (e.g., a signal provided by a force sensor), a signal indirectly related to the pitch force (such as the amount of energy provided to a pitch actuator) (e.g., a signal provided by a torque, current, voltage, or pressure sensor), or any other signal having a value related to the pitch force. The pitch monitoring process may sample the one or more time-domain signals indicative of the pitch force and store the sampled values, or "samples," in a memory as discrete time-domain signals suitable for analysis. For example, the pressure from one or more pitch actuators 42 of the pitch system 56 may be sampled at regular intervals (e.g., at a sampling frequency f that is above the Nyquist limit for any subsequent spectral analysis). S ) for sampling. Typical sampling frequency f SIt is possible to have a frequency greater than or equal to 50 Hz, which would potentially allow analysis of spectral content up to at least 25 Hz.

[0067] Samples indicative of pitch forces, or "pitch force samples," can be stored in memory local to wind turbine 10 where the data was generated and can also be uploaded to a central database for analysis. Thus, the pitch monitoring process enables measurement of time-varying characteristics of one or more signals indicative of pitch forces. These characteristics can include spectral content associated with one or more pitch axes and pitch actuators. For example, the pressure in the chambers of hydraulic cylinders of wind turbine 10 blades can be sampled and stored.

[0068] The spectral density of the signal indicative of the pitch force can be analyzed over a predetermined frequency band, such as a band between 0.2 Hz and 20 Hz. As used herein, the term "spectral density" refers to the relationship between the amplitude and frequency of a signal in the frequency domain, i.e., the power or energy distribution of the signal relative to frequency. The amplitude of the frequency domain signal can correspond to the amount of power or energy in the signal at the frequency in question or within a frequency range centered on the frequency in question.

[0069] Other operating parameters may also be measured, sampled, and stored for wind turbine 10. As described in more detail below, these operating parameters may be indexed to the pitch force samples and used to determine which pitch force samples to select for analysis and which pitch force samples to exclude from the analysis, e.g., ignore or discard.

[0070] Figures 7 to 10 Graphs 130 to 133 are depicted, which include the pitch position φ (graphs 130 and 131 ) and the pitch force F for each of the three blades 20 of the wind turbine 10 . p The exemplary amplitude versus frequency curves 140 to 146, 150-156 of (graphs 132 and 133) are for three blades of a wind turbine subjected to a wind speed V of approximately 18 meters per second (m / s) to 20 m / s. w and has a generating capacity of approximately 2.2 megawatts (MW). Each of graphs 130 through 133 includes a respective vertical axis 160-163 corresponding to amplitude and a horizontal axis 170-173 corresponding to frequency. The vertical axes 160, 161 of graphs 130, 131 are in degrees, the vertical axes 162-163 of graphs 132, 133 are in Newtons, and the horizontal axes 170-173 are in Hertz (Hz).

[0071] The spectral density of the pitch position φ shown by the graphs 130 and 131 is generated from the position versus time data using a fast Fourier transform (FFT) using a 4 second sampling window (graphs 140-142) and a 600 second sampling window (graphs 143-145). The pitch system in question includes a system having a frequency response f SD = 100 Hz servo drive that adjusts the pitch of each blade 20 based on the pitch command signal received from wind turbine controller 52 and pitch position signal 102. Curve 146 is an exemplary alarm threshold for the pitch position spectral density generated using a 4 second sampling window.

[0072] Pitch force spectral density shown in curves 132 and 133 is generated from force versus time data using FFT using a 4-second sampling window (curves 150-152) and a 600-second sampling window (curves 153-155) for the pitch system described above with respect to curves 130 and 131. Curve 156 is an exemplary alarm threshold for the pitch force spectral density generated using a 4-second sampling window.

[0073] It has been determined that relatively short sampling windows (e.g., 4 seconds) are generally useful for determining spectral content of position or force data at frequencies above 3 Hz, and particularly frequencies in the range of 3 Hz to 7 Hz. In contrast, relatively long sampling windows (e.g., 600 seconds) are generally useful for determining spectral content of frequencies below 3 Hz, such as frequencies associated with the rotation of the rotor 16. Frequencies corresponding to the rate at which the rotor 16 is rotating can be viewed as amplitude peaks centered around approximately 0.25 Hz (corresponding to a peak 180 of the fundamental frequency or first harmonic of rotation), 0.50 Hz (corresponding to a peak 181 of the second harmonic of rotation), 0.75 Hz (corresponding to a peak 182 of the third harmonic of rotation), 1.0 Hz (corresponding to a peak 183 of the fourth harmonic of rotation), and 1.25 Hz (corresponding to a peak 184 of the fifth harmonic of rotation).

[0074] It has further been determined that energy in a spectral region 190, generally centered around 3.0 Hz, is associated with vibrations at the blade edge. Frequencies generated by edge vibrations are particularly visible in curves 150-152 of pitch force graph 132, where distinct regions 190, 192 of increased amplitude are observed at 3.0 Hz and from 4.5 Hz to 7.0 Hz. This energy is believed to be generated by the blade's trailing edge and manifests itself as relatively high amplitudes in spectral region 190, centered around 3.0 Hz, and as broad peaks in spectral region 192, which begins at approximately 4.5 Hz and ends at approximately 7.0 Hz. The relatively high spectral density in these spectral regions 190, 192 can be associated with pitch actuation eigenfrequencies. The relatively high spectral density in these spectral regions 190, 192 can also be associated with the concurrence of hydraulic cylinder and blade inertia eigenfrequencies and blade eigenfrequencies.

[0075] The alarm threshold levels 146, 156 can be determined empirically, i.e., by collecting data about one or more wind turbines (e.g., the specific wind turbines to be monitored) that are known to be operating properly and under normal operating conditions. The alarm thresholds can then be set at sufficiently high limits to avoid alarms under these conditions, thereby preventing unnecessary downtime due to false alarms. Generally, short sampling window data may be useful for monitoring wind turbine 10, while long sampling window data may be useful for characterizing wind turbine operation and setting alarm limits.

[0076] In an embodiment of the present invention, frequency binning may be used to analyze the spectral density generated by the pitch force related data. Figure 11 A graph 132 illustrating an exemplary frequency binning is illustrated for the purpose of describing certain binning techniques that can be used for data analysis. In this exemplary embodiment, frequency bins are defined to include a low frequency bin 200 covering frequencies between 1 Hz and 4 Hz, a medium frequency bin 201 covering frequencies between 4 Hz and 8 Hz, and a high frequency bin 202 covering frequencies above 8 Hz. It should be understood that these frequency bins are merely exemplary and that the present invention is not limited to any particular size or number of frequency bins. Thus, any alternative frequency bins may be defined, for example, one bin per Hz or even overlapping frequency bins.

[0077] In an embodiment of the present invention, the maximum amplitude, average amplitude, and minimum amplitude of the spectral density can be determined for each frequency bin. For example, lines 203 and 204 indicate the maximum value in each of bins 200-202, respectively. The maximum amplitude, average amplitude, and minimum amplitude can also be determined for a plurality of "operating points" 206-210, each corresponding to a harmonic of the rotation of the rotor 16. For example, for a frequency bin with a rotation time t R= 4 seconds (i.e., a 360 degree rotation every four seconds) of the rotor 16, the frequency of the operating points 206-210 may be f x =x1 / t r 、x2 / t r 、x3 / t r 、x4 / t r 、x5 / t r , where x1 to x5 are integers (eg, 1, 2, 3, 4, 5). The operating point may be determined relative to the rotor speed for each time step so that if the angular speed of the rotor changes, the operating points 206-210 shift in frequency.

[0078] The maximum, minimum, and mean values ​​for each bin and each operating point can be determined and used to trigger an alarm to avoid controlling the bin value by averaging over time for a moving time window (e.g., 600 seconds). The maximum, minimum, and mean values ​​can each be compared to a corresponding limit, and an alarm is triggered if at least one of these values ​​exceeds (e.g., rises above or falls below) its limit.

[0079] Figure 12 An exemplary user interface 250 is illustrated, including a plurality of windows 252-255, each displaying data related to the operation of wind turbine 10. Each window includes graphs 260-263 showing values ​​of monitored parameters of wind turbine 10 versus time, and data boxes 268-271 showing statistical values ​​(e.g., maximum, minimum, mean, and standard deviation) associated with the data represented by the corresponding graphs 260-263. Exemplary parameters include: wind speed graph 260 showing wind speed versus time, graph 276 showing power output of wind turbine 10 versus time, graph 278 showing power output of wind turbine 10 versus time, pitch position graph 262 showing pitch position φ of each of three blades versus time, and pitch pressure graph 263 showing pitch pressure (e.g., front chamber pressure) of each of three blades versus time, graphs 286-288 showing pitch pressure (e.g., front chamber pressure) of each of three blades versus time.

[0080] Figures 7 to 11 The illustrated frequency domain data may be based on time domain data collected from sensors in wind turbine 10 and displayed by graphs 260-263 of user interface 250. Supervision limits may be set to make supervision robust against false alarms. Initially, a 600-second sampling window value may be used to generate a baseline for supervision values. For example, alarm limits may be set to trigger an alarm if the maximum amplitude level of an operating point or frequency band exceeds an alarm threshold. Exemplary alarm thresholds may be provided by:

[0081]

[0082] Among them, T alarm is a threshold of the measured or calculated amplitude, and β is a scaling value that depends on the parameter being monitored. Exemplary values ​​for the scaling value may be β=125 for the operating point and β=500 for the frequency band. Typically, the lowest frequency value in the frequency band will produce the highest alarm threshold, and this trend is reflected by the 1 / f factor. The alarm threshold may be active for a period exceeding the longest sampling window (e.g., ≥600 seconds). In response to the monitored parameter exceeding the alarm threshold, an alarm may be detected and supervision may be triggered.

[0083] The hydraulic actuator is used to generate the pitch force F p In embodiments of the present invention, the frequency content of the hydraulic fluid pressure versus time can be used to provide information about system resonance and blade stability. This data can be obtained, for example, by determining the frequency content versus time data of the pressure taken from a single chamber of the hydraulic cylinder. It has been determined that the sampling frequency f S =100 Hz to sample the sensor data to provide sufficient resolution to enable analysis of frequencies up to at least 10 Hz. This frequency content can provide an indication of controller stability, wear or fatigue of hardware components, condition of the hydraulic fluid (e.g., high air content variation frequency measurement), blade status, etc.

[0084] Changes in the amplitude of the monitored frequency or operating point can indicate changes in system performance and the robustness of the system design. For example, if the blade and pitch system each have a resonance at approximately the same frequency, the resonance may produce a peak in the spectral density near the resonance. It has been determined that, in general, frequency characteristics depend on certain fundamental characteristics of the wind turbine and can therefore provide useful information about how the wind turbine is performing.

[0085] The process of determining the spectral density can be performed using many different methods. For example, binning can be performed using a digital bandpass filter in the time domain, or using an FFT to convert the sampled signal from the time domain to the frequency domain. In any case, as can be seen by the difference in smoothness between a 4-second sampling window and a 600-second sampling window, the sampling window should be selected to maximize the utility of the resulting frequency data.

[0086] Frequency analysis of pitch force data can provide improved knowledge of components and systems at low cost. Where sensors already exist, the only cost may be an increase in the processing load on the controller or other computing device performing the analysis. However, much of this processing can occur during times when the processing load from operating the wind turbine is low, thereby limiting the impact on the computing devices in the wind turbine. p Additional parameters that are related to or otherwise affect the spectral content of the pressure data include: absolute direction of the wind, bandwidth of the control loop, wind speed, pitch output, and resonances in the blade edges.

[0087] Frequency analysis of the pressure, force, and position data provides a spectral fingerprint of wind turbine 10 under different operating conditions. This spectral fingerprint can enable the identification of problems in specific portions of pitch system 56 (e.g., the control loop) or other components of wind turbine 10 (such as the gearbox). For example, it has been determined that vibrations occurring in the 0.5 Hz to 3 Hz range are typically generated by pitch drive system 40. Therefore, changes in frequency content within this range can indicate wear or failure of components in pitch drive system 40.

[0088] Some sources of vibration in the blades of a wind turbine may depend on which rotation sector the blade is in, as well as other operating parameters of the wind turbine. For example, a failed pitch bearing may only cause certain vibrations in the pitch system when the blade in question passes through a certain rotation sector due to the effect of gravity. By configuring the sampling window to select pitch force samples collected from the blade when the blade is in one or more specified rotation sectors, embodiments of the present invention can isolate these vibration sources. In a preferred embodiment, the sampling window can also be adjusted based on other operating conditions so that only samples collected under conditions known to provide good data are used to determine the condition of the wind turbine. The sampling window can also be configured differently depending on the component being monitored in order to optimize the system's ability to determine the condition of the component in question. For example, by selecting samples that correspond to operating conditions that produce certain vibrations in the component.

[0089] Figure 13 A rotor plane 300 is shown perpendicular to the axis of rotation of the rotor 16, through which the blades 20 of the rotor 16 rotate. The position of each blade 20 in the rotor plane 300 can be determined by the azimuth angle θ of the blade. blade The azimuth angle of the blade is θ blade is relative to the reference angle θ ref In the embodiment shown, the reference angle θ ref It is defined by a vector 302 originating from the axis of rotation and pointing downwardly towards the rotor plane 300 and having an azimuth angle θblade The value of ref However, it should be understood that the reference angle θ ref The position and azimuth angle θ blade The direction of the increasing value is arbitrary, and any reference angle and rotation direction can be used.

[0090] The rotor plane 300 can be divided into a plurality of sectors 304-307 (e.g., four sectors), each defined by a range of azimuth angles θ. In the depicted embodiment, the sectors 304-307 include: a sector 304 having an azimuth angle between θ1 and θ2 (θ1 < θ < θ2), a sector 305 having an azimuth angle between θ2 and θ3 (θ2 < θ < θ3), a sector 306 having an azimuth angle between θ3 and θ4 (θ3 < θ < θ4), and a sector 307 having an azimuth angle between θ4 and θ1 (θ4 < θ < θ1), wherein for sector 304, θ1 = 0 degrees, and for sector 307, θ1 = 360 degrees. Although Figure 13 Four sectors are shown, but it will be understood that embodiments of the present invention are not limited to a particular number or size of sectors, and the rotor plane 300 may be divided into any number of sectors, each having any size.

[0091] The angular position of the rotor 16 can be determined by a device configured to measure the azimuth angle θ of the rotor 16 or the rotor azimuth angle θ rotor Because the azimuth angle θ of each blade 20, or the blade azimuth angle θ blade Relative to the rotor azimuth angle θ rotor is fixed, so the sector in which each blade 20 is located at a given time can be based on the rotor azimuth angle θ rotor to confirm.

[0092] For example, for the exemplary rotor plane 300 above, the data received from the sensors in the blades can be parsed into windows, with each window corresponding to one of the sectors 304-307 that the blade passes through during each complete 360-degree rotation. The data corresponding to each window can then be converted from the time domain to the frequency domain to isolate the frequency content of the signal for each sector. A different spectral fingerprint can then be applied to each sector to increase the resolution and accuracy of the monitoring process.

[0093] The origins of peaks in these spectral fingerprints may include resonant frequencies in blades 20 or components thereof, as well as resonances in the pitch drive system, which may depend on the pitch or angular position of rotor 16. Where the resonances of different components or systems align in frequency, they may constructively reinforce each other and thereby cause problems in the operation of the wind turbine.

[0094] To control oscillations in wind turbine 10, pitch controller 70 can be configured to implement a resonance control algorithm that suppresses resonance in response to detecting excess energy at a specific frequency or frequency band. Embodiments of this system can also be used to test wind turbine models. For example, each model being tested can be used to predict the spectral density that will be generated by an operating wind turbine. Models that accurately predict spectral densities similar to those measured can then be considered accurate models.

[0095] As mentioned above, including the indicated pitch force F p Samples of the discrete time-domain signal can be selected for analysis or excluded from analysis based on conditions at the time of sampling. For example, whether to sample during a period when a parameter exceeds a threshold. This selection can be performed by generating a sampling window that selects which pitch force samples to analyze based on the desired conditions.

[0096] Figure 14 The blade azimuth angle θ is shown. blade Graph 320 of blade azimuth angle versus time, graph 322, and sampling window graph 324 including graph 326 of sampling window S(t). Graph 320 includes graphs 326 of blade azimuth angle versus time. blade 328, and a horizontal axis 330, corresponding to time. Graph 324 includes a vertical axis 332, corresponding to whether a sample is selected for analysis of the discrete time domain signal (S=1) or excluded from analysis of the discrete time domain signal (S=0), and a horizontal axis 334, corresponding to time. Horizontal axes 330, 334 are aligned so that sampling window curve 326 illustrates sampling window S(t) versus blade azimuth angle θ. blade Graph 320 includes horizontal dashed lines 336-339, each of which corresponds to an azimuth angle θ threshold value that defines a boundary between sectors 304-307 of rotor plane 300 during which samples are selected for analysis and sectors 304-307 of rotor plane 300 during which samples are excluded from analysis. Vertical dashed lines 344-355 indicate times t1 to t2 when the azimuth angle θ of blade 20 crosses (i.e., exceeds) one of these threshold values. 12 , and illustrates how this defines a sampling window S(t), as shown by the sampling window curve 326.

[0097] Figure 15 Shown from Figure 14 The sampling window graph 324 and the wind speed graph 360. The wind speed graph 360 includes: w The vertical axis 362 corresponds to time t, the horizontal axis 364 corresponds to wind speed V w366 of the graph 360 versus time t. The horizontal axis 364 of the graph 360 is aligned with the horizontal axis 334 of the graph 324. The horizontal dashed line 368 indicates a wind speed threshold, above which wind speeds are considered too high to provide good data. Although not shown, embodiments of the present invention may also include a wind speed threshold, below which wind speeds are considered too low to provide good data. In either case, when the wind speed V w When a wind speed threshold is exceeded, samples can be excluded from the analysis via a sampling window S(t).

[0098] In the depicted example, vertical dashed lines 370, 372 indicate wind speed V w Time t when wind speed threshold is exceeded 13 , t 14 Wind speed V w At time t 13 exceeds the wind speed threshold, and at time t 14 falls below the threshold. Because at time t 13 , the blade 20 is in a rotation sector from which the sampling window S(t) has excluded sampling, so the wind speed V exceeding the wind speed threshold w There is no direct effect on the sampling window S(t). However, as shown in the sampling window curve 326 at time t4 and t 14 As shown in the gray area 374 between the blade 20 and the time t4, when the blade 20 exceeds the blade azimuth angle threshold corresponding to the time t4, the wind speed V w Still higher than the wind speed threshold, the sampling window curve 326 remains at 0 until the wind speed V w Time t below the wind speed threshold 15 .

[0099] Figure 16 Shown from Figure 15 The sampling window graph 324 and the power output graph 380 are shown. The power output graph 380 includes: wt The vertical axis 382 corresponds to time t, the horizontal axis 384 corresponds to the power output P wt 386 versus time t. The horizontal axis 384 of the graph 380 is aligned with the horizontal axis 334 of the graph 324, and the horizontal dashed line 388 indicates the power output threshold above which the sampling window S(t) excludes samples. The vertical dashed lines 390, 392 indicate the time t at which the power output of the wind turbine 10 exceeds the threshold. 15 , t 16 .

[0100] In the depicted example, the power output at time t 15 exceeds the power output threshold, and at time t16 falls below the threshold. Because at time t 15 , the blade 20 is in a rotation sector that typically includes samples in the analysis of discrete time domain signals, so the power output exceeding the power output threshold has a direct impact on the sampling window S(t). As can be seen from the sampling window curve 324 at time t 15 As seen in the gray area 394 between t and t3, when the power output of wind turbine 10 exceeds the power output threshold, sampling window S(t) is modified to exclude samples from the analysis of the discrete time domain signal. 16 Before falling back below the power output threshold, the blade 20 has exceeded the blade azimuth angle threshold corresponding to time t3, so the sampling window curve 326 remains at 0 until time t when the blade 20 enters the next sector to be analyzed. 14 .

[0101] Figure 17 Shown from Figure 16 324, and a pitch position graph 400. Pitch position graph 400 includes a vertical axis 402 corresponding to the pitch position φ of blade 20, a horizontal axis 404 corresponding to time t, and a graph 406 of pitch position φ versus time. Horizontal axis 406 of graph 400 is aligned in time with horizontal axis 334 of graph 324, and dashed horizontal line 408 indicates a pitch position threshold above which sampling window S(t) will exclude samples. Dashed vertical line 410 indicates the time t at which the pitch position φ of blade 20 exceeds the pitch position threshold. 17 .

[0102] In the depicted example, at time t 17 , the pitch position φ exceeds the pitch position threshold. Because at time t 17 , the blade 20 is in a rotation sector that typically includes samples in the analysis of discrete time domain signals, so the pitch position φ exceeding the pitch position threshold has a direct impact on the sampling window S(t). As can be seen from the sampling window curve 324 at time t 17 With t 11 The gray area 412 between t 12 The grey area 414 thereafter shows that when the pitch position φ of the blade 20 exceeds the pitch position threshold, the sampling window S(t) is modified so as to exclude samples from the analysis of the discrete time domain signal.

[0103] Figure 18 Shown from Figures 14 to 17 The sampling window graph 324 of FIG. 4 and the pitch force graph 420 . The pitch force graph 420 includes: the pitch force F being applied to the blade 20p The corresponding vertical axis 422, the horizontal axis 424 corresponding to time t, and the pitch force F p versus time graph 426. Horizontal axis 424 of graph 400 is aligned with horizontal axis 334 of graph 324, and vertical dashed lines 428-438 indicate when sampling window graph 326 changes value between "1" (selecting samples) and "0" (excluding samples).

[0104] The sampling window S(t) represented by curve 326 shows that the samples selected for analysis include those before t1, at time t2 and at time t 15 Between, at time t 14 Between time t5, between time t6 and time t7, between time t8 and time t9, and between time t 10 With time t 17 The portion of the pitch force signal represented by curve 426 that falls within the portion of the sampling window S(t) where samples are selected for analysis is shown as bold line segments 450-455. These line segments 450-455 correspond to the period where all selection parameters are met such that the value of the sampling window S(t) is "1."

[0105] The sampling window function S(t) can be implemented as a logical AND function with inputs according to a plurality of logical conditions, each of which defines a sampling window. θ (t)×S V (t)×S P (t)×S φ (t), where the blade azimuth angle θ blade When the blade azimuth angle threshold is exceeded, the blade azimuth angle sampling window S θ (t)=0, and when the blade azimuth angle θ blade When the blade azimuth angle is within the threshold, S θ (t)=1. Similarly, when the wind speed V w When the wind speed threshold is exceeded, the wind speed sampling window S V (t)=0, and when the wind speed V w When the wind speed is within the threshold, S V (t) = 1. When the power output P of the wind turbine 10 wt When the power output threshold is exceeded, the power output sampling window S P (t) = 0, and when the power output P of the wind turbine 10 wt When the power output is within the threshold, S P (t) = 1. When the pitch position φ of the blade 20 exceeds the pitch position threshold, the pitch position sampling window S φ(t) = 0, and when the pitch position φ of the blade 20 is within the pitch position threshold, S φ (t)=1.

[0106] It will be appreciated that other embodiments may use different combinations of operating parameters to generate sampling windows. Furthermore, in addition to the values ​​of the parameters, the rates of change of the parameter values ​​or the integrated parameter values ​​over a time interval may be used to generate sampling windows. Thus, the present invention is not limited to sampling windows based on any particular type or combination of operating parameters, or to sampling windows based on any particular type or combination of functions applied to the operating parameters.

[0107] Now refer to Figure 19 , the embodiments of the present invention described above, or portions thereof, may be implemented using one or more computer devices or systems, such as an exemplary computer 460. The computer 460 may include a processor 462, a memory 464, an input / output (I / O) interface 466, and a human-machine interface (HMI) 468. The computer 460 may also be operably coupled to one or more external resources 470 via a network 472 or the I / O interface 466. The external resources may include, but are not limited to, servers, databases, mass storage devices, peripheral devices, cloud-based network services, or any other resources that may be used by the computer 460.

[0108] Processor 462 may include one or more devices selected from the group consisting of a microprocessor, a microcontroller, a digital signal processor, a microcomputer, a central processing unit, a field programmable gate array, a programmable logic device, a state machine, a logic circuit, an analog circuit, a digital circuit, or any other device that manipulates signals (analog or digital) based on operating instructions stored in memory 464. Memory 464 may include a single memory device or multiple memory devices, including but not limited to read-only memory (ROM), random access memory (RAM), volatile memory, non-volatile memory, static random access memory (SRAM), dynamic random access memory (DRAM), flash memory, cache memory, or a data storage device such as a hard drive, an optical drive, a tape drive, a volatile or non-volatile solid-state device, or any other device capable of storing data.

[0109] The processor 462 may operate under the control of an operating system 474 resident in the memory 464. The operating system 474 may manage computer resources so that computer program code, embodied as one or more computer software applications, such as the application 476 resident in the memory 464, may have instructions executed by the processor 462.

[0110] I / O interface 466 may provide a machine interface that operably couples processor 462 to other devices and systems, such as external resources 470 or network 472. Applications 476 may also have program code executed by one or more external resources 470, or otherwise rely on functionality or signals provided by other systems or network components external to computer 460.

[0111] Database 480 may reside in memory 464 and may be used to collect and organize data used by the various systems and modules described herein. Database 480 may include data and supporting data structures that store and organize the data.

[0112] In general, the routines executed to implement embodiments of the present invention (whether implemented as part of an operating system or as a specific application, component, program, object, module, or sequence of instructions, or a subset thereof) may be referred to herein as "computer program code" or simply "program code." Program code typically includes computer-readable instructions that reside at various times in various memories and storage devices in a computer and, when read and executed by one or more processors in a computer, cause the computer to perform the operations or elements necessary to implement various aspects of the embodiments of the present invention.

[0113] The program code specifically implemented in any application / module described herein can be distributed individually or collectively as a computer program product in a variety of different forms. In particular, the program code can be distributed using a computer-readable storage medium having computer-readable program instructions thereon, which are used to cause a processor to perform various aspects of the embodiments of the present invention.

[0114] In some alternative embodiments, the functions, actions, or operations specified in a flow chart, sequence diagram, or block diagram can be reordered, processed serially, or processed concurrently according to embodiments of the present invention. Moreover, any one of a flow chart, sequence diagram, or block diagram can include more or fewer frames than those illustrated in accordance with embodiments of the present invention. It should also be understood that any combination of the individual frames in the block diagram or flow chart, or the frames in the block diagram or flow chart, can be implemented by a dedicated hardware-based system configured to perform a specified function or action, or performed by a combination of dedicated hardware and computer instructions.

[0115] Although the entire invention has been illustrated by the description of various embodiments, and although these embodiments have been described in considerable detail, it is not the intention of the applicant to define or in any way limit the scope of the appended claims to such details. Additional advantages and modifications will readily occur to those skilled in the art. Therefore, the invention in its broader aspects is not limited to the specific details, representative apparatus and methods, and illustrative examples shown and described. Therefore, changes may be made to these details without departing from the spirit or scope of the applicant's overall inventive concept.

Claims

1. A system for monitoring operation of a wind turbine (10), the wind turbine comprising a rotor (16) having blades (20) and rotating in a rotor plane (300) having a plurality of sectors (304-307), the system comprising: one or more processors (462); as well as a memory (464) coupled to the one or more processors (462) and comprising program code (476) that, when executed by the one or more processors (462), causes the system to: receiving a time domain signal (426) indicative of a pitch force being applied to the blade (20); Determining a first spectral density (150) of the time domain signal (426) by: sampling the time domain signal (426) to generate a discrete time domain signal; selecting a plurality of samples within a sampling window S(t) from the discrete time domain signal, the sampling window S(t) corresponding to a first time period when the blade (20) is in a selected sector of the plurality of sectors (304-307); and transforming the plurality of samples from the time domain to the frequency domain; determining a condition of the wind turbine (10) based on the frequency content of the first spectral density (150); as well as In response to the condition indicating a problem with the wind turbine (10), an alarm signal is generated.

2. The system according to claim 1, wherein: The time domain signal (426) indicates the pressure of a fluid in a chamber (90) of a hydraulic actuator (72) of a pitch drive device that controls the pitch of the blade (20).

3. The system according to claim 1, wherein: The plurality of samples are selected such that each of the plurality of samples corresponds to a wind speed within a predetermined wind speed range (V w ) or the wind speed (V w ) is within a predetermined rate of change within the wind speed range.

4. The system according to claim 1, wherein: The samples are selected such that each sample of the plurality of samples corresponds to a pitch position (φ) within a predetermined pitch position range or a rate of change of the pitch position (φ) within a predetermined rate of change of the pitch position range.

5. The system according to claim 1, wherein: The samples are selected such that each sample in the plurality of samples corresponds to a power output (P wt ) or the power output (P wt ) is within a predetermined rate of change of the power output range.

6. The system according to claim 1, wherein: The rotor (16) rotates in a rotor plane (300) having a plurality of sectors (304-307), the sampling window (S(t)) is one of at least two sampling windows (S(t)), the samples are selected such that each sample of the plurality of samples is within each sampling window (S(t)) of the at least two sampling windows (S(t)), and the at least two sampling windows (S(t)) are selected from the following: A first sampling window (S) corresponding to a first period of time when the blade (20) is in a selected sector of the plurality of sectors (304-307) θ (t)); With the wind speed within the predetermined wind speed range (V w ) or the wind speed (V w ) is within a predetermined change rate of the wind speed range corresponding to the second sampling window (S V (t)); A third sampling window (S) corresponding to a pitch position (φ) within a predetermined pitch position range or a rate of change of the pitch position (φ) within a predetermined rate of change of the pitch position range. φ (t)); and The power output (P) of the wind turbine (10) is within a predetermined power output range. wt ) or the power output (P wt ) is within a predetermined rate of change of the power output range corresponding to the fourth sampling window (S P (t)).

7. The system according to claim 1, wherein: The samples are selected such that each sample in the plurality of samples corresponds to a load on the blade (20) within a predetermined load range.

8. The system according to claim 1, wherein: The program code (476) causes the system to determine a condition of the wind turbine (10) based on the frequency content of the first spectral density (150) by: The first spectral density (150) is compared to a second spectral density determined for the wind turbine (10) during a second time period when the wind turbine (10) is known to have been operating normally.

9. The system according to claim 1, wherein: The program code (476) causes the system to determine a condition of the wind turbine (10) based on the first spectral density (150) by: defining at least one frequency bin (200-202) covering a portion of said first spectral density (150); determining one or more of a maximum amplitude, an average amplitude, and a minimum amplitude of the portion of the first spectral density (150) covered by the at least one frequency bin (200-202); comparing the one or more of the maximum amplitude, the average amplitude, and the minimum amplitude of the portion of the first spectral density (150) to corresponding alarm thresholds (156); as well as If one or more of the maximum amplitude, the average amplitude, and the minimum amplitude exceeds its corresponding alarm threshold (156), an alarm is triggered.

10. The system according to claim 9, wherein: The at least one frequency bin (200-202) is one frequency bin of a plurality of frequency bins (200-202), each of the plurality of frequency bins covering a different portion of the first spectral density (150), and the determining step, the comparing step, and the triggering step are performed for each frequency bin of the plurality of frequency bins (200-202).

11. The system according to claim 1, wherein: The program code (476) causes the system to determine a condition of the wind turbine (10) based on the first spectral density (150) by: defining at least one operating point (206-210) having a frequency corresponding to a harmonic of the rotation of the rotor (16); determining one or more of a maximum amplitude, an average amplitude, and a minimum amplitude of the at least one operating point (206-210); comparing said one or more of said maximum amplitude, said average amplitude, and said minimum amplitude of said at least one operating point (206-210) to corresponding alarm thresholds (156); as well as If one or more of the maximum amplitude, the average amplitude, and the minimum amplitude of the at least one operating point (206-210) exceeds the alarm threshold (156), an alarm is triggered.

12. The system according to claim 11, wherein The at least one operating point (206-210) is one operating point of a plurality of operating points (206-210), each of the plurality of operating points corresponding to a different harmonic of the rotation of the rotor (16), and the determining step, the comparing step, and the triggering step are performed for each operating point of the plurality of operating points (206-210).

13. The system according to claim 1, wherein: The program code (476) further causes the system to: In response to the first spectral density (150) including a frequency component having a magnitude above a resonance threshold, a resonance control algorithm is enabled that suppresses resonance corresponding to the frequency component.

14. A method for monitoring operation of a wind turbine (10), the wind turbine comprising a rotor (16) having blades (20) and rotating in a rotor plane (300) having a plurality of sectors (304-307), the method comprising the steps of: receiving a time domain signal (426) indicative of a pitch force being applied to the blade (20); Determining a first spectral density (150) of the time domain signal (426) by: sampling the time domain signal (426) to generate a discrete time domain signal; selecting a plurality of samples within a sampling window S(t) from the discrete time domain signal, the sampling window S(t) corresponding to a first time period when the blade (20) is in a selected sector of the plurality of sectors (304-307); and transforming the plurality of samples from the time domain to the frequency domain; determining a condition of the wind turbine (10) based on the frequency content of the first spectral density (150); as well as In response to the condition indicating a problem with the wind turbine (10), an alarm signal is generated.

Images