Based on its actual operation, the system uses the cumulative load histogram to operate the wind turbine.

Abstract

A method for operating a wind turbine includes determining one or more load and travel metrics or functions for one or more components of the wind turbine during operation. The method also includes using the one or more load and travel metrics during wind turbine operation to generate at least one distribution of cumulative load data for the one or more components. Furthermore, the method includes applying a life model of the one or more components to at least one distribution of the cumulative load data to determine the actual accumulated damage to the one or more components of the wind turbine to date. Moreover, the method includes implementing corrective actions on the wind turbine based on the accumulated damage.

Description

Technical Field

[0001] This disclosure generally relates to wind turbines, and more particularly to systems and methods for operating wind turbines using cumulative load histograms based on their actual operation. Background Technology

[0002] Wind power is considered one of the cleanest and most environmentally friendly energy sources available today, and wind turbines are receiving increasing attention in this area. A modern wind turbine typically consists of a tower, generator, gearbox, nacelle, and one or more rotor blades. The rotor blades are the primary components used to convert wind energy into electrical energy. For example, the blades typically have an airfoil cross-sectional profile, such that during operation, airflow across the blade creates a pressure difference between its sides. Therefore, lift, directed from the pressure side towards the suction side, acts on the blade. This lift generates torque on the main rotor shaft, which is connected to the generator to produce electricity.

[0003] Typically, wind turbines are designed to operate at a fixed nominal power output for a predetermined or anticipated operating life. For example, a typical wind turbine is designed for a 20-year lifespan. However, in many cases, this anticipated overall operating life is finite or based on the anticipated fatigue life of one or more of the wind turbine components. As used herein, the lifespan consumption or operating consumption of a wind turbine (which may include fatigue or extreme loads, wear, and / or other life parameters) generally refers to the lifespan of the wind turbine or its components that has been consumed or exhausted by previous operations. Therefore, for conventional wind turbines, various preventative maintenance actions are generally scheduled at predetermined intervals throughout the wind turbine's lifespan to prevent accelerated lifespan consumption that could occur without such maintenance actions.

[0004] However, the cost of such maintenance actions and the associated downtime are significant drivers of the total lifecycle cost of wind turbines and therefore should be optimized. Furthermore, wind turbines with higher operating volumes are more susceptible to under-maintenance and are at greater risk of unplanned low-quality events. Similarly, wind turbines with lower operating volumes are more susceptible to over-maintenance.

[0005] As an example, wind turbine components (such as rotor blades, towers, pitch bearings, gearboxes, etc.) have physics-based life models used during the design and site selection phases to ensure the components survive the expected lifespan of operation. Such models use component-specific input parameters and a specific life load histogram determined through simulated operation. Accordingly, the model is designed to capture complex and nonlinear mechanisms by which these loads act on and damage the physical materials of the component. The model can then output a metric of the predicted damage or reliability (probability of failure) over the turbine's lifespan.

[0006] The model output can then be compared to a threshold to determine whether the damage output is acceptable for the planned operation of the wind turbine. Generally, the threshold considers material strength, acceptable risk, and possible mitigation factors to account for uncertainties in the simulation and modeling. If the model output is less than the threshold, turbine operation is considered safe and acceptable. Using such models, the threshold for acceptable operation based on the model output can be translated into a damage threshold based on the damage also generated by the model.

[0007] Furthermore, wind turbines are simulated to determine their suitability for specific applications. Complex simulations cover the aeroelastic behavior of the entire wind turbine under controlled wind conditions expected throughout its design life. These simulations directly generate the load data needed to create load histograms used in component life models.

[0008] Therefore, improved systems and methods for operating wind turbines (e.g., actual operating data) are welcome in the art. Consequently, this disclosure relates to systems and methods for operating wind turbines using cumulative load histograms based on their actual operation rather than simulation data. Summary of the Invention

[0009] Aspects and advantages of the invention will be set forth in part in the description which follows, or may be apparent from the description, or may be learned by practicing the invention.

[0010] In one aspect, this disclosure relates to a method for operating a wind turbine. The method includes determining one or more load and travel metrics or functions relating to one or more components of the wind turbine during operation. The method also includes using one or more load and travel metrics during wind turbine operation to generate at least one distribution of cumulative load data relating to the one or more components, at least partially. Furthermore, the method includes applying a life model of the one or more components (such as a physics-based model, a statistical model, or a combination thereof) to at least one distribution of the cumulative load data to determine the actual cumulative damage to date relating to the one or more components of the wind turbine. Moreover, the method includes implementing corrective actions on the wind turbine based on the cumulative damage.

[0011] In one embodiment, load metrics relating to one or more components of a wind turbine may include, for example, bearing load, tower load, rotor blade load, drivetrain load, shaft load, and thermal load. In another embodiment, travel metrics may include, for example, time, angular travel, the number of times a component is started and stopped, the number of times a component travels an angular distance between reversals, or the number of stress cycles experienced by one or more components under a specific load.

[0012] In another embodiment, the components may include, for example, pitch bearings, yaw bearings, towers, gearboxes, generators, rotor blades, rotors, hubs, shafts, converters, fans, or nacelles, or any sub-components of these main components.

[0013] In additional embodiments, the function of one or more load and travel metrics may include average, minimum, maximum, standard deviation, median, quantile, etc.

[0014] In several embodiments, determining one or more load and travel metrics for one or more components of a wind turbine during operation may include measuring one or more load and travel metrics for one or more components of a wind turbine via one or more sensors or processors, or estimating one or more load and travel metrics for one or more components of a wind turbine via one or more processors.

[0015] In another embodiment, the method may include real-time estimation of one or more load and travel metrics for one or more components of a wind turbine. In a particular embodiment, the method may also include using machine learning to estimate one or more load and travel metrics for one or more components of a wind turbine.

[0016] In one embodiment, the processor may include a physics-based system model for estimating one or more load and travel metrics for one or more components of a wind turbine.

[0017] In some embodiments, the method may include determining multiple load and travel metrics for one or more components of the wind turbine during operation.

[0018] In yet another embodiment, at least one distribution used to at least partially generate cumulative load data for one or more components during wind turbine operation may include using one or more load and travel metrics to at least partially generate at least one cumulative load histogram for one or more components. For example, the use of histograms is advantageous compared to storing time-series data because storage requirements do not increase significantly over time. As used herein, distribution generally refers to any data structure, such as a histogram, data binning, or hash, that accumulates and represents the frequency and / or count of numerical data.

[0019] In another embodiment, generating at least one cumulative load histogram for one or more components using one or more load and travel metrics may include: defining the range and binregion of the at least one cumulative load histogram; defining a first load metric among a plurality of load metrics and a first travel metric among a plurality of travel metrics for the at least one cumulative load histogram; and during operation of the wind turbine, populating the at least one cumulative load histogram with the first load metric and the first travel metric by adding the first load metric and the first travel metric to the binregion of the cumulative load histogram.

[0020] In an embodiment, the method may further include adaptively changing at least one of the range or box region based on ongoing operational data of the wind turbine.

[0021] In an additional embodiment, the method may include applying a lifetime or damage model of the component to multiple cumulative load histograms collected over different time periods.

[0022] In another embodiment, the method may include processing the cumulative load histogram before applying a lifetime or damage model of the component.

[0023] In several embodiments, processing the cumulative load histogram before applying a lifetime model to the component may include redefining the range or bin area of ​​the cumulative load histogram, smoothing or recalibrating the cumulative load histogram to produce a more accurate estimate of historical load estimates, or scaling the cumulative load histogram to at least one of the desired durations predicted by the lifetime model.

[0024] In certain embodiments, corrective actions on wind turbines based on damage accumulation may include shutting down the wind turbine, idling the wind turbine, changing the wind turbine's power output, torque, speed, or other control parameters, and / or scheduling one or more preventative maintenance actions.

[0025] In another aspect, this disclosure relates to a system for operating a wind turbine. The system includes a controller configured to perform multiple operations, including, but not limited to: determining one or more load and travel metrics or functions relating to one or more components of the wind turbine during operation of the wind turbine; generating at least one cumulative load histogram relating to the one or more components using the one or more load and travel metrics, at least partially; applying a lifetime or damage model of the one or more components to at least one distribution of the cumulative load data to determine the actual cumulative damage to date relating to the one or more components of the wind turbine; and performing corrective actions on the wind turbine based on the cumulative damage.

[0026] It should be understood that the system can also be configured to have any of the features described herein.

[0027] Technical Solution 1. A method for operating a wind turbine, the method comprising:

[0028] During operation of the wind turbine, determine one or more load and stroke metrics or functions relating to one or more components of the wind turbine;

[0029] During the operation of the wind turbine, the one or more load and travel metrics are used to generate at least one distribution of cumulative load data for the one or more components, in at least partially.

[0030] The life model of the one or more components is applied to the at least one distribution of the cumulative load data to determine the actual cumulative damage to date for the one or more components of the wind turbine; and

[0031] Corrective actions are performed on the wind turbine based on the accumulated damage.

[0032] Technical Solution 2. The method according to Technical Solution 1, wherein the one or more load measures relating to one or more components of the wind turbine include at least one of bearing load, tower load, rotor blade load, transmission load, shaft load, and thermal load, and wherein the one or more travel measures include at least one of time, angular travel, the number of times the component travels a certain angular distance between directional reversals, or the number of times the one or more components undergo stress cycles under a specific load.

[0033] Technical Solution 3. The method according to Technical Solution 1, wherein the one or more components include at least one of a pitch bearing, a yaw bearing, a tower, a gearbox, a generator, rotor blades, a rotor, a hub, a shaft, a converter, or a nacelle.

[0034] Technical Solution 4. The method according to Technical Solution 1, wherein the function of the one or more load and travel measures further includes at least one of the average value, minimum value, maximum value, standard deviation, quantile or median.

[0035] Technical Solution 5. The method according to Technical Solution 1, wherein determining the one or more load and stroke metrics for one or more components of the wind turbine during operation further includes:

[0036] Measuring at least one of the one or more load and travel measurements of one or more components of the wind turbine via one or more sensors or processors, or estimating at least one of the one or more load and travel measurements of one or more components of the wind turbine via one or more processors.

[0037] Technical Solution 6. The method according to Technical Solution 5 further includes real-time estimation of the one or more load and travel metrics for one or more components of the wind turbine.

[0038] Technical Solution 7. The method according to Technical Solution 5 further includes using machine learning to estimate the one or more load and travel metrics for one or more components of the wind turbine.

[0039] Technical Solution 8. The method according to Technical Solution 5, wherein the processor includes a physics-based system model for estimating the one or more load and travel metrics for one or more components of the wind turbine.

[0040] Technical Solution 9. The method according to Technical Solution 1 further includes determining multiple load and stroke measurements for one or more components of the wind turbine during operation of the wind turbine.

[0041] Technical Solution 10. The method according to Technical Solution 1, wherein generating at least one distribution of cumulative load data for the one or more components using the one or more load and travel metrics during operation of the wind turbine further includes generating at least one cumulative load histogram for the one or more components using the one or more load and travel metrics.

[0042] Technical Solution 11. The method according to Technical Solution 10, wherein generating the at least one cumulative load histogram for the one or more components using the one or more load and travel metrics further comprises:

[0043] Define the range and bin area of ​​the at least one cumulative load histogram;

[0044] Define a first load metric among the plurality of load metrics and a first travel metric among the plurality of travel metrics for the at least one cumulative load histogram;

[0045] During the operation of the wind turbine, the first load metric and the first travel metric are used to populate the at least one cumulative load histogram by adding the first load metric and the first travel metric to the box area of ​​the at least one cumulative load histogram.

[0046] Technical Solution 12. The method according to Technical Solution 11 further includes adaptively changing at least one of the range or the box region based on the ongoing operating data of the wind turbine.

[0047] Technical Solution 13. The method according to Technical Solution 10 further includes applying the lifetime model of the one or more components to multiple cumulative load histograms collected over different time periods.

[0048] Technical Solution 14. The method according to Technical Solution 10 further includes processing the at least one cumulative load histogram before applying the lifetime model of the one or more components.

[0049] Technical Solution 15. The method according to Technical Solution 14, wherein processing the at least one cumulative load histogram before applying the lifetime model of the one or more components further includes redefining the range or bin area of ​​the at least one cumulative load histogram, smoothing or recalibrating the at least one cumulative load histogram to produce a more accurate estimate of historical load estimates, or scaling the at least one cumulative load histogram to at least one of the expected durations anticipated by the lifetime model.

[0050] Technical Solution 16. The method according to Technical Solution 1, wherein performing the corrective action on the wind turbine based on the accumulated damage further includes shutting down the wind turbine, idling the wind turbine, changing the control parameters of the wind turbine, or scheduling one or more preventive maintenance actions.

[0051] Technical Solution 17. A system for operating a wind turbine, the system comprising:

[0052] A controller configured to perform a plurality of operations, the plurality of operations including:

[0053] During operation of the wind turbine, determine one or more load and stroke metrics or functions relating to one or more components of the wind turbine;

[0054] Use the one or more load and travel metrics to at least partially generate at least one cumulative load histogram for the one or more components;

[0055] The life model of the one or more components is applied to at least one distribution of the cumulative load data to determine the actual cumulative damage to date for the one or more components of the wind turbine; and

[0056] Corrective actions are performed on the wind turbine based on the accumulated damage.

[0057] Technical Solution 18. The system according to Technical Solution 17, wherein the one or more load measures relating to one or more components of the wind turbine include at least one of bearing load, tower load, rotor blade load, transmission load, and shaft load, and wherein the one or more travel measures include at least one of time, angular travel, the number of times a component travels a certain angular distance between directional reversals, or the number of times the one or more components undergo stress cycles under a specific load.

[0058] Technical Solution 19. The system according to Technical Solution 17, wherein generating the at least one cumulative load histogram for the one or more components using the one or more load and travel metrics further includes:

[0059] Define the range and bin area of ​​the at least one cumulative load histogram;

[0060] Define a first load metric among the plurality of load metrics and a first travel metric among the plurality of travel metrics for the at least one cumulative load histogram;

[0061] During the operation of the wind turbine, the first load metric and the first travel metric are used to populate the at least one cumulative load histogram by adding the first load metric and the first travel metric to the box area of ​​the at least one cumulative load histogram.

[0062] Technical Solution 20. The system according to Technical Solution 19, wherein the plurality of operations further includes adaptively changing at least one of the range or the box region based on the ongoing operation data of the wind turbine.

[0063] These and other features, aspects, and advantages of the invention will become more readily understood with reference to the following description and the appended claims. The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and, together with the description, explain the principles of the invention. Attached Figure Description

[0064] The complete and feasible disclosure of the present invention (including its best mode) for those skilled in the art is set forth in the description with reference to the accompanying drawings, in which:

[0065] Figure 1 A perspective view showing one embodiment of a wind turbine according to the present disclosure;

[0066] Figure 2 A simplified interior view of one embodiment of the nacelle of a wind turbine according to the present disclosure is shown;

[0067] Figure 3 The illustration may include in Figure 1A schematic diagram of one embodiment of suitable components within the turbine controller of a wind turbine is shown in the figure;

[0068] Figure 4 A wind farm with multiple wind turbines is shown in accordance with this disclosure;

[0069] Figure 5 A block diagram illustrating an embodiment of an odometer-based control system for controlling a wind turbine connected to a power grid, according to the present disclosure;

[0070] Figure 6 A flowchart illustrating one embodiment of a method for operating a wind turbine according to the present disclosure;

[0071] Figure 7 A block diagram of one embodiment of a system for operating a wind turbine using a damage / life odometer, according to the present disclosure, is shown.

[0072] Figure 8 A block diagram of one embodiment of a system for generating an online life odometer for pitch bearings according to the present disclosure is shown.

[0073] Figure 9 A diagram illustrating an embodiment of fatigue accumulation over time with and without odometer-based control activated, according to the present disclosure;

[0074] Figure 10 The diagram illustrates an embodiment of fatigue accumulation over time with and without odometer-based control, according to this disclosure, particularly showing the lifespan extension provided by the odometer-based control; and

[0075] Figure 11 An embodiment of a cumulative load histogram according to this disclosure is shown. Detailed Implementation

[0076] Reference will now be made in detail to embodiments of the invention, one or more examples of which are illustrated in the accompanying drawings. Each example is provided by way of explanation rather than limitation of the invention. Indeed, it will be apparent to those skilled in the art that various modifications and variations may be made to the invention without departing from the scope or spirit thereof. For example, a feature shown or described as part of one embodiment may be used with another embodiment to produce yet another embodiment. Therefore, it is intended that the invention cover such modifications and variations and their equivalents as fall within the scope of the appended claims.

[0077] Generally, this disclosure relates to odometer-based supervisory control of wind turbines. For example, in embodiments, when a wind turbine is operating, a fatigue / damage odometer can be used, for instance, to estimate and track cumulative damage to turbine components relative to their failure modes. Supervisory control regulates the operation of the wind turbine to achieve long-term operational objectives. Operational objectives may include maximizing energy or revenue output while keeping extreme loads and cumulative damage metrics within limits. Alternatively, operational objectives may include minimizing one or more cumulative damage metrics while keeping extreme loads and other cumulative damage metrics within limits. Such fatigue / damage odometers can be developed, for example, using histograms populated with actual operating data of the wind turbine.

[0078] This disclosure offers numerous advantages not present in the prior art. For example, the use of histograms is beneficial because storage requirements do not increase significantly over time. Furthermore, this disclosure can be implemented within turbine controller hardware, on a different computer at the wind turbine site, or on a network-connected computer at the wind farm or other nearby or remote locations. Additionally, this disclosure can utilize readily available / existing operational data and does not necessarily require the collection of new or additional data (new or additional sensors may be utilized if desired). Moreover, this disclosure is applicable to any wind turbine, regardless of model, design, size, or manufacturer. Furthermore, the systems and methods of this disclosure enable the full utilization of mechanical design and load margins for each wind turbine based on how each turbine actually operates and the conditions a particular turbine actually experiences.

[0079] Now refer to the attached diagram, Figure 1 This is a perspective view showing one embodiment of a wind turbine 10 configured to implement control technology according to this disclosure. As shown, the wind turbine 10 generally includes a tower 12 extending from a support surface 14, a nacelle 16 mounted on the tower 12, and a rotor 18 coupled to the nacelle 16. The rotor 18 includes a rotatable hub 20 and at least one rotor blade 22 coupled to and extending outward from the hub 20. For example, in the illustrated embodiment, the rotor 18 includes three rotor blades 22. However, in an alternative embodiment, the rotor 18 may include more or fewer than three rotor blades 22. Each rotor blade 22 may be spaced apart around the hub 20 to allow rotation of the rotor 18 so that kinetic energy can be converted from wind energy into usable mechanical energy, and subsequently into electrical energy. For example, the hub 20 may be rotatably coupled to a generator positioned within the nacelle 16. Figure 2 This allows for the generation of electrical energy.

[0080] The wind turbine 10 may also include a wind turbine controller 26 centralized within the nacelle 16. However, in other embodiments, the controller 26 may be located within any other component of the wind turbine 10 or at a location outside the wind turbine. Furthermore, the controller 26 may be communicatively coupled to any number of components of the wind turbine 10 to control the operation of such components and / or perform corrective actions. Accordingly, the controller 26 may include a computer or other suitable processing unit. Therefore, in several embodiments, the controller 26 may include suitable computer-readable instructions that, when implemented, configure the controller 26 to perform various functions, such as receiving, sending, and / or executing wind turbine control signals.

[0081] Therefore, controller 26 can be generally configured to control various operating modes of wind turbine 10 (e.g., start-up or shutdown sequence), adjust control parameters of wind turbine 10 to change power output, and / or control various components of wind turbine 10. For example, controller 26 can be configured to control the blade pitch or pitch angle (i.e., the angle that determines the viewpoint of rotor blade 22 relative to the direction of the wind) of each of the rotor blades 22 to control the power output generated by wind turbine 10 by adjusting the angular position of at least one rotor blade 22 relative to the wind. For example, by transmitting appropriate control signals to the pitch drive or pitch adjustment mechanism (not shown) of wind turbine 10, controller 26 can control the pitch angle of rotor blade 22 by rotating rotor blade 22 individually or simultaneously about pitch axis 28.

[0082] Now for reference Figure 2 , shown in Figure 1 The diagram shows a simplified internal view of one embodiment of the nacelle 16 of the wind turbine 10. As shown, a generator 24 may be coupled to a rotor 18 to generate electricity from the rotational energy generated by the rotor 18. For example, as shown in the illustrated embodiment, the rotor 18 may include a rotor shaft 34 coupled to a hub 20 for rotation therewith. The rotor shaft 34 is in turn rotatably coupled to a generator shaft 36 of the generator 24 via a gearbox 38. As generally understood, in response to rotation of the rotor blades 22 and the hub 20, the rotor shaft 34 may provide a low-speed, high-torque input to the gearbox 38. The gearbox 38 may then be configured to convert the low-speed, high-torque input into a high-speed, low-torque output to drive the generator shaft 36 and thus the generator 24.

[0083] Each rotor blade 22 may also include a pitch adjustment mechanism 32 configured to rotate each rotor blade 22 about its pitch axis 28. Furthermore, each pitch adjustment mechanism 32 may include a pitch drive motor 40 (e.g., any suitable electric, hydraulic, or pneumatic motor), a pitch drive gearbox 42, and a pitch drive pinion 44. In such embodiments, the pitch drive motor 40 may be coupled to the pitch drive gearbox 42 such that the pitch drive motor 40 applies mechanical force to the pitch drive gearbox 42. Similarly, the pitch drive gearbox 42 may be coupled to the pitch drive pinion 44 for rotation therewith. The pitch drive pinion 44 may in turn rotatably engage with a pitch bearing 46 connected between the hub 20 and the corresponding rotor blade 22, such that rotation of the pitch drive pinion 44 causes rotation of the pitch bearing 46. Therefore, in such embodiments, the rotation of the pitch drive motor 40 drives the pitch drive gearbox 42 and the pitch drive pinion 44, thereby rotating the pitch bearing 46 and the rotor blades 22 about the pitch axis 28. Similarly, the wind turbine 10 may include one or more yaw drive mechanisms 66 communicatively coupled to the controller 26, wherein each pitch drive mechanism 66 is configured to change the angle of the nacelle 16 relative to the wind (e.g., by engaging the yaw bearing 68 of the wind turbine 10).

[0084] Now for reference Figure 3 This diagram illustrates a block diagram of one embodiment of suitable components that may be included within a controller, according to various aspects of this disclosure. It should be understood that... Figure 3 Various components of the controller can be applied to any suitable controller, including, for example, the turbine controller 26, the field-level controller 56, and / or the supervisory controller 102 described herein.

[0085] As shown, the controller may include one or more processors 58 and associated memory devices 60, configured to perform a variety of computer-implemented functions (e.g., performing the methods, steps, calculations, etc. disclosed herein). As used herein, the term "processor" refers not only to an integrated circuit known in the art as contained in a computer, but also to a controller, microcontroller, microcomputer, programmable logic controller (PLC), application-specific integrated circuit, and other programmable circuits. Additionally, memory devices 60 may generally include memory elements, including but not limited to computer-readable media (e.g., random access memory (RAM)), computer-readable non-volatile media (e.g., flash memory), floppy disks, compact disc read-only memory (CD-ROM), magneto-optical disks (MOD), digital versatile discs (DVDs), and / or other suitable memory elements.

[0086] Additionally, the controller may include a communication module 62 to facilitate communication between the controller and various components of the wind turbine 10. For example, the communication module 62 may include a sensor interface 64 (e.g., one or more analog-to-digital converters) to allow signals transmitted by one or more sensors 65, 66, 67 to be converted into signals that can be understood and processed by the controller. It should be understood that sensors 65, 66, 67 can be communicatively connected to the communication module 62 using any suitable means. For example, such as... Figure 3 As shown, sensors 65, 66, and 67 are connected to sensor interface 64 via a wired connection. However, in other embodiments, sensors 65, 66, and 67 may be connected to sensor interface 64 via a wireless connection, such as by using any suitable wireless communication protocol known in the art. Accordingly, processor 58 may be configured to receive one or more signals from sensors 65, 66, and 67.

[0087] The sensors 65, 66, and 67 of the wind turbine 10 can be any suitable sensor configured to measure any operating conditions and / or wind parameters at or near the wind turbine. For example, sensors 65, 66, and 67 may include blade sensors for measuring the pitch angle of one of the rotor blades 22 or for measuring the load acting on one of the rotor blades 22; generator sensors for monitoring the generator (e.g., torque, rotational speed, acceleration, and / or power output); and / or various wind sensors for measuring various wind parameters. Additionally, sensors 65, 66, and 67 may be located near the ground, on the nacelle, or on a weather mast near the wind turbine.

[0088] It should also be understood that any other number or type of sensors may be employed and in any location. For example, sensors may be analog sensors, digital sensors, optical / visual sensors, accelerometers, pressure sensors, angle-of-attack sensors, vibration sensors, MIMU sensors, fiber optic systems, temperature sensors, wind sensors, sonic detection and ranging (SODAR) sensors, infrared lasers, light detection and ranging (LIDAR) sensors, radiometers, pitot tubes, radiosonde anemometers, and / or any other suitable sensors. It should be understood that various sensors, as used herein and referred to by the term "monitoring" and its variations indicating wind turbines, may be configured to provide direct or indirect measurements of the monitored parameters. Thus, sensors 65, 66, and 67 may, for example, be used to generate signals associated with the monitored parameters, which can then be utilized by the controller to determine the actual conditions.

[0089] As mentioned, processor 58 is configured to perform any of the steps of the method according to this disclosure. For example, processor 58 may be configured to determine operating parameters of wind turbine 10. As used herein, "operating parameters" generally refers to the number of operating seconds, minutes, hours, etc., of wind turbine 10 and / or its various components operating under various operating parameters and / or specific conditions. Such operating parameters that may be considered or tracked may include, for example, one or more of the following: power output, torque, pitch angle, load conditions, generator speed, rotor speed, wind direction, air density, turbulence intensity, gusts, wind shear, wind speed, updraft, yaw, pitch, or temperature. Moreover, operating data may include sensor data, historical wind turbine operating data, historical wind farm operating data, historical maintenance data, historical quality problems, or combinations thereof. Therefore, processor 58 may also be configured to record and store operating parameters in memory 60 for later use. For example, processor 58 may store operating parameters in one or more lookup tables (LUTs). Moreover, operating parameters may be stored in the cloud.

[0090] Now for reference Figure 4 The systems and methods described herein can also be combined with the wind farm controller 56 of the wind farm 50. As shown, the wind farm 50 may include a plurality of wind turbines 52, including the aforementioned wind turbine 10. For example, as shown in the illustrated embodiment, the wind farm 50 includes twelve wind turbines, including wind turbine 10. However, in other embodiments, the wind farm 50 may include any other number of wind turbines, such as fewer than twelve or more than twelve wind turbines. In one embodiment, the controller 26 of the wind turbine 10 may be communicatively coupled to the field controller 56 via a wired connection, such as by connecting the controller 26 via a suitable communication link 57 (e.g., a suitable cable). Alternatively, the controller 26 may be communicatively coupled to the field controller 56 via a wireless connection, such as by using any suitable wireless communication protocol known in the art.

[0091] In several embodiments, one or more of the wind turbines 52 in the wind farm 50 may include multiple sensors for monitoring various operating parameters / conditions of the wind turbines 52. For example, as shown, one of the wind turbines 52 includes a wind sensor 54 configured to measure wind speed, such as an anemometer or any other suitable device. As generally understood, wind speed can vary significantly throughout the wind farm 50. Therefore, the wind sensor 54 may allow monitoring of local wind speed at each wind turbine 52. Additionally, the wind turbines 52 may also include additional sensors 55. For example, the sensor 55 may be configured to monitor the electrical characteristics of the output of the generator of each wind turbine 52, such as a current sensor, voltage sensor, temperature sensor, or a power monitor that monitors power output directly based on current and voltage measurements. Alternatively, the sensor 55 may include any other sensors that can be used to monitor the power output of the wind turbines 52. It should also be understood that the wind turbines 52 in the wind farm 50 may include any other suitable sensors known in the art for measuring and / or monitoring wind conditions and / or wind turbine conditions.

[0092] Now for reference Figure 5-11 According to aspects of this disclosure, various features of several embodiments of an odometer-based control (OBC) system 100 and method 200 for operating a wind turbine (such as wind turbine 10) are presented. More specifically, Figure 5 A block diagram of one embodiment of a system 100 for controlling a wind turbine 10 according to this disclosure is shown. For example, as shown, the OBC system 100 includes a turbine controller 26 and a supervisory controller 102 communicatively coupled to the turbine controller 26. Specifically, as shown, the turbine controller 26 may be part of an internal control loop 103 that includes a wind turbine control system. In such embodiments, the wind turbine 10 may include one or more sensors, such as electrical components and accelerometers, read by the turbine controller 26. Therefore, the turbine controller 26 is configured to send control signals to actuators such as blade pitch motors and nacelle yaw motors. The actuators affect the dynamic behavior of the turbine. Therefore, the turbine controller 26 is configured to operate the wind turbine 10 to generate electricity while preventing undesirable or destructive behavior. Furthermore, the turbine controller 26 may have a number of operating parameters that can be set to change or regulate turbine performance. During turbine operation, operating parameters typically remain constant or change in a particular manner; however, as will be described herein, the OBC system 100 can modify some of these parameters using optimized strategies or control functions 116, as described below. Such operating parameters may include, for example, torque setpoint, speed setpoint, thrust limit, and / or parameters controlling when active pitch control of rotor imbalance is enabled. Furthermore, the turbine controller 26 is configured to determine a state estimate 105 of the wind turbine 10, which is explained in more detail below.

[0093] Still referencing Figure 5 The supervisory controller 102 may include an intermediate control loop 104 and an external control loop 106. The intermediate control loop 104, or supervisory parameter control loop, is configured to improve turbine control and performance by changing turbine controller operating parameters based on estimates of conditions such as wind conditions or grid conditions. Specifically, as shown, the intermediate control loop 104 may include a condition estimator module 108, a dynamic function module 110, one or more damage odometers 118, and a power estimator module 120.

[0094] Therefore, in some embodiments, the condition estimator module 108 is configured to estimate or predict the values ​​of the condition parameters described herein based on the state estimate 105 and possibly external measurements. These condition parameters are generally referred to herein as the current condition 112. In other words, the current condition 112 generally refers to the current estimate or prediction of the condition parameters in the set of condition parameters, which is described in more detail below.

[0095] Additionally, as shown, damage odometer 118 is configured to estimate the damage levels of various wind turbine components based on the condition estimate 105 of the wind turbine 10. Generally, each damage level corresponds to a specific component and a specific failure mode of that component. Accordingly, the value of the damage level represents cumulative damage. Examples include blade root fatigue, tower base fatigue, and pitch bearing fatigue, but any number of damage levels can be generated and considered. In wind turbine design and site selection, damage (generally fatigue) limits are typically determined to ensure safe and reliable operation. Therefore, these limits may be based on damage models of specific building materials, manufacturing quality, and stress cycle counts, as well as Goodman or similar damage curves.

[0096] Furthermore, the damage odometers described herein can utilize the history of state estimates from turbine controller 26 and models of wind turbine 10 and its components to determine the damage to the turbine based on actual operation of the turbine during its elapsed lifespan. Therefore, the damage level for each component can be determined within a damage level set. In some embodiments, different damage odometers may exist for different parts of wind turbine 10, and multiple damage odometers may exist for a single part or component of wind turbine 10, each associated with a different failure mode or wear mechanism. For example, many damage odometers may be associated with crack propagation and fatigue failure modes of structural materials. In some embodiments, the damage odometers may be implemented based on state estimates 105 from turbine controller 26, where the state estimates include instantaneous loads and forces on wind turbine 10. In alternative embodiments, the damage odometers may also be based on specialized sensors such as strain gauges (not shown). Moreover, in addition to the damage level, the damage odometers may also generate uncertainty levels for each damage parameter used by function design module 114.

[0097] In another embodiment, damage assessment can utilize a fusion of damage odometer readings and validated diagnostic algorithms, based on real-time operating signals or a condition monitoring system. For example, if internal signals from the pitch motor or generator acceleration signals indicate an operational anomaly, the damage assessment will be high, even if the odometer reading is low. Therefore, when addressing the same failure mode, the OBC system 100 can use both to assess real-time damage.

[0098] Additionally, in this embodiment, the power estimator module 120 is configured to calculate statistics on the power output of the wind turbine 10. In some embodiments, this calculation may simply be the cumulative energy generated, but it may also include other cumulative statistics or statistics calculated for each dynamic control interval. Thus, the power statistics 136 can be used to externally evaluate turbine performance or to learn the power output performance of the wind turbine 10.

[0099] Still referencing Figure 5 The current condition 112 from the condition estimator module 108 can then be used by the dynamic function module 110 to set the operating parameters for the next dynamic control interval. The condition estimator module 108 can also simply generate an estimate (not a prediction) of the current condition parameters. This is sufficient if the dynamic control interval is relatively short, such as 30 seconds or one minute. The condition estimator module 108 can also generate predictions of the condition parameters for the next dynamic control interval. Predicting future values ​​of the condition parameters becomes more desirable when the dynamic control interval is longer.

[0100] Therefore, the dynamic function module 110 is configured to change the operating parameters of the turbine controller 26 based on the current condition 112. In some embodiments, the dynamic function module 110 may include a lookup operation, wherein the current condition is received and used to determine the operating parameters according to a strategy table. Therefore, in some embodiments, the dynamic function module 110 changes the operating parameters for each dynamic control interval. As used herein, a dynamic control interval generally refers to the duration for which the dynamic function module 110 sets the operating parameters.

[0101] Furthermore, in certain embodiments, the external control loop 106 or the odometer-based control loop also includes a function design module 114 for generating a strategy or control function 116. Additionally, the function design module 114 is configured to update the control function 116 based on the current fatigue damage state, anticipated future conditions, the turbine's system model, the planned operating time domain, and the possible anticipated future values ​​of the generated power.

[0102] In an embodiment, for example, control function 116 defines the relationship between the set of conditional parameters and multiple operating parameters of the wind turbine 10. More specifically, control function 116 may be a mapping from conditional parameters in the set of conditional parameters to operating parameters in the set of operating parameters, which is essentially performed by dynamic function module 110. Thus, in an embodiment, control function 116 may be a lookup table, an interpolation function, or have any other functional form. In one embodiment, control function 116 may be a constant strategy, whereby the operating parameters are the same for any value of the conditional parameters. This is equivalent to having no intermediate loops. Another simple example of control function 116 is an example that specifies a higher rated torque (for greater power) when the air density is below a threshold and the wind speed is less than 15 m / s. But generally, control function 116 is an arbitrary function of the current condition 112 and can be complex. Therefore, in many embodiments, control function 116 may be determined by an optimization process in the function design module 114 box.

[0103] Furthermore, in some embodiments, the control function 116 may be defined as discrete values ​​of condition parameters, and the current condition 112 may be continuous. In this case, the dynamic function module 110 may select the closest entry in the control function 116 based on the current condition 112 to determine the operating parameters, or the dynamic function module 110 may interpolate the operating parameters. Accordingly, the dynamic function module 110 receives the control function 116 from the function design module 114. Moreover, the dynamic function module 110 determines the operating parameters and sends them to the turbine controller 26 to dynamically control the wind turbine 10 within multiple dynamic control intervals based on the current condition 112 and the control function 116.

[0104] Still referencing Figure 5 The function design module 114 is configured to determine the control function 116 through an optimization process, for example, using a conditional distribution 122, a model 124 of the operational behavior of the wind turbine 10, the design life 126 of the wind turbine 10 (e.g., the total planned operating time of the wind turbine 10), the elapsed life 128 of the wind turbine 10 (e.g., the amount of time the wind turbine 10 has been operating up to this point in its design life), one or more damage limits 130, one or more damage levels 134 (e.g., from a damage odometer), or a future value discount 132 (e.g., future monetary value used for value optimization). This information represents the current state of the wind turbine 10 and the expected future conditions that the wind turbine 10 will experience, as well as a model of its operational behavior. Given these inputs, there are various optimization methods available to generate the control function 116.

[0105] As used herein, a conditional distribution generally refers to the expected future distribution of conditional parameters in a set of conditional parameters. In one embodiment, for example, the conditional distribution is a joint probability distribution of the conditional parameters. Thus, in an embodiment, if the set of conditional parameters consists of wind speed and turbulence intensity, the conditional distribution could be a joint distribution of these two values. For practical reasons, the conditional distribution could be changed to an independent distribution for each conditional parameter, or some conditional parameters could be assumed to be constant.

[0106] In some embodiments, a conditional distribution can be established for the wind turbine and wind farm before commissioning using, for example, meteorological mast data. Furthermore, the conditional distribution can be fixed and invariant while the wind turbine 10 is operating, or it can adaptively learn and update over time.

[0107] In another embodiment, the model 124 of the operating behavior of the wind turbine 10 may include a table of condition parameters, operating parameters, power, and / or damage (referred to herein as COPD table 124), i.e., a model of the power generated and the damage caused to the wind turbine 10, depending on the condition parameters and / or operating parameters. Thus, in such embodiments, model 124 may be a representation of the turbine's operating behavior, modeling various aspects of the turbine system response required by the OBC system 100. In one embodiment, for example, the model may be a mapping from condition parameters and operating parameters to expected power statistics 136 that the wind turbine 10 will generate and the expected increments in the damage values ​​caused to turbine components during a dynamic control interval. In other words, if wind conditions and grid conditions are known and the operating parameters to be selected are known, then the model / table provides the amount of power that the wind turbine 10 will generate and the amount of damage that the wind turbine 10 will accumulate during a dynamic control interval. In such embodiments, the expected power statistics 136 of the wind turbine 10 may include the power output, power factor, power stability, etc., of the wind turbine 10.

[0108] Several methods can be used to determine a model of the operational behavior of the wind turbine 10. For example, in one embodiment, a model of the operational behavior of the wind turbine 10 can be determined using simulations of the wind turbine 10, a design of experiments (DOE) process performed by the wind turbine 10, adaptive learning as the wind turbine 10 operates, and / or a combination thereof.

[0109] Furthermore, in this embodiment, the function design module 114 is configured to use the design life duration when designing the control function 116 to ensure that the expected damage incurred during the design life is within the damage limit. It should be understood that the OBC system 100 can be used for newly installed wind turbines as well as upgrades to existing wind turbines. Therefore, the design life and damage limit can be appropriately scaled.

[0110] In an additional embodiment, the discounted future value generally refers to the discount applied to revenue, and optionally to the negative value of expenses at future times. This can be simply represented as a discount rate. The discounted future value can also be more generally represented as a specific discount of revenue and expenses at future points in time. The discounted future value can optionally be used by the function design module 114 to optimize value, particularly with respect to net present value.

[0111] Now for reference Figure 6 This diagram illustrates a flowchart of one embodiment of a method 200 for operating a wind turbine according to aspects of this disclosure. In a particular embodiment, for example, method 200 may be used to generate one or more of the damage odometers 118 described herein. Method 200 is described herein as being implemented using at least one of the wind turbines 52 of, for example, the wind turbine 10 of the wind farm 50 described above. However, it should be understood that the disclosed method 200 may be implemented using any other suitable wind turbine currently known in the art or developed in the future. Additionally, although Figure 6 The steps performed in a particular order are depicted for illustrative and explanatory purposes, but the methods described herein are not limited to any particular order or arrangement. Those skilled in the art will understand, using the disclosure provided herein, that the various steps of the methods can be omitted, rearranged, combined, and / or modified in various ways.

[0112] As shown at (202), method 200 includes determining load and travel metrics or functions for one or more components of the wind turbine 10 during operation. For example, in one embodiment, method 200 may include collecting load and travel metric measurements from one or more sensors, such as strain gauges or similar sensors. In an alternative embodiment, method 200 may include estimating the load and travel metrics, for example, via a processor. In a particular embodiment, the processor may include a physics-based load estimator as part of a control system. Such estimators are configured to produce real-time estimates of the load and travel metrics required by a component life model. In another embodiment, the load may also be estimated via a heuristically or machine learning-trained function. Thus, the measured or estimated load and travel metrics reflect the actual operation of the wind turbine 10, rather than operations assumed before commissioning. Therefore, this disclosure can eliminate or significantly reduce modeling errors in simulations.

[0113] If the load metric is obtained from a control-oriented estimator, a smoothing step can be performed to refine the load metric. In such embodiments, the smoother retrospectively improves the estimator's performance using later observations. An example of a smoother is Kalman smoothing, which is a backward version of the Kalman filter. Smoothing can be particularly useful in cases of missing data or estimator reset.

[0114] Furthermore, in such embodiments, to implement a smoother, the online fatigue odometer 118 described herein can store measurement data from the current time up to a specified time window. The time window (typically a few seconds) can be selected to cover the transient dynamics of the wind turbine 10. Accordingly, the measurement data can be smoothed in a rolling time-domain manner within the time window. If desired, the stored data can then be discarded after smoothing. The time window can also be reset in the event of data loss or estimator reset.

[0115] As shown at (204), method 200 includes generating at least one distribution of cumulative load data for one or more components during operation of the wind turbine 10 using load and travel metrics. For example, in one embodiment, the distribution of cumulative load data for a component may be at least one cumulative load histogram for the component. In another embodiment, the distribution of cumulative load data may be any kind of data binning or distribution mechanism, bucketing mechanism, or hash data structure method. Thus, in such embodiments, the cumulative load histogram is based on the actual operation of the wind turbine 10 and uses the aforementioned measured / estimated load and travel metrics instead of predicted loads from simulations. In particular embodiments, the specific set of histograms required and used may depend on the component and component life model.

[0116] Furthermore, in some embodiments, each histogram may have a travel metric and a load metric. In such embodiments, the load metric may include, for example, bearing load, tower load, rotor blade load, drivetrain load, shaft load, thermal load, etc. (estimated bending moments or torques of turbine components such as the main shaft, blade roots, or tower base). In another embodiment, the travel metric may include, for example, time, angular travel, the number of times a component travels an angular distance between directional reversals, or the number of stress cycles experienced by one or more components under a specific load.

[0117] In such embodiments, the bending torque may be in a single direction or may be a composite torque in multiple directions. In another embodiment, statistics of the load time series (such as standard deviation, mean, median, minimum, maximum, quantiles, the difference between the minimum and maximum load measures during stress cycles or a certain interval) may also be used as load measures to form a histogram. In one example, when the load measure is the difference between the minimum and maximum bending moments during stress cycles, the stroke measure may be a count of the number of occurrences of that variation. In yet another embodiment, the load measure may be a scalar or a vector. For example, in an embodiment, bending moments in two orthogonal directions may be used together as the load measure. In this case, the histogram may be a two-dimensional histogram.

[0118] In one embodiment, for example, the original one-second load measurement showing the change in load over time may no longer be available. In such embodiments, the distribution of the load measurement results can be estimated from statistics to recapture the change. For example, a truncated Gaussian distribution with the same mean, standard deviation, minimum, and maximum values ​​for the measurement results can be used. Such an estimated distribution can be used to fill time periods of up to, for example, 10 minutes in a histogram to capture the change in load measurement results. Other types of distribution models can also be used to better capture the empirical data distribution based on actual load distribution data.

[0119] In another embodiment, load and travel metrics can be received and used to cumulatively populate a histogram. The histogram can be predefined by the range and bin regions of the load metrics. In such embodiments, as each measurement is received, the travel metric is added to the appropriate bin. In another embodiment, the histogram can also be more complex and adaptively change in real time, such that its range or bin regions are modified based on accumulated data or other requirements. In certain embodiments, the bin regions can have non-uniform bin sizes, for example, with higher resolution where the component life model is more sensitive to variations in load. In another embodiment, the bin regions can be uniform in size. In some embodiments, the bin width and range can be set to match the expected or required inputs of an established damage calculation process.

[0120] In another embodiment, the bin regions may have variable widths and can be adjusted over time to match load metric data. Furthermore, the cumulative load histogram may be a standard density (or frequency-based or cumulative) histogram, or a cumulative probability mass function over discrete values. Additionally, data accumulation in the cumulative load histogram may be deterministic or random, with some of the accumulated probabilities becoming alternating bin regions.

[0121] In another embodiment, a single histogram can be accumulated for each desired histogram. In an alternative embodiment, multiple histograms can be accumulated, with each histogram representing a different external condition. For example, there may be one histogram for below-rated operation and one histogram for above-rated operation. This may be necessary if the fatigue life model differs under different external conditions. This may also be necessary if the component life model has different accuracies under different operating conditions. This may be necessary if the component or software changes.

[0122] In additional embodiments, the preprocessing step of determining whether histogram data should be included or excluded can be based on a variety of sources, which may include, for example, condition-based classification data characterizing the operating state or condition of wind turbine 10 and other wind turbines. For example, filtering these states may be desirable in some cases where the estimator is known to have low predictive power or is characterized by excessive dispersion. Moreover, in cases where data is unavailable or when data is corrupted, signals from other wind turbines on the wind farm can be used as a proxy for wind turbine 10. Crowd-based or collective averaging or known weighting functions based on field layout and wake-up conditions can also be used to correct for the average proxy replacement signal.

[0123] Still referencing Figure 6 As shown at (206), method 200 includes applying a component life model (also referred to herein as a component life model) to the distribution of accumulated load data to determine the actual accumulated damage to the components of the wind turbine 10 to date. For example, in one embodiment, the component life model may be applied to a set of histograms using actual partial accumulated histograms instead of predicted full-life histograms. Moreover, as an example, the life model may be a physics-based model, a statistical model, or a combination thereof. The resulting damage output represents the accumulated damage to the turbine component to date (based on its actual operation). This damage may be used directly as part of the turbine control mechanism or to guide condition-based maintenance of the wind turbine 10. In another embodiment, the component damage model may be applied to histograms collected over different time periods. In this mode, accumulated histograms are created over time periods, the component damage model is applied to the histograms, and the output fatigue damage is recorded over the time periods. This process may then be repeated. The damage recorded for each time period may be summed to determine the total damage.

[0124] Corrections to the damage accumulation / output determined by the component life model can also be applied to its output. In such embodiments, these corrections improve damage estimation and correct for biases. Corrections can be based on physical understanding, observed behavior, engineering safety factors, and / or any other suitable parameters.

[0125] Furthermore, in embodiments, the cumulative load histogram may also be post-processed before applying the component lifetime model. In such embodiments, processing the cumulative load histogram before applying the component model may include re-defining the range or bin area of ​​the cumulative load histogram, smoothing or recalibrating the cumulative load histogram to produce a more accurate estimate of historical load estimates, or scaling the cumulative load histogram to at least one of the desired durations predicted by the lifetime model.

[0126] As shown at (208), method 200 includes assessing actual damage accumulation, for example, to determine whether wind turbine 10 has exceeded its design life. In such embodiments, the damage accumulation or output may be divided by a damage threshold to produce a percentage of the design life used. Thus, in such embodiments, damage or damage accumulation rates may be compared between wind turbines to prioritize inspections or investigations targeting anomalous behavior.

[0127] Still referencing Figure 6 As shown at (210), method 200 includes performing corrective actions on the wind turbine 10 based on damage accumulation. For example, in certain embodiments, corrective actions may include shutting down the wind turbine 10, idling the wind turbine 10, changing the control parameters of the wind turbine 10 (such as power output, speed, torque, etc.), scheduling one or more preventive maintenance actions, and / or any other suitable measures.

[0128] In some embodiments, a corrective action can be determined by comparing the actual damage accumulation with one or more damage thresholds. In another embodiment, the corrective action can be determined by mapping the actual damage accumulation to at least one of the following: failure / survival probability, economic cost, simulation of future turbine operation, etc. In yet another embodiment, the corrective action can be determined by a machine learning model that receives the actual damage accumulation as input.

[0129] In particular, the damage or fatigue odometer described herein can be used for operational optimization of the wind turbine 10 via supervised control. Through numerous control system parameters and options, the wind turbine 10 can generate more energy with more accumulated damage or less energy with less accumulated damage. In one embodiment, for example, it may be desirable to maximize energy output while taking into account accumulated damage and damage limits and the risks arising from additional damage. In such embodiments, the supervised control system may make this trade-off and control of the wind turbine 10 in part based on inputs from the fatigue / damage odometer 118. This control action may be further accomplished in conjunction with many other constraints on turbine operation, such as extreme loads and electrical system limits.

[0130] In an additional embodiment, the fatigue / damage odometer 118 can also be used to schedule maintenance and inspection of wind turbine components. Additionally, the fatigue / damage odometer 118 can be used to determine whether the wind turbine 10 can continue to operate safely beyond its design life, or what mitigation measures are needed to do so.

[0131] In another embodiment, a plurality of histogram-based fatigue odometers 118 may be used on the wind turbine 10. In such embodiments, each odometer may be specific to a component and failure mode for which a life model exists. For turbine control and condition-based maintenance, separate odometers for each component may be used. For other purposes, separate odometers may also be combined into a single metric that covers all failure modes of turbine components or all components.

[0132] In embodiments, the cumulative load histogram described herein may be one-dimensional, but it should be understood that the cumulative load histogram may be two-dimensional or multi-dimensional, where the component life model requires input as a joint distribution of multiple aspects of one or more load metrics. For example, a component life model for a portion of the structure of a wind turbine 10 may require a two-dimensional joint distribution of load cycle amplitude and load cycle average, where load cycles are determined by a load cycle algorithm such as rainflow counting. In this case, a two-dimensional histogram of load cycle amplitude and load cycle average can be used.

[0133] Now for reference Figure 7 A block diagram of one embodiment of a conventionally constructed design-time damage odometer 300 is provided for comparison with an online fatigue odometer 302 according to this disclosure. As shown and previously discussed, the design-time damage odometer 300 uses simulation 304, while the online fatigue odometer 302 uses data 306 collected during the operation of the wind turbine 10. Thus, as shown at 308 and 310 respectively, the design-time damage odometer 300 also uses a full-life histogram (e.g., assuming a 20-year lifespan), while the online fatigue odometer 302 uses a histogram filled with cumulative loads to date. Thus, the design-time damage odometer 300 and the fatigue odometer use their respective but different histograms 308, 310 to determine damage 312 or the design life (DLU) 314 used to date. As shown at 316 and 318, the design-time damage odometer 300 can then compare the damage level 312 with a threshold and determine whether the damage level 312 is acceptable. In contrast, the online fatigue odometer 302 uses DLU 314 to develop chart 320, which can be used to assess whether the damage to the wind turbine 10 is consistent with the predicted level, or whether the damage is higher or lower than such values.

[0134] Now for reference Figure 8This illustration shows a schematic diagram of one embodiment of an example system 400 for generating an online life odometer for a pitch bearing (PB) according to the present disclosure. More specifically, as shown, system 400 includes two options for determining pitch bearing damage. First, as shown, a cumulative load histogram can be transferred from controller 402 to cloud 404, where a pitch bearing damage model 406 can be applied. Second, as shown, the pitch bearing model 406 can be applied to controller 402, and the damage level can be transferred to cloud 404. In an alternative embodiment, the pitch bearing model 406 can also be applied to controller 402 and used at controller 402 (not shown).

[0135] Now for reference Figure 9 and Figure 10 Various diagrams illustrating embodiments of fatigue accumulation over time are shown. In particular, Figure 9 Figure 500 shows the cumulative fatigue over time with 502 activated and 504 not activated, according to this disclosure. Figure 10 Figure 600 illustrates an embodiment of fatigue accumulation over time with odometer-based control activated (602) and deactivated (604) according to this disclosure, and in particular shows the lifespan extension (606) provided by the odometer-based control.

[0136] Figure 11 A graph illustrating one embodiment of a cumulative load histogram 700 according to this disclosure is shown. Specifically, as illustrated, the cumulative load histogram 700 shows how a truncated Gaussian distribution can fit the statistics of the mean, standard deviation, minimum, and maximum values ​​over a time period. This produces an approximation of the distribution of the original measurement data, which can be used to accumulate a histogram over the time period.

[0137] This written description uses examples to disclose the invention, including the best mode, and also enables any person skilled in the art to practice the invention, including making and using any device or system and performing any incorporated methods. The patentability of the invention is defined by the claims and may include other examples that would occur to a person skilled in the art. Such other examples are intended to be within the scope of the claims if they include structural elements that are not different from the literal language of the claims, or if they include equivalent structural elements that are not substantially different from the literal language of the claims.

Claims

1. A method for operating a wind turbine, the method comprising: During operation of the wind turbine, multiple load metrics and multiple stroke metrics are determined for one or more components of the wind turbine; During the operation of the wind turbine, at least one distribution is generated, using the plurality of load metrics and the plurality of travel metrics, to at least partially generate cumulative load data for the one or more components, including: Define the range and bin area of ​​at least one cumulative load histogram; Define a first load metric among the plurality of load metrics and a first travel metric among the plurality of travel metrics for the at least one cumulative load histogram; During the operation of the wind turbine, the at least one cumulative load histogram is populated using the first load metric and the first travel metric by adding the first load metric and the first travel metric to the bin area of ​​the at least one cumulative load histogram; and The range or at least one of the box area is adaptively changed based on the ongoing operational data of the wind turbine. The life model of the one or more components is applied to the at least one distribution of the cumulative load data to determine the actual cumulative damage to date for the one or more components of the wind turbine; and Corrective actions are performed on the wind turbine based on the accumulated damage.

2. The method according to claim 1, wherein, The plurality of load measures relating to one or more components of the wind turbine include at least one of bearing load, tower load, wind turbine impeller blade load, transmission load, shaft load, and thermal load, and wherein the plurality of travel measures include at least one of time, angular travel, the number of times a component travels an angular distance between reversals, or the number of times the one or more components undergo stress cycles under a specific load.

3. The method according to claim 1, wherein, The one or more components include at least one of a pitch bearing, a yaw bearing, a tower, a gearbox, a generator, a wind turbine blade, a wind turbine hub, a shaft, a converter, or a nacelle.

4. The method according to claim 1, wherein, The one or more components include a fan impeller.

5. The method according to claim 1, wherein, The plurality of load metrics further include at least one of the average, minimum, maximum, standard deviation, quantile, or median of the load, and the plurality of travel metrics further include at least one of the average, minimum, maximum, standard deviation, quantile, or median of the travel.

6. The method according to claim 1, wherein, Determining the plurality of load metrics and the plurality of travel metrics with respect to one or more components of the wind turbine during operation of the wind turbine also includes: The plurality of load metrics and the plurality of travel metrics are measured with respect to one or more components of the wind turbine via one or more sensors or processors, or at least one of the plurality of load metrics and the plurality of travel metrics is estimated with respect to one or more components of the wind turbine via one or more processors.

7. The method of claim 6, further comprising real-time estimation of the plurality of load metrics and the plurality of travel metrics with respect to one or more components of the wind turbine.

8. The method of claim 6, further comprising using machine learning to estimate the plurality of load metrics and the plurality of travel metrics for one or more components of the wind turbine.

9. The method according to claim 6, wherein, The processor includes a physics-based system model for estimating the plurality of load metrics and the plurality of travel metrics for one or more components of the wind turbine.

10. The method of claim 1, further comprising applying the lifetime model of the one or more components to multiple cumulative load histograms collected over different time periods.

11. The method of claim 1, further comprising processing the at least one cumulative load histogram prior to applying a lifetime model of the one or more components.

12. The method according to claim 11, wherein, Processing the at least one cumulative load histogram before applying the lifetime model of the one or more components also includes relimiting the range or bin region of the at least one cumulative load histogram, smoothing the at least one cumulative load histogram to produce a more accurate historical load estimate than the estimate before smoothing, recalibrating the at least one cumulative load histogram to produce a more accurate historical load estimate than the estimate before recalibration, or scaling the at least one cumulative load histogram to at least one of the desired durations predicted by the lifetime model.

13. The method according to claim 1, wherein, Performing the corrective action on the wind turbine based on the accumulated damage also includes shutting down the wind turbine, idling the wind turbine, changing the control parameters of the wind turbine, or scheduling one or more preventive maintenance actions.

14. A system for operating a wind turbine, the system including a controller configured to perform a plurality of operations, the plurality of operations including: During operation of the wind turbine, multiple load metrics and multiple stroke metrics are determined for one or more components of the wind turbine; Using the plurality of load metrics and the plurality of travel metrics to generate at least one cumulative load histogram for at least one of the one or more components, including: Define the range and bin area of ​​the at least one cumulative load histogram; Define a first load metric among the plurality of load metrics and a first travel metric among the plurality of travel metrics for the at least one cumulative load histogram; During the operation of the wind turbine, the at least one cumulative load histogram is populated using the first load metric and the first travel metric by adding the first load metric and the first travel metric to the bin area of ​​the at least one cumulative load histogram; and The range or at least one of the box area is adaptively changed based on the ongoing operational data of the wind turbine. The life model of the one or more components is applied to the at least one cumulative load histogram to determine the actual cumulative damage to date for the one or more components of the wind turbine; and Corrective actions are performed on the wind turbine based on the accumulated damage.

15. The system according to claim 14, wherein, The plurality of load measures relating to one or more components of the wind turbine include at least one of bearing load, tower load, turbine impeller blade load, transmission load, and shaft load, and wherein the plurality of travel measures include at least one of time, angular travel, the number of times a component travels an angular distance between directional reversals, or the number of times the one or more components undergo stress cycles under a specific load.

Images