Systems and methods for operating a power generation asset

Abstract

Systems and methods for operating a power generation asset are provided. Accordingly, at least one external data set indicative of a plurality of variables affecting performance of the power generation asset is received by a controller. The controller also receives at least one operational data set indicative of performance of the power generation asset. A plurality of production-evaluation models for the power generation asset are generated and trained based on the data sets. Performance predictions are then generated for each of a plurality of model-variable combinations, and a control action is implemented based on one of the performance predictions.

Description

Technical Field

[0001] This invention relates generally to power generation assets, and more particularly to systems and methods for operating power generation assets by generating performance predictions based on various model-variable combinations. Background Technology

[0002] As disclosed herein, power generation assets can take many forms and may include those relying on renewable and / or non-renewable energy sources. Those relying on renewable energy are generally considered to be among the cleanest and most environmentally friendly energy sources currently available. Wind turbines, for example, have received increasing attention in this regard. A modern wind turbine typically comprises a tower, generator, gearbox, nacelle, and one or more rotor blades. The nacelle includes a rotor assembly coupled to the gearbox and then to the generator. The rotor assembly and gearbox are mounted on a base support frame located within the nacelle. The rotor blades capture the kinetic energy of the wind using the known airfoil principle. The rotor blades transfer this kinetic energy as rotational energy, thereby rotating a shaft that couples the rotor blades to the gearbox, or, if no gearbox is used, directly to the generator. The generator then converts the mechanical energy into electrical energy, which can be transmitted to a converter and / or transformer housed within the tower and subsequently dispatched to the public grid. Modern wind power systems typically take the form of wind farms with multiple wind turbine generators operable to supply power to a transmission system that supplies electricity to the grid.

[0003] Typically, what is desired is a forecast of the expected power generation from a power generation asset. For example, a forecast can serve as the basis for a production guarantee agreement. Typically, such forecasts are accomplished using conventional methods that consider a linear relationship between variables such as wind speed and the energy production of the power generation asset. This variable is often retrospectively modeled on a monthly basis. This conventional approach can lead to significant variations in output estimates, especially with small sample sizes. Therefore, what may be desired is a more accurate forecast of the performance of the power generation asset.

[0004] In view of the foregoing, the art is constantly seeking new and improved systems and methods for operating power generation assets based on performance forecasts. Summary of the Invention

[0005] Many 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 practice of the invention.

[0006] On one hand, this invention discloses a method for operating a power generation asset. The method may include steps a) to f). Therefore, the method may include receiving, via a controller, at least one external dataset for a sampling period from at least one source separate from the power generation asset. The external dataset may indicate multiple variables affecting the performance of the power generation asset. The method may also include receiving, via the controller, at least one operational dataset for the power generation asset for the sampling period. The operational dataset may indicate the performance of the power generation asset. The controller may also generate multiple production-evaluation models for the power generation asset. The production-evaluation models may be trained via the external dataset and the operational dataset to correlate the performance of the power generation asset as a function of multiple variables. Furthermore, the method may include generating performance predictions for each of multiple model-variable combinations within a prediction implementation period via the controller. The model-variable combinations may include multiple combinations of each production-evaluation model and multiple variables. Therefore, each performance prediction may include a power generation prediction and its confidence interval. Furthermore, based on one of the performance predictions, the controller may implement control actions.

[0007] In one embodiment, the multiple variables may include at least data indicating wind speed and wind direction at sampling intervals of the sampling period.

[0008] In another embodiment, the variables may also include data indicating at least one of time correlation, temperature, atmospheric pressure, air density, wind shear, wind veer, and turbulence intensity.

[0009] In another embodiment, generating multiple production-evaluation models may further include generating a statistical algorithm or machine learning algorithm for each of the multiple production-evaluation models, the statistical algorithm or machine learning algorithm being configured to determine an optimal transfer function between at least two of the multiple variables and the performance of the power generation asset.

[0010] In yet another embodiment, the external dataset and the operational dataset can be generated at a first sampling interval and a second sampling interval, respectively. Therefore, generating multiple production-evaluation models may further include generating a first portion of multiple production-evaluation models based on the external dataset and the operational dataset with the first sampling interval. Furthermore, a second portion of multiple production-evaluation models can be generated based on the external dataset and the operational dataset with the second sampling interval. The second sampling interval may have a higher frequency than the first sampling interval.

[0011] In one embodiment, the external dataset may include a modeling environmental dataset that indicates multiple environmental variables affecting power generation assets.

[0012] In another embodiment, the external dataset may include an environmental dataset assembled from a group of power generation systems. Therefore, the method may include receiving indications of each of a plurality of variables from each power generation system in the group via a controller at each sampling interval of the sampling period. The controller may then combine the plurality of variables received from each of the plurality of power generation systems in the group to generate a consistent environmental dataset that indicates a plurality of consistent environmental variables affecting the performance of the power generation asset.

[0013] In another embodiment, the external dataset may include multiple environmental measurements collected by a weather mast.

[0014] In yet another embodiment, training multiple production-evaluation models may also include training multiple production-evaluation models using multiple environmental variables obtained via environmental sensors of the power generation assets.

[0015] In one embodiment, at least one of the external dataset and the operational dataset may further include at least one anomalous input for a sampling interval relative to the sampling period. Therefore, the controller can generate interpolated values ​​for the anomalous input for the sampling interval via an interpolation algorithm.

[0016] In another embodiment, the power generation asset may include a wind turbine.

[0017] In another embodiment, the power generation asset may include multiple power generation systems.

[0018] In yet another embodiment, power generation assets may include wind farms, solar power facilities, and / or hybrid power generation facilities.

[0019] In one embodiment, the power generation asset may be one of a plurality of power generation assets. In this embodiment, steps a) to e) may be repeated for each of the plurality of power generation assets. Furthermore, the controller may establish a hierarchical order for each of the plurality of power generation assets based on the desired performance prediction characteristics. The implementation of control actions may be based on this hierarchical order.

[0020] In another embodiment, implementing control actions may include upgrading at least one of a plurality of power generation assets. Therefore, an upgrade threshold may be established, corresponding to a percentage increase in the performance forecast of a power generation asset relative to the performance of the power generation asset as indicated by an operational dataset. This percentage increase may be attributable to a prospective upgrade of the power generation asset. Thus, control actions may include upgrading the power generation capacity of power generation assets that have a percentage increase in the performance forecast greater than the upgrade threshold.

[0021] In another embodiment, implementing control actions may include performing a diagnostic process on at least one of a plurality of power generation assets. Therefore, a diagnostic threshold may be established, which may indicate the percentage by which the performance of a power generation asset, as shown in the operational dataset, falls short of performance predictions. Thus, control actions may include performing root cause analysis to determine the root cause of the percentage shortfall.

[0022] On one hand, this invention discloses a system for operating power generation assets. The system includes at least one sensor operatively coupled to the power generation assets. Furthermore, the system includes a controller communicatively coupled to the sensor. The controller includes at least one processor configured to perform multiple operations. These multiple operations may include, but are not limited to, receiving external datasets and operational datasets, generating and training multiple production-evaluation models, generating performance predictions for multiple model-variable combinations, and implementing control actions based on the performance predictions as described herein. However, in other embodiments, the multiple operations may include any of the methods, steps, and / or features described herein.

[0023] These and other features, aspects, and advantages of the invention will become better understood with reference to the following description and the appended claims. Embodiments of the invention are illustrated in conjunction with the accompanying drawings, which are incorporated in and form part of this specification, and serve to explain the principles of the invention together with the description. Attached Figure Description

[0024] The specification describes a complete and practiceable disclosure of the invention for those skilled in the art, including its preferred mode, and includes reference to the accompanying drawings, wherein:

[0025] Figure 1 The illustration shows a perspective view of an embodiment of a power generation asset configured as a wind turbine according to the present invention;

[0026] Figure 2 The illustration shows a perspective and interior view of an embodiment of the nacelle of a wind turbine according to the present invention;

[0027] Figure 3 The illustration shows multiple wind turbines configured into a wind farm according to the present invention;

[0028] Figure 4 The diagram illustrates a schematic representation of an embodiment of the controller disclosed in this invention.

[0029] Figure 5 The diagram illustrates a schematic of an embodiment of the control logic of a system for operating power generation assets disclosed in this invention.

[0030] Figure 6The illustration depicts the relationship between two variables in the operational dataset and at least one external dataset according to the present invention; and

[0031] Figure 7 The illustration shows a tabular representation of the performance predictions for a single power generation asset using multiple model-variable combinations disclosed in this invention.

[0032] The repeated use of reference numerals in this specification and drawings is intended to represent the same or similar features or elements in this invention. Detailed Implementation

[0033] Reference will now be made in detail to embodiments of the invention, one or more of which are illustrated in the accompanying drawings. Each example is provided as an explanation of the invention and not as a limitation thereof. In fact, it will be apparent to those skilled in the art that various modifications and variations can be made in the invention without departing from the scope or spirit of the invention. For example, features illustrated and described as part of one embodiment may be used in conjunction with another embodiment to produce yet another embodiment. Therefore, it is intended that the invention cover such modifications and variations falling within the scope of the appended claims and their equivalents.

[0034] Unless otherwise specified herein, the terms “connection,” “fixation,” “attachment,” etc., refer to both direct connection, fixation, or attachment, and indirect connection, fixation, or attachment via one or more intermediate components or features.

[0035] In general, this invention discloses machine learning and model-based analytics for operating power generation assets. Specifically, the invention may include systems and methods that facilitate the generation of multiple different performance predictions based on various model-variable combinations over a specified time period. Therefore, various machine learning algorithms can be employed to generate diverse models that can simulate the performance of power generation assets using different modeling approaches. For example, these models may reflect linear or nonlinear relationships between the performance of power generation assets and various variables that may affect their performance. Thus, various models can generate corresponding performance predictions based on various variables, combinations of variables, and / or sampling intervals. Each performance prediction may include a power generation prediction and a confidence interval for the prediction.

[0036] It should be understood that the variables described here refer to a specific set of values ​​corresponding to a particular factor, rather than its specific measurement. These variables can correspond to various environmental factors, such as wind speed, wind direction, wind shear, temperature, air density, humidity level, or other similar factors. Therefore, these models may reflect linear or nonlinear relationships between the performance of power generation assets and selection factors. For example, one model-variable combination may reflect the relationship between performance and wind speed, while another model-variable combination may reflect the relationship between performance and both wind speed and wind direction.

[0037] As an illustration, the systems and methods disclosed herein can generate and employ model-variable combinations as shown below, from A) to J).

[0038]

[0039] sheet .

[0040] Each of the model-variable combinations A) through J) can output different performance predictions for the power generation asset. Each different performance prediction can represent a different expected power generation value and a different range of expected deviations relative to the predicted value. For example, determining the performance prediction based on a two-variable, nonlinear model (e.g., combination (A)) with an hourly sampling interval may result in a relatively small range of expected deviations. In contrast, determining the performance prediction based on a single-variable, nonlinear model (e.g., combination (I)) with a monthly sampling interval may result in greater predicted performance, but also a larger range of expected deviations relative to combination (A). Since various implementation strategies may require different levels of performance prediction fidelity, the model-variable combination on which the control action is based and the corresponding performance prediction can be selected based on certain operational considerations. In other words, the optimal model-variable combination for one implementation strategy may differ from that for another.

[0041] It should be understood that various performance forecasts can be used in a variety of applications. For example, various performance forecasts can be used to form, say, production agreements / guarantees related to power purchase agreements. When considering such agreements, it may be desirable to choose a model-variable combination that minimizes the uncertainty surrounding the predicted power generation relative to the remaining performance forecasts, and therefore has a relatively high confidence level in the predicted power generation.

[0042] As another example, various performance forecasts can be used to analyze the rationale for anticipated upgrades to power generation assets. When used in this way, it may be desirable to choose a model-variable combination that outputs the maximum predicted power generation with minimal uncertainty.

[0043] As a further example, various model-variable combinations can be used to analyze multiple power generation assets. Performance predictions for each asset can facilitate asset ranking. Based on the ranking, assets suitable for meeting operational objectives can be selected. Furthermore, the ranking may reveal performance deficiencies in certain assets. Therefore, diagnostic systems can be implemented to identify the root causes of these deficiencies relative to the remaining assets, facilitating their correction.

[0044] Furthermore, various performance forecasts can be used to determine at least one operating mode for power generation assets. For example, various performance forecasts can be used to determine the optimal or desired thrust limit, target rotor speed, and / or rated power.

[0045] Now refer to the attached diagram, Figure 1 The illustration shows a perspective view of one embodiment of a power generation asset 100 disclosed in this invention. As shown, the power generation asset 100 can be configured as a wind turbine 114. In another embodiment, the power generation asset 100 can be configured, for example, as a solar power generation asset, a hydroelectric power plant, a fossil fuel generator, and / or a hybrid power generation asset.

[0046] For example Figure 3 In the embodiment illustrated, power generation asset 100 can be constructed as power generation facility 142. For example... Figure 3 As shown, in one embodiment, the power generation facility 142 may be configured as a wind farm; however, in another embodiment, the power generation facility 142 may be configured as a solar power generation facility and / or a hybrid power generation facility.

[0047] In another embodiment, the power generation asset 100 may be configured as a plurality of power generation systems 146. Subsystem 148 may be configured as a wind turbine 114, a solar power generation asset, a hydroelectric power plant, a fossil fuel generator, a hybrid power generation asset, or a combination thereof.

[0048] Refer again Figure 1 When constructed as a wind turbine 114, the power generation asset 100 typically includes a tower 102 extending from a support surface 104, a nacelle 106 mounted on the tower 102, and a rotor 108 coupled to the nacelle 106. The rotor 108 may include a rotatable hub 110 and at least one rotor blade 112 coupled to and extending outward from the hub 110. For example, in this illustrated embodiment, the rotor 108 includes three rotor blades 112. However, in another embodiment, the rotor 108 may include more or fewer than three rotor blades 112. Each rotor blade 112 may be spaced around the hub 110 to allow rotation of the rotor 108, thereby enabling kinetic energy to be converted from wind into usable mechanical energy, and subsequently into electrical energy. For example, the hub 110 may be rotatably coupled to a generator 118 located within the nacelle 106. Figure 2 This allows for the generation of electrical energy.

[0049] The power generation asset 100 may also include a controller 200. When configured as a wind turbine 114, the controller 200 may be configured as a turbine controller centralized within the nacelle 106. However, in other embodiments, the controller 200 may be located within any other component of the wind turbine 100 or at a location outside the wind turbine. Furthermore, the controller 200 may be communicatively coupled to any number of components of the power generation asset 100 to control those components. Therefore, the controller 200 may include a computer or other suitable processing unit. Thus, in several embodiments, the controller 200 may include appropriate computer-readable instructions that, when implemented, configure the controller 200 to perform various functions, such as receiving, sending, and / or executing wind turbine control signals.

[0050] Now for reference Figure 2 , showed Figure 1 The diagram shows a simplified internal view of one embodiment of the nacelle 106 of the wind turbine 114. As shown, a generator 118 may be coupled to a rotor 108 to generate electricity from the rotational energy produced by the rotor 108. For example, as shown in the illustrated embodiment, the rotor 108 may include a rotor shaft 122 coupled to a hub 110 for rotation therewith. The rotor shaft 122 may be rotatably supported by a main bearing 144. The rotor shaft 122 is, in turn, rotatably coupled to a high-speed shaft 124 of the generator 118 via a gearbox 126 connected to a base plate support frame 136. As generally understood, the rotor shaft 122 may provide a low-speed, high-torque input to the gearbox 126 in response to rotation of the rotor blades 112 and the hub 110. The gearbox 126 may then be configured to convert the low-speed, high-torque input into a high-speed, low-torque output to drive the high-speed shaft 124, and thus the generator 118.

[0051] Each rotor blade 112 may also include a pitch control mechanism 120 configured to rotate each rotor blade 112 about its pitch axis 116. Each pitch control mechanism 120 may include a pitch drive motor 128, a pitch drive gearbox 130, and a pitch drive pinion 132. In this embodiment, the pitch drive motor 128 may be coupled to the pitch drive gearbox 130 to apply mechanical force to the pitch drive gearbox 130. Similarly, the pitch drive gearbox 130 may be coupled to the pitch drive pinion 132 to rotate therewith. The pitch drive pinion 132 may in turn be rotatably engaged with a pitch bearing 134 connected between the hub 110 and the corresponding rotor blade 112, such that rotation of the pitch drive pinion 132 causes rotation of the pitch bearing 134. Therefore, in this embodiment, the rotation of the pitch drive motor 128 drives the pitch drive gearbox 130 and the pitch drive pinion 132, thereby rotating the pitch bearing 134 and the rotor blades 112 about the pitch axis 116.

[0052] Similarly, the wind turbine 114 may include one or more yaw drive mechanisms 138 communicatively coupled to the controller 200, and each yaw drive mechanism 138 is configured to change the angle of the nacelle 106 relative to the wind (e.g., by engaging the yaw bearing 140 of the wind turbine 114). It should be understood that the controller 200 may guide the yaw of the nacelle 106 and / or the pitch of the rotor blades 112 to aerodynamically orient the wind turbine 114 relative to the wind acting on it, thereby promoting power generation.

[0053] In several embodiments, the power generation asset 100 may include at least one environmental sensor 156 for monitoring at least one environmental condition affecting the power generation asset 100. In one embodiment, the environmental sensor 156 may be, for example, a wind vane, an anemometer, a lidar sensor, a thermometer, a barometer, or any other suitable sensor. Thus, the environmental sensor 156 may collect data indicating wind direction, wind speed, wind shear, gusts, wind direction change, atmospheric pressure, pressure gradient, and / or temperature. In at least one embodiment, the environmental sensor 156 may be mounted on the nacelle 106 located downwind of the rotor 108.

[0054] It should be understood that the environmental sensor 156 may include a sensor network and may be located remotely from the power generation asset 100. For example, in one embodiment, the environmental sensor 156 may be configured as a weather mast 150.

[0055] In addition, the power generation asset 100 may include one or more operating sensors 158. The operating sensors 158 may be configured to detect the performance of the power generation asset 100 in response to environmental conditions. The operating sensors 158 may be configured to monitor multiple parameters associated with the performance and / or health of components of the power generation asset 100. For example, the operating sensors 158 may monitor parameters related to vibration, audio signals, visual indications, angular position, rotational speed, bending moment, power consumption, power generation, temperature, and / or other suitable parameters.

[0056] In one embodiment, the operating sensor 158 may be, for example, a speed sensor operatively coupled to the controller 200. For instance, the operating sensor 158 may be directed at the rotor shaft 122 of the power generation asset 100 (e.g., a wind turbine 114). The operating sensor 158 may collect data indicating the rotational speed and / or rotational position of the rotor shaft 122, and therefore the rotor 108, in the form of rotor speed and / or rotor azimuth angle. In one embodiment, the operating sensor 158 may be an analog tachometer, a direct current (DC) tachometer, an alternating current (AC) tachometer, a digital tachometer, a contact tachometer, a non-contact tachometer, or a time-frequency tachometer.

[0057] Still referencing Figure 2 In one embodiment, the operational sensor 158 may be configured to collect data indicating the response of components of the power generation asset 100 to environmental conditions or other loads. For example, the operational sensor 158 may be configured to monitor electrical parameters of the output of the power generation asset 100. Thus, the operational sensor 158 may be a current sensor, voltage sensor, temperature sensor, power sensor, and / or frequency meter that monitors the electrical output of the power generation asset 100.

[0058] As further illustrated, in one embodiment, the operational sensor 158 may be configured as a strain gauge to detect tensile loads on a component (e.g., rotor 108). In another embodiment, the operational sensor 158 may include at least one of an accelerometer, photoelectric sensor, acoustic sensor, transducer, lidar system, vibration sensor, force sensor, rate sensor, piezoelectric sensor, position sensor, inclinometer, and / or torque sensor. In one embodiment, the operational sensor 158 may be configured, for example, to collect sensor data indicating at least one of the following: nacelle acceleration, vibration of tower 102, bending of rotor shaft 122, acoustic signature of power generation asset 100, and optical sensor blockage due to the passage of rotor blades 112, discontinuities of rotor blades 112, horizontal and vertical deflection of rotor 108, and / or acceleration of rotor 108.

[0059] It should also be understood that, as used herein, the term "monitoring" and its variations refer to the various sensors of the power generation asset 100 that can be configured to provide direct or indirect measurements of the monitored parameters. Thus, the sensors described herein can, for example, be used to generate signals associated with the monitored parameters, which can then be utilized by the controller 200 to determine the state or response of the power generation asset 100.

[0060] Now for reference Figure 3 The power generation asset 100 is configured as a power generation facility 142 (e.g., a wind farm). As illustrated, the power generation facility may include multiple power generation systems 148 as described herein (e.g., wind turbines 114). For example, as shown in the illustrated embodiment, the power generation facility 142 may include twelve power generation systems 148. However, in other embodiments, the power generation facility 142 may include any other number of power generation systems 148, such as fewer than twelve or more than twelve power generation systems 148. It should be understood that the power generation facility 142 may be coupled to the controller 200 and / or POI 152 via a communication link 154.

[0061] Now for reference Figure 4-7This illustrates various aspects of multiple embodiments of a system 300 for operating a power generation asset 100 according to the present invention. For example, as described herein, system 300 can be used to operate the aforementioned wind turbine 114. However, it should be understood that the disclosed system 300 can be used for any other power generation asset 100 having any suitable construction. Furthermore, although... Figure 5 The steps are described in a specific order for illustrative and discussion purposes, but the methods and steps 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 changed in various ways.

[0062] like Figure 4 Specifically, a schematic diagram of one embodiment of suitable components that may be included within controller 200 is shown. For example, as shown, controller 200 may include one or more processors 206 configured to perform various computer-implemented functions (e.g., performing the methods, steps, calculations, etc. disclosed herein and storing related data) and associated storage devices 208. Furthermore, controller 200 may also include a communication module 210 to facilitate communication between controller 200 and power generation asset 100 and its components. Additionally, communication module 210 may include a sensor interface 212 (e.g., one or more analog-to-digital converters) to allow signals transmitted from one or more sensors (e.g., environmental sensor 156 and / or operational sensor 158) to be converted into signals that can be understood and processed by processor 206. It should be understood that any suitable means can be used to communicatively connect sensors to communication module 210. For example, as Figure 4 As shown, the sensors can be connected to the sensor interface 212 via a wired connection. However, in other embodiments, sensors 156 and 158 can be connected to the sensor interface 212 via a wireless connection (e.g., using any suitable wireless communication protocol known in the art). Furthermore, the communication module 210 can also be operatively connected to an operation state control module 214 configured to perform control actions.

[0063] In one embodiment, controller 200 may be configured as an asset controller and may be integrated with power generation asset 100. For example, controller 200 may be configured as a turbine controller, field controller, and / or other similar controller configured to direct the operation of power generation asset 100. In another embodiment, controller 200 may include a distributed network of computing devices. In this embodiment, one of the distributed computing devices may be integrated with power generation asset 100, while additional computing devices may be located remotely from the power generation asset, such as at a design or manufacturing facility.

[0064] As used herein, the term "processor" refers not only to integrated circuits as used in the art and included in a computer, but also to controllers, microcontrollers, microcomputers, programmable logic controllers (PLCs), special purpose integrated circuits, and other programmable circuits. Additionally, storage device 208 may typically include storage 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, read-only compact optical discs (CD-ROMs), magneto-optical discs (MODs), digital versatile optical discs (DVDs), and / or other suitable storage elements. Such storage device 208 may typically be configured to store appropriate computer-readable instructions that, when implemented by processor 206, configure controller 200 to perform various functions, including but not limited to: generating multiple production-evaluation models corresponding to at least one external dataset and at least one operational dataset; training the production-evaluation models; and generating performance predictions for each of multiple model-variable combinations to implement the control actions described herein, as well as various other suitable computer-implemented functions.

[0065] For details, please refer to the following: Figure 5 In one embodiment, the controller 200 of system 300 may be configured to receive at least one external dataset 302 for a sampling period 304 from at least one source separate from the power generation asset 100. The external dataset 302 may indicate multiple variables 306. The multiple variables 306 may affect the performance of the power generation asset 100. The controller 200 may also be configured to receive at least one operational dataset 308 for the power generation asset 100 for the sampling period 304. The operational dataset 308 may indicate the performance of the power generation asset 100, for example, in response to the variables 306. Furthermore, the controller 200 may generate multiple production-evaluation models 310 for the power generation asset 100. As shown at 312, the controller 200 may then train (e.g., via machine learning) the multiple production-evaluation models 310 via the external and operational datasets 302, 308. Training the multiple production-evaluation models 310 may correlate the performance of the power generation asset 100 as a function of the multiple variables 306. Once multiple production-evaluation models 310 are trained, the controller 200 can generate performance predictions 314 for each of the multiple model-variable combinations 318 during the prediction implementation cycle 316. Figure 7 The multiple model-variable combination 318 may include multiple combinations of each of the production assessment model 310 and the multiple variables 306. Performance predictions 314 may each include a power generation prediction 320 and a confidence interval 322. Furthermore, the controller 200 may implement control actions 324 based on one of the performance predictions 314.

[0066] In one embodiment, the multiple variables 306 of the external dataset 302 may include at least two independent variables 306. Therefore, the performance of the power generation asset 100 (e.g., historical power generation 326) Figure 6 )) can be correlated as a function of two independent variables 306. For example Figure 6 As shown, the correlation function between the performance of power generation asset 100 and two independent variables 306 can be visualized as a three-dimensional graph. Since the production-assessment model 310 can be trained based on the correlation between performance (e.g., historical power generation 326) and the two independent variables 306, performance prediction 314 can be based on the predictive impact of the two independent variables 306 on the predicted performance of power generation asset 100. It should be understood that the controller 200's use of at least two variables 306 can facilitate a higher fidelity performance prediction 314 than could be obtained using only a single variable 306.

[0067] It should also be understood that variable 306 refers to a specific set of values ​​corresponding to a specific factor affecting the performance of power generation asset 100, rather than a specific measurement of that factor at a specific sampling interval. Therefore, each of the multiple model-variable combinations 318 can reflect a linear or nonlinear relationship between the performance of the power generation asset and selected factors modeled by a specific production-evaluation model 310. For example, one model-variable combination 318 can reflect the relationship between performance and wind speed 328, while another model-variable combination 318 can reflect the relationship between performance and both wind speed 328 and wind direction 330.

[0068] It should also be understood that using more than two variables 306 can further refine the correlation function, thereby increasing the fidelity of the performance prediction 314. While increasing the fidelity of the performance prediction 314, using more than two variables 306 can also increase the complexity of its data collection and / or analysis. Therefore, it may be desirable to balance the benefits attributable to the increased fidelity of the performance prediction 314 with the increased complexity of its data collection and / or analysis.

[0069] like Figure 6As illustrated, in one embodiment, multiple variables 306 of the external dataset 302 include wind speed 328 and wind direction 330, which affect the performance of the power generation asset 100. Therefore, the performance of the power generation asset 100 in response to wind speed 328 and wind direction 330 (e.g., historical power generation 326) can be reflected by the operational dataset 308. The simultaneous correlation between both wind speed 328 and wind direction 330 and the historical performance of the power generation asset 100 can facilitate performance predictions 314 with higher fidelity than can be obtained using only one of wind speed 328 or wind direction 330. For example, a wind turbine 114 may encounter a specific wind speed 328 from a first direction and may generate a first amount of power. However, when the same wind speed 328 comes from a different direction, the wind turbine 114 may generate a second, smaller amount of power. This could be attributed, for example, to turbulent airflow (such as wake effects), obstacles, operational limits, and / or other conditions. Therefore, developing a performance prediction 314 based on only one of wind speed 328 or wind direction 330, compared to a performance prediction 314 based on at least wind speed 328 and wind direction 330, would reduce the accuracy and / or confidence of the performance prediction 314.

[0070] In one embodiment, the plurality of variables 306 may further include at least one additional variable 306, which may further define the correlation between the performance of the power generation asset 100 and the external dataset 302. Therefore, in one embodiment, the plurality of variables 306 may include at least three variables 306. For example, the plurality of variables 306 may include wind speed 328, wind direction 330, and at least one additional variable 306. In one embodiment, the additional variable 306 may be a temporal correlation 332. The temporal correlation 332 may correlate the performance of the power generation asset 100 with a defined time period (e.g., daytime, nighttime, season, or other similar metric). For example, the temporal correlation 332 may illustrate variations in the performance of the power generation asset 100 during certain periods of the year (e.g., seasons that may experience increased wind speeds). In an additional embodiment, the additional variable 306 may include measurements of wind shear 334, turbulence intensity 336, and / or other variables (e.g., temperature, air pressure, air density, humidity level, wind shear, wind direction, turbulence intensity, etc.) that may affect the performance of the power generation asset 100.

[0071] In one embodiment, sampling period 304 may include multiple sampling intervals at which data, including external dataset 302 and / or operational dataset 308, can be collected. The sampling intervals can be selected to develop external and / or operational datasets 302, 308 with a sufficient number of historical data points to support the training of production-evaluation model 310. For example, sampling intervals may be established at a monthly frequency. In an additional embodiment, each sampling interval may correspond to 24 hours. In yet another additional embodiment, the sampling interval may have a duration of less than or equal to 60 minutes (e.g., a 10-minute sampling interval).

[0072] Still referencing Figure 5 In one embodiment, the external and operational datasets 302 and 308 can each be generated with a first sampling interval 338 and at least a second sampling interval 340. In one embodiment, the second sampling interval 340 can have a higher frequency / sampling rate relative to the first sampling interval 338. For example, in one embodiment, the first sampling interval 338 can correspond to a monthly sampling interval, the second sampling interval 340 can correspond to a daily sampling interval, and the third sampling interval can correspond to an hourly sampling interval, with the external and operational datasets 302 and 308 corresponding to each sampling interval. In this embodiment, various production-evaluation models 310 can be trained for the corresponding variable 306 at each sampling interval (as shown in Table 1). In other words, in one embodiment, a first part 342 of the production-evaluation model 310 can be generated based on the external dataset 302 and operational dataset 308 with the first sampling interval 338. In this embodiment, a second part 344 of the production-evaluation model 310 can be generated based on the external dataset 302 and operational dataset 308 with the second sampling interval 340.

[0073] It should be understood that, within the same sampling period 304 duration, differences in the number of data points generated at each sampling interval may lead to differences in the production-evaluation model 310, and thus may lead to variations in the performance prediction 314 for each model-variable combination 318.

[0074] In one embodiment, external dataset 302 and / or operational dataset 308 may include at least one anomalous input for the sampling interval of sampling period 304. The anomalous input may be a quality issue with a portion of the external and / or operational datasets 302, 308. For example, the anomalous input may be due to inaccurate anemometer wind speed measurements, limited operating conditions of power generation asset 100, icing of rotor blades 112, and / or missing power generation data for the sampling interval.

[0075] The impact of anomalous inputs can be mitigated by interpolating them. Therefore, in one embodiment, controller 200 may employ an interpolation algorithm to generate interpolated values ​​346 for anomalous inputs at sampling intervals. For example, in one embodiment, controller 200 may use operational dataset 308 to learn the power generation characteristics (e.g., power curves) of power generation asset 100. The controller can then use the learned features to estimate / input power generation values ​​for missing power generation data.

[0076] In one embodiment, external dataset 302 may include a modeled environmental dataset. The modeled environmental dataset may indicate multiple environmental variables that may affect power generation asset 100. For example, external dataset 302 may be developed using environmental models, such as modern research and applied retrospective analysis, the MERRA-2 model, or other similar reanalysis methods.

[0077] In another embodiment, the external dataset 302 may include multiple environmental measurements collected by at least one meteorological mast 150 (e.g., a weather mast). The weather mast 150 may be mounted near the power generation asset 100 and may serve as a mounting location for environmental sensors 156. Thus, the weather mast 150 may include anemometers, wind vanes, barometers, hydrometers, thermometers, and / or other similar meteorological instruments configured to collect measurements of variables 306 that indicate the performance of the power generation asset 100.

[0078] In another embodiment, the external dataset 302 may include an environmental dataset assembled from a group of power generation systems 148 of the power generation asset 100. Therefore, the controller 200 may receive indications of each of a plurality of variables 306 from each power generation system 148 in the group at each sampling interval of the sampling period 304. The controller may then combine the plurality of variables 306 received from each power generation system 148 to generate a consistent environmental dataset that indicates a plurality of consistent environmental variables 306 affecting the performance of the power generation asset 100. For example, in an embodiment involving a plurality of wind turbines 114, the controller 200 may receive yaw setpoint indications from each wind turbine 114. Since the wind turbines 114 are generally optimized to be aligned parallel to the wind direction 330, the combined yaw setpoints of the specified group of wind turbines 114 can provide a consistent indication of the dominant wind direction 330 affecting the power generation asset 100. U.S. Patent Application No. 17 / 027789 provides a more comprehensive description of using data obtained from a group of power generation systems 148 to determine environmental conditions affecting the power generation asset 100. Therefore, U.S. Patent Application No. 17 / 027789, filed September 22, 2020, entitled “Systems and Methods for Controlling a Wind Turbine,” is incorporated herein in its entirety for all purposes.

[0079] Still referencing Figure 5In one embodiment, controller 200 may generate and train multiple production-evaluation models 310 for power generation asset 100. In one embodiment, generating production-evaluation models 310 may include generating a statistical algorithm or machine learning algorithm for each of the multiple production-evaluation models 310 and training the algorithm via external and operational datasets 302, 308. Thus, the statistical algorithm or machine learning algorithm may be configured to determine an optimal transfer function 341 between at least two of a plurality of variables 306 and the performance of power generation asset 100 as reflected by operational dataset 308.

[0080] For example, in one embodiment, stepwise regression can be used to generate and train at least one of the production-evaluation models 310. Typically, stepwise regression adds or removes one feature at a time to obtain the optimal regression model without overfitting. Furthermore, stepwise regression typically has two variations, including forward and backward regression, both of which are within the scope and spirit of the invention. For example, forward stepwise regression is a stepwise process of building a model by continuously adding predictor variables. At each step, models with and without potential predictor variables are compared, and a larger model is accepted only if it results in a significantly better fit to the data. Alternatively, backward stepwise regression starts with a model containing all predictor variables and removes terms that are not statistically significant in terms of modeling the response variable.

[0081] Another statistical method that can be used in one embodiment to generate and train at least one of the production-evaluation models 310 is the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm. Typically, the LASSO algorithm minimizes the probability of a negative logarithm under the constraint that the sum of the absolute values ​​of the coefficients is less than a constant. Yet another statistical algorithm that can be used to generate and train at least one of the production-evaluation models 310 is the M5 Prime (M5P) algorithm, a tree-based regression algorithm that is effective in many domains. For example, while stepwise regression might produce a single, globally linear model for the data, a tree-based regression algorithm performs logistic tests on the features to form a tree structure. Typically, the M5P algorithm uses a linear regression model at each node of the tree to provide a more specialized model.

[0082] In another embodiment, other machine learning methods that can be used to generate and train various production-evaluation models 310 may include, for example, Gaussian process models, random forest models, neural networks, deep neural networks, and / or support vector machines. Furthermore, one or more production-evaluation models 310 may be developed by the controller 200 using a generalized additive model (GAM). In one embodiment, the GAM may implement a packing and boosting method. In another embodiment, the GAM may implement a spline method. It should be understood that the system 300 may employ a combination of statistical algorithms or machine learning algorithms disclosed herein to determine a model-variable combination 318 that generates the most desired performance prediction 314 for a given operational objective.

[0083] In addition to external dataset 302, in one embodiment, controller 200 may be configured to receive multiple environmental variables 348 from environmental sensors 156 of power generation asset 100. Controller 200 can then incorporate the multiple environmental variables 348, such as those monitored by the environmental sensors 156 of power generation asset 100, into the training of production-evaluation model 310. It should be understood that the perception of multiple environmental variables 348 by power generation asset 100 affects various setpoints of power generation asset 100 and thus its performance. Therefore, including multiple environmental variables 348 monitored by power generation asset 100 in the training of production-evaluation model 310 facilitates the refinement of the relevant functions.

[0084] In one embodiment, after training the production-evaluation model 310, the production-evaluation model 310 can be tested to determine its accuracy. For ease of testing, a portion of the external and operational datasets 302 and 308 corresponding to specific sampling periods can be excluded from the dataset used in training the production-evaluation model 310. The controller 200 can then use the retained portion of the external dataset 302 to model the performance of the power generation asset 100 under the conditions indicated by the retained portion. The modeled performance can then be compared with the retained portion of the operational dataset 308 to determine the accuracy of the production-evaluation model 310.

[0085] like Figure 7 As shown, each model-variable combination 318 can generate different performance predictions 314 for the power generation asset 100. Each performance prediction 314 may differ in power generation prediction 320 and / or confidence interval 322. Differences in performance prediction 314 can be attributed to modeling differences, variable selection, and / or the sampling interval of the sampling period 304. Therefore, the optimal model-variable combination 318 can be selected from multiple model-variable combinations 318 based on the operational objectives and / or operational constraints of the power generation asset 100.

[0086] As an illustration, when system 300 is used to predict the energy production of power generation asset 100 in order to form an energy guarantee agreement, the optimal model-variable combination 318 can be the model-variable combination 318 with the narrowest confidence interval 322. In such an embodiment, the confidence interval 322 can indicate that the deviation between the power generation forecast 320 and the actual power generation of power generation asset 100 is smaller than the range achievable using other model-variable combinations 318. In other words, the selected optimal model-variable combination 318 can indicate that the actual power generation of the power generation asset will not deviate significantly from the power generation forecast 320. Therefore, the energy guarantee agreement can be formed based on the power generation forecast 320.

[0087] As further illustrative, in one embodiment, performance prediction 314 may be used to determine whether sufficient benefits can be achieved from upgrading the power generation asset 100 to justify the cost of the upgrade. In this embodiment, the optimal model-variable combination 318 may be the model-variable combination 318 with the maximum power generation prediction 320 and the narrowest confidence interval 322. In other words, in this embodiment, if the prediction is accompanied by a relatively wide confidence interval 322, the optimal model-variable combination 318 may not be the model-variable combination 318 with the maximum total power generation prediction 320.

[0088] Refer again Figure 5 In one embodiment, controller 200 may implement at least one control action 324 based on performance prediction 314. For example, in one embodiment, control action 324 may include generating an alarm. The generation of an alarm may facilitate the scheduling of maintenance events to address performance prediction 314, which is less than expected and / or includes a significant degree of potential variability. Therefore, an alarm may include audible signals, visual signals, notifications, system inputs, and / or any other system that identifies the possibility that performance expectations for power generation asset 100 may not be met, and thus identifies potential faults within power generation asset 100. It should be understood that control action 324 as described herein may also include any appropriate commands or constraints imposed on controller 200. For example, in one embodiment, control action 324 may include temporarily lowering the rating of power generation asset 100. Furthermore, in one embodiment, control action 324 may include restricting the operation of at least one component of power generation asset 100. For example, control action 324 may restrict the pitch of rotor blades 112 and / or the yaw of the nacelle 106 of wind turbine 114.

[0089] In one embodiment, a power generation asset may be one of a plurality of power generation assets 160. For example, in one embodiment, a power generation asset 100 may be one of a plurality of wind farms, and each wind farm constitutes a different power generation asset 100. In another embodiment, a power generation asset 100 may be a plurality of power generation systems 146 corresponding to a portion of a power generation facility (e.g., a wind farm), such that the power generation facility may include a plurality of power generation systems 146 (e.g., various groups of power generation systems 148).

[0090] In one embodiment, power generation asset 100 is one of a plurality of power generation assets 160, and the method disclosed herein can be repeated for each of the plurality of power generation assets 160. Repeating the steps of the method disclosed herein for each power generation asset 100 can generate a performance prediction 314 for each of the plurality of power generation assets 160. Then, in one embodiment, controller 200 can establish a ranking order 350 for each of the plurality of power generation assets 160 based on desired performance prediction characteristics. Desired performance prediction characteristics may correspond to power generation prediction 320, confidence interval 322, or a combination thereof.

[0091] In one embodiment, the implementation of control action 324 may be based on a hierarchy order 350. For example, the hierarchy order 350 may indicate which of a plurality of power generation assets 160 is most likely to meet the power generation guarantee. Therefore, relative to the remaining plurality of power generation assets 160, the identified power generation asset, which may be the highest-ranking power generation asset, may be prioritized to meet the required power production.

[0092] In one embodiment, implementing control action 324 may involve upgrading at least a portion of a plurality of power generation assets 160. Therefore, the production-evaluation model 310 can be modified / altered to take into account proposed changes / upgrades to power generation assets 100. Thus, if the considered upgrade is executed, performance prediction 314 can reflect the expected performance of each of the plurality of power generation assets 160.

[0093] In embodiments where control action 324 involves upgrading at least a portion of multiple power generation assets 160, the determination of the power generation asset 100 to be upgraded can be based at least in part on a hierarchy 350 reflecting the expected performance of performance prediction 314. Therefore, an upgrade threshold 352 can be established. The upgrade threshold 352 can correspond to a percentage increase in performance prediction 314 relative to the current performance (e.g., power generation) of the power generation asset 100, as shown in the operational dataset 308. This percentage increase can be attributed to a prospective upgrade of the power generation asset 100. As shown at 354, an increase in predicted performance greater than the upgrade threshold 352 can be identified. When the increase in predicted performance exceeds the upgrade threshold 352, as shown at 356, the corresponding power generation asset 100 can be upgraded.

[0094] In one embodiment, implementing control action 324 may include performing a diagnostic process on at least one of the plurality of power generation assets 160. In this embodiment, a diagnostic threshold 358 may be established. The diagnostic threshold 358 may indicate a percentage deficiency of the performance of power generation asset 100 relative to the performance prediction 314 of the remainder of the plurality of power generation assets 160. For example, a percentage deficiency may indicate a percentage deficiency relative to the median or other statistical measure of the performance prediction 314 of the plurality of power generation assets 160. Thus, as shown at 360, a predicted percentage deficiency greater than the threshold can be detected.

[0095] In an embodiment where the percentage deviation of the performance prediction 314 for power generation asset 100 exceeds a diagnostic threshold 358, root cause analysis 362 may be performed. Root cause analysis 362 may attempt to determine the root cause of the percentage insufficiency. This root cause analysis 362 is described more fully in U.S. Patent Application No. 17 / 032218. Therefore, for all purposes, U.S. Patent Application No. 17 / 032218, filed September 25, 2020, entitled “Systems and Methods for Operating a PowerGenerating Asset,” is incorporated herein by reference in its entirety.

[0096] Furthermore, those skilled in the art will recognize the interchangeability of various features from different embodiments. Similarly, the various method steps and features described, as well as other known equivalents for each such method and feature, can be mixed and matched by those skilled in the art to construct additional systems and techniques based on the principles disclosed herein. It should be understood, of course, that not all such objects or advantages described above may necessarily be achieved according to any particular embodiment. Therefore, for example, those skilled in the art will recognize that the systems and techniques described herein may be implemented or performed in a manner that achieves or optimizes one or more of the advantages taught herein, without necessarily achieving other objects or advantages taught or suggested herein.

[0097] 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 combined methods. The patentable scope 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 fall 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.

[0098] Other aspects of the invention are provided by the subject matter of the following provisions:

[0099] Clause 1. A method for operating a power generation asset, the method comprising: a) receiving, via a controller, at least one external dataset for a sampling period from at least one source separate from the power generation asset, the external dataset indicating multiple variables affecting the performance of the power generation asset; b) via the controller, receiving at least one operational dataset for the power generation asset for the sampling period, the at least one operational dataset indicating the performance of the power generation asset; c) via the controller, generating multiple production-evaluation models for the power generation asset; d) via the controller, training the multiple production-evaluation models, via the at least one external dataset and the at least one operational dataset, to correlate the performance of the power generation asset as a function of multiple variables; e) via the controller, generating a performance prediction for the implementation period for each of multiple model-variable combinations, wherein the multiple model-variable combinations include multiple combinations of each production-evaluation model and multiple variables, wherein each of the performance predictions includes a power generation prediction and a confidence interval; and f) via the controller, implementing a control action based on one of the performance predictions.

[0100] Clause 2. The method according to Clause 1, wherein the plurality of variables includes at least data indicating wind speed and wind direction at sampling intervals of the sampling period.

[0101] Clause 3. The method according to any of the preceding clauses, wherein the plurality of variables further includes data indicating at least one of time correlation, temperature, air density, wind shear, wind direction shift, and turbulence intensity.

[0102] Clause 4. The method according to any of the preceding clauses, wherein generating the plurality of production-evaluation models further includes generating a statistical algorithm or machine learning algorithm for each of the plurality of production-evaluation models, the statistical algorithm or machine learning algorithm being configured to determine an optimal transfer function between at least two of the plurality of variables and the performance of the power generation asset.

[0103] Clause 5. The method according to any of the preceding clauses, wherein the at least one external dataset and the at least one operational dataset are each generated with a first sampling interval and a second sampling interval, and wherein generating the plurality of production-evaluation models further comprises: generating a first portion of the plurality of production-evaluation models based on the at least one external dataset and the at least one operational dataset having a first sampling interval; and generating a second portion of the plurality of production-evaluation models based on the at least one external dataset and the at least one operational dataset having a second sampling interval, wherein the second sampling interval has a higher frequency relative to the first sampling interval.

[0104] Clause 6. The method according to any of the preceding clauses, wherein the at least one external dataset includes a modeled environmental dataset indicating multiple environmental variables affecting the power generation asset.

[0105] Clause 7. The method according to any of the preceding clauses, wherein the at least one external dataset comprises an environmental dataset assembled from a group of power generation systems, the method further comprising: receiving via the controller an indication of each of a plurality of variables from each of the power generation systems in the group at each sampling interval of the sampling period; and combining via the controller the plurality of variables received from each of the plurality of power generation systems in the group to generate a consistent environmental dataset of a plurality of consistent environmental variables indicating the performance of the power generation asset.

[0106] Clause 8. The method according to any of the preceding clauses, wherein the at least one external dataset comprises multiple environmental measurements collected by a weather mast.

[0107] Clause 9. The method according to any of the preceding clauses, wherein training the plurality of production-assessment models further includes training the plurality of production-assessment models via a plurality of environmental variables obtained via environmental sensors of the power generation assets.

[0108] Clause 10. The method according to any of the preceding clauses, wherein at least one of the at least external dataset and the at least one operational dataset further includes at least one anomalous input for a sampling interval of the sampling period, the method further comprising: generating an interpolated value for the anomalous input for the sampling interval via an interpolation algorithm of the controller.

[0109] Clause 11. The method according to any of the preceding clauses, wherein the power generation asset includes a wind turbine.

[0110] Clause 12. The method according to any of the preceding clauses, wherein the power generation asset comprises a plurality of power generation systems.

[0111] Clause 13. The method according to any of the preceding clauses, wherein the power generation asset includes one of a wind farm, a solar power generation facility and a hybrid power generation facility.

[0112] Clause 14. The method according to any of the preceding clauses, wherein the power generation asset is one of a plurality of power generation assets, the method further comprising: repeating steps a) to e) for each of the plurality of power generation assets; and establishing a hierarchical order for each of the plurality of power generation assets via the controller based on desired performance prediction characteristics, wherein the implementation of the control action is based on the hierarchical order.

[0113] Clause 15. The method according to any of the preceding clauses, wherein implementing the control action includes upgrading at least one of the plurality of power generation assets, the method further comprising: establishing an upgrade threshold corresponding to a percentage increase in the performance forecast relative to the performance of the power generation asset as shown in the at least one operational dataset, the percentage increase being attributable to an anticipated upgrade of the power generation asset; and upgrading the power generation capacity of the at least one power generation asset having a percentage increase in the performance forecast greater than the upgrade threshold.

[0114] Clause 16. The method according to any of the preceding clauses, wherein implementing the control action includes performing a diagnostic process on at least one of the plurality of power generation assets, the method further comprising: establishing a diagnostic threshold indicating a percentage deficiency of the performance of the at least one power generation asset as shown by the at least one operational dataset relative to the predicted performance of the plurality of power generation assets; and performing a root cause analysis to determine the root cause of the percentage deficiency.

[0115] Clause 17. A system for operating a power generation asset, the system comprising: at least one sensor operatively coupled to the power generation asset; and a controller communicatively coupled to the at least one sensor, the controller including at least one processor configured to perform a plurality of operations, the plurality of operations including: a) receiving from at least one source separate from the power generation asset for a sampling period, the external dataset indicating a plurality of variables affecting the performance of the power generation asset, the plurality of variables including at least data indicating wind speed and wind direction at sampling intervals of the sampling period; b) receiving at least one operation for the power generation asset for the sampling period. The dataset, wherein the at least one operational dataset indicates the performance of the power generation asset; c) generating multiple production-assessment models for the power generation asset; d) training the multiple production-assessment models via the at least one external dataset and the at least one operational dataset to correlate the performance of the power generation asset as a function of multiple variables; e) generating performance predictions for each of the multiple model-variable combinations for the implementation period, wherein the multiple model-variable combinations include multiple combinations of each production-assessment model and multiple variables, wherein each of the performance predictions includes a power generation prediction and a confidence interval; and f) implementing control actions based on one of the performance predictions.

[0116] Clause 18. A system pursuant to any of the preceding clauses, wherein the plurality of variables further includes data indicating at least one of time correlation, wind shear, and turbulence intensity.

[0117] Clause 19. A system according to any of the preceding clauses, wherein the at least one external dataset and the at least one operational dataset are each generated at a first sampling interval and a second sampling interval, and wherein generating the plurality of production-evaluation models further comprises: generating a first portion of the plurality of production-evaluation models based on the at least one external dataset and the at least one operational dataset having a first sampling interval; and generating a second portion of the plurality of production-evaluation models based on the at least one external dataset and the at least one operational dataset having a second sampling interval, wherein the second sampling interval has a higher frequency relative to the first sampling interval.

[0118] Clause 20. A system pursuant to any of the preceding clauses, wherein implementing the control action includes implementing a diagnostic process, the method further comprising: establishing a diagnostic threshold indicating a percentage shortfall in the performance of the power generation asset relative to the performance prediction, as shown by the at least one operational dataset; and performing root cause analysis to determine the root cause of the percentage shortfall.

Claims

1. A method for operating a power generation asset, the method comprising: a) Receive, via a controller, at least one external dataset for a sampling period from at least one source separate from the power generation asset, the external dataset indicating multiple variables affecting the performance of the power generation asset; b) Receive, via the controller, at least one operational dataset for the power generation asset for the sampling period, the at least one operational dataset indicating the performance of the power generation asset; c) Generate multiple production-evaluation models for the power generation asset via the controller; d) The plurality of production-evaluation models are trained via the controller, via the at least one external dataset and the at least one operational dataset, to correlate the performance of the power generation assets as a function of the plurality of variables; e) via the controller, generate performance forecasts for the implementation period for each of a plurality of model-variable combinations, wherein the plurality of model-variable combinations includes multiple combinations of each of the production-evaluation models and the plurality of variables, wherein each of the performance forecasts includes a power generation forecast and a confidence interval; and f) Implement control actions via the controller based on one of the performance predictions. Wherein, the at least one external dataset and the at least one operational dataset are each generated at a first sampling interval and a second sampling interval, and wherein generating the plurality of production-evaluation models further includes: The first part of the plurality of production-evaluation models is generated based on the at least one external dataset having a first sampling interval and the at least one operational dataset; and A second part of the plurality of production-evaluation models is generated based on the at least one external dataset and the at least one operational dataset having a second sampling interval, wherein the second sampling interval has a higher frequency relative to the first sampling interval.

2. The method according to claim 1, wherein, The plurality of variables includes at least data indicating wind speed and wind direction at sampling intervals of the sampling period.

3. The method according to claim 2, wherein, The variables also include data indicating at least one of time correlation, temperature, air density, wind shear, wind direction shift, and turbulence intensity.

4. The method according to claim 1, wherein, Generating the plurality of production-evaluation models also includes generating a statistical algorithm or machine learning algorithm for each of the plurality of production-evaluation models, the statistical algorithm or machine learning algorithm being configured to determine an optimal transfer function between at least two of the plurality of variables and the performance of the power generation asset.

5. The method according to claim 1, wherein, The at least one external dataset includes a modeled environmental dataset that indicates multiple environmental variables affecting the power generation asset.

6. The method according to claim 1, wherein, The at least one external dataset includes an environmental dataset assembled from a set of electronic systems, and the method further includes: The controller receives indications from each of a plurality of variables of each of the generating electronic systems in the group at each sampling interval of the sampling period; and The controller combines multiple variables received from each of the power generation systems in the group to generate a consistent environmental dataset that indicates multiple consistent environmental variables affecting the performance of the power generation assets.

7. The method according to claim 1, wherein, The at least one external dataset includes multiple environmental measurements collected by the meteorological mast.

8. The method according to claim 1, wherein, Training the multiple production-evaluation models also includes training the multiple production-evaluation models using multiple environmental variables obtained via environmental sensors of the power generation assets.

9. The method according to claim 1, wherein, At least one of the at least external dataset and the at least one operational dataset further includes at least one anomalous input for the sampling interval of the sampling period, and the method further includes: The interpolation algorithm of the controller generates interpolated values ​​for the abnormal input at the sampling interval.

10. The method according to claim 1, wherein, The power generation assets include wind turbines.

11. The method according to claim 1, wherein, The power generation assets include multiple power generation systems.

12. The method according to claim 1, wherein, The power generation assets include one of the following: wind farms, solar power facilities, and hybrid power generation facilities.

13. The method according to claim 1, wherein, The power generation asset is one of a plurality of power generation assets, and the method further includes: Repeat steps a) to e) for each of the plurality of power generation assets, and Based on the desired performance prediction characteristics, a hierarchical order is established for each of the plurality of power generation assets via the controller, wherein the implementation of the control actions is based on the hierarchical order.

14. The method according to claim 13, wherein, Implementing the control action includes upgrading at least one of the plurality of power generation assets, and the method further includes: Establish an upgrade threshold corresponding to a percentage increase in the predicted performance relative to the performance of the power generation asset as shown by the at least one operational dataset, where the percentage increase is attributable to the anticipated upgrade of the power generation asset; and Upgrade the power generation capacity of the at least one power generation asset, wherein the at least one power generation asset has a percentage increase in the predicted performance greater than the upgrade threshold.

15. The method according to claim 13, wherein, Implementing the control action includes performing a diagnostic process on at least one of the plurality of power generation assets, and the method further includes: Establish a diagnostic threshold indicating the percentage shortfall in the performance of the at least one power generation asset relative to the performance predictions of the plurality of power generation assets, as shown by the at least one operational dataset; and Perform root cause analysis to determine the root causes of the inadequacy percentage.

16. A system for operating power generation assets, the system comprising: At least one sensor is operatively connected to the power generation asset; as well as A controller communicatively coupled to the at least one sensor, the controller including at least one processor configured to perform a plurality of operations, the plurality of operations including: a) Receive at least one external dataset for a sampling period from at least one source separate from the power generation asset, the external dataset indicating multiple variables affecting the performance of the power generation asset, the multiple variables including at least data indicating wind speed and wind direction at sampling intervals of the sampling period; b) Receive at least one operational dataset for the power generation asset for the sampling period, the at least one operational dataset indicating the performance of the power generation asset; c) Generate multiple production-evaluation models for the power generation assets; d) Train the plurality of production-evaluation models via the at least one external dataset and the at least one operational dataset to correlate the performance of the power generation assets as a function of multiple variables; e) Generate performance forecasts for the implementation period for each of a plurality of model-variable combinations, wherein the plurality of model-variable combinations includes multiple combinations of each of the production-evaluation models and the plurality of variables, wherein each of the performance forecasts includes a power generation forecast and a confidence interval; and f) Implement control actions based on one of the performance predictions. Wherein, the at least one external dataset and the at least one operational dataset are each generated at a first sampling interval and a second sampling interval, and wherein generating the plurality of production-evaluation models further includes: The first part of the plurality of production-evaluation models is generated based on the at least one external dataset having a first sampling interval and the at least one operational dataset; and A second part of the plurality of production-evaluation models is generated based on the at least one external dataset and the at least one operational dataset having a second sampling interval, wherein the second sampling interval has a higher frequency relative to the first sampling interval.

17. The system according to claim 16, wherein, The variables also include data indicating at least one of time correlation, wind shear, and turbulence intensity.

18. The system according to claim 16, wherein, Performing the control action includes performing a diagnostic process, and the plurality of operations further include: Establish a diagnostic threshold indicating the percentage shortfall in the performance of the power generation asset relative to the performance prediction, as shown by the at least one operational dataset; and Perform root cause analysis to determine the root causes of the inadequacy percentage.

Images