Wind turbine power production prediction method and system

Abstract

A method for predicting power production of a wind turbine (10) includes obtaining local weather data at an area of the wind turbine over a first time period, obtaining ice formation data of ice formed on blades (30) of the wind turbine (10) over a second time period, obtaining power production data of power output of the wind turbine over a third time period, and predicting power production of the wind turbine at a future time point based on the obtained power production data, the obtained ice formation data, the obtained local weather data as well as current weather forecast data for the future time point and current ice formation data.

Description

[0001] WIND TURBINE POWER PRODUCTION PREDICTION METHOD AND SYSTEM

[0002] Field

[0003] The technology relates to the field of wind energy and especially wind turbine power production. It focuses on the prediction and optimization of power output from wind turbines under various weather conditions, including ice formation on the turbine blades.

[0004] Background

[0005] Wind turbines are a popular and effective source of renewable energy, harnessing the power of the wind to generate electricity. However, the power output of a wind turbine is highly dependent on weather conditions, particularly wind speed and temperature. Accurate prediction of power production from a wind turbine is crucial for efficient management of the electrical grid, for optimizing the operation of the wind turbine itself as well as for offering a correct amount of power from wind turbines in power trading markets.

[0006] One of the challenges in predicting wind turbine power output is the impact of ice formation on the blades of the turbine. Ice accumulation on the blades can significantly reduce the efficiency of the wind turbine, leading to a decrease in power production. In cold and icy conditions, it is essential to accurately predict the effects of ice formation on the power output of the wind turbine.

[0007] Several methods have been proposed in the prior art for predicting ice build-up on wind turbines and forecasting the yield of wind farms during icing conditions. For example, US patent application US2020 / 0386206 describes a method that uses sensors, such as vibration or acceleration sensors, to detect ice build-up on the wind turbine. The method also involves obtaining meteorological data of the wind farm and feeding this data, along with the ice build-up data, into a machine learning model (MLM) to create a prediction model. The MLM is then used to predict future ice build-up and the probability of having to slow down or switch off the wind turbine due to ice accumulation. European patent application EP3524813 also discusses the prediction of ice build-up on wind turbines based on historical ice build-up information and meteorological data to build up an MLM for ice prediction.

[0008] However, the prior art has several shortcomings. For example, the accuracy of weather forecasting algorithms is often limited, leading to potential errors in power production predictions. Furthermore, current prediction models may not adequately account for the impact of ice accumulation on the aerodynamics and weight of wind turbine blades, which can lead to underestimating the reduction in power output due to the additional drag and weight caused by ice. There is also a lack of historical data regarding turbine performance in cold and icy conditions, which is necessary to train and validate the models. Without sufficient historical data, the models may not be able to accurately predict the effects of such conditions on power output. Finally, the prediction models might not be sufficiently tailored to local weather patterns and microclimates, which can vary significantly and have a profound impact on wind turbine performance, especially in areas prone to rapid icing.

[0009] In light of these challenges, there is a need for an improved method and system for predicting power production from a wind turbine, taking into account weather conditions, including ice formation on the blades.

[0010] Summary

[0011] According to a first aspect of the disclosure, a method for predicting power production of a wind turbine is disclosed. This method comprises obtaining local weather data at the area of the wind turbine over a first time period. The method also involves obtaining ice formation data of ice formed on the blades of the wind turbine over a second time period and / or the ice formation data being current ice formation data, wherein information on the ice formation data originates from one or more ice formation sensors positioned at the wind turbine. Furthermore, the method includes obtaining power production data of the power output of the wind turbine over a third time period. The power production of the wind turbine at a future time point is then predicted based on the obtained power production data, the obtained ice formation data, the obtained local weather data as well as current weather forecast data for the future time point. This method allows for more accurate predictions of power production, which can lead to more efficient operation and maintenance of the wind turbine as well as a better knowledge on how to act on power trading markets. Especially, as the ice formation data originates from ice formation sensors positioned at the wind turbine, the method has data on ice that is actually formed on blades of the wind turbine, not just any prognosis of how much ice it may be formed on blades at a certain weather situation. Hereby, the claimed method for predicting power production of a wind turbine has shown to provide reliable predictability.

[0012] Optionally in some examples, the method further comprises obtaining de-icing system operational status data regarding the power status of a de-icing system for de-icing the blades of the wind turbine over a fourth time period. The prediction of the power production of the wind turbine at the future time point is also based on the obtained deicing system operational status data and current de-icing system operational status data. This additional data can provide valuable information about the impact of the deicing system on the power production of the wind turbine.

[0013] Optionally in some examples, the method further comprises obtaining weather forecast data for the area of the wind turbine, the weather forecast data being for a fifth time period. The prediction of the power production of the wind turbine at the future time point is also based on the obtained weather forecast data. This allows for the consideration of future weather conditions in the prediction of power production, which can further improve the accuracy of the prediction. This also allows for determining how weather forecasts influenced the power production at the future time for which the weather forecast was intended, including for example how well the weather forecast matched the actual weather at the future time point.

[0014] Optionally in some examples, the prediction of the power production is performed by a machine learning model using the obtained data. This can allow for more complex and accurate predictions, as the machine learning model can learn from the obtained data and improve its predictions over time. Optionally in some examples, the local weather data comprises one or more of wind speed, temperature, wind direction, precipitation amount, precipitation type, and atmospheric humidity. This provides a comprehensive overview of the local weather conditions, which can have a significant impact on the power production of the wind turbine.

[0015] Optionally in some examples, the one or more ice formation sensors comprise a camera, the information on the ice formation data is images captured by the camera of the blades of the wind turbine, and the ice formation data is determined by image processing of the images captured by the camera. This provides a direct and accurate way of determining ice formation on the blades, which can significantly affect the power production of the wind turbine. Determining ice formation on blades by image processing of images captured by the camera has turned out to have a very high reliability. In one embodiment, the camera may be a camera for visible light, i.e. for the wavelength range around 380 - 780 nm. In another embodiment, which may or may not be used together with the visible light camera, the camera may be a camera for infrared light, i.e. for the wavelength range around 800 - 1700 nm. By illuminating the blades of the wind turbine with monochromatic light in this wavelength range, from e.g. one or more lasers, at e.g., 980, 1310 and / or 1550 nm (for which there are commercially available lasers), image processing can be used to detect ice formation. This may be done in a similar way as done for the visible light case. Further, image processing and classification algorithms may be used to detect ice formation.

[0016] Optionally, in some examples, the one or more ice formation sensors comprise one or more vibration sensors that sense vibrations of the blades of the wind turbine. Further, the information on the ice formation data is data on the vibrations sensed, and the ice formation data is determined based on the data on the sensed vibrations. The one or more vibration sensors may be mounted in the blades. The one or more vibration sensors may be accelerometers. By filtrating away the rotating movement of the blades, the vibrations in the blades caused by outer factors such as wind or ice / snow formation can be isolated. When ice / snow is gathered on the blades, the frequency of vibrations is changed, which change in frequency can be determined using a frequency analysis of the sensed vibrations. How much the frequency changes depends on the amount, type and placement of ice / snow, which means the ice formation can be determined from the frequency of the isolated vibrations. Also, in some use cases the amplitude of the vibration changes depending on amount of ice / snow and placement of ice / snow on the blades, so also the amplitude of the isolated vibrations can be taken into account for determining the ice formation on the blades.

[0017] Optionally, in some examples, the one or more ice formation sensors comprise one or more microphones. Further, the information on the ice formation data is data on sound originating from the blades and registered by the one or more microphones. Further, the ice formation data is determined based on the data on sound registered by the one or more microphones. The one or more microphones may be positioned in the nacelle or in the blades. As sound is a kind of vibration, a similar principle as for the vibration sensors may be used. When ice / snow is gathered on the blades, the frequency of sound is changed. By using a Fast Fourier Transform (FFT) on the sound signals, the different frequencies of the sound detected by the microphones are obtained. Changes in those frequencies can be connected to the formation of ice and snow. It is advantageous to use such sound frequency analysis together with a camera system, such as with visible light or infrared light, to verify how the different sound frequencies relate to a certain type of ice / snow formation and to build reliable models on that.

[0018] Optionally in some examples, each obtained local weather data, ice formation data, power production data, de-icing system operational status data, and weather forecast data is time-stamped. This allows for a clear and organized collection of data, which can facilitate the analysis and prediction process.

[0019] Optionally in some examples, the first time period, the second time period, the third time period, the fourth time period, and the fifth time period are at least one month, more preferably at least four months. This ensures a comprehensive collection of data over a significant period of time, which can improve the accuracy of the predictions.

[0020] Optionally in some examples, the local weather data, the ice formation data, the power production data, and the de-icing system operational status data each comprise at least 1000 data points, preferably at least 10000, most preferably at least 20000, distributed over the respective first, second third and fourth time period. This large amount of data can provide a more detailed and accurate picture of the conditions affecting the power production of the wind turbine.

[0021] Optionally in some examples, the weather forecast data comprises at least 200 data points spread out over the fifth time period. This provides a detailed forecast of the weather conditions, which can be for accurate power production predictions.

[0022] Optionally in some examples, the prediction of the power production at the future time point is further based on the obtained power production data and current power production data. This allows the prediction to take into account the current power production of the wind turbine, which can improve the accuracy of the prediction.

[0023] According to a second aspect of the disclosure, a system for predicting power produced by a wind turbine is disclosed. The system comprises a control unit configured to perform the method as described above. This system can provide a practical and efficient way of implementing the method, and can facilitate the obtaining, analysis, and prediction of data.

[0024] Optionally in some examples, the system further comprises a statistical database configured to store the obtained local weather data, the obtained ice formation data, the obtained power production data, and optionally the obtained de-icing system operational status data and the obtained weather forecast data. This database can provide a centralized and organized storage of the obtained data, which can facilitate the analysis and prediction process.

[0025] Optionally in some examples, the system further comprises an image processor configured to be arranged at the wind turbine and configured to detect ice formation on the blades of the wind turbine using image recognition algorithms to analyze images captured by a camera. This can provide a direct and accurate way of determining ice formation on the blades, which can significantly affect the power production of the wind turbine. Optionally in some examples, the system further comprises a camera configured to be arranged at the wind turbine for capturing images of the blades of the wind turbine. This allows for the direct observation of the blades, which can provide valuable information about ice formation and other conditions affecting the power production of the wind turbine.

[0026] Optionally in some examples, the control unit comprises a memory, a processor, and a network interface for obtaining the data and predicting the power production and, optionally, communicating with a statistical database. This provides the necessary hardware for the control unit to perform its functions, and can facilitate the obtaining, analysis, and prediction of data.

[0027] Brief Description of the Drawings

[0028] Examples are described in more detail below with reference to the appended drawings. Figure 1 is a schematic representation of a wind turbine system equipped with the power prediction method and system.

[0029] Figure 2 is a flowchart illustrating the steps of the power prediction method.

[0030] Figure 3 is a detailed diagram of the control unit and its components used in the power prediction system.

[0031] Detailed Description

[0032] The detailed description set forth below provides information and examples of the disclosed technology with sufficient detail to enable those skilled in the art to practice the disclosure.

[0033] Figure 1 shows a schematic representation of a wind turbine system equipped with a power prediction method and system according to embodiments. The wind turbine 10 is shown with its main components including blades 30, hub 35, and nacelle 20. The wind turbine 10 is equipped with a de-icing system 80, which is arranged at the blades 30 for de-icing purposes. The wind turbine 10 also includes one or more ice formation sensors 40 positioned at the wind turbine and used for collecting information that is to be used for determining ice formation data, or for obtaining ice formation data directly. The wind turbine 10 also comprises a local processor 50 for processing the information collected by the ice formation sensors into ice formation data. Alternatively, there is no such local processor 50, wherein the one or more ice formation sensors 40 communicate the information on the ice formation data to a control unit 60 that determines the ice formation data based on the information on the ice formation data . The wind turbine 10 also includes a control unit 60. The control unit 60 is responsible for obtaining the data, including the ice formation data, and predicting the power production, and optionally, communicating with a statistical database 70. The statistical database 70 is configured to store the obtained local weather data, the obtained ice formation data, the obtained power production data, and optionally the obtained deicing system operational status data and the obtained weather forecast data.

[0034] Figure 2 is a flowchart illustrating the steps of the power prediction method. The method starts with obtaining 102 local weather data at an area of the wind turbine 10 over a first time period. The method then involves obtaining 104 ice formation data of ice formed on the blades 30 of the wind turbine 10 over a second time period. Alternatively, or in combination with the obtaining of ice formation data over the second time period, the obtained ice formation data may be current ice formation data. Further, information on the ice formation data originates from one or more ice formation sensors 40 positioned at the wind turbine. The method also includes obtaining 106 power production data of power output of the wind turbine 10 over a third time period. The method may also comprise obtaining 108 de-icing system operational status data regarding power status of a de-icing system 80 for de-icing the blades 30 of the wind turbine 10, the de-icing system operational status data being determined over a fourth time period. The method may also comprise obtaining 110 weather forecast data for the area of the wind turbine 10, the weather forecast data being for a fifth time period. The method further includes predicting 112 power production of the wind turbine 10 at a future time point based on the obtained power production data, the obtained ice formation data, the obtained local weather data as well as current weather forecast data for the future time point. The predicting 112 may also be based on the obtained de-icing system operational data and / or the obtained weather forecast data. Figure 3 is a detailed diagram of the control unit 60 and its components used in the power prediction system. The control unit 60 includes a memory 67, a processor 65, and a network interface 68. The memory 67 is used for storing the obtained data. The processor 65 is used for processing the obtained data and predicting the power production. The network interface 68 is a communication interface used for obtaining the data and, optionally, communicating with the statistical database 70.

[0035] 1 . Component Details. The disclosure includes several components that work together to predict the power production of a wind turbine 10. These components include a system for predicting power produced by a wind turbine 10, a statistical database 70, a control unit 60, , and other components that are part of the wind turbine 10 itself. Each of these components has specific features and functions that contribute to the overall operation of the disclosure.

[0036] 1.1. System For Predicting Power Produced by A Wind Turbine. The system for predicting power produced by a wind turbine 10 is a comprehensive system that obtains and analyzes various types of data to predict the power production of the wind turbine 10. The system includes several components, including a statistical database 70 and a control unit 60, ice formation sensors 40 and a local processor 50 . The system is designed to obtain data over specific time periods, analyze the obtained data, and predict the power production of the wind turbine 10 based on the analyzed data.

[0037] 1.1.1. Statistical Database. The statistical database 70 is a component of the system for predicting power produced by a wind turbine 10. The statistical database 70 is configured to store various types of obtained data, including local weather data, ice formation data, power production data, de-icing system operational status data, and weather forecast data. The data obtained is obtained over time, preferably at a large number of different time points. In some implementations, the statistical database 70 may store each obtained data with a timestamp, providing a chronological record of the data. The statistical database 70 may also store a large number of data points for each type of data, providing a comprehensive dataset for analysis. The large number of data points at least partly relates to different time points of observance. 1.1.2. Control Unit. The control unit 60 is another component of the system for predicting power produced by a wind turbine 10. The control unit 60 is responsible for obtaining the data, storing the obtained data in the statistical database 70, and predicting the power production of the wind turbine 10 based on the obtained and stored data. The control unit 60 includes several sub-components, including a memory 67, a processor 65, and a network interface 68.

[0038] 1 .1 .2.1 . Memory. The memory 67 is a sub-component of the control unit 60. The memory 67 may be used for storing the obtained data before it is stored in the statistical database 70. The memory 67 may be a volatile or non-volatile memory, and it may have a large storage capacity to store a large number of data points for each type of data. The memory 67 may also store a computer program 69 comprising instructions, which, when executed by the processor 65 causes the control unit 60 to perform the obtaining and storing of obtained data as well as the prediction of the power production of the wind turbine.

[0039] 1 .1 .2.2. Processor. The processor 65 is another sub-component of the control unit 60. The processor 65 is used for processing the obtained data and predicting the power production of the wind turbine 10. In some examples, the processor 65 may use a defined algorithm or model to analyze the obtained data and predict the power production. The processor 65 may also be configured to perform machine learning techniques to improve the accuracy of the power production prediction.

[0040] 1 .1 .2.3. Network Interface. The network interface 68 is a further sub-component of the control unit 60. The network interface 68 is used for obtaining the data from various sources, such as weather stations, the ice formation sensors, power output meters, de-icing systems, and weather forecast systems. The network interface 68 may also be used for communicating with the statistical database 70, allowing the control unit 60 to store the obtained data in the statistical database 70 and retrieve the stored data for analysis. The network interface 68 may be a communication interface enabling communication with the different sources. 1.1.3. Ice formation sensor(s). The one or more ice formation sensors 40 are positioned at, in and / or on the wind turbine 10 to obtain information for determining the ice formation data, i.e. information that is received at the wind turbine at the time when ice formation is to be determined. The information are information on the ice formation data, i.e. information that can be processed into data that shows the formation of ice at that time. The one or more ice formation sensors 40 can be any kind of sensor that can be used to collect information that describes the amount of ice formation on the blades 30 either directly or indirectly. In one embodiment, the one or more ice formation sensors 40 comprises a camera. The camera is arranged at the wind turbine 10, specifically on the hub 35 or the nacelle 20, or alternatively at a position outside of the wind turbine 10 where it can supervise the blades 30. The camera is used for capturing images of the blades 30 of the wind turbine 10. These images are then processed by an image processor to detect ice formation on the blades 30. In some examples, the camera may be a high-resolution camera capable of capturing detailed images of the blades 30, allowing for accurate detection of ice formation. The camera may be arranged to capture images in one or more of different wavelength ranges such as visible light, infrared, ultraviolet etc. In another embodiment, the one or more ice formation sensors 40 comprise one or more vibration sensors arranged to sense vibrations of the blades 30 of the wind turbine 10. The vibration sensors may be mounted in one or more of the blades 30. Data on the sensed vibrations are sent to a local processor 50 for being processed into ice formation data. In another embodiment, the one or more ice formation sensors 40 comprise one or more microphones that senses sound originating from the blades 30. The one or more microphones may be mounted for example in the nacelle 20. Data on the sensed sound are sent to a local processor 50 for being processed into ice formation data.

[0041] 1.1.4. Local Processor. The local processor 50 is a component of the system for determining ice formation data based on the information on ice formation data collected by the one or more ice formation sensors 40 The local processor 50 is arranged at the wind turbine 10, preferably in the nacelle 20 or in a tower of the wind turbine 10. The local processor 50 determines the ice formation data that is obtained by the control unit 60, which is used in predicting the power production of the wind turbine 10. In case the one or more ice formation sensors 40 comprises a camera, the local processor 50 comprises an image processor that uses image recognition algorithms to analyze images captured by the camera, which is also arranged at the wind turbine 10. The image recognition algorithms that the image processor uses may be one or more of neural network algorithms, image segmentation algorithms, image edge detection algorithms, and other image recognition algorithms known by a person skilled in the art of image recognition. In case the one or more ice formation sensors 40 comprise one or more vibration sensors, the local processor 50 determines the ice formation data based on the sensed vibrations, by isolating the vibrations originating from ice formation on the blades from other vibrations and analyzing the frequency and possibly also the amplitude of the isolated vibrations. In case the one or more ice formation sensors 40 comprise one or more microphones, the local processor 50 determines the ice formation data based on the data on the sensed sound received from the one or more microphones, by using e.g. FFT on the sound signals to obtain the frequencies of the sensed sound, and analyzing changes in the obtained frequencies that refers to ice formation on the blades.

[0042] 1 .2. Wind Turbine. The wind turbine 10 is the subject of the power production prediction. The wind turbine 10 includes several components, including a tower, blades 30, a hub 35, a nacelle 20, and a de-icing system 80. Each of these components has specific features and functions that contribute to the operation of the wind turbine 10 and the power production prediction.

[0043] 1 .2.1 . De-Icing System. The de-icing system 80 is a component of the wind turbine 10. The de-icing system 80 is preferably arranged at the blades 30 of the wind turbine 10 and is used for de-icing the blades 30. The de-icing system 80 may include a heating element or other de-icing mechanism that removes ice from the blades 30, allowing the wind turbine 10 to operate efficiently even in icy conditions. In some implementations, the de-icing system 80 may provide de-icing system operational status data to the control unit 60, which is used in predicting the power production of the wind turbine 10. The de-icing system operational status data may include the on / off status of the de-icing system 80 and, if applicable, a power level at which the de-icing system 80 was operating.

[0044] 1 .2.2. Blades. The blades 30 are a prominent component of the wind turbine 10. The blades 30 are arranged to the hub 35 of the wind turbine 10 and are responsible for capturing the wind energy and converting it into mechanical energy. The blades 30 are also the subject of ice formation, which can affect the efficiency of the wind turbine 10. In some examples, the wind turbine 10 may have at least two, preferably three blades 30. The blades 30 may be made of a durable material capable of withstanding harsh weather conditions, including high wind speeds and icy conditions. The blades 30 may also be designed with a specific shape and size to maximize the capture of wind energy.

[0045] 1 .2.3. Hub. The hub 35 is another component of the wind turbine 10. The hub 35 is arranged at the wind turbine 10 and is where the blades 30 are attached. The hub 35 and the blades 30 are rotatably arranged to the nacelle 20, allowing the blades 30 to rotate and capture wind energy. The hub 35 may be designed to securely hold the blades 30 and to facilitate the rotation of the blades 30. In some examples, the hub 35 may also include a mechanism for adjusting the angle of the blades 30 to optimize the capture of wind energy.

[0046] 1.2.4. Nacelle. The nacelle 20 is a further component of the wind turbine 10. The nacelle 20 is arranged to the wind turbine 10 so that it extends substantially horizontally when the wind turbine has been erected. The nacelle 20 is arranged on top of the tower. The nacelle 20 may house components such as a generator and a gearbox, which are used to convert the mechanical energy captured by the blades 30 into electrical energy. The nacelle 20 may also comprise the local processor 50. The nacelle 20 may be designed to protect these components from the weather and to facilitate the operation of the wind turbine 10.

[0047] 2. Method for Predicting Power Production of a Wind Turbine. The method for predicting power production of a wind turbine 10 involves several steps, including the obtaining of various types of data and the prediction of power production based on the obtained data. The method is designed to accurately predict the power production of the wind turbine 10, taking into account various factors that can affect the power production.

[0048] 2.1 . Obtaining of Local Weather Data. The first step of the method involves obtaining 102 local weather data at an area of the wind turbine 10. The local weather data is determined over a first time period. The local weather data may include one or more of wind speed, temperature, wind direction, precipitation amount, precipitation type, and atmospheric humidity. Each obtained local weather data is time-stamped, providing a chronological record of the weather conditions at the area of the wind turbine 10. The local weather data provides valuable information about local weather conditions that directly or indirectly affect the power production of the wind turbine 10.

[0049] 2.1 .1 . Time Period and Data Points for Weather Data Obtaining. The local weather data is obtained 102 over a first time period, which is at least one month, more preferably at least four months. A longer time period of at least four months is preferred to ensure a more comprehensive dataset. The local weather data comprises at least 1000 data points, preferably at least 10000, most preferably at least 20000. The data points are spread out over the first time period, providing a detailed record of the weather conditions over the first time period.

[0050] 2.2. Obtaining of Ice Formation Data. The second step of the method involves obtaining 104 ice formation data of ice formed on the blades 30 of the wind turbine 10. The ice formation data may be determined over a second time period. The second time period may be the same time period as the first time period, or it may be a different time period. The ice formation data may be current ice formation data. The ice formation data is determined based on information originating from one or more ice formation sensors 40 positioned at the wind turbine. According to an embodiment, the ice formation data is determined by image processing of images captured of the blades 30 of the wind turbine 10. According to another embodiment, the ice formation data is determined by analyzing vibrations sensed by one or more vibration sensors positioned on the blades of the wind turbine. According to another embodiment, the ice formation data is determined by analyzing sound detected by one or more microphones positioned at the wind turbine. According to an embodiment, each obtained ice formation data is time-stamped, providing a chronological record of the ice formation on the blades 30. The ice formation data provides valuable information about the ice conditions on the blades 30, which can affect the efficiency of the wind turbine 10. The obtained ice formation data may be the data obtained over the second time period. The obtained data may also or alternatively be current ice formation data. “Current” here means latest obtained ice formation data from the wind turbine. When predicting the power production at the future time point, the historical data obtained over the second time period and / or the current ice formation data may be taken into consideration.

[0051] 2.2.1. Time Period and Data Points for Ice Formation Data Obtaining. As mentioned, the ice formation data may be obtained 104 over a second time period, which is at least one month, more preferably at least four months. A longer time period of at least four months is preferred to ensure a more comprehensive dataset. The ice formation data comprises at least 1000 data points, preferably at least 10000, most preferably at least 20000. The data points are spread out over the second time period, providing a detailed record of the ice formation on the blades 30 over the second time period.

[0052] 2.3. Obtaining of Power Production Data. The third step of the method involves obtaining 106 power production data of power output of the wind turbine 10. The power production data is determined over a third time period. Each obtained power production data is time-stamped, providing a chronological record of the power output of the wind turbine 10. The power production data provides valuable information about the power output of the wind turbine 10, which is the subject of the power production prediction. The third time period may be the same time period as the first time period and / or the second time period or it may be a different time period.

[0053] 2.3.1. Time Period and Data Points for Power Production Data Obtaining. The power production data is obtained 106 over a third time period, which is at least one month, more preferably at least four months. A longer time period of at least four months is preferred to ensure a more comprehensive dataset. The power production data comprises at least 1000 data points, preferably at least 10000, most preferably at least 20000. The data points are spread out over the third time period, providing a detailed record of the power output of the wind turbine 10 over the third time period.

[0054] 2.4. Obtaining of De-Icing System Operational Status Data. The fourth step of the method involves obtaining 108 de-icing system 80 operational status data regarding power status of a de-icing system 80 for de-icing the blades 30 of the wind turbine 10 over a fourth time period. Each obtained de-icing system 80 operational status data is time-stamped, providing a chronological record of the operation of the de-icing system 80. The de-icing system 80 operational status data provides valuable information about the operation of the de-icing system 80, which can affect the efficiency of the wind turbine 10 and the power production prediction. The fourth time period may be the same time period as the first time period, the second time period and / or the third time period or it may be a different time period. In this process, the control unit 60 obtains the operational status data of the de-icing system 80 over a fourth time period. This data includes information about the power status of the de-icing system 80, such as whether it is on or off and, if applicable, the power level at which it is operating. The data is obtained at various time points throughout the fourth time period, providing a detailed record of the operation of the de-icing system 80. This data is valuable for predicting the power production of the wind turbine 10, as the operation of the de-icing system 80 can affect the efficiency of the wind turbine 10. For example, if the de-icing system 80 is operating at a high-power level, this could indicate that there is a significant amount of ice on the blades 30, which could reduce the efficiency of the wind turbine 10 and therefore its power production. Conversely, if the de-icing system 80 is off, this could indicate that there is little or no ice on the blades 30, which could increase the efficiency of the wind turbine 10 and therefore its power production.

[0055] 2.4.1. Time Period and Data Points for De-Icing System Data Obtaining. The de-icing system operational status data is obtained 108 over a fourth time period, which is at least one month, more preferably at least four months. A longer time period of at least four months is preferred to ensure a more comprehensive dataset. The de-icing system operational status data comprises at least 1000 data points, preferably at least 10000, most preferably at least 20000. The data points are spread out over the fourth time period, providing a detailed record of the operation of the de-icing system 80 over the fourth time period. 2.5. Obtaining of Weather Forecast Data. The fifth step of the method involves obtaining 110 weather forecast data for the area of the wind turbine 10. The weather forecast data is for a fifth time period. Each weather forecast data is a weather forecast valid for less than 24 hours later than a time point when the weather forecast data was determined. The weather forecast data provides valuable information about the expected weather conditions at the area of the wind turbine 10, which can affect the power production prediction. The fifth time period may be the same time period as the first time period, the second time period, the third time period and / or the fourth time period or it may be a different time period.

[0056] 2.5.1 . Time Period and Data Points for Weather Forecast Data Obtaining. The weather forecast data is obtained 110 over a fifth time period, which is at least one month, more preferably at least four months. The weather forecast data comprises at least 200 data points. The data points are spread out over the fifth time period, providing a detailed record of the expected weather conditions over the fifth time period.

[0057] 2.6. Predicting Power Production. The final step of the method involves predicting 112 power production of the wind turbine 10 at a future time point based on the obtained power production data, the obtained ice formation data, the obtained local weather data as well as current weather forecast data for the future time point. The predicting 112 of the power production may be performed by a machine learning model (MLM) using the data obtained in the statistical database 70. The predicting 112 of the power production may also be based on the obtained de-icing system 80 operational status data and current de-icing system 80 operational status data. “Current de-icing system operational status data” here means latest available de-icing system operational status data. The prediction 112 of power production may be based on any type of statistical algorithms, such as decision tree algorithms, nearest neighbor algorithms, support vector machines, etc. In this process, the control unit 60 uses the obtained data to predict the power production of the wind turbine 10 at a future time point. This prediction takes into account various factors that can affect the power production, such as the local weather conditions, the ice formation on the blades 30, the operation of the de-icing system 80, and the current weather forecast. By considering these factors, the control unit 60 can provide a more accurate prediction of the power production. This prediction can be used for various purposes, such as planning the operation of the wind turbine 10, managing the supply of electricity in a power grid, and trading electricity in the energy market.

[0058] 3. Operational Process. The operational process of the method for predicting power production of a wind turbine 10 involves several steps, including data obtaining and power production prediction. These steps are performed by the system for predicting power produced by a wind turbine 10, which includes several components, such as a control unit 60, a statistical database 70, one or more ice formation sensors 40 situated at the wind turbine 10 and a local processor 50.

[0059] 3.1. Data Obtaining Process. The data obtaining process involves obtaining various types of data, including local weather data, ice formation data, power production data, de-icing system operational status data, and weather forecast data. Each type of data is obtained over a specific time period and comprises a large number of data points. The data is obtained by the control unit 60 using its network interface 68 and is stored in the statistical database 70. The data obtaining process provides a comprehensive dataset for the power production prediction.

[0060] 3.1 .1 . Process of Obtaining Weather Data. The process of obtaining 102 weather data involves obtaining local weather data at an area of the wind turbine 10 over a first time period. The local weather data includes one or more of wind speed, temperature, wind direction, precipitation amount, precipitation type, and atmospheric humidity. The local weather data is obtained by the control unit 60 using its network interface 68 and is stored in the statistical database 70 with a timestamp.

[0061] 3.1.2. Process of Obtaining Ice Formation Data. The process of obtaining 104 ice formation data involves obtaining ice formation data of ice formed on the blades 30 of the wind turbine 10. The ice formation data is determined from information received by one or more ice formation sensors 40 positioned at the wind turbine 10. According to an embodiment, the ice formation data may be determined by an image processor, which uses image recognition algorithms to analyze images captured by a camera. The ice formation data is obtained by the control unit 60 using its network interface 68 and may be stored in the statistical database 70 with a timestamp.

[0062] 3.1.3. Process of Obtaining Power Production Data. The process of obtaining 106 power production data involves obtaining power production data of power output of the wind turbine 10 over a third time period. The power production data is obtained by the control unit 60 using its network interface 68 and is stored in the statistical database 70 with a timestamp.

[0063] 3.1.4. Process of Obtaining De-Icing System Data. The process of obtaining 108 deicing system operational status data involves obtaining de-icing system operational status data regarding power status of a de-icing system 80 for de-icing the blades 30 of the wind turbine 10 over a fourth time period. The de-icing system operational status data is obtained by the control unit 60 using its network interface 68 and is stored in the statistical database 70 with a timestamp.

[0064] 3.1.5. Process of Obtaining Weather Forecast Data. The process of obtaining 110 weather forecast data involves obtaining weather forecast data for the area of the wind turbine 10 for a fifth time period. Each weather forecast data is a weather forecast valid for less than 24 hours later than a time point when the weather forecast data was determined. The weather forecast data is obtained by the control unit 60 using its network interface 68 and is stored in the statistical database 70 with a timestamp. The weather forecast data may comprise one or more of wind speed, temperature, wind direction, precipitation amount, precipitation type, and atmospheric humidity.

[0065] 3.2. Power Production Prediction Process. The power production prediction process involves predicting 112 power production of the wind turbine 10 at a future time point based on the obtained power production data, the obtained ice formation data, the obtained local weather data as well as current weather forecast data for the future time point. “Current weather forecast data” may be interpreted as the latest weather forecast data available. The current weather forecast data is for a geographical area covering the position of the wind turbine. The power production prediction may be performed by the control unit 60 using a machine learning model (MLM) that analyzes the data obtained in the statistical database 70. The power production prediction provides a prediction of the power output of the wind turbine 10 at the future time point, which can be used for various purposes, such as power grid management and power trading. The future time point may be for example between 10 hours and 72 hours later, preferably less than 24 hours.

[0066] 4. Description of Examples of the Disclosure. The disclosure provides several examples of the method for predicting power production of a wind turbine 10. These examples illustrate the obtaining of various types of data and the prediction of power production based on the obtained data.

[0067] 4.1 . Example of Weather Data Obtaining. In one example, the local weather data is obtained 102 at an area of the wind turbine 10 over a first time period of four months. The local weather data includes wind speed, temperature, wind direction, precipitation amount, precipitation type, and atmospheric humidity. Each obtained local weather data is time-stamped, providing a chronological record of the weather conditions at the area of the wind turbine 10. The local weather data comprises 20000 data points, providing a comprehensive dataset for analysis. The data points are spread out over the first time period, providing a detailed record of the weather conditions over the first time period.

[0068] 4.1.1. Example of Ice Formation Data Obtaining. In another example, the ice formation data is obtained 104 over a second time period of four months. The ice formation data is obtained from information on ice formation obtained by one or more ice formation sensors 40 arranged at the wind turbine. The one or more ice formation sensors 40 may comprise a camera that captures images of the blades 30 of the wind turbine 10, from which images an image processor determines the ice formation data using image recognition algorithms to analyze the images captured by the camera. Alternatively, the one or more ice formation sensors 40 may comprise one or more vibration sensors that sense vibrations of the blades of the wind turbine. The vibration sensor may be mounted in any of the blades. The local processor 50 determines the ice formation data based on the sensed vibrations, by isolating the vibrations originating from ice formation on the blades and analyzing the frequency and possibly also the amplitude of data of the isolated vibrations. The one or more ice formation sensors 40 may comprise one or more microphones that senses sound originating from the blades, i.e. collects data on sensed sound. The microphone(s) may be mounted in the nacelle. The local processor 50 determines the ice formation data based on the collected data on the sensed sound, by using an FFT on the sound signals to obtain the frequencies of the sensed sound. When ice / snow is formed on the blades, the frequency composition of sound from the wind turbine changes. Such changes can be connected to the formation of ice / snow on the blades, based e.g., on previous knowledge of frequencies of sound at certain amounts and extension of ice formation. Each obtained ice formation data is time- stamped, providing a chronological record of the ice formation on the blades 30. The ice formation data comprises 20000 data points, providing a comprehensive dataset for analysis. The data points are spread out over the second time period, providing a detailed record of the ice formation on the blades 30 over the second time period. The ice formation data may also or alternatively comprise current ice formation data.

[0069] 4.2. Example of Power Production Data Obtaining. In a further example, the power production data is obtained 106 over a third time period of four months. The power production data is obtained by the control unit 60 using its network interface 68. Each obtained power production data is time-stamped, providing a chronological record of the power output of the wind turbine 10. The power production data comprises 20000 data points, providing a comprehensive dataset for analysis. The data points are spread out over the third time period, providing a detailed record of the power output of the wind turbine 10 over the third time period.

[0070] 4.2.1 . Example of De-Icing System Data Obtaining. In yet another example, the deicing system operational status data is obtained 108 over a fourth time period of four months. The de-icing system operational status data includes the on / off status and, if applicable, a power level at which the de-icing system 80 was operating. Each obtained de-icing system operational status data is time-stamped, providing a chronological record of the operation of the de-icing system 80. The de-icing system operational status data comprises 20000 data points, providing a comprehensive dataset for analysis. The data points are spread out over the fourth time period, providing a detailed record of the operation of the de-icing system 80 over the fourth time period.

[0071] 4.2.2. Example of Weather Forecast Data Obtaining. In a final example, the weather forecast data is obtained 110 for the area of the wind turbine 10 for a fifth time period of four months. Each weather forecast data is a weather forecast valid for less than 24 hours later than a time point when the weather forecast data was determined. The weather forecast data comprises 200 data points. The data points are spread out over the fifth time period, providing a detailed record of the expected weather conditions over the fifth time period. Each obtained weather forecast data is time- stamped, providing a chronological record of the weather forecasts for the area of the wind turbine 10.

[0072] 5. Potential Applications. The method and system for predicting power production of a wind turbine 10 have several potential applications. These applications include use in wind energy production, weather forecasting, ice detection and de-icing operations, and power grid management.

[0073] 5.1. Application in Wind Energy Production. In the field of wind energy production, the method and system for predicting power production of a wind turbine 10 can be used to optimize the operation of wind turbines 10. By accurately predicting the power production of a wind turbine 10, operators can plan and manage the operation of the wind turbine 10 more effectively. For example, operators can adjust the operation of the wind turbine 10 based on the predicted power production, such as adjusting the angle of the blades 30 to maximize the capture of wind energy. Operators can also plan maintenance activities based on the predicted power production, such as scheduling maintenance activities when the predicted power production is low.

[0074] 5.1.1. Application in Weather Forecasting. In the field of weather forecasting, the method and system for predicting power production of a wind turbine 10 can be used to improve the accuracy of weather forecasts. By obtaining and analyzing local weather data and weather forecast data, the method and system can provide valuable insights into the local weather conditions and trends. These insights can be used to improve the accuracy of weather forecasts, which can benefit various sectors, such as agriculture, construction, and transportation.

[0075] 5.2. Application in Ice Detection and De-Icing Operations. In the field of ice detection and de-icing operations, the method and system for predicting power production of a wind turbine 10 can be used to detect ice formation on the blades 30 of the wind turbine 10 and to manage de-icing operations. By obtaining and analyzing ice formation data and de-icing system operational status data, the method and system can detect ice formation on the blades 30 and predict the need for de-icing operations. This can help to maintain the efficiency of the wind turbine 10 and to prevent damage to the blades 30 caused by ice accumulation.

[0076] 5.2.1 . Application in Power Grid Management. In the field of power grid management, the method and system for predicting power production of a wind turbine 10 can be used to manage the supply of electricity in a power grid. By accurately predicting the power production of a wind turbine 10, operators can plan the supply of electricity in the power grid more effectively. For example, operators can adjust the supply of electricity from other sources based on the predicted power production of the wind turbine 10. Operators can also plan the storage and distribution of electricity based on the predicted power production, such as storing excess electricity when the predicted power production is high and distributing stored electricity when the predicted power production is low. Further, in electrical power trading, what is sold at a current time point is what is produced at a future time point, e.g. tomorrow. Therefore, it is important for the wind turbine owner to have as good prognosis as possible for what the wind turbine can produce at the future time point.

[0077] The terminology used herein is for the purpose of describing particular aspects only and is not intended to be limiting of the disclosure. As used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items. It will be further understood that the terms "comprises," "comprising," "includes," and / or "including" when used herein specify the presence of stated features, integers, actions, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, actions, steps, operations, elements, components, and / or groups thereof.

[0078] It will be understood that, although the terms first, second, etc., may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element without departing from the scope of the present disclosure.

[0079] Relative terms such as "below" or "above" or "upper" or "lower" or "horizontal" or "vertical" may be used herein to describe a relationship of one element to another element as illustrated in the Figures. It will be understood that these terms and those discussed above are intended to encompass different orientations of the device in addition to the orientation depicted in the Figures. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element, or intervening elements may be present. In contrast, when an element is referred to as being "directly connected" or "directly coupled" to another element, there are no intervening elements present.

[0080] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It will be further understood that terms used herein should be interpreted as having a meaning consistent with their meaning in the context of this specification and the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

[0081] It is to be understood that the present disclosure is not limited to the aspects described above and illustrated in the drawings; rather, the skilled person will recognize that many changes and modifications may be made within the scope of the present disclosure and appended claims. In the drawings and specification, there have been disclosed aspects for purposes of illustration only and not for purposes of limitation, the scope of the disclosure being set forth in the following claims.

Claims

CLAIMS1. A method for predicting power production of a wind turbine (10), the method comprising: obtaining (102) local weather data at an area of the wind turbine (10), the local weather data being determined over a first time period; obtaining (104) ice formation data of ice formed on blades (30) of the wind turbine (10), the ice formation data being determined over a second time period, and / or the ice formation data being current ice formation data, wherein information on the ice formation data originates from one or more ice formation sensors (40) positioned at the wind turbine (10); obtaining (106) power production data of power output of the wind turbine (10), the power production data being determined over a third time period; and predicting (112) power production of the wind turbine (10) at a future time point based on the obtained power production data, the obtained ice formation data, the obtained local weather data as well as current weather forecast data for the future time point.

2. The method according to claim 1 , further comprising obtaining (108) de-icing system operational status data regarding power status of a de-icing system (80) for de-icing the blades (30) of the wind turbine (10), the de-icing system operational status data being determined over a fourth time period, wherein the predicting (112) of the power production of the wind turbine (10) at the future time point is also based on the obtained de-icing system operational status data and current de-icing system operational status data.

3. The method according to claims 1 or 2, further comprising obtaining (110) weather forecast data for the area of the wind turbine (10), the weather forecast data being for a fifth time period, wherein the predicting (112) of the power production of the wind turbine (10) at the future time point is also based on the obtained weather forecast data.

4. The method according to any one of claims 1 to 3, wherein the predicting (112) of the power production is performed by a machine learning model (MLM) using the obtained data.

5. The method according to any one of claims 1 to 4, wherein the one or more ice formation sensors (40) comprise a camera, wherein the information on the ice formation data is images captured by the camera of the blades (30) of the wind turbine (10), and wherein the ice formation data is determined by image processing of the images captured by the camera.

6. The method according to any one of claims 1 to 5, wherein the one or more ice formation sensors (40) comprise one or more vibration sensors that sense vibrations of the blades (30), wherein the information on the ice formation data is data on the vibrations sensed, and wherein the ice formation data is determined based on the data on the sensed vibrations.

7. The method according to any one of claims 1 to 6, wherein the one or more ice formation sensors (40) comprise one or more microphones, wherein the information on the ice formation data is data on sound originating from the blades (30) and registered by the one or more microphones, and wherein the ice formation data is determined based on the data on the sound registered by the one or more microphones.

8. The method according to any one of claims 1 to 7, wherein the first time period, the second time period, the third time period, the fourth time period, and the fifth time period are at least one month, more preferably at least four months.

9. The method according to any one of claims 1 to 8, wherein the local weather data, the ice formation data, the power production data, and the de-icing system operational status data each comprise at least 1000 data points, preferably at least 10000, most preferably at least 20000, distributed over the respective first, second third and fourth time period.

10. The method according to any one of claims 1 to 9, wherein the weather forecast data comprises at least 200 data points spread out over the fifth time period.

11. The method according to any one of claims 1 to 10, wherein the predicting (112) of the power production at the future time point is further based on the obtained power production data and current power production data.

12. A system for predicting power produced by a wind turbine (10), the system comprising: a control unit (60) configured to perform the method according to any one of claims 1 to 11.

13. The system according to claim 12, further comprising a statistical database (70) configured to store the obtained local weather data, the obtained ice formation data, the obtained power production data, and optionally the obtained de-icing system operational status data and the obtained weather forecast data.

14. The system according to claims 12 or 13, further comprising one or more ice formation sensors (40) configured to be positioned at the wind turbine for obtaining information on the ice formation data and a local processor (50) for determining the ice formation data based on the information on the ice formation data obtained by the one or more ice formation sensors (40).

15. The system according to any one of claims 12 to 14, wherein the control unit (60) comprises a memory (67), a processor (65), and a network interface (68) for obtaining the data and predicting the power production and, optionally, communicating with a statistical database (70).

Images