Wind power plant wind resource simulation method and simulation equipment

Abstract

The invention discloses a wind power plant wind resource simulation method and simulation equipment, and relates to the field of wind resource simulation. The method comprises the following steps: collecting wind resource data in a wind power plant in a short time, wherein the collected data serve as real-time short-time wind data; s2, data discrimination and classification; the classification model is used for classifying the real-time short-time wind data, and according to classification, the judgment model is used for judging the wind condition type corresponding to the real-time short-time wind data and outputting the wind condition type; s3, generating data; and S2, the wind regime type and the real-time short-time wind data in the step S2 serve as input, simulated long-time wind data are generated through a generation model corresponding to the wind regime type, and the simulated long-time wind data are within the error range compared with wind resource data collected for a long time in the wind power plant. According to the method, validity judgment and wind regime classification can be automatically completed on the short-term wind resource data, and the long-term wind resource data matched with the short-term wind resource data are generated based on the classification result so that a wind power equipment simulation platform can carry out simulation testing.

Description

Technical Field

[0001] This invention relates to the field of wind resource simulation, and in particular to a wind farm wind resource simulation method and simulation equipment. Background Technology

[0002] With the continuous expansion of wind energy utilization, the number of wind farms has grown rapidly, and their construction layout has become increasingly diversified, covering a wide range of complex geographical environments, from flat plains to plateaus, mountains, coastlines, and even the sea. This has led to significant differences in wind resource characteristics, including low-density winds in plateaus, strong turbulent winds in mountains, highly corrosive winds at sea, and turbulent winds in complex terrain. These different types of wind resources not only place higher demands on the performance of wind power equipment but also make the data acquisition and processing of wind resources more complex. Therefore, wind resource simulation is an essential step in the initial testing of wind power equipment.

[0003] Currently, the main method for wind resource simulation is long-term wind resource data acquisition, followed by processing of the acquired data to serve as simulation data. Examples include Chinese Patent Publication No. CN118246820A, "A Method and System for Wind Resource Assessment and Power Generation Calculation in Upstream and Downstream Wind Farms," ​​and Chinese Patent Publication No. CN120449766B, "A Method and System for Constructing Equivalent Simulation Models for Multi-Type Wind Farms." These methods require long-term (e.g., one month or longer) wind resource data acquisition, which presents challenges such as high acquisition costs, reliance on dedicated acquisition systems, and complex data preprocessing. Given the inherent problems of long-term wind resource data acquisition, the industry has considered using short-term wind resource data (short-term refers to several days of wind data). However, due to the limited data duration and insufficient representation of fluctuation characteristics, this approach cannot fully support the testing and verification of wind power equipment performance. Therefore, using short-term wind resource data as input for wind resource simulation still presents technical challenges that need to be addressed in the industry. Summary of the Invention

[0004] The purpose of this invention is to provide a wind farm wind resource simulation method and simulation equipment to address the problems mentioned above.

[0005] The technical solution adopted in this invention is as follows: A wind farm wind resource simulation method, comprising the following steps: S1: Data Acquisition; Collecting wind resource data in a short period of time at the wind farm; the collected data is called real-time short-term wind data. S2: Data discrimination and classification; Classify real-time short-term wind data using a classification model, and based on the classification, use a judgment model to determine the wind condition type corresponding to the real-time short-term wind data and output it. S3: Data generation; Using the wind condition type and real-time short-term wind data in step S2 as input, simulated long-term wind data is generated using the generation model of the corresponding wind condition type. The simulated long-term wind data is within the error range compared to the wind resource data collected over a long period of time in the wind farm.

[0006] Furthermore, in step S2, the real-time short-term wind data is preprocessed before being input into the classification model.

[0007] Furthermore, in step S2, the classification model establishment includes the following steps: A1: Obtain historical short-term wind data for the wind farm, preprocess the historical short-term wind data, and label the preprocessed historical short-term wind data with wind condition type tags. A2: The classification model is trained based on CNN by taking the preprocessed historical short-term wind data as input and the output of various wind condition types and their corresponding probabilities, until the loss function converges.

[0008] Furthermore, in step S2, after classifying the real-time short-term wind data using a classification model, the following steps are taken to determine the wind condition type corresponding to the real-time short-term wind data using a judgment model and output the result: B1: Obtain the probability corresponding to various wind conditions and determine the maximum probability value; B2: Determine if the maximum probability value is greater than the set threshold; If the probability value is greater than the set threshold, and the wind condition type corresponding to the maximum probability value is not an invalid wind type, then output the wind condition type corresponding to the maximum probability value. If the maximum probability value is not greater than the set threshold, or if the wind condition type corresponding to the maximum probability value is an invalid wind type, then the data is considered unreasonable and discarded.

[0009] Further, in step S3, a generative model is established based on GAN, where GAN includes a generator and a discriminator, and the establishment includes the following steps: C1: Take the wind condition type and the corresponding historical short-term wind data as input to the generator, and the generator outputs simulated long-term wind resource data. C2: Introduce fluid dynamics physical constraints into the simulated long-term wind resource data obtained in step C1 to obtain physical losses; C3: Use real long-term wind resource data and simulated long-term wind resource data obtained in step C1 as input to the discriminator, and output the discriminator to counteract the loss. C4: The physical loss in step C2 is added to the adversarial loss in step C3 to form the total loss. The total loss is fed back to the generator and discriminator to correct the mapping relationship in the generator and discriminator. C5: Repeat steps C1-C4 until the total loss converges, completing the establishment of the generative model.

[0010] Furthermore, in addition to the unusable wind type, the wind condition type also includes at least one or more of the following wind condition types: Sea breeze; plateau breeze; mountain breeze; plain breeze.

[0011] Furthermore, wind resource data includes at least one or more of the following: wind speed, wind direction, turbulence intensity, and ambient temperature.

[0012] A wind farm wind resource simulation device, applied to the aforementioned wind farm wind resource simulation method, includes a data acquisition module, a classification module, and a generation module, wherein: The data acquisition module is used to collect real-time short-term wind data. Its output end is connected to the input end of the classification module, and the collected real-time short-term wind data is input into the classification module. The classification module contains a classification model and a judgment model, and its output is connected to the input of the generation module. The generation module contains generation models corresponding to various wind condition types.

[0013] Furthermore, it also includes an access module, the input of which is connected to the output of the generation module, and the output of which is used to connect to the wind power equipment simulation platform.

[0014] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: This invention can automatically determine the validity of short-term wind resource data and classify wind conditions, and generate matching long-term wind resource data based on the classification results for simulation testing by wind power equipment simulation platforms. This can significantly reduce testing costs and improve the coverage and effectiveness of wind resource simulation. Attached Figure Description

[0015] The present invention will be described by way of example and with reference to the accompanying drawings, wherein: Figure 1 This is a flowchart of the method steps disclosed in this invention; Figure 2 A flowchart for establishing the classification model disclosed in this invention; Figure 3 This is a flowchart of the judgment model disclosed in this invention; Figure 4 A flowchart is provided for establishing the generative model disclosed in this invention. Detailed Implementation

[0016] In the description of this specification, it should be noted that if terms such as "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," or "outer" appear to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product is in use, they are only for the convenience of describing this specification and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this specification.

[0017] Furthermore, the use of terms such as "horizontal" or "vertical" in this specification does not imply that the component must be absolutely horizontal or suspended, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal relative to "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0018] In the description of this specification, it should also be noted that, unless otherwise expressly specified and limited, the terms “set up,” “install,” “connect,” and “link” should be interpreted broadly. For example, a link can be a fixed link, a detachable link, or an integral link; it can be a mechanical link or an electrical link; it can be a direct link or an indirect link through an intermediate medium; it can be a connection within two components.

[0019] Example 1 like Figures 1-4 As shown, to address the technical problems mentioned in the background section, a wind farm wind resource simulation method is proposed. Taking a specific wind farm as an example, the method includes the following steps: S1: Data Acquisition; Collect wind resource data in the wind farm for a short period of time. The collected data is called real-time short-term wind data. Specifically, a lidar anemometer and a temperature and humidity sensor are installed in the wind farm. The input end of the data acquisition module is connected to the lidar anemometer and the temperature and humidity sensor to continuously collect real-time short-term wind data of the wind farm for 5 days, including wind speed, wind direction, turbulence intensity and ambient temperature data.

[0020] After obtaining real-time short-term wind data, operations such as removing invalid data are performed on the real-time short-term wind data to complete the data processing and ensure the validity of the real-time short-term wind data.

[0021] S2: Data discrimination and classification; The preprocessed real-time short-term wind data is classified using a classification model, and the wind condition type corresponding to the real-time short-term wind data is determined and output based on the classification using a judgment model; For this wind farm, the threshold is set to 0.75; The probability vectors output by the classification model are as follows: Unusable wind: 0.02; Sea breeze: 0.85; Plateau wind: 0.03; Mountain wind: 0.035; Plain wind: 0.065; The wind condition type output is thus sea breeze.

[0022] like Figure 3 As shown, in this step, after classifying the real-time short-term wind data using a classification model, the judgment model is used to determine the wind condition type corresponding to the real-time short-term wind data and output the result. B1: Obtain the probability corresponding to various wind conditions and determine the maximum probability value; B2: Determine if the maximum probability value is greater than the set threshold; If the probability value is greater than the set threshold, and the wind condition type corresponding to the maximum probability value is not an invalid wind type, then output the wind condition type corresponding to the maximum probability value. If the maximum probability value is not greater than the set threshold, or if the wind condition type corresponding to the maximum probability value is an invalid wind type, then the data is considered unreasonable and discarded.

[0023] S3: Data Generation; Using the wind condition type and real-time short-term wind data from step S2 as input, simulated long-term wind data is generated using the corresponding wind condition type generation model. Specifically, the generation model receives a wind condition type labeled "sea breeze," calls a generation model specifically trained for "coastal monsoon climate" characteristics, uses the real-time short-term wind data acquired over these 5 days as input to the generation model, and outputs simulated long-term wind data that conforms to the climatological and physical laws of the wind farm and lasts for up to two weeks, along with the "sea breeze" label.

[0024] The simulated long-term wind data has been verified in practice, as follows: Data verification: The wind farm obtained simulated long-term wind data using the wind farm wind resource simulation method described above. The data was then compared with half-month wind resource data obtained from the wind farm's meteorological station. The comparison revealed that the mean absolute error (MAE) of the average wind speed was less than 3%, and the mean error of turbulence intensity (TI) within the observation period was less than 8%. Furthermore, the wind speed power spectral density (PSD) of the generated data completely covered the inertial sub-region (0.01Hz-1Hz), which is consistent with the theoretical characteristics of this frequency band.

[0025] Testing and Verification: Simulated long-term wind data obtained from the wind farm using the aforementioned wind farm wind resource simulation method was input into the wind power equipment simulation platform for simulation testing. This data was then compared with half-month operating data of the actual wind power equipment at the wind farm. The comparison revealed that the equivalent fatigue load (DEL) deviation of key components (such as blade root bending moment and tower base bending moment) was ≤5%, and the deviation of half-month power generation was less than or equal to 4%. Furthermore, simulated long-term wind data obtained from the wind farm using the aforementioned wind farm wind resource simulation method was input into the wind power equipment simulation platform for simulation testing. This data was then compared with half-month wind resource data obtained from the wind farm's meteorological station. The comparison showed that the operational consistency was over 90%.

[0026] In summary, practical verification shows that the simulated long-term wind data is within the error range compared to wind resource data collected over a long period of time in wind farms, and the generated simulated long-term wind data has high accuracy and matching degree.

[0027] Example 2 The following implementation method is proposed for establishing the classification model and generative model used in Example 1.

[0028] like Figure 2 As shown, the establishment of a classification model includes the following steps: A1: Obtain historical short-term wind data for the wind farm, preprocess the historical short-term wind data, and label the preprocessed historical short-term wind data with wind condition type tags. A2: The classification model is trained using preprocessed historical short-term wind data as input, with multiple wind condition types and their corresponding probabilities as outputs. The model is trained based on CNN (Convolutional Neural Network) until the loss function converges, thus completing the establishment of the classification model.

[0029] It is known that by using CNN and training it with a large amount of precisely labeled short-term wind data, it is possible to adaptively extract deep features from time-series data and achieve high-precision classification of wind conditions, including unusable wind resources, plateau winds, sea winds and other wind condition categories.

[0030] like Figure 4 As shown, the generative model is built based on GAN, which includes a generator and a discriminator, and includes the following steps: C1: Take the wind condition type and the corresponding historical short-term wind data as input to the generator, and the generator outputs simulated long-term wind resource data. C2: Introduce fluid dynamics physical constraints into the simulated long-term wind resource data obtained in step C1 to obtain physical losses; C3: Use real long-term wind resource data and simulated long-term wind resource data obtained in step C1 as input to the discriminator, and output the discriminator to counteract the loss. C4: The physical loss in step C2 is added to the adversarial loss in step C3 to form the total loss. The total loss is fed back to the generator and discriminator to correct the mapping relationship in the generator and discriminator. C5: Repeat steps C1-C4 until the total loss converges, completing the establishment of the generative model.

[0031] In the process of establishing the generative model, the total loss is introduced with physical constraints based on fluid dynamics, which makes the generated simulated long-term wind resource data not only close to the real data in terms of statistical distribution, but also more reliable in terms of physical laws.

[0032] Example 3 A wind farm wind resource simulation device, applied to the aforementioned wind farm wind resource simulation method, includes a data acquisition module, a classification module, and a generation module, wherein: The data acquisition module is used to collect real-time short-term wind data. Its output end is connected to the input end of the classification module, and the collected real-time short-term wind data is input into the classification module. The classification module contains a classification model and a judgment model, and its output is connected to the input of the generation module. The generation module contains generation models corresponding to various wind condition types.

[0033] Furthermore, it also includes an access module, the input of which is connected to the output of the generation module, and the output of which is used to connect to the wind power equipment simulation platform.

[0034] This invention is not limited to the specific embodiments described above. The invention extends to any new feature or combination disclosed in this specification, as well as any new method or process step or combination disclosed herein.

Claims

1. A method for simulating wind resources in a wind farm, characterized in that: Includes the following steps: S1: Data Acquisition; Collecting wind resource data in a short period of time at the wind farm; the collected data is called real-time short-term wind data. S2: Data discrimination and classification; Classify real-time short-term wind data using a classification model, and based on the classification, use a judgment model to determine the wind condition type corresponding to the real-time short-term wind data and output it. S3: Data generation; Using the wind condition type and real-time short-term wind data from step S2 as input, simulated long-term wind data is generated using the generation model corresponding to the wind condition type.

2. The wind farm wind resource simulation method according to claim 1, characterized in that: In step S2, the real-time short-term wind data is preprocessed before being input into the classification model.

3. The wind farm wind resource simulation method according to claim 1, characterized in that: In step S2, the classification model establishment includes the following steps: A1: Obtain historical short-term wind data for the wind farm, preprocess the historical short-term wind data, and label the preprocessed historical short-term wind data with wind condition type tags. A2: The classification model is trained based on CNN by taking the preprocessed historical short-term wind data as input and the output of various wind condition types and their corresponding probabilities, until the loss function converges.

4. The wind farm wind resource simulation method according to claim 1, characterized in that: In step S2, after classifying the real-time short-term wind data using a classification model, the following steps are taken to determine the wind condition type corresponding to the real-time short-term wind data using a judgment model and output the result: B1: Obtain the probability corresponding to various wind conditions and determine the maximum probability value; B2: Determine if the maximum probability value is greater than the set threshold; If the probability value is greater than the set threshold, and the wind condition type corresponding to the maximum probability value is not an invalid wind type, then output the wind condition type corresponding to the maximum probability value. If the maximum probability value is not greater than the set threshold, or if the wind condition type corresponding to the maximum probability value is an invalid wind type, then the data is considered unreasonable and discarded.

5. The wind farm wind resource simulation method according to claim 1, characterized in that: In step S3, a generative model is built based on GAN, which includes a generator and a discriminator. The building process includes the following steps: C1: Take the wind condition type and the corresponding historical short-term wind data as input to the generator, and the generator outputs simulated long-term wind resource data. C2: Introduce fluid dynamics physical constraints into the simulated long-term wind resource data obtained in step C1 to obtain physical losses; C3: Use real long-term wind resource data and simulated long-term wind resource data obtained in step C1 as input to the discriminator, and output the discriminator to counteract the loss. C4: The physical loss in step C2 is added to the adversarial loss in step C3 to form the total loss. The total loss is fed back to the generator and discriminator to correct the mapping relationship in the generator and discriminator. C5: Repeat steps C1-C4 until the total loss converges, completing the establishment of the generative model.

6. The wind farm wind resource simulation method according to any one of claims 1-5, characterized in that: In addition to the unusable wind type, the wind condition type shall have at least one or more of the following wind condition types: Sea breeze; plateau breeze; mountain breeze; plain breeze.

7. The wind farm wind resource simulation method according to any one of claims 1-5, characterized in that: Wind resource data includes at least one or more of the following: wind speed, wind direction, turbulence intensity, and ambient temperature.

8. A wind farm wind resource simulation device, applied to the wind farm wind resource simulation method according to any one of claims 1-7, characterized in that: It includes a data acquisition module, a classification module, and a generation module, among which: The data acquisition module is used to collect real-time short-term wind data. Its output end is connected to the input end of the classification module, and the collected real-time short-term wind data is input into the classification module. The classification module contains a classification model and a judgment model, and its output is connected to the input of the generation module. The generation module contains generation models corresponding to various wind condition types.

9. The wind farm wind resource simulation equipment according to claim 8, characterized in that: It also includes an access module, whose input is connected to the output of the generation module, and whose output is used to connect to the wind power equipment simulation platform.

Images