Spatial soft measurement method, device and equipment of wind condition data and storage medium

Abstract

The present disclosure relates to a spatial soft measurement method and device for wind condition data, equipment and storage medium. The method comprises: obtaining current hub rear measured wind condition corresponding to a target unit; using a pre-trained wind condition spatial soft measurement model to perform wind condition spatial soft measurement processing on the current hub rear measured wind condition, to obtain current hub front measured wind condition corresponding to the target unit, and taking the current hub front measured wind condition as a spatial soft measurement result. Thus, for a unit without a nacelle type laser wind measurement radar, the non-real inflow wind condition of the target domain collected by the unit is not directly used for wake control and performance evaluation, but the non-real inflow wind condition is processed into the real inflow wind condition of the source domain before the wake control and performance evaluation, thereby improving the accuracy of the wake control and optimizing the effect of the performance evaluation, and finally ensuring the power generation capacity of the wind farm.

Description

Technical Field

[0001] This disclosure relates to the field of new energy power generation technology, and in particular to a spatial soft measurement method, device, equipment and storage medium for wind condition data. Background Technology

[0002] With the development of new energy power generation technologies, more and more regions are using wind turbines for wind power generation. In the process of wind power generation, it is often necessary to conduct wake control and efficiency assessment of wind farms in order to ensure the power generation capacity of wind farms as much as possible.

[0003] In related technologies, considering the cost of wind farm construction, anemometers or wind vanes are generally used to acquire Supervisory Control and Data Acquisition (SCADA) data, i.e., nacelle SCADA data, which is then used as input data for wake control and performance evaluation. However, nacelle SCADA data is back-hub wind data affected by the wind turbine rotor (i.e., not the actual incoming wind data). This data deviates from the back-hub wind data unaffected by the wind turbine rotor (i.e., the actual incoming wind data). This discrepancy affects the accuracy and effectiveness of wake control and performance evaluation based on nacelle SCADA data, ultimately compromising the wind farm's power generation capacity. Summary of the Invention

[0004] To solve the above-mentioned technical problems, or at least partially solve them, this disclosure provides a spatial soft measurement method, apparatus, device, and storage medium for wind condition data.

[0005] In a first aspect, this disclosure provides a spatial soft measurement method for wind condition data, the method comprising:

[0006] The current measured wind conditions behind the hub of the target unit are obtained. The target unit is not equipped with a nacelle-type laser wind measuring radar. The current measured wind conditions behind the hub are non-real incoming wind conditions affected by the wind turbine of the target unit. The spatial domain corresponding to the current measured wind conditions behind the hub is the target domain.

[0007] Using a pre-trained wind condition spatial soft measurement model, the measured wind condition behind the current turbine hub is processed by wind condition spatial soft measurement to obtain the measured wind condition in front of the current turbine hub corresponding to the target turbine. The measured wind condition in front of the current turbine hub is used as the spatial soft measurement result. The measured wind condition in front of the current turbine hub is the real incoming wind condition that is not affected by the turbine of the target turbine. The spatial domain corresponding to the measured wind condition in front of the current turbine hub is the source domain. Furthermore, the measured wind condition in front of the current turbine hub is used to perform wake control and efficiency evaluation for the wind farm where the target turbine is located.

[0008] Secondly, this disclosure provides a spatial soft measurement device for wind condition data, the device comprising:

[0009] The current measured wind conditions behind the hub module is used to acquire the current measured wind conditions behind the hub corresponding to the target unit. The target unit is not equipped with a nacelle-type laser wind measuring radar, and the current measured wind conditions behind the hub are non-real incoming wind conditions affected by the wind turbine of the target unit.

[0010] The wind condition spatial soft measurement processing module is used to perform wind condition spatial soft measurement processing on the current measured wind condition behind the turbine hub using a pre-trained wind condition spatial soft measurement model, to obtain the current measured wind condition in front of the turbine hub corresponding to the target turbine, and to use the current measured wind condition in front of the turbine hub as the spatial soft measurement result. The current measured wind condition in front of the turbine hub is the real incoming wind condition that is not affected by the turbine of the target turbine. The spatial domain corresponding to the current measured wind condition in front of the turbine hub is the source domain. Furthermore, the current measured wind condition in front of the turbine hub is used to perform wake control and efficiency evaluation on the wind farm where the target turbine is located.

[0011] Thirdly, embodiments of this disclosure also provide an apparatus, the apparatus comprising:

[0012] One or more processors;

[0013] Storage device for storing one or more programs.

[0014] When one or more programs are executed by one or more processors, the one or more processors implement the methods provided in the first aspect.

[0015] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method provided in the first aspect.

[0016] The technical solution provided in this disclosure has the following advantages compared with the prior art:

[0017] This disclosure discloses a spatial soft measurement method, apparatus, device, and storage medium for wind condition data. It acquires the current measured wind condition behind the hub of a target turbine, where the target turbine is not equipped with a nacelle-type laser wind-measuring radar. The current measured wind condition behind the hub is a non-real incoming wind condition affected by the turbine's rotor, and the spatial domain corresponding to the current measured wind condition behind the hub is the target domain. Using a pre-trained spatial soft measurement model, the current measured wind condition behind the hub is processed to obtain the current measured wind condition in front of the hub corresponding to the target turbine. The current measured wind condition in front of the hub is used as the spatial soft measurement result. The current measured wind condition in front of the hub is a real incoming wind condition unaffected by the turbine's rotor, and the spatial domain corresponding to the current measured wind condition in front of the hub is the source domain. Furthermore, the current measured wind condition in front of the hub is used for wake control and performance evaluation of the wind farm where the target turbine is located. By using the above method, for wind turbines that are not equipped with nacelle-type laser wind radar, instead of directly using the non-real incoming wind conditions in the target domain collected by the turbine for wake control and performance evaluation, the non-real incoming wind conditions are processed into real incoming wind conditions in the source domain before wake control and performance evaluation are performed. This improves the accuracy of wake control and optimizes the effect of performance evaluation, ultimately ensuring the power generation capacity of the wind farm. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0019] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A schematic flowchart illustrating a spatial soft measurement method for wind condition data provided in an embodiment of this disclosure;

[0021] Figure 2 A flowchart illustrating a method for training a wind condition spatial soft measurement model provided in an embodiment of this disclosure;

[0022] Figure 3 This is a schematic diagram of the structure of a preset network provided in an embodiment of the present disclosure;

[0023] Figure 4 A schematic diagram of the structure of a spatial soft measurement device for wind condition data provided in an embodiment of this disclosure;

[0024] Figure 5 This is a schematic diagram of the structure of a spatial soft measurement device for wind condition data provided in an embodiment of this disclosure. Detailed Implementation

[0025] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0026] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0027] In related technologies, to acquire real incoming wind condition data for wind turbines, nacelle-mounted LiDAR (LiDAR) radars are installed on individual turbines to measure the actual incoming wind conditions. However, this approach cannot obtain real incoming wind condition data for all turbines in the entire wind farm. To address this issue, related technologies have proposed a nacelle wind speed transfer function, specifically a piecewise linear function to transfer the wind speed measured by the anemometer on the top of the nacelle to the actual incoming wind speed. However, a simple piecewise linear function cannot accurately transfer the wind speed measured by the anemometer on the top of the nacelle to the actual incoming wind speed, and it is not adaptable to the transfer of other wind condition data (such as wind direction data), exhibiting poor reliability and lacking universality.

[0028] To ensure the accuracy and effectiveness of wake control and performance evaluation, embodiments of this disclosure provide a spatial soft measurement method, device, and storage medium for wind condition data that can eliminate wake effects.

[0029] Below, firstly, in combination with Figures 1 to 3 This disclosure describes a spatial soft measurement method for wind condition data provided in an embodiment.

[0030] Figure 1 A schematic flowchart of a spatial soft measurement method for wind condition data provided in an embodiment of this disclosure is shown.

[0031] In this embodiment of the disclosure, Figure 1 The spatial soft measurement method for wind condition data shown can be executed by a spatial soft measurement device for wind condition data. This device can be an electronic device or a server. The electronic device can be, but is not limited to, a fixed terminal such as a smartphone, laptop, or desktop computer. The server can be a cloud server or server cluster, or other device with storage and computing capabilities. This disclosure uses an electronic device as the execution subject for detailed explanation.

[0032] like Figure 1 As shown, the spatial soft measurement method for this wind condition data may include the following steps.

[0033] S110. Obtain the current measured wind conditions behind the hub corresponding to the target unit. The target unit is not equipped with a nacelle-type laser wind measuring radar. The current measured wind conditions behind the hub are non-real incoming wind conditions affected by the wind turbine of the target unit. The spatial domain corresponding to the current measured wind conditions behind the hub is the target domain.

[0034] In this embodiment, considering the cost of wind farm construction, the wind turbines in the wind farm do not need to be equipped with nacelle-type laser wind radar. Instead, the wind conditions at the current hub of the target turbine can be measured using the installed anemometers or wind vanes.

[0035] The target unit is any unit in the wind farm that does not have a nacelle-type laser wind measuring radar installed, but the target unit can be equipped with a wind vane and anemometer.

[0036] It is understandable that the current measured wind conditions behind the wheel hub refer to the nacelle SCADA data. The target domain is a spatial domain that does not include lidar data. When directly using the nacelle SCADA data in the target domain for wake control and performance evaluation, the accuracy and reliability are poor.

[0037] In some embodiments, when an anemometer is installed on the target unit, the current measured wind conditions behind the hub are non-real incoming wind speeds affected by the rotor of the target unit. Correspondingly, the wind condition space soft measurement model is the wind speed space soft measurement model, and the current measured wind conditions in front of the hub are real incoming wind speeds not affected by the rotor of the target unit.

[0038] In other embodiments, when the target unit is equipped with a wind vane, the current measured wind conditions behind the hub are non-real incoming wind directions affected by the wind turbine of the target unit. Correspondingly, the wind condition space soft measurement model is a wind direction space soft measurement model, and the current measured wind conditions in front of the hub are real incoming wind directions not affected by the wind turbine of the target unit.

[0039] S120. Using a pre-trained wind condition spatial soft measurement model, perform wind condition spatial soft measurement processing on the current measured wind condition behind the hub to obtain the current measured wind condition in front of the hub corresponding to the target unit. The current measured wind condition in front of the hub is used as the spatial soft measurement result. The current measured wind condition in front of the hub is the real incoming wind condition that is not affected by the wind turbine of the target unit. Furthermore, the current measured wind condition in front of the hub is used to perform wake control and efficiency evaluation for the wind farm where the target unit is located.

[0040] It is understandable that there is a certain deviation between the measured wind conditions behind the hub (i.e., the non-real incoming wind conditions) and the measured wind conditions in front of the hub (i.e., the real incoming wind conditions) which are not affected by the wind turbine rotor. Directly using the non-real incoming wind condition sequence for wake control and performance evaluation results in poor accuracy and effectiveness. To improve the accuracy of transmitting the measured wind conditions from behind the hub to in front of the hub, this embodiment uses a pre-trained wind space soft measurement model to transform the current measured wind conditions behind the hub into the current measured wind conditions in front of the hub, thus obtaining the real incoming wind conditions. These real incoming wind conditions are then used to perform wake control and performance evaluation on the wind farm where the target turbine is located, which helps improve the accuracy and effectiveness of wake control and performance evaluation.

[0041] It is understandable that the current measured wind conditions in front of the wheel hub refer to the lidar data collected by the nacelle-type lidar wind measurement radar. That is, the source domain is the spatial domain containing lidar data. When directly using the lidar data corresponding to the source domain for wake control and performance evaluation, accuracy and reliability can be guaranteed.

[0042] Optionally, the wind condition spatial soft measurement model includes, but is not limited to, Domain Adversarial Neural Networks (DANNs), Domain Separating Neural Networks (DSNs), and other types of networks.

[0043] In this embodiment of the disclosure, optionally, S120 specifically includes: performing feature extraction processing on the current measured wind conditions behind the hub based on the feature extractor in the wind condition space soft measurement model to obtain the current wind condition features behind the hub; performing feature fitting processing on the current wind condition features behind the hub based on the feature fitter in the wind condition space soft measurement model to obtain the current fitted wind conditions behind the hub, and using the current fitted wind conditions behind the hub as the current measured wind conditions in front of the hub corresponding to the target unit.

[0044] To improve the processing performance of the wind condition spatial soft measurement model, the measured wind conditions behind the current hub can be pre-cleaned and interpolated to obtain the pre-processed measured wind conditions behind the current hub. The wind condition spatial soft measurement model can then perform wind condition spatial soft measurement processing on the pre-processed measured wind conditions behind the current hub to obtain the measured wind conditions in front of the current hub corresponding to the target unit.

[0045] Therefore, by utilizing the feature extractor and feature fitter in the wind condition spatial soft measurement model, the current measured wind condition behind the wheel hub is sequentially processed for feature extraction and feature fitting, thereby transforming the non-real incoming wind condition corresponding to the target domain into the real incoming wind condition corresponding to the source domain, thus realizing the transfer of wind condition data in different spaces.

[0046] This disclosure discloses a spatial soft measurement method for wind data, which acquires the current measured wind conditions behind the hub of a target turbine. The target turbine is not equipped with a nacelle-type laser wind-measuring radar, and the current measured wind conditions behind the hub are non-real incoming wind conditions affected by the turbine's rotor. The spatial domain corresponding to the current measured wind conditions behind the hub is the target domain. A pre-trained spatial soft measurement model is used to process the current measured wind conditions behind the hub, obtaining the current measured wind conditions in front of the hub corresponding to the target turbine. The current measured wind conditions in front of the hub are used as the spatial soft measurement result. The current measured wind conditions in front of the hub are real incoming wind conditions unaffected by the turbine's rotor, and the spatial domain corresponding to the current measured wind conditions in front of the hub is the source domain. Furthermore, the current measured wind conditions in front of the hub are used for wake control and performance evaluation of the wind farm where the target turbine is located. By using the above method, for wind turbines that are not equipped with nacelle-type laser wind radar, instead of directly using the non-real incoming wind conditions in the target domain collected by the turbine for wake control and performance evaluation, the non-real incoming wind conditions are processed into real incoming wind conditions in the source domain before wake control and performance evaluation are performed. This improves the accuracy of wake control and optimizes the effect of performance evaluation, ultimately ensuring the power generation capacity of the wind farm.

[0047] In another embodiment of this disclosure, the wind condition spatial soft measurement model can be trained using the first historical wind conditions behind the hub corresponding to the target unit, the historical wind conditions before the hub corresponding to the relocated unit, and the second historical wind conditions behind the hub of the relocated unit. The training process of the wind condition spatial soft measurement model will be explained in detail below.

[0048] Figure 2 A schematic flowchart of a wind condition spatial soft measurement model training method provided in an embodiment of this disclosure is shown.

[0049] like Figure 2 As shown, the training method for the wind condition spatial soft measurement model includes the following steps.

[0050] S210. Obtain the first historical measured wind conditions after the hub corresponding to the target unit, wherein the first historical measured wind conditions after the hub are non-real incoming wind conditions affected by the wind turbine of the target unit, and the spatial domain corresponding to the first historical measured wind conditions after the hub is the target domain.

[0051] It is understandable that the first historical measured wind conditions after the hub refer to the nacelle SCADA data. The target domain is a spatial domain that does not include lidar data. In other words, the first historical measured wind conditions after the hub are the nacelle SCADA data in the target domain. Furthermore, the first historical measured wind conditions after the hub are affected by the wind turbine of the target unit and become non-real incoming wind conditions.

[0052] S220. From the multiple reference turbines included in the wind farm where the target turbine is located, determine the migration turbine corresponding to the target turbine, and obtain the historical measured wind conditions before the hub and the second historical measured wind conditions after the hub corresponding to the migration turbine. The historical measured wind conditions before the hub are the real incoming wind conditions that are not affected by the turbine of the migration turbine, and the spatial domain corresponding to the historical measured wind conditions before the hub is the source domain. The second historical measured wind conditions after the hub are the non-real incoming wind conditions that are affected by the turbine of the target turbine, and the spatial domain corresponding to the second historical measured wind conditions after the hub is the target domain.

[0053] In this embodiment, to train the wind condition spatial soft measurement model corresponding to the target turbine, nacelle-mounted laser wind-measuring radars are installed on individual turbines in the wind farm where the target turbine is located. These individual turbines with nacelle-mounted laser wind-measuring radars are used as reference turbines. Additionally, anemometers and wind vanes can be installed on the reference turbines. To make the trained wind condition spatial soft measurement model more consistent with the target turbine, turbines with a high correlation to the target turbine can be selected from the reference turbines as the corresponding migration turbines. The nacelle-mounted laser wind-measuring radars installed on the migration turbines are used to collect historical measured wind conditions in front of the turbine hub, and the anemometers and wind vanes installed on the migration turbines are used to collect second historical measured wind conditions behind the turbine hub.

[0054] In this embodiment of the disclosure, optionally, S220 specifically includes: obtaining the historical reference wind conditions corresponding to each reference unit, wherein the historical reference wind conditions are non-real incoming wind conditions affected by the wind turbine of the corresponding reference unit; calculating the correlation coefficient between the target unit and each reference unit based on the first historical measured wind conditions after the hub and the historical reference wind conditions corresponding to each reference unit; and selecting the unit with the largest correlation coefficient from multiple reference units as the migration unit of the target unit.

[0055] It is understandable that the historical reference wind conditions are not the actual incoming wind conditions. That is, the historical reference wind conditions belong to the historical post-hub measured wind conditions corresponding to the target domain. At the same time, the first historical post-hub measured wind conditions also belong to the historical post-hub measured wind conditions corresponding to the target domain. Therefore, in the target domain, based on the historical reference wind conditions and the first historical post-hub measured wind conditions, the correlation coefficient between the target unit and each reference unit is calculated.

[0056] The correlation coefficient is used to characterize the degree of correlation between the target unit and the reference unit. Optionally, the correlation coefficient includes, but is not limited to, the Spearman coefficient, the Pearson coefficient, and the maximum information coefficient.

[0057] Understandably, a higher correlation coefficient indicates a greater correlation between the target unit and the reference unit. Using the historical measured wind conditions before and after the hub corresponding to the reference unit, a wind space soft measurement model that better matches the target unit can be trained. Conversely, a lower correlation coefficient indicates a lower correlation between the target unit and the reference unit. Using the historical measured wind conditions before and after the hub corresponding to the reference unit, a wind space soft measurement model that doesn't quite match the target unit can be trained. To train a wind space soft measurement model that best matches the target unit, the unit with the highest correlation coefficient from multiple reference units is selected as the transfer unit for the target unit. This facilitates subsequent training of the target unit's wind space soft measurement model based on the historical measured wind conditions before and after the hub corresponding to the transfer unit.

[0058] S230. Using the first historical wind conditions behind the wheel hub, the historical wind conditions in front of the wheel hub, and the second historical wind conditions behind the wheel hub, the preset network is iteratively trained to obtain a wind condition spatial soft measurement model.

[0059] In some embodiments, when an anemometer is installed on the target unit, the measured wind conditions after the first historical hub are non-true incoming wind speeds affected by the rotor of the target unit, and the measured wind conditions after the second historical hub are non-true incoming wind speeds affected by the rotor of the relocated unit. Accordingly, the wind condition space soft measurement model is the wind speed space soft measurement model, and the measured wind conditions before the historical hub are true incoming wind speeds not affected by the rotor of the relocated unit.

[0060] In other embodiments, when the target unit is equipped with a wind vane, the measured wind conditions behind the historical hub are non-true incoming wind directions affected by the wind turbine of the target unit, and the measured wind conditions behind the second historical hub are non-true incoming wind directions affected by the wind turbine of the relocated unit. Accordingly, the wind condition space soft measurement model is a wind direction space soft measurement model, and the measured wind conditions behind the historical hub are true incoming wind directions not affected by the wind turbine of the relocated unit.

[0061] In this embodiment of the disclosure, optionally, S230 specifically includes:

[0062] S2301. Use the feature extractor in the preset network to perform feature extraction processing on the first historical wind conditions behind the wheel hub and the second historical wind conditions behind the wheel hub respectively, to obtain the first historical wind conditions behind the wheel hub corresponding to the first historical wind conditions behind the wheel hub and the second historical wind conditions behind the wheel hub corresponding to the second historical wind conditions behind the wheel hub.

[0063] S2302. Use the feature fitter in the preset network to perform feature fitting on the second historical hub wind condition features to obtain the historical hub-front fitted wind condition corresponding to the migrated unit.

[0064] S2303. Using the domain discriminator in the preset network, perform domain discrimination processing on the first historical wind condition feature behind the wheel hub and the second historical wind condition feature behind the wheel hub respectively to obtain the source domain fitting label corresponding to the first historical wind condition feature behind the wheel hub and the target domain fitting label corresponding to the second historical wind condition feature behind the wheel hub.

[0065] S2304. Based on the historical wind conditions fitted in front of the wheel hub, the historical measured wind conditions in front of the wheel hub, the source domain fitting labels, the target domain fitting labels, the pre-determined source domain true labels, and the pre-determined target domain true labels, the preset network is iteratively trained to obtain the wind condition spatial soft measurement model.

[0066] Specifically, S2304 includes: calculating a first loss value based on historical pre-hub fitted wind conditions and historical pre-hub measured wind conditions; calculating a second loss value based on source domain fitted labels, target domain fitted labels, source domain true labels, and target domain true labels; inverting the second loss value using a gradient inversion layer in a preset network to obtain an inverted loss value; and iteratively training the feature extractor, feature fitter, and domain discriminator together based on the first loss value, the second loss value, and the inverted loss value until the first loss value and the second loss value meet a preset stopping condition to obtain a wind speed spatial soft measurement model.

[0067] For ease of understanding, see Figure 3 The diagram shows the structure of a pre-defined network, which includes a feature extractor, a feature fitter, and a domain discriminator. Combined with... Figure 3The training process of the wind condition spatial soft measurement model is explained as follows: First, the first historical wind condition after the hub (x1) and the second historical wind condition after the hub (x2) are used as input data for the wind condition spatial soft measurement model. Based on the feature extractor of the wind condition spatial soft measurement model, the first historical wind condition after the hub (f1) is extracted from the first historical wind condition after the hub (x1), and the second historical wind condition after the hub (f2) is extracted from the second historical wind condition after the hub (x2). Then, based on the feature fitter in the preset network, the second historical wind condition after the hub (f2) is subjected to feature fitting processing to obtain the historical wind condition before the hub (y') corresponding to the migrated unit. At the same time, based on the domain discriminator in the preset network, the source domain fitting label (labled1') of the first historical wind condition after the hub (f1) and the target domain fitting label (labled2') of the second historical wind condition after the hub (f2) are determined. Further, based on the historical wind condition before the hub (y') and the historical wind condition before the hub (y), The first loss value, LossLy, is calculated. Simultaneously, a second loss value, LossLd, is calculated based on the source domain fitted label labled1', the target domain fitted label labled2', the source domain true label labled1, and the target domain true label labled2. To achieve unsupervised transfer of data from the source domain to the target domain, a gradient inversion layer is added before the feature extractor and the domain discriminator. The gradient inversion layer is used to invert the second loss value, LossLd, to obtain the inverted loss value (-LossLd). This ensures that as the model trains, the domain discriminator cannot correctly distinguish whether the received information is a source domain sample or a target domain sample. Finally, based on the first loss value, the second loss value, and the inverted loss value, the feature extractor, feature fitter, and domain discriminator are jointly iteratively trained until both the first and second loss values ​​tend to stabilize, reaching the preset stopping condition, thus obtaining the wind speed spatial soft measurement model.

[0068] It should be noted that the first loss value, LossLy, is used to calculate... and Two gradient values, the second loss value LossLd is used to calculate The gradient value and the inversion loss value (-LossLd) are used to calculate The gradient value is then adjusted jointly based on the first loss value, the second loss value, and the reversal loss value. An equal gradient value is chosen so that when both the first and second loss values ​​tend to stabilize, Once stability is achieved, the training process continues, resulting in a spatial soft measurement model of wind speed.

[0069] To ensure the effectiveness of model training, preprocessing operations such as cleaning and interpolation can be performed on the measured wind conditions behind the first historical wheel hub, the measured wind conditions before the historical wheel hub, and the measured wind conditions behind the second historical wheel hub to obtain preprocessed training samples. Then, the preprocessed training samples are normalized and divided into a target training set and a target test set according to a set ratio (such as 8:2 or 7:3).

[0070] To ensure the training accuracy of the model, after obtaining the wind condition spatial soft measurement model, the model can be tested using a target test set. During the testing process, the model's evaluation index is calculated to evaluate the wind condition spatial soft measurement model based on the evaluation index.

[0071] Optionally, evaluation metrics may include, but are not limited to, accuracy and precision.

[0072] Therefore, by using the historical measured wind conditions in front of the wheel hub in the source domain, the second historical measured wind conditions behind the wheel hub in the target domain, and the first historical measured wind conditions behind the wheel hub in the target domain, a preset network is trained to obtain a transfer model from the non-real incoming wind conditions in the target domain to the real incoming wind conditions in the source domain. This results in a wind space soft measurement model, which facilitates the subsequent use of the wind space soft measurement model to process the current measured wind conditions behind the wheel hub in the target domain into the current measured wind conditions in front of the wheel hub in the source domain. Then, wake control and performance evaluation are performed based on the current measured wind conditions in front of the wheel hub in the source domain, thereby improving the accuracy of wake control and optimizing the performance evaluation effect.

[0073] This disclosure also provides a spatial soft measurement device for wind condition data to implement the above-described spatial soft measurement method for wind condition data. The following is in conjunction with... Figure 4 The following explanation is provided. In this embodiment, the spatial soft measurement device for wind condition data can be an electronic device. This electronic device may include a mobile terminal, tablet computer, or other device with communication capabilities.

[0074] Figure 4 A schematic diagram of the structure of a spatial soft measurement device for wind condition data provided in an embodiment of this disclosure is shown.

[0075] like Figure 4 As shown, the spatial soft measurement device 400 for wind condition data may include:

[0076] The current measured wind conditions behind the hub module 410 is used to acquire the current measured wind conditions behind the hub corresponding to the target unit. The target unit is not equipped with a nacelle-type laser wind measuring radar. The current measured wind conditions behind the hub are non-real incoming wind conditions affected by the wind turbine of the target unit. The spatial domain corresponding to the current measured wind conditions behind the hub is the target domain.

[0077] The wind condition spatial soft measurement processing module 420 is used to perform wind condition spatial soft measurement processing on the current measured wind condition behind the hub using a pre-trained wind condition spatial soft measurement model, to obtain the current measured wind condition in front of the hub corresponding to the target unit, and to use the current measured wind condition in front of the hub as the spatial soft measurement result. The current measured wind condition in front of the hub is the real incoming wind condition that is not affected by the wind turbine of the target unit. The spatial domain corresponding to the current measured wind condition in front of the hub is the source domain. Furthermore, the current measured wind condition in front of the hub is used to perform wake control and efficiency evaluation on the wind farm where the target unit is located.

[0078] This disclosure discloses a spatial soft measurement device for wind condition data. It acquires the measured wind condition behind the current hub of a target turbine, where the target turbine is not equipped with a nacelle-type laser wind-measuring radar. The measured wind condition behind the current hub is a non-real incoming wind condition affected by the turbine's rotor, and the spatial domain corresponding to the measured wind condition behind the current hub is the target domain. Using a pre-trained spatial soft measurement model, the measured wind condition behind the current hub is processed to obtain the measured wind condition in front of the current hub corresponding to the target turbine. This measured wind condition in front of the current hub is used as the spatial soft measurement result. The measured wind condition in front of the current hub is a real incoming wind condition unaffected by the turbine's rotor, and the spatial domain corresponding to the measured wind condition in front of the current hub is the source domain. Furthermore, the measured wind condition in front of the current hub is used for wake control and performance evaluation of the wind farm where the target turbine is located. By using the above method, for wind turbines that are not equipped with nacelle-type laser wind radar, instead of directly using the non-real incoming wind conditions in the target domain collected by the turbine for wake control and performance evaluation, the non-real incoming wind conditions are processed into real incoming wind conditions in the source domain before wake control and performance evaluation are performed. This improves the accuracy of wake control and optimizes the effect of performance evaluation, ultimately ensuring the power generation capacity of the wind farm.

[0079] In some embodiments, the wind condition space soft measurement processing module 420 includes:

[0080] The first feature extraction unit is used to perform feature extraction processing on the current measured wind conditions behind the wheel hub based on the feature extractor in the wind condition spatial soft measurement model, so as to obtain the current wind condition features behind the wheel hub.

[0081] The first feature fitting unit is used to perform feature fitting processing on the current wind condition features behind the hub based on the feature fitter in the wind condition space soft measurement model, to obtain the current fitted wind condition behind the hub, and to use the current fitted wind condition behind the hub as the current measured wind condition in front of the hub corresponding to the target unit.

[0082] In some embodiments, when an anemometer is installed on the target unit, the measured wind conditions behind the current hub are non-real incoming wind speeds affected by the wind turbine of the target unit. Correspondingly, the wind condition space soft measurement model is a wind speed space soft measurement model, and the measured wind conditions in front of the current hub are real incoming wind speeds not affected by the wind turbine of the target unit.

[0083] When the target unit is equipped with a wind vane, the measured wind conditions behind the hub are non-real incoming wind directions affected by the wind turbine of the target unit. Correspondingly, the wind condition space soft measurement model is a wind direction space soft measurement model, and the measured wind conditions in front of the hub are real incoming wind directions not affected by the wind turbine of the target unit.

[0084] In some embodiments, the device further includes:

[0085] The first historical measured wind conditions acquisition module is used to acquire the first historical measured wind conditions behind the hub corresponding to the target unit. The first historical measured wind conditions behind the hub are non-real incoming wind conditions affected by the wind turbine of the target unit, and the spatial domain corresponding to the first historical measured wind conditions behind the hub is the target domain.

[0086] The relocation unit determination module is used to determine the relocation unit corresponding to the target unit from multiple reference units included in the wind farm where the target unit is located, and to obtain the historical measured wind conditions before the hub and the second historical measured wind conditions after the hub corresponding to the relocation unit. The historical measured wind conditions before the hub are the real incoming wind conditions that are not affected by the wind turbine of the relocation unit, and the spatial domain corresponding to the historical measured wind conditions before the hub is the source domain. The second historical measured wind conditions after the hub are the non-real incoming wind conditions that are affected by the wind turbine of the target unit, and the spatial domain corresponding to the second historical measured wind conditions after the hub is the target domain.

[0087] The model training module is used to iteratively train a preset network using the first historical wind conditions behind the wheel hub, the historical wind conditions in front of the wheel hub, and the second historical wind conditions behind the wheel hub, to obtain the wind condition spatial soft measurement model.

[0088] In some embodiments, the migration unit determination module is specifically used for:

[0089] Obtain the historical reference wind conditions corresponding to each reference unit, wherein the historical reference wind conditions are non-real incoming wind conditions affected by the wind turbine of the corresponding reference unit;

[0090] Based on the first historical wind conditions after the hub and the historical reference wind conditions corresponding to each reference unit, the correlation coefficient between the target unit and each reference unit is calculated.

[0091] The unit with the highest correlation coefficient among the multiple reference units is selected as the migration unit for the target unit.

[0092] In some embodiments, the model training module includes:

[0093] The second feature extraction unit is used to perform feature extraction processing on the first historical wind conditions behind the wheel hub and the second historical wind conditions behind the wheel hub using the feature extractor in the preset network, so as to obtain the first historical wind conditions behind the wheel hub corresponding to the first historical wind conditions behind the wheel hub and the second historical wind conditions behind the wheel hub corresponding to the second historical wind conditions behind the wheel hub.

[0094] The second feature fitting unit is used to perform feature fitting processing on the second historical hub wind condition feature using the feature fitter in the preset network to obtain the historical hub wind condition fitting for the migrating unit.

[0095] The domain discrimination unit is used to perform domain discrimination processing on the first historical wind condition feature after the wheel hub and the second historical wind condition feature after the wheel hub using the domain discriminator in the preset network, so as to obtain the source domain fitting label corresponding to the first historical wind condition feature after the wheel hub and the target domain fitting label corresponding to the second historical wind condition feature after the wheel hub.

[0096] An iterative training unit is used to iteratively train the preset network based on the historical wind conditions fitted in front of the wheel hub, the historical measured wind conditions in front of the wheel hub, the source domain fitted labels, the target domain fitted labels, the pre-determined source domain true labels, and the pre-determined target domain true labels, to obtain the wind condition spatial soft measurement model.

[0097] In some embodiments, the iterative training unit is specifically used for:

[0098] Based on the historical fitted wind conditions in front of the wheel hub and the historical measured wind conditions in front of the wheel hub, the first loss value is calculated;

[0099] Based on the source domain fitted label, the target domain fitted label, the source domain true label, and the target domain true label, calculate the second loss value;

[0100] The second loss value is inverted using the gradient inversion layer in the preset network to obtain the inverted loss value of the second loss value.

[0101] Based on the first loss value, the second loss value, and the inversion loss value, the feature extractor, the feature fitter, and the domain discriminator are jointly iteratively trained until the first loss value and the second loss value meet a preset stopping condition, thereby obtaining the wind speed spatial soft measurement model.

[0102] In some embodiments, when an anemometer is installed on the target unit, the measured wind conditions after the first historical hub are non-true incoming wind speeds affected by the rotor of the target unit, and the measured wind conditions after the second historical hub are non-true incoming wind speeds affected by the rotor of the relocated unit. Accordingly, the wind condition spatial soft measurement model is a wind speed spatial soft measurement model, and the measured wind conditions before the historical hub are true incoming wind speeds not affected by the rotor of the relocated unit.

[0103] When the target unit is equipped with a wind vane, the measured wind conditions behind the historical hub are non-true incoming wind directions affected by the wind turbine of the target unit, and the measured wind conditions behind the second historical hub are non-true incoming wind directions affected by the wind turbine of the relocated unit. Accordingly, the wind condition spatial soft measurement model is a wind direction spatial soft measurement model, and the measured wind conditions behind the historical hub are true incoming wind directions not affected by the wind turbine of the relocated unit.

[0104] It should be noted that, Figure 4 The spatial soft measurement device 400 that displays the wind condition data can perform... Figures 1 to 3 The various steps in the method embodiment shown are implemented. Figures 1 to 3 The various processes and effects in the methods or system embodiments shown are not described in detail here.

[0105] Figure 5 A schematic diagram of the structure of a spatial soft measurement device for wind condition data provided in an embodiment of this disclosure is shown.

[0106] like Figure 5 As shown, the spatial soft measurement device for wind condition data may include a processor 501 and a memory 502 storing computer program instructions.

[0107] Specifically, the processor 501 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0108] Memory 502 may include a large-capacity storage for information or instructions. For example, and not limitingly, memory 502 may include a hard disk drive (HDD), a floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 502 may include removable or non-removable (or fixed) media. Where appropriate, memory 502 may be internal or external to the integrated gateway device. In a particular embodiment, memory 502 is a non-volatile solid-state memory. In a particular embodiment, memory 502 includes read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (Electrically Programmable ROM, EPROM), an electrically erasable programmable PROM (EEPROM), an electrically alterable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0109] The processor 501 reads and executes computer program instructions stored in the memory 502 to perform the steps of the spatial soft measurement method for wind data provided in this embodiment of the disclosure.

[0110] In one example, the spatial soft measurement device for the wind condition data may also include a transceiver 503 and a bus 504. Wherein, as Figure 5 As shown, the processor 501, memory 502 and transceiver 503 are connected via bus 504 and communicate with each other.

[0111] Bus 504 may include hardware, software, or both. For example, and not limitingly, a bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industrial Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 504 may include one or more buses. Although specific buses are described and illustrated in the embodiments of this application, this application considers any suitable bus or interconnection.

[0112] The following are embodiments of a computer-readable storage medium provided in this disclosure. This computer-readable storage medium and the spatial soft measurement method for wind data in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the computer-readable storage medium, please refer to the embodiments of the spatial soft measurement method for wind data described above.

[0113] This embodiment provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a spatial soft measurement method for wind condition data.

[0114] Of course, the computer-executable instructions provided in the embodiments of this disclosure are not limited to the above-described method operations, but can also perform related operations in the spatial soft measurement method for wind condition data provided in any embodiment of this disclosure.

[0115] Based on the above description of the implementation methods, those skilled in the art can clearly understand that this disclosure can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer cloud platform (which may be a personal computer, server, or network cloud platform, etc.) to execute the spatial soft measurement method for wind condition data provided in the various embodiments of this disclosure.

[0116] Note that the above description is merely a preferred embodiment and the technical principles employed in this disclosure. Those skilled in the art will understand that this disclosure is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this disclosure. Therefore, although this disclosure has been described in detail through the above embodiments, it is not limited to the above embodiments. Many other equivalent embodiments may be included without departing from the concept of this disclosure, and the scope of this disclosure is determined by the scope of the appended claims.

Claims

1. A spatial soft measurement method for wind condition data, characterized in that, include: The current measured wind conditions behind the hub of the target unit are obtained. The target unit is not equipped with a nacelle-type laser wind measuring radar. The current measured wind conditions behind the hub are non-real incoming wind conditions affected by the wind turbine of the target unit. The spatial domain corresponding to the current measured wind conditions behind the hub is the target domain. Using a pre-trained wind condition spatial soft measurement model, the measured wind condition behind the current hub is processed by wind condition spatial soft measurement to obtain the measured wind condition in front of the current hub corresponding to the target unit. The measured wind condition in front of the current hub is used as the spatial soft measurement result. The measured wind condition in front of the current hub is the real incoming wind condition that is not affected by the wind turbine of the target unit. The spatial domain corresponding to the measured wind condition in front of the current hub is the source domain. Furthermore, the measured wind condition in front of the current hub is used to perform wake control and efficiency evaluation for the wind farm where the target unit is located. Before obtaining the measured wind conditions at the current hub corresponding to the target unit, the method further includes: The first historical measured wind conditions after the hub of the target unit are obtained. The first historical measured wind conditions after the hub are non-real incoming wind conditions affected by the wind turbine of the target unit. The spatial domain corresponding to the first historical measured wind conditions after the hub is the target domain. From the multiple reference turbines included in the wind farm where the target turbine is located, the migration turbine corresponding to the target turbine is determined, and the historical measured wind conditions before the hub and the second historical measured wind conditions after the hub corresponding to the migration turbine are obtained. The historical measured wind conditions before the hub are the real incoming wind conditions that are not affected by the rotor of the migration turbine, and the spatial domain corresponding to the historical measured wind conditions before the hub is the source domain. The second historical measured wind conditions after the hub are the non-real incoming wind conditions that are affected by the rotor of the target turbine, and the spatial domain corresponding to the second historical measured wind conditions after the hub is the target domain. The wind conditions spatial soft measurement model is obtained by iteratively training the preset network using the first historical wind conditions behind the wheel hub, the historical wind conditions in front of the wheel hub, and the second historical wind conditions behind the wheel hub.

2. The method according to claim 1, characterized in that, The step of using a pre-trained wind condition spatial soft measurement model to perform wind condition spatial soft measurement processing on the current measured wind conditions behind the hub to obtain the current measured wind conditions in front of the hub corresponding to the target unit includes: Based on the feature extractor in the wind condition space soft measurement model, the current measured wind condition behind the wheel hub is processed to extract features and obtain the current wind condition features behind the wheel hub. Based on the feature fitter in the wind condition space soft measurement model, the current wind condition features behind the hub are subjected to feature fitting processing to obtain the current fitted wind condition behind the hub, and the current fitted wind condition behind the hub is used as the current measured wind condition in front of the hub corresponding to the target unit.

3. The method according to claim 1 or 2, characterized in that, When the target unit is equipped with an anemometer, the measured wind conditions behind the current hub are non-real incoming wind speeds affected by the wind turbine of the target unit. Correspondingly, the wind condition space soft measurement model is a wind speed space soft measurement model, and the measured wind conditions in front of the current hub are real incoming wind speeds not affected by the wind turbine of the target unit. When the target unit is equipped with a wind vane, the measured wind conditions behind the hub are non-real incoming wind directions affected by the wind turbine of the target unit. Correspondingly, the wind condition space soft measurement model is a wind direction space soft measurement model, and the measured wind conditions in front of the hub are real incoming wind directions not affected by the wind turbine of the target unit.

4. The method according to claim 1, characterized in that, The step of determining the relocation unit corresponding to the target unit from multiple reference units included in the wind farm where the target unit is located includes: Obtain the historical reference wind conditions corresponding to each reference unit, wherein the historical reference wind conditions are non-real incoming wind conditions affected by the wind turbine of the corresponding reference unit; Based on the first historical wind conditions after the hub and the historical reference wind conditions corresponding to each reference unit, the correlation coefficient between the target unit and each reference unit is calculated. The unit with the highest correlation coefficient among the multiple reference units is selected as the migration unit for the target unit.

5. The method according to claim 1, characterized in that, The step of iteratively training a preset network using the first historical wind conditions behind the wheel hub, the first historical wind conditions before the wheel hub, and the second historical wind conditions behind the wheel hub to obtain the wind condition spatial soft measurement model includes: The feature extractor in the preset network is used to perform feature extraction processing on the first historical wind condition behind the wheel hub and the second historical wind condition behind the wheel hub respectively, to obtain the first historical wind condition behind the wheel hub corresponding to the first historical wind condition behind the wheel hub and the second historical wind condition behind the wheel hub corresponding to the second historical wind condition behind the wheel hub. The feature fitter in the preset network is used to perform feature fitting on the second historical hub wind condition features to obtain the historical hub-pre-fit wind condition corresponding to the migrating unit. Using the domain discriminator in the preset network, the first historical wind condition feature after the wheel hub and the second historical wind condition feature after the wheel hub are respectively subjected to domain discrimination processing to obtain the source domain fitting label corresponding to the first historical wind condition feature after the wheel hub and the target domain fitting label corresponding to the second historical wind condition feature after the wheel hub. Based on the historical wind conditions fitted in front of the wheel hub, the historical wind conditions measured in front of the wheel hub, the source domain fitted labels, the target domain fitted labels, the pre-determined source domain true labels, and the pre-determined target domain true labels, the preset network is iteratively trained to obtain the wind condition spatial soft measurement model.

6. The method according to claim 5, characterized in that, The method involves iteratively training the preset network based on the historical pre-wheel hub fitted wind conditions, the historical pre-wheel hub measured wind conditions, the source domain fitted labels, the target domain fitted labels, the pre-determined source domain true labels, and the pre-determined target domain true labels to obtain the wind condition spatial soft measurement model, including: Based on the historical fitted wind conditions in front of the wheel hub and the historical measured wind conditions in front of the wheel hub, the first loss value is calculated; Based on the source domain fitted label, the target domain fitted label, the source domain true label, and the target domain true label, calculate the second loss value; The second loss value is inverted using the gradient inversion layer in the preset network to obtain the inverted loss value of the second loss value. Based on the first loss value, the second loss value, and the inversion loss value, the feature extractor, the feature fitter, and the domain discriminator are jointly iteratively trained until the first loss value and the second loss value meet a preset stopping condition, thereby obtaining the wind condition spatial soft measurement model.

7. The method according to any one of claims 1 or 4 to 6, characterized in that, When the target unit is equipped with an anemometer, the measured wind conditions after the first historical hub are non-true incoming wind speeds affected by the rotor of the target unit, and the measured wind conditions after the second historical hub are non-true incoming wind speeds affected by the rotor of the relocated unit. Accordingly, the wind condition space soft measurement model is a wind speed space soft measurement model, and the measured wind conditions before the historical hub are true incoming wind speeds not affected by the rotor of the relocated unit. When the target unit is equipped with a wind vane, the measured wind conditions behind the historical hub are non-true incoming wind directions affected by the wind turbine of the target unit, and the measured wind conditions behind the second historical hub are non-true incoming wind directions affected by the wind turbine of the relocated unit. Accordingly, the wind condition spatial soft measurement model is a wind direction spatial soft measurement model, and the measured wind conditions behind the historical hub are true incoming wind directions not affected by the wind turbine of the relocated unit.

8. A spatial soft measurement device for wind condition data, characterized in that, include: The current measured wind conditions behind the hub module is used to acquire the current measured wind conditions behind the hub corresponding to the target unit. The target unit is not equipped with a nacelle-type laser wind measuring radar. The current measured wind conditions behind the hub are non-real incoming wind conditions affected by the wind turbine of the target unit. The spatial domain corresponding to the current measured wind conditions behind the hub is the target domain. The wind condition spatial soft measurement processing module is used to perform wind condition spatial soft measurement processing on the current measured wind condition behind the hub using a pre-trained wind condition spatial soft measurement model, to obtain the current measured wind condition in front of the hub corresponding to the target unit, and to use the current measured wind condition in front of the hub as the spatial soft measurement result. The current measured wind condition in front of the hub is the real incoming wind condition that is not affected by the wind turbine of the target unit. The spatial domain corresponding to the current measured wind condition in front of the hub is the source domain. Furthermore, the current measured wind condition in front of the hub is used to perform wake control and efficiency evaluation on the wind farm where the target unit is located. The spatial soft measurement device for the wind condition data also includes: The first historical measured wind conditions behind the hub module is used to acquire the first historical measured wind conditions behind the hub corresponding to the target unit before acquiring the current measured wind conditions behind the hub corresponding to the target unit. The first historical measured wind conditions behind the hub are non-real incoming wind conditions affected by the wind turbine of the target unit, and the spatial domain corresponding to the first historical measured wind conditions behind the hub is the target domain. The relocation unit determination module is used to determine the relocation unit corresponding to the target unit from multiple reference units included in the wind farm where the target unit is located, and to obtain the historical measured wind conditions before the hub and the second historical measured wind conditions after the hub corresponding to the relocation unit. The historical measured wind conditions before the hub are the real incoming wind conditions that are not affected by the wind turbine of the relocation unit, and the spatial domain corresponding to the historical measured wind conditions before the hub is the source domain. The second historical measured wind conditions after the hub are the non-real incoming wind conditions that are affected by the wind turbine of the target unit, and the spatial domain corresponding to the second historical measured wind conditions after the hub is the target domain. The model training module is used to iteratively train a preset network using the first historical wind conditions behind the wheel hub, the historical wind conditions in front of the wheel hub, and the second historical wind conditions behind the wheel hub, to obtain the wind condition spatial soft measurement model.

9. A device, characterized in that, include: processor; Memory, used to store executable instructions; The processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the method of any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The storage medium stores a computer program that, when executed by a processor, causes the processor to implement the method described in any one of claims 1-7.

Images