A method for calculating the hub height wind speed of an offshore wind farm
By combining floating lidar and spaceborne SAR data deployed in offshore wind farm areas, a machine learning model was established, solving the problems of accuracy and economy in measuring wind speed at hub height in offshore wind farms, and achieving high-precision wind speed estimation and assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- THREE GORGES GRP ZHEJIANG ENERGY INVESTMENT CO LTD
- Filing Date
- 2026-02-28
- Publication Date
- 2026-06-05
AI Technical Summary
Existing technologies struggle to achieve simultaneous large-scale coverage, high vertical accuracy, and cost-effective wind turbine hub height and wind speed measurements in offshore wind farms. Spaceborne SAR has insufficient accuracy, fixed offshore wind measurement towers are expensive and difficult to maintain, and numerical weather prediction models have large errors in complex areas.
By deploying multiple floating lidars in the target sea area and combining them with spaceborne SAR data, a training sample set was established and a machine learning model was trained to capture the complex nonlinear mapping relationship between sea surface wind speed and wheel hub height. The model was then used to calculate the wind speed at wheel hub height.
It achieves high-precision and economical calculation of hub height and wind speed, improves the accuracy and reliability of data, and supports micro-site selection, power generation assessment and operation and maintenance optimization of offshore wind farms.
Smart Images
Figure CN122151087A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of offshore wind resource assessment technology, and in particular to a method for calculating wind speed at hub height in offshore wind farms. Background Technology
[0002] In the field of offshore wind resource assessment, accurately obtaining wind speed at the hub height of wind turbines is a core prerequisite for carrying out micro-site selection, power generation assessment and operation and maintenance optimization of wind farms. However, existing technologies have significant limitations and cannot simultaneously meet the requirements of large-scale coverage, high vertical accuracy and economic efficiency.
[0003] The existing technologies have the following problems: 1. Spaceborne SAR has all-weather, all-day operation capabilities and can provide high-resolution wind speed distribution at a height of 10m above the sea surface, down to the kilometer level. However, its absolute accuracy is only about ±1.5~2.0m / s, and due to limitations in the observation principle, it cannot directly extrapolate the 10m wind speed at the sea surface to the height of the wind turbine hub (usually tens to over one hundred meters), making it difficult to meet the height dimension accuracy requirements of practical wind farm applications. 2. Fixed offshore wind measurement towers can accurately measure wind speeds within a height range of 10m~150m with high vertical accuracy, but they suffer from high cost, low deployment density (difficult to cover large sea areas), and high maintenance difficulty and cost in the offshore environment, making it impossible to achieve full-area, high-density monitoring of wind resources in the target sea area. 3. Current numerical weather prediction models can output three-dimensional wind field data, but in areas with complex nearshore topography and extreme weather conditions such as tropical cyclones, the wind field simulation error is significant, making it impossible to provide reliable high-precision wind speed data support for offshore wind farms. Summary of the Invention
[0004] The main objective of this invention is to provide a method for calculating the hub height and wind speed of offshore wind farms, which solves the problems of limited space coverage of spaceborne SAR, high maintenance difficulty and cost in the offshore environment, and large errors in wind farm simulation in the existing technology.
[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for calculating wind speed at hub height in offshore wind farms, comprising the following steps: S1: Acquire raw spaceborne SAR data for the target sea area, and use geophysical model functions to invert and obtain sea surface height and wind speed. , where (x,y) are spatial coordinates and t is time; S2: Deploy multiple floating lidar units in the wind farm area of the target sea area to measure the three-dimensional wind speed profile within a predetermined height range in real time. Where z is the altitude, For measuring time, Location of the equipment; S3: Wind speed at the sea surface height With the aforementioned three-dimensional wind speed profile Spatiotemporal matching is performed to establish a training sample set, where the input features include the sea surface height and wind speed. Atmospheric stability parameter L, sea surface roughness The output label is the wind speed at the predetermined target height. ; S4: Train the prediction model f based on the training sample set, so that the wind speed at the predetermined target height is... ; S5: Using the aforementioned prediction model f, the total sea surface height and wind speed within the target sea area retrieved by SAR are calculated. Input the prediction model f to calculate the wind speed at the predetermined target height. Generate a wind speed map at the predetermined target height for the target sea area.
[0006] In the preferred embodiment, S2 includes: The multiple floating lidars are distributed in the wind farm area to cover the corresponding locations of the target sea area, and the predetermined height range covers the z value corresponding to the predetermined target height; The real-time measurement of the three-dimensional wind speed profile By using multiple floating lidars to collect wind speed components and direction information at preset time intervals at each z-height, the corresponding... Time and The three-dimensional wind speed profile at the location And record it for subsequent spatiotemporal matching.
[0007] In the preferred embodiment, step S3, establishing a training sample set, includes: Select the sea surface height and wind speed in the spatiotemporal proximity. With the aforementioned three-dimensional wind speed profile The three-dimensional wind speed profile The wind speed at the predetermined target altitude zh is used as the output label. ; From the three-dimensional wind speed profile Alternatively, auxiliary data can be used to obtain the atmospheric stability parameter L and the sea surface roughness at the corresponding location. The input features that form each spatiotemporal matching sample include the sea surface height and wind speed. The atmospheric stability parameter L, the sea surface roughness .
[0008] In the preferred embodiment, S4 includes: The input features and output labels of the training sample set are input into the prediction model f for supervised training. The prediction model f uses a neural network structure to capture the wind speed at sea surface height. Wind speed at the predetermined target altitude The mapping relationship; Iteratively optimize the parameters of the prediction model f until the input features are output by the prediction model f and match the predetermined target height wind speed with the output label. .
[0009] In the preferred embodiment, S5 includes: The sea surface height and wind speed at each (x, y, t) location in the entire region. Obtain the corresponding atmospheric stability parameter L and the sea surface roughness. ; The overall sea surface height wind speed The atmospheric stability parameter L, the sea surface roughness The wind speed at the predetermined target height is obtained by feeding the input features into the prediction model f. Based on this, a wind speed map of the predetermined target height for the target sea area is generated.
[0010] In a preferred embodiment, the multiple floating lidar units are distributed across the wind farm area to cover corresponding locations in the target sea area, including: Multiple preset locations (xi, yi) are determined within the wind farm area, and the floating lidar is deployed at each of the preset locations to ensure the accuracy of the three-dimensional wind speed profile. The spatial range covering the target sea area; After the floating lidar is activated, it continuously collects the three-dimensional wind speed profile within the predetermined height range. The acquired data is aligned with the spaceborne SAR observation time t for spatiotemporal matching.
[0011] In the preferred embodiment, from the three-dimensional wind speed profile Alternatively, auxiliary data can be used to obtain the atmospheric stability parameter L and the sea surface roughness at the corresponding location. ,include: According to the three-dimensional wind speed profile The atmospheric stability parameter L is calculated based on the wind speed gradient at different z-heights. Based on the sea surface height and wind speed With the aforementioned three-dimensional wind speed profile Determine the sea surface roughness And the atmospheric stability parameter L is compared with the sea surface roughness. Associated with the training samples at the corresponding (x,y,t) positions.
[0012] In the preferred embodiment, the prediction model f undergoes supervised training, including: The training sample set is divided into a training subset and a validation subset, and the prediction model f is optimized using the training subset. The difference between the output of the prediction model f and the output label is evaluated for the validation subset. If the difference is less than a preset threshold, the prediction model f is confirmed to have completed training and will be used for subsequent global sea surface height wind speed calculations. The calculation.
[0013] In the preferred embodiment, in step S3, the atmospheric stability parameter L is obtained in the following way: The data is calculated based on the temperature profile and wind speed data measured by the floating lidar, or obtained from the numerical weather prediction model. The sea surface roughness Obtain it through the following methods: Based on the wind speed at a height of 10 meters above the sea surface It is obtained by calculating using the Charnock relation.
[0014] In the preferred embodiment, the determination of multiple preset locations within the wind farm area ( Specifically, it includes the following steps: The target sea area is divided into several sub-regions of similar size to ensure that the horizontal distance from which each floating lidar reaches the farthest point in its sub-region does not exceed a preset representative radius, which is the maximum distance acceptable for subsequent spatial interpolation. Within the wind farm planning area, excluding areas along ship routes and areas already covered by cables, a list of locations suitable for deploying floating lidar was compiled. Select one point from the list of deployable points as a reference point, and determine the remaining deployment points in sequence along the prevailing wind direction of the target sea area and in a direction perpendicular to the prevailing wind direction, so that the distance between adjacent points matches the representative radius, forming a blind-spot-free coverage network. Through on-site surveys, the water depth conditions, mooring feasibility, and communication signal coverage of each initially determined location were verified. Minor adjustments were made to locations that did not meet the deployment requirements, ultimately locking in the preset coordinates of each floating lidar unit. ), and complete the determination of deployment locations.
[0015] This invention provides a method for estimating wind speed at hub height in offshore wind farms. It involves acquiring sea surface wind speed from satellite-borne SAR data for the target sea area and simultaneously deploying at least three floating lidar units in the wind farm area to measure the three-dimensional wind speed profile at high altitude. The two data sources are spatiotemporally matched, and a training sample set is constructed using the matched data. A machine learning model is trained using sea surface wind speed, atmospheric stability, and sea surface roughness as input features, and hub height wind speed as the output label to capture the complex nonlinear mapping relationship between the two. Using the trained model, the SAR sea surface wind speed data across the entire area is batch-estimated to hub height, ultimately generating a hub height wind resource map covering the entire target sea area. This data source approach achieves organic complementarity between the isal sea surface wind field and the point-like vertical profile, improving data accuracy, estimating calculation precision, and enhancing the accuracy and reliability of wind speed estimation from sea surface to hub height. Attached Figure Description
[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a schematic diagram of the calculation method of the present invention; Figure 2 This is a schematic diagram of the calculation system structure of the present invention; Figure 3 This is a comparative schematic diagram of the fixed wind measurement tower and the floating lidar group of the present invention; Figure 4 This is a schematic diagram of the wind speed at a predetermined target altitude in the target sea area of this invention. Detailed Implementation
[0017] Example 1 like Figure 1-4 As shown, a method for calculating wind speed at hub height in offshore wind farms includes the following steps: S1: Acquire raw spaceborne SAR data for the target sea area, and use geophysical model functions to invert and obtain sea surface height and wind speed. , where (x,y) are spatial coordinates and t is time.
[0018] S2: Deploy multiple floating lidar units in the wind farm area of the target sea area to measure the three-dimensional wind speed profile within a predetermined height range in real time. Where z is the altitude, t i To measure the time, ( x i ,y i () indicates the location of the device.
[0019] S3: Wind speed at sea level With three-dimensional wind speed profile Spatiotemporal matching was performed to establish a training sample set, where input features included sea surface height and wind speed. Atmospheric stability parameter L, sea surface roughness The output label is the wind speed at the predetermined target height. .
[0020] S4: Train the prediction model f based on the training sample set to ensure the wind speed at the predetermined target height. .
[0021] S5: Using prediction model f, the total sea surface height and wind speed within the target sea area retrieved from SAR are calculated. Input the prediction model f to calculate the wind speed at the predetermined target height. Generate a wind speed map of the target sea area at the predetermined target height.
[0022] In this embodiment, the predetermined target height is the height of the wind turbine hub.
[0023] In step S1, a virtual typical sea area is simulated as a rectangular region, 50 km east-west and 30 km north-south, with a grid resolution of 1 km, totaling 50 × 30 = 1500 pixels. The data source is a single image in Sentinel-1IW mode (publicly available and free). An example of the inversion result (first 5 pixels) is shown in Table 1.
[0024] Table 1. Example data table of data source inversion results
[0025] This embodiment acquires sea surface wind speed from spaceborne SAR inversion in the target sea area and simultaneously deploys no fewer than three floating lidars in the wind farm area to measure the three-dimensional wind speed profile at high altitude. The two data sources are spatiotemporally matched, and a training sample set is constructed using the spatiotemporally matched data. With sea surface wind speed, atmospheric stability, and sea surface roughness as input features and hub height wind speed as the output label, a machine learning model is trained to capture the complex nonlinear mapping relationship between the two. Using the trained model, the SAR sea surface wind speed data of the entire area is batch extrapolated to hub height, and finally a hub height wind resource map covering the entire target sea area is generated. At the data source level, the isometric sea surface wind field and the point-like vertical profile are organically complementary, which improves the accuracy of the data, enhances the extrapolation accuracy, and improves the accuracy and reliability of the wind speed extrapolation from the sea surface to hub height.
[0026] In the preferred embodiment, step S2 includes: Multiple floating lidar units are distributed across the wind farm area to cover corresponding locations in the target sea area, with a predetermined height range covering the z-value corresponding to the predetermined target height. Real-time measurement of three-dimensional wind speed profile By using multiple floating lidars to collect wind speed components and direction information at preset time intervals at each z-height, a three-dimensional wind speed profile at time ti and position (xi,yi) is obtained. And record it for subsequent spatiotemporal matching.
[0027] In this embodiment, by determining the preset location to deploy lidar and ensuring coverage of the target sea area, the spatial deviation of the samples caused by the concentrated deployment of lidar points is avoided, so that the training samples can reflect the wind field characteristics of different areas of the target sea area and improve the model's adaptability to the estimation of wind speed at different locations across the entire area.
[0028] In this embodiment, four floating lidars, numbered L1# to L4#, are deployed to mark virtual typical locations. They are arranged according to the principle of "one center + three controls", as shown in Table 2.
[0029] Table 2. Relevant data for 4 floating lidar units
[0030] In the preferred scheme, preset locations are determined to deploy the lidar and ensure coverage of the target sea area. Multiple preset locations are determined within the wind farm area. Specifically: Step 1: Divide the target sea area into several sub-regions of similar size, ensuring that the horizontal distance from each lidar to the farthest point within a sub-region does not exceed its representative radius (i.e., the maximum acceptable distance for subsequent spatial interpolation).
[0031] Step 2: Within the planned area of the wind farm, avoid shipping routes and existing cables, and select a list of suitable deployment points.
[0032] Step 3: Select one point from the list as a baseline, and determine the remaining points sequentially along the prevailing wind direction and vertical direction, ensuring that the spacing between adjacent points matches the representative radius to form a blind-spot-free network. Deploy one unit near the geometric center of the site to represent the average wind resources across the entire site; deploy one unit at each end along the main wind direction (prevailing wind direction ±15°) to capture wind speed variations along the path; if the site length is >15km, add one unit every 10km to ensure that the horizontal distance between any location and the nearest radar is ≤10km.
[0033] Step 4: Conduct on-site reconnaissance to confirm water depth, mooring conditions, and communication signals. Make minor adjustments to the remaining infeasible locations, and finally lock in the preset coordinates to complete the deployment plan.
[0034] In the preferred embodiment, multiple floating lidar units are distributed across the wind farm area to cover corresponding locations in the target sea area, including: Determine multiple preset locations within the wind farm area ( Floating lidar units are deployed at predetermined locations to ensure the accuracy of three-dimensional wind speed profiles. The spatial range covering the target sea area; After activation, the floating lidar continuously collects three-dimensional wind speed profiles within a predetermined height range. The collected data is aligned with the spaceborne SAR observation time t for spatiotemporal matching.
[0035] like Figure 3 As shown in the diagram, a comparison between a fixed wind measurement tower and a floating lidar array intuitively demonstrates the technical limitations of fixed towers, such as being single-point, expensive, and difficult to maintain. In this embodiment, the floating lidar array has the characteristics of being multi-point, inexpensive, and mobile, further improving the rationality and feasibility of using a floating lidar array.
[0036] This embodiment uses a cluster of floating lidar arrays deployed in a region to replace or reduce the use of fixed wind measurement towers, and automatically calculates key atmospheric parameters using SAR remote sensing data and lidar measured data. Compared with costly and fixed wind measurement towers, floating lidar significantly reduces the hardware investment and operation and maintenance costs of wind resource assessment, while maintaining the vertical accuracy of the data; the automated process for acquiring atmospheric parameters further improves the overall engineering feasibility and efficiency of the technology.
[0037] In the preferred scheme, step S3, establishing a training sample set, includes: Select sea surface height and wind speed in the spatial and temporal vicinity With three-dimensional wind speed profile , to create a three-dimensional wind speed profile The wind speed at the predetermined target height z_h is used as the output label. ; From the three-dimensional wind speed profile Alternatively, auxiliary data can be used to obtain the atmospheric stability parameter L and sea surface roughness at the corresponding location. The input features that form each spatiotemporal matching sample include sea surface height and wind speed. Atmospheric stability parameter L, sea surface roughness .
[0038] In the preferred scheme, from the three-dimensional wind speed profile Alternatively, auxiliary data can be used to obtain the atmospheric stability parameter L and sea surface roughness at the corresponding location. ,include: Based on the three-dimensional wind speed profile Calculate the atmospheric stability parameter L using wind speed gradients at different z-heights. Based on wind speed at sea level With three-dimensional wind speed profile Determine sea surface roughness And the atmospheric stability parameter L is correlated with the sea surface roughness Associated with the training samples at the corresponding (x,y,t) positions.
[0039] As shown in Table 3, radar data and remote sensing impact data are spatiotemporally matched, with a time limit of ≤30 min and a spatial limit of ≤3 km (i.e., ≤3 grids). Table 3 provides 5 sets of virtual inputs / outputs.
[0040] Table 3. Virtual Input / Output Data Table for Spatiotemporal Matching of Radar Data and Remote Sensing Impact Data
[0041] This embodiment constructs a training sample set using spatiotemporally matched data, and uses sea surface wind speed, atmospheric stability, and sea surface roughness as input features, and hub height wind speed as the output label to train a machine learning model to capture the complex nonlinear mapping relationship between the two. This solves the problem of the linear assumption limitation of traditional logarithmic / power-law extrapolation. The machine learning model (such as a neural network) adaptively learns and expresses the complex physical processes in the ocean-atmosphere boundary layer, significantly improving the accuracy and reliability of wind speed extrapolation from sea surface to hub height, and is suitable for wind resource assessment with extremely high accuracy requirements.
[0042] In the preferred scheme, in step S3, the atmospheric stability parameter L is obtained in the following way: The data is calculated based on temperature profiles and wind speed data measured by floating lidar, or obtained from numerical weather prediction models. Sea surface roughness Obtain it through the following methods: Based on wind speed at a height of 10 meters above sea level It is obtained by calculating using the Charnock relation.
[0043] The preferred solution includes: The sea surface height and wind speed at each location (x, y, t) across the entire region. Take the corresponding atmospheric stability parameter L and sea surface roughness ; The overall sea surface wind speed Atmospheric stability parameter L, sea surface roughness The features are fed into the prediction model f to obtain the wind speed at the predetermined target height. Based on this, a wind speed map of the predetermined target altitude for the target sea area is generated.
[0044] In the preferred embodiment, step S4 includes: The input features and output labels of the training sample set are input into the prediction model f for supervised training. The prediction model f uses a neural network structure to capture sea surface height and wind speed. Wind speed at the predetermined target altitude The mapping relationship; Iteratively optimize the parameters of the prediction model f until the input features match the predetermined target height wind speed of the output label through the prediction model f. .
[0045] In the preferred scheme, the prediction model f undergoes supervised training, including: The training sample set is divided into a training subset and a validation subset, and the prediction model f is optimized using the training subset. The difference between the output of the prediction model f and the output label is evaluated based on the validation subset. If the difference is less than a preset threshold, the prediction model f is considered to have completed training and will be used for subsequent global sea surface height and wind speed calculations. The calculation.
[0046] In this embodiment, a trained model is used to extrapolate SAR sea surface wind speed data across the entire region to hub height in batches, ultimately generating a hub height wind resource map covering the entire target sea area with high resolution and high accuracy. The output map can be directly applied to the micro-site selection of offshore wind farms, accurate assessment of power generation, and optimization of operation and maintenance strategies, providing key data support for investment decisions and operation management.
[0047] It is recommended to illustrate the entire methodology with practical examples and data: For example, in practical applications: S1: Acquire raw spaceborne SAR data (such as a single image from Sentinel-1 IW mode) for the target sea area, and use geophysical model functions to invert and obtain sea surface height and wind speed. , where (x,y) are relative grid coordinates (e.g., 50km east-west, 30km north-south, 1km resolution), and t is the SAR observation time (e.g., 2024-03-01T02:30:00). S2: Deploy XX floating lidar units to measure the three-dimensional wind speed profile within a predetermined height range in real time. Where z is the altitude, For measuring time, For the device's relative grid coordinates (e.g., L1# is at (25,15)); S3: Wind speed at sea level With three-dimensional wind speed profile Spatiotemporal matching was performed to establish a training sample set, where input features included sea surface height and wind speed. For example, 8.2 m / s), atmospheric stability parameter L is (for example, 80 m), sea surface roughness (e.g., 0.3 mm), the output label is the wind speed at the predetermined target height. (e.g., wind speed of 9.5 m / s at a hub height of 100m). S4: Train the prediction model f to achieve the wind speed at the predetermined target height. ; S5: Using prediction model f, the total sea surface height and wind speed within the target sea area retrieved from SAR are calculated. Input the prediction model f to calculate the wind speed at the predetermined target height. For example, if a certain grid point is 11.2 m / s, generate a wind speed map of the target sea area at the predetermined target height.
[0048] like Figure 4 The image shows a wind speed map at a predetermined target height in the target sea area. Different identifiers (such as color and numerical value) are used to present the hub height wind speed data of each grid point in the entire target sea area, clearly showing the spatial distribution differences of hub height wind speed in this sea area.
[0049] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.
Claims
1. A method for calculating wind speed at hub height in offshore wind farms, characterized in that, Includes the following steps: S1: Acquire raw spaceborne SAR data for the target sea area, and use geophysical model functions to invert and obtain sea surface height and wind speed. , where (x,y) are spatial coordinates and t is time; S2: Deploy multiple floating lidar units in the wind farm area of the target sea area to measure the three-dimensional wind speed profile within a predetermined height range in real time. Where z is the altitude, For measuring time, Location of the equipment; S3: Wind speed at the sea surface height With the aforementioned three-dimensional wind speed profile Spatiotemporal matching is performed to establish a training sample set, where the input features include the sea surface height and wind speed. Atmospheric stability parameter L, sea surface roughness The output label is the wind speed at the predetermined target height. ; S4: Train the prediction model f based on the training sample set, so that the wind speed at the predetermined target height is... ; S5: Using the aforementioned prediction model f, the total sea surface height and wind speed within the target sea area retrieved by SAR are calculated. Input the prediction model f to calculate the wind speed at the predetermined target height. Generate a wind speed map at the predetermined target height for the target sea area.
2. The method for calculating wind speed at hub height in offshore wind farms according to claim 1, characterized in that, The S2 includes: The multiple floating lidars are distributed in the wind farm area to cover the corresponding locations of the target sea area, and the predetermined height range covers the z value corresponding to the predetermined target height; The real-time measurement of the three-dimensional wind speed profile By using multiple floating lidars to collect wind speed components and direction information at preset time intervals at each z-height, the corresponding... Time and The three-dimensional wind speed profile at the location And record it for subsequent spatiotemporal matching.
3. The method for calculating wind speed at hub height in offshore wind farms according to claim 1, characterized in that, S3 establishes a training sample set, including: Select the sea surface height and wind speed in the spatiotemporal proximity. With the aforementioned three-dimensional wind speed profile The three-dimensional wind speed profile The wind speed at the predetermined target altitude zh is used as the output label. ; From the three-dimensional wind speed profile Alternatively, auxiliary data can be used to obtain the atmospheric stability parameter L and the sea surface roughness at the corresponding location. The input features that form each spatiotemporal matching sample include the sea surface height and wind speed. The atmospheric stability parameter L, the sea surface roughness .
4. The method for calculating wind speed at hub height in offshore wind farms according to claim 1, characterized in that, The S4 includes: The input features and output labels of the training sample set are input into the prediction model f for supervised training. The prediction model f uses a neural network structure to capture the wind speed at sea surface height. Wind speed at the predetermined target altitude The mapping relationship; Iteratively optimize the parameters of the prediction model f until the input features are output by the prediction model f and match the predetermined target height wind speed with the output label. .
5. The method for calculating wind speed at hub height in offshore wind farms according to claim 1, characterized in that, The S5 includes: The sea surface height and wind speed at each (x, y, t) location in the entire region. Obtain the corresponding atmospheric stability parameter L and the sea surface roughness. ; The overall sea surface height wind speed The atmospheric stability parameter L, the sea surface roughness The wind speed at the predetermined target height is obtained by feeding the input features into the prediction model f. Based on this, a wind speed map of the predetermined target height for the target sea area is generated.
6. The method for calculating wind speed at hub height in offshore wind farms according to claim 2, characterized in that, The multiple floating lidar units are distributed in the wind farm area to cover the corresponding locations of the target sea area, including: Determine multiple preset locations within the wind farm area ( The floating lidar is deployed at each of the preset locations to ensure the three-dimensional wind speed profile. The spatial range covering the target sea area; After the floating lidar is activated, it continuously collects the three-dimensional wind speed profile within the predetermined height range. The acquired data is aligned with the spaceborne SAR observation time t for spatiotemporal matching.
7. The method for calculating wind speed at hub height in offshore wind farms according to claim 3, characterized in that, From the three-dimensional wind speed profile Alternatively, auxiliary data can be used to obtain the atmospheric stability parameter L and the sea surface roughness at the corresponding location. ,include: According to the three-dimensional wind speed profile The atmospheric stability parameter L is calculated based on the wind speed gradient at different z-heights. Based on the sea surface height and wind speed With the aforementioned three-dimensional wind speed profile Determine the sea surface roughness And the atmospheric stability parameter L is compared with the sea surface roughness. Associated with the training samples at the corresponding (x,y,t) positions.
8. The method for calculating wind speed at hub height in offshore wind farms according to claim 4, characterized in that, The prediction model f undergoes supervised training, including: The training sample set is divided into a training subset and a validation subset, and the prediction model f is optimized using the training subset. The difference between the output of the prediction model f and the output label is evaluated for the validation subset. If the difference is less than a preset threshold, the prediction model f is confirmed to have completed training and will be used for subsequent global sea surface height wind speed calculations. The calculation.
9. The method for calculating wind speed at hub height in offshore wind farms according to claim 1, characterized in that, In step S3, the atmospheric stability parameter L is obtained in the following way: The data is calculated based on the temperature profile and wind speed data measured by the floating lidar, or obtained from the numerical weather prediction model. The sea surface roughness Obtain it through the following methods: Based on the wind speed at a height of 10 meters above the sea surface It is obtained by calculating using the Charnock relation.
10. The method for calculating wind speed at hub height in offshore wind farms according to claim 6, characterized in that, The determination of multiple preset locations within the wind farm area ( Specifically, it includes the following steps: The target sea area is divided into several sub-regions of similar size to ensure that the horizontal distance from which each floating lidar reaches the farthest point in its sub-region does not exceed a preset representative radius, which is the maximum distance acceptable for subsequent spatial interpolation. Within the wind farm planning area, excluding areas along ship routes and areas already covered by cables, a list of locations suitable for deploying floating lidar was compiled. Select one point from the list of deployable points as a reference point, and determine the remaining deployment points in sequence along the prevailing wind direction of the target sea area and in a direction perpendicular to the prevailing wind direction, so that the distance between adjacent points matches the representative radius, forming a blind-spot-free coverage network. Through on-site surveys, the water depth conditions, mooring feasibility, and communication signal coverage of each initially determined location were verified. Minor adjustments were made to locations that did not meet the deployment requirements, ultimately locking in the preset coordinates of each floating lidar unit. ), and complete the determination of deployment locations.