Offshore wind plant water and sediment operation and maintenance monitoring method based on sky-ground hydraulic engineering integration

Abstract

The invention provides an offshore wind plant water and sediment operation and maintenance monitoring method based on sky-ground hydraulic engineering integration. The method comprises the steps of multi-source data acquisition and preprocessing, suspended sediment inversion model construction, fan wake flow monitoring and difference analysis, and water and sediment environment risk assessment and visual output. According to the method, the problems of low efficiency and high cost of a traditional monitoring method can be solved, and large-range, high-frequency and low-cost monitoring of the water and sand environment of the offshore wind plant is realized; the suspended sediment inversion precision is improved, errors are reduced through multi-source data fusion, and the fan wake effect is accurately monitored; the influence of different foundation types and tidal conditions on water and sediment disturbance is quantified, the foundation scouring risk is pre-warned, a water and sediment change risk level map is generated by constructing an integrated monitoring system, and a scientific basis is provided for safe operation and maintenance of a wind power plant.

Description

Technical Field

[0001] This invention relates to the field of offshore wind farm environmental monitoring technology, specifically to an integrated air-ground-hydraulic monitoring method for offshore wind farm water and sediment operation and maintenance, aiming to achieve comprehensive, efficient and accurate monitoring of the water and sediment environment of offshore wind farms. Background Technology

[0002] Currently, water and sediment environment monitoring at offshore wind farms mainly relies on traditional methods and remote sensing technology. Traditional methods include on-site sampling (collecting water samples using water samplers and analyzing suspended sediment concentrations using laboratory gravimetric methods) and monitoring at local hydrological stations. In recent years, remote sensing technology has been increasingly applied to large-scale suspended sediment retrieval, such as monitoring using Landsat series satellites (Landsat-8 / 9) or Sentinel-2 satellite data. There are already monitoring cases at offshore wind farms in Shanghai Lingang, Fujian Changle, and Jiangsu Dafeng, both domestically and internationally.

[0003] The existing technology has the following drawbacks: Traditional methods have significant limitations: on-site sampling is costly, infrequent, and has limited spatial coverage, making it difficult to achieve long-term dynamic monitoring, and is greatly affected by weather and sea conditions.

[0004] Insufficient integration of remote sensing technologies: Existing remote sensing monitoring relies heavily on single data sources and lacks coordination with field data and UAV monitoring, resulting in limited inversion accuracy (e.g., model coefficient of determination R). 2 Only 0.6-0.7, the root mean square error is relatively high.

[0005] Inadequate monitoring of wind turbine wake: The wake extraction method is simple and does not take into account changes in tidal direction and differences in wind turbine foundation type, resulting in incomplete disturbance assessment. For example, wind turbine foundation scouring may cause "self-diggering" phenomenon, but existing technology is unable to quantify its risk.

[0006] Systemic risk assessment is lacking: An integrated monitoring system of "sky, ground, and water" has not been formed, making it impossible to provide risk level maps of water and sediment changes, which makes it difficult to support wind farm operation and maintenance decisions. Summary of the Invention

[0007] The main objective of this invention is to provide a monitoring method for water and sediment operation and maintenance in offshore wind farms based on an integrated air-ground-hydraulic system. Based on the integrated monitoring concept of "air-ground-hydraulic system," and combining satellite remote sensing, on-site monitoring, UAV data, and engineering parameters, this method addresses the following technical problems: 1. Overcome the problems of low efficiency and high cost of traditional monitoring methods, and realize large-scale, high-frequency and low-cost monitoring of water and sediment environment in offshore wind farms.

[0008] 2. Improve the accuracy of suspended sediment inversion by reducing errors through multi-source data fusion.

[0009] 3. Accurately monitor the wake effect of wind turbines, quantify the impact of different foundation types and tidal conditions on water and sediment disturbance, and provide early warning of foundation scour risks.

[0010] 4. Construct an integrated monitoring system to generate risk level maps of water and sediment changes, providing a scientific basis for the safe operation and maintenance of wind farms.

[0011] The present invention achieves the above objectives through the following technical solutions: A method for monitoring and maintaining water and sediment in offshore wind farms based on integrated air-ground-water hydraulic engineering includes: Multi-source data acquisition and preprocessing steps: Multispectral image data of the target sea area is acquired through a satellite remote sensing platform. Images with cloud coverage below a preset threshold are selected and subjected to radiometric correction and cloud removal. Simultaneously, on-site water sample collection and high-resolution UAV image acquisition were carried out in the wind farm area to determine the measured values ​​of suspended sediment concentration and ensure the spatiotemporal consistency of the images. Construct an integrated dataset that matches time and space, combining remote sensing reflectance data with measured suspended sediment concentration data; Steps for constructing a suspended sediment inversion model: Based on the correlation analysis between multispectral band combinations and measured suspended sediment concentration data, band combinations that meet the preset correlation threshold are selected. At least two types of mathematical inversion models were constructed using measured data, and the optimal model was selected as the suspended sediment concentration inversion model through cross-validation. Steps for monitoring and analyzing the difference in wind turbine wake: A multi-level annular buffer zone was established with the wind turbine foundation as the center and sector-shaped analysis units were divided. The wake influence area was identified by combining tidal flow data and the wake intensity index was calculated. The wake intensity index was introduced as a correction parameter in the suspended sediment inversion model. The differences in wake characteristics of wind turbines with different foundation structure types under high tide and low tide conditions were analyzed separately, and the correction parameters of the suspended sediment inversion model were dynamically adjusted according to the foundation structure type. Steps for water and sediment environmental risk assessment and visualization output: The suspended sediment concentration inversion model was used to perform long-term suspended sediment concentration inversion on images before and after the wind farm construction, generating a risk level distribution map of dynamic changes in water and sediment. Integrating underwater topographic monitoring data, the system assesses the risks of sediment migration, submarine cable exposure, and wind turbine foundation scour, and outputs a comprehensive risk assessment report on the water and sediment environment that includes spatiotemporal evolution characteristics.

[0012] The present invention provides a method for monitoring and maintaining water and sediment in offshore wind farms based on integrated air-ground-hydraulic engineering, which includes the following steps when acquiring multispectral image data of the target sea area through a satellite remote sensing platform: Simultaneously acquire Sentinel-2 and Landsat-8 / 9 images, which must cover at least the target wind farm's sea area and a surrounding 5km buffer zone; The automatic cloud detection algorithm filters images with cloud coverage less than a preset threshold and removes invalid data with strip noise, missing rows, or abnormal radiation. Atmospheric correction was performed using a DSF algorithm improved based on the dark target method. An improved Acolite declouding algorithm is applied to generate a binary cloud mask, and neighborhood interpolation is used to repair the residual cloud shadow areas. Orthorectification was performed using the RPC model, and the image was projected onto the WGS84 coordinate system; Spatial resolution fusion was performed on Sentinel-2 and Landsat-8 / 9 images, and a daily 10m resolution reflectance dataset was generated using the STARFM temporal fusion model. Construct a data cube that combines blue, green, red, near-infrared, and short-wave infrared bands.

[0013] According to the present invention, a method for monitoring and maintaining water and sediment in offshore wind farms based on integrated air-ground-hydraulic engineering includes the following steps when conducting on-site water sample collection and high-resolution UAV image acquisition: At the sampling site pre-set within the target wind farm sea area, water samples were collected underwater using a vertical water sampler. The sampling site covered a 500m radius around the wind turbine foundation and key areas of the tidal current channel. The collected water samples were sealed and stored in brown glass containers, placed in an insulated transport device, and then analyzed in the laboratory. During the laboratory analysis, vacuum filtration was performed using a glass fiber membrane, and the concentration of suspended sediment was calculated. Using a multi-rotor drone equipped with a visible-near-infrared camera, orthophotos of the area covered by the sampling were acquired within a set time period before and after sampling.

[0014] According to the present invention, a method for monitoring the operation and maintenance of offshore wind farms based on integrated air-ground-hydraulic engineering is provided, which unifies the image data obtained by satellite remote sensing platform and projected onto the WGS84 coordinate system after orthorectification, and the orthorectified image data obtained by UAV, into the same geographic coordinate reference frame. Based on the specific time of on-site water sample collection, satellite remote sensing image data and UAV image data acquired within a certain time window before and after the sampling time were selected; For satellite remote sensing data: data cubes containing combinations of blue, green, red, near-infrared, and shortwave infrared bands will be constructed from Sentinel-2 and Landsat-8 / 9 images. Based on the target wind farm sea area and surrounding area, reflectance data of the corresponding area will be extracted to form a subset of satellite remote sensing reflectance data. For UAV imagery data: Band extraction is performed on high-resolution orthophotos acquired by UAVs equipped with visible light-near infrared cameras to obtain reflectivity information that corresponds to or is similar to satellite remote sensing bands, and a subset of UAV remote sensing reflectivity data is constructed. By fusing subsets of satellite remote sensing reflectance data and UAV remote sensing reflectance data, and using a pixel-based method, the data is resampled and integrated according to the spatial resolution differences of different data sources to form a unified remote sensing reflectance dataset.

[0015] According to the present invention, a method for monitoring and maintaining water and sediment in offshore wind farms based on integrated air-ground-hydraulic engineering is provided, which organizes the suspended sediment concentration data obtained by laboratory analysis and calculation from each sampling station according to the spatial location information of the sampling station. Spatial coordinate information is assigned to the suspended sediment concentration data of each sampling station so that it can correspond spatially with the remote sensing reflectance data; Based on the spatial coordinates of the sampling sites, the corresponding pixels or regions are found in the unified remote sensing reflectance dataset, and the measured suspended sediment concentration data are correlated with the remote sensing reflectance data of that pixel or region. For UAV imagery data, the sampling stations are matched to the corresponding pixels; for satellite remote sensing data, the matching is performed by determining the pixel or the average value of multiple pixels where the sampling station is located, based on its spatial resolution. Establish a data fusion table or database to store the matched remote sensing reflectance data and measured suspended sediment concentration data, forming an integrated dataset with spatiotemporal matching. The dataset contains the remote sensing reflectance information and measured suspended sediment concentration value corresponding to each sampling point, as well as the corresponding spatiotemporal information.

[0016] According to the present invention, a method for monitoring and maintaining water and sediment in offshore wind farms based on integrated air-ground hydraulic engineering is provided. The specific steps for constructing the floating sediment concentration inversion model include: From the red, green, blue, near-infrared and short-wave infrared bands of the band combination data cube, all possible dual-band ratios, three-band combinations and normalized difference indices are selected as candidate feature variables. Calculate the Pearson correlation coefficient between each characteristic variable and the measured suspended sediment concentration, and select band combinations that meet the conditions as candidate variables for modeling; Variance inflation factor analysis was performed on the selected band combinations to remove multicollinear variables with VIF>10; Construct at least three mathematical models using the selected feature variables: Linear regression model: SSC = a·X + b, where X is the feature variable, and a and b are the regression coefficients; Exponential regression model: SSC = a·e (b·X) ; Log-regression model: SSB = a·ln(X) + b; Alternatively, a support vector regression model based on machine learning could be used, employing a radial basis function kernel. The actual test dataset was divided into training and test sets, and the model performance was evaluated using 5-fold cross-validation. Using the coefficient of determination, root mean square error, and mean absolute error as evaluation indicators, the model with the best overall performance is selected as the final inversion model. Validate the optimal model using an independent validation dataset. When the validation set meets the preset conditions, perform the following correction steps: Auxiliary environmental parameters are introduced as secondary feature variables; A segmented modeling strategy was adopted, and sub-models were constructed according to the concentration range of the measured suspended sediment concentration. Gaussian process regression is used to correct the model residuals to generate the final composite model.

[0017] According to the present invention, a method for monitoring and maintaining sediment in offshore wind farms based on integrated air-ground-hydraulic engineering is provided. The specific steps for wake region identification and wake intensity calculation include: A double-layered circular buffer zone is constructed with the center of a single wind turbine as the center. The annular region is divided into N=32 sector units according to the azimuth angle; Obtain the power flow direction data at the sampling time and determine the sector unit number corresponding to the mainstream direction; With the fan-shaped unit in the mainstream direction as the center, k units are extended to both sides to form a candidate wake region; Within the candidate area, the fan-shaped unit with the highest suspended sediment concentration was selected as the core wake zone, and its two adjacent fan-shaped units were selected as the transition wake zones. Within the vertical orientation of the mainstream direction, two symmetrical sector units at a distance from the center of the wind turbine are selected as background reference areas; The wake intensity index ΔSSC is calculated using the following formula: ΔSSC = SSC w - (SSC b +α·σ b ) Among them, SSC wSSC is the mean SSC value in the core wake region; b σ is the mean SSC value of the background region. b α is the standard deviation of the SSC in the background region; α is the empirical coefficient.

[0018] According to the present invention, a water and sediment operation and maintenance monitoring method for offshore wind farms based on integrated air-ground-hydraulic engineering is provided, which divides the high tide and low tide conditions according to the measured data of the tide level station. Under each operating condition, the following data were collected: Measured values ​​of suspended sediment concentration in the wake region of wind turbines with different foundation structure types C obs ; The wake intensity index ΔSSC corresponds to the region. For each combination of working conditions and foundation structure type, establish correction parameters. γ Dynamic mapping table: The support vector machine algorithm is used to train the working condition-structure type identification model. The input parameters are power flow direction, current velocity, water depth, and foundation type. The output is the corresponding... γ value; Embed a working condition identification module in the monitoring system to automatically match the current working condition with the basic type; Call the parameter correction library or SVM model to obtain real-time γ The value is substituted into the correction formula to update the suspended sediment inversion results.

[0019] According to the present invention, a method for monitoring and maintaining water and sediment in offshore wind farms based on integrated air-ground-hydraulic engineering is provided. This method utilizes a suspended sediment concentration inversion model to perform long-term time-series suspended sediment concentration inversion on images before and after wind farm construction, and includes the following steps: Geometric correction and radiometric normalization were performed on multiple satellite remote sensing images before and after the wind farm construction to eliminate sensor differences and atmospheric effects. Based on the time difference threshold between the actual sampling time and the image acquisition time, and the spatial matching threshold between the spatial sampling point and the image pixel, the spatiotemporally synchronized image data and the measured data are filtered. The preprocessed images are input into the suspended sediment concentration inversion model, and the suspended sediment concentration values ​​of each period of images are calculated pixel by pixel to generate a long-term concentration dataset covering at least 1 year before the wind farm construction and at least 3 years after the construction. Trend analysis was performed on long-term time-series datasets, and the Mann-Kendall test was used to identify regions of significant concentration changes. The four concentration thresholds are set according to the "Marine Functional Zoning Water Quality Standards"; Based on the trend analysis results, the areas with excessive concentrations are divided into areas of continuous deterioration, areas of fluctuating exceedances, and areas of temporary exceedances, and a dynamic risk level distribution map including the time dimension is generated by overlaying these areas. Kriging interpolation is used to spatially interpolate the missing data to ensure the spatial continuity of the risk level map.

[0020] According to the present invention, a method for monitoring and maintaining water and sediment in offshore wind farms based on integrated air-ground-hydraulic engineering is provided, which integrates underwater topographic data and long-term suspended sediment concentration data obtained by a multibeam echo sounder to construct a three-dimensional water and sediment environment database. Slope analysis of topographic data, combined with changes in suspended sediment concentration gradients, identifies areas of active sediment migration; Based on measured data of cable burial depth and sediment erosion rate, a cable exposure early warning model is established:

[0021] in, For cable burial depth, For flushing rate; Combining the basic flow type and local flow velocity, CFD numerical simulation is used to correct the scour depth prediction formula:

[0022] in, Basic type coefficients, This is the actual flow rate. The critical scouring velocity. For runtime, It is an experience index; By overlaying dynamic risk level distribution maps, cable exposure risk maps, and foundation scour risk maps, a comprehensive risk heat map is generated using GIS spatial overlay analysis. Extract the area proportion of each risk level region, the time series of typical risk events, and the risk evolution trend, and automatically generate a 3D visualization report containing spatiotemporal evolution characteristics, supporting dynamic display at multiple scales by year / quarter / month.

[0023] Therefore, compared with the existing technology, the water and sediment operation and maintenance monitoring method and system for offshore wind farms based on integrated air-ground-hydraulic engineering proposed in this invention has the following beneficial effects: 1. This invention constructs an integrated "sky-ground-hydraulic" monitoring system, effectively integrating multi-source data resources such as satellite remote sensing, on-site monitoring, UAV data, and engineering parameters. This not only significantly expands the monitoring range, achieving comprehensive coverage of the water and sediment environment of offshore wind farms, but also significantly reduces reliance on costly on-site sampling. Compared to traditional monitoring methods, the implementation cost of this invention is reduced by approximately 50% or more, while simultaneously improving monitoring frequency and timeliness, providing strong data support for the operation and maintenance management of wind farms.

[0024] 2. The multi-source data fusion model adopted in this invention constructs a high-precision suspended sediment concentration inversion model by comprehensively analyzing satellite remote sensing images, field sampling data, and UAV images. This model can significantly reduce inversion errors, controlling the root mean square error (RMSE) to below 6 mg / L, which is far superior to traditional single remote sensing methods (RMSE is often greater than 10 mg / L). This makes the monitoring results more accurate and reliable, providing a solid data foundation for water and sediment environmental management and risk assessment of wind farms.

[0025] 3. The wake monitoring method of this invention can accurately quantify the disturbance effect of wind turbines and, combined with changes in tidal current direction, effectively identify the impact of tidal differences on the water and sediment environment. Through real-time monitoring and data analysis, this invention can provide early warning of wind turbine foundation scour risks, providing a scientific basis for wind farm operation and maintenance decisions, enabling wind farms to respond promptly to potential safety hazards and ensuring the safe and stable operation of wind power facilities.

[0026] 4. The water and sediment change risk level map generated by this invention integrates multi-dimensional information such as suspended sediment concentration, topographic changes, and engineering parameters, providing comprehensive data support for the entire life cycle operation and maintenance of wind farms. From the initial construction phase to the operation and maintenance phase, this invention can provide continuous and stable monitoring services, helping operation and maintenance personnel to understand the water and sediment environment of the wind farm in a timely manner and formulate scientific and reasonable operation and maintenance plans.

[0027] 5. The technical solution of this invention is not only applicable to current offshore wind farm monitoring needs, but also possesses strong adaptability and scalability. With the continuous development of the offshore wind power industry, the technology of this invention can be easily extended to wind farm monitoring projects in other sea areas, providing technical support for broader offshore wind power development. Simultaneously, the technical architecture and data processing methods of this invention also possess strong flexibility and customizability, enabling personalized adjustments and optimizations based on the actual needs of different wind farms.

[0028] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0029] Figure 1 This is a flowchart of an embodiment of an offshore wind farm water and sediment operation and maintenance monitoring method based on the integrated sky-ground-hydraulic engineering of the present invention.

[0030] Figure 2 This is a schematic diagram illustrating the principle of constructing a suspended sediment inversion model in an embodiment of an integrated air-ground-hydraulic monitoring method for offshore wind farm water and sediment operations.

[0031] Figure 3 This is a schematic diagram of the band combination in an embodiment of the water and sediment operation and maintenance monitoring method for offshore wind farms based on the integration of air, ground and hydraulic engineering of the present invention.

[0032] Figure 4 This is a schematic diagram of model training and verification data in an embodiment of the water and sediment operation and maintenance monitoring method for offshore wind farms based on the integrated sky-ground-hydraulic engineering of the present invention.

[0033] Figure 5 This is a schematic diagram of wake sampling in an embodiment of the water and sediment operation and maintenance monitoring method for offshore wind farms based on the integration of air, ground and hydraulic engineering of the present invention.

[0034] Figure 6 This is a schematic diagram of the spatial distribution of the multi-year average concentration of suspended sediment after the wind farm was built (2019-2024) in an embodiment of the "Air-Ground-Hydraulic Integrated Monitoring Method for Operation and Maintenance of Offshore Wind Farms" of the present invention.

[0035] Figure 7 This is a schematic diagram illustrating the monitoring of water and sediment concentration distribution in the wake of wind turbine foundations in an embodiment of an integrated air-ground-hydraulic monitoring method for offshore wind farms. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0037] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0038] An Example of a Monitoring Method for Water and Sediment Operation and Maintenance in Offshore Wind Farms Based on Integrated Sky-Ground Hydraulic Engineering See Figures 1 to 7 This embodiment provides a method for monitoring and maintaining water and sediment in offshore wind farms based on an integrated air-ground-hydraulic system, including: Multi-source data acquisition and preprocessing steps: Multispectral image data of the target sea area is acquired through a satellite remote sensing platform. Images with cloud coverage below a preset threshold are selected and subjected to radiometric correction and cloud removal. Simultaneously, on-site water sample collection and high-resolution UAV image acquisition were carried out in the wind farm area to determine the measured values ​​of suspended sediment concentration and ensure the spatiotemporal consistency of the images. Construct an integrated dataset that matches time and space, combining remote sensing reflectance data with measured suspended sediment concentration data; Steps for constructing a suspended sediment inversion model: Based on the correlation analysis between multispectral band combinations and measured suspended sediment concentration data, band combinations that meet the preset correlation threshold are selected. At least two types of mathematical inversion models were constructed using measured data, and the optimal model was selected as the suspended sediment concentration inversion model through cross-validation. Steps for monitoring and analyzing the difference in wind turbine wake: A multi-level annular buffer zone was established with the wind turbine foundation as the center and sector-shaped analysis units were divided. The wake influence area was identified by combining tidal flow data and the wake intensity index was calculated. The wake intensity index was introduced as a correction parameter in the suspended sediment inversion model. The differences in wake characteristics of wind turbines with different foundation structure types under high tide and low tide conditions were analyzed separately, and the correction parameters of the suspended sediment inversion model were dynamically adjusted according to the foundation structure type. Steps for water and sediment environmental risk assessment and visualization output: The suspended sediment concentration inversion model was used to perform long-term suspended sediment concentration inversion on images before and after the wind farm construction, generating a risk level distribution map of dynamic changes in water and sediment. Integrating underwater topographic monitoring data, the system assesses the risks of sediment migration, submarine cable exposure, and wind turbine foundation scour, and outputs a comprehensive risk assessment report on the water and sediment environment that includes spatiotemporal evolution characteristics.

[0039] In this embodiment, the process of acquiring multispectral image data of the target sea area via a satellite remote sensing platform includes the following steps: Simultaneously acquire Sentinel-2 and Landsat-8 / 9 images, which must cover at least the target wind farm's sea area and a surrounding 5km buffer zone; Images with cloud coverage less than a preset threshold are filtered out using automatic cloud detection algorithms (such as Fmask or MAJA), and invalid data with strip noise, missing rows, or abnormal radiation are removed. Atmospheric correction is performed using a modified DSF (Dark Spectrum Fitting) algorithm based on the dark target method, which includes: Aerosol optical thickness was jointly retrieved from Sentinel-2 images using the 620nm, 842nm, and 1610nm bands. The Landsat-8 / 9 imagery was processed using a combination of 655nm, 865nm, and 1609nm bands, and a neighborhood smoothing constraint was introduced. An improved Acolite declouding algorithm is applied to generate a binary cloud mask, and neighborhood interpolation is used to repair the residual cloud shadow areas. Orthorectification is performed using the RPC model, projecting the image onto the WGS84 coordinate system, achieving a planar accuracy better than one pixel. Spatial resolution fusion was performed on Sentinel-2 and Landsat-8 / 9 images, and a daily 10m resolution reflectance dataset was generated using the STARFM temporal fusion model. Construct a data cube that combines blue, green, red, near-infrared, and short-wave infrared bands.

[0040] In this embodiment, the following steps are included when conducting on-site water sample collection and high-resolution UAV image acquisition: At the sampling sites pre-set within the target wind farm sea area, vertical water samplers were used to collect water samples with a volume of 300mL to 1L at a depth of 0.3m to 0.7m underwater. The sampling sites covered a 500m radius around the wind turbine foundation and key areas of the tidal current channel. The collected water samples were sealed and stored in brown glass containers, placed in an insulated transport device at 4℃~10℃, and laboratory analysis was completed within 4 hours after sampling. Using a multi-rotor drone equipped with a visible light-near infrared camera, orthophotos of the sampling area were acquired within 15 minutes before and after sampling, with the flight altitude controlled between 50m and 200m and the spatial resolution of the images better than 5cm. In laboratory analysis, a glass fiber filter membrane with a pore size of 0.4μm~0.5μm was used for vacuum filtration. The filter membrane drying temperature was set at 35℃~55℃, and the mass weighing accuracy after drying reached 0.1mg. The suspended sediment concentration is calculated using the formula SSC=(M2-M1) / V×1000, where M1 is the initial mass of the filter membrane (g), M2 is the total mass of the dried filter membrane and sediment (g), and V is the volume of the filtered water sample (mL). The unit of the calculation result is mg / L.

[0041] Image data obtained through satellite remote sensing platforms and orthorectified and projected onto the WGS84 coordinate system, as well as orthophoto data obtained by UAVs, are unified under the same geographic coordinate reference frame to ensure spatial consistency.

[0042] Based on the specific time of on-site water sample collection, satellite remote sensing image data and UAV image data acquired within a certain time window before and after the sampling time are selected to ensure data matching in the time dimension.

[0043] For satellite remote sensing data: data cubes containing combinations of blue, green, red, near-infrared, and shortwave infrared bands will be constructed from Sentinel-2 and Landsat-8 / 9 images. Based on the target wind farm sea area and surrounding area, reflectance data of the corresponding area will be extracted to form a subset of satellite remote sensing reflectance data. For UAV imagery data: Band extraction is performed on high-resolution orthophotos acquired by UAVs equipped with visible light-near infrared cameras to obtain reflectivity information that corresponds to or is similar to satellite remote sensing bands, and a subset of UAV remote sensing reflectivity data is constructed.

[0044] By fusing subsets of satellite remote sensing reflectance data and UAV remote sensing reflectance data, and using a pixel-based method, the data is resampled and integrated according to the spatial resolution differences of different data sources to form a unified remote sensing reflectance dataset.

[0045] The suspended sediment concentration data obtained from laboratory analysis and calculation of each sampling site were organized according to the spatial location information of the sampling sites.

[0046] The suspended sediment concentration data of each sampling station is assigned accurate spatial coordinate information so that it can correspond spatially with the remote sensing reflectance data.

[0047] Based on the spatial coordinates of the sampling sites, the corresponding pixels or regions are found in the unified remote sensing reflectance dataset, and the measured suspended sediment concentration data are correlated with the remote sensing reflectance data of that pixel or region.

[0048] For UAV imagery data, due to its high spatial resolution, sampling stations can be accurately matched to the corresponding pixels; for satellite remote sensing data, matching is performed by determining the pixel or the average value of multiple pixels where the sampling station is located, based on its spatial resolution.

[0049] Establish a data fusion table or database to store the matched remote sensing reflectance data and measured suspended sediment concentration data, forming an integrated dataset with spatiotemporal matching. The dataset contains remote sensing reflectance information (reflectance values ​​of each band) and measured suspended sediment concentration values ​​corresponding to each sampling point, as well as the corresponding spatiotemporal information (time, spatial coordinates, etc.).

[0050] In this embodiment, the steps for constructing the floating sediment concentration inversion model specifically include: From the red, green, blue, near-infrared and shortwave infrared bands of the band combination data cube, all possible dual-band ratios (R1 / R2), three-band combinations (R1×R2 / R3), and normalized difference index (NDI=(R1-R2) / (R1+R2)) are selected as candidate feature variables, where R1, R2, and R3 are the surface reflectance of different bands; Calculate the Pearson correlation coefficient between each characteristic variable and the measured suspended sediment concentration (SSC), and select band combinations that satisfy |r|≥0.65 as candidate variables for modeling; Variance inflation factor (VIF) analysis was performed on the selected band combinations to remove multicollinear variables with VIF > 10; Construct at least three mathematical models using the selected feature variables: Linear regression model: SSC = a·X + b, where X is the feature variable, and a and b are the regression coefficients; Exponential regression model: SSC = a·e (b·X) ; Log-regression model: SSB = a·ln(X) + b; Alternatively, a machine learning-based support vector regression (SVR) model could be used, employing a radial basis function (RBF). The actual test dataset was divided into training and test sets in a 7:3 ratio, and the model performance was evaluated using 5-fold cross-validation. The coefficient of determination (R) 2 The evaluation indicators are ≥0.85), root mean square error (RMSE≤15mg / L), and mean absolute error (MAE≤10mg / L). The model with the best overall performance is selected as the final inversion model. The optimal model is validated using an independent validation dataset (not the measured data used for modeling), when the validation set R... 2 If the concentration is <0.8 or RMSE>20 mg / L, perform the following correction procedure: Auxiliary environmental parameters (such as tide level and wind speed) are introduced as secondary characteristic variables; A segmented modeling strategy was adopted, and sub-models were constructed according to the SSC concentration range (e.g., 0-50 mg / L, 50-200 mg / L). Gaussian process regression (GPR) is applied to the model residuals to generate the final composite model.

[0051] Specifically, based on the correlation analysis between remote sensing band combinations and measured SSCs, band combinations with a correlation coefficient higher than 0.6 (such as (R+G) / (B+G+N), R... 2 / (G B), R / B, etc., are used to construct subsequent inversion models.

[0052] Based on the aforementioned band combinations with high correlation coefficients, 70% of the measured data was used to construct linear, exponential, and logarithmic inversion models, as shown in Table 1: Table 1: Inversion Model and Band Combinations

[0053] Select R among them 2 For the top four models, the remaining measured data are used for accuracy evaluation. The accuracy evaluation results show that among them, Model 3 (the expression is: y = 0.0863 e6.4814x, x = (R + G) / (B + G + N)) has the best comprehensive performance in terms of indicators such as R², RMSE, and MAPD, and is finally selected as the suspended sediment concentration inversion model of this embodiment, as shown in Table 2: Table 2: Selection of suspended sediment concentration inversion model

[0054] In this embodiment, the steps of wake area identification and wake intensity calculation specifically include: Taking the center of a single fan as the center of a circle, construct a double-layer circular buffer zone, where the inner ring radius R1 = 80m to 120m, preferably 100m, and the outer ring radius R2 = 300m to 400m, preferably 350m; Divide the circular area into N = 32 sector units according to the azimuth angle, and the central angle of each unit θ = 360° / N = 11.25°; Obtain the tidal current direction data at the sampling moment, which can be measured by ADCP or output by a numerical model, and determine the sector unit number corresponding to the main flow direction; Taking the sector unit in the main flow direction as the center, expand k units (k = 1 to 3) to both sides to form a candidate wake area, where the k value is dynamically adjusted according to the rated power of the fan: When P ≤ 2MW, k = 1; When 2MW < P ≤ 5MW, k = 2; When P > 5MW, k = 3; Select the sector unit with the highest suspended sediment concentration (SSC) in the candidate area as the core wake area, and its two adjacent sector units as the transition wake area; Within the range of ±90° in the perpendicular direction of the main flow direction, select 2 symmetric sector units at a distance of 250m to 300m from the fan center as the background reference area; When the standard deviation σ of the SSC in the background area > 15mg / L, remove the outliers and re-select until σ ≤ 15mg / L; Calculate the wake intensity index ΔSSC, and the formula is: ΔSSC = SSC w - (SSC b + α·σ b ) Where, SSC w is the average value of SSC in the core wake area (mg / L); SSC b is the average value of SSC in the background area (mg / L); σb α is the standard deviation of SSC in the background area (mg / L); α is an empirical coefficient, ranging from 0.5 to 1.5, adjusted according to the water type: high turbidity water in estuaries: α = 1.2 to 1.5; clear water near the shore: α = 0.5 to 0.8.

[0055] When ΔSSC>0, it is determined to be an effective wake, and its spatial influence range is defined as follows: Core area: R1~R3 ​​(R3=R2×0.7); Diffusion region: R3~R2; When the angle φ between the tidal current direction and the prevailing wind direction is greater than 45°, the radius of the wake region is corrected according to R2'=R2×(1-0.2·sinφ); When the tidal change rate |dh / dt|>0.5m / h, pause the wake intensity calculation and mark the data quality level as "suspicious".

[0056] Based on the measured data from the tide gauge station, the operating conditions are divided into high tide (tide level rise rate > 0.05 m / h) and low tide (tide level fall rate > 0.05 m / h); Under each operating condition, the following data were collected: Measured values ​​of suspended sediment concentration in the wake region of wind turbines with different foundation structure types (monopile / jacket / gravity type) C obs ; The wake intensity index ΔSSC for the corresponding region; For each combination of working conditions and foundation structure type (e.g., high tide-monopile, low tide-jacket structure), establish correction parameters. γ Dynamic mapping table: The support vector machine (SVM) algorithm is used to train the working condition-structure type identification model. The input parameters are power flow direction, current velocity, water depth, and foundation type. The output is the corresponding... γ value; Embed a working condition identification module in the monitoring system to automatically match the current working condition with the basic type; Call the parameter correction library or SVM model to obtain real-time γ The value is substituted into the correction formula to update the suspended sediment inversion results; The corrected results are verified for error. When the relative error is greater than 15%, the parameter re-optimization process is triggered.

[0057] In this embodiment, a suspended sediment concentration inversion model is used to perform long-term time-series suspended sediment concentration inversion on images before and after wind farm construction, including the following steps: Geometric correction and radiometric normalization were performed on multiple satellite remote sensing images before and after the wind farm construction to eliminate sensor differences and atmospheric effects. Based on the time difference threshold between the actual sampling time and the image acquisition time, and the spatial matching threshold between the spatial sampling point and the image pixel, the spatiotemporally synchronized image data and the measured data are filtered. The preprocessed images are input into the suspended sediment concentration inversion model, and the suspended sediment concentration values ​​of each period of images are calculated pixel by pixel to generate a long-term concentration dataset covering at least 1 year before the wind farm construction and at least 3 years after the construction. Trend analysis was performed on long-term time-series datasets, and the Mann-Kendall test was used to identify regions of significant concentration changes. According to the "Marine Functional Zoning Water Quality Standard", four concentration thresholds are set (Class I ≤10mg / L, Class II 10-20mg / L, Class III 20-50mg / L, Class IV >50mg / L). Based on the trend analysis results, the areas with excessive concentrations are divided into areas of continuous deterioration, areas of fluctuating exceedances, and areas of temporary exceedances, and a dynamic risk level distribution map including the time dimension is generated by overlaying these areas. Kriging interpolation is used to spatially interpolate the missing data to ensure the spatial continuity of the risk level map.

[0058] A three-dimensional water and sediment environment database is constructed by integrating underwater topographic data and long-term suspended sediment concentration data obtained by a multibeam bathymetry system. Slope analysis was performed on the topographic data, and areas with a slope greater than 15° were marked as high-risk areas. The changes in suspended sediment concentration gradients were combined to identify areas with active sediment migration. Based on measured data of cable burial depth and sediment erosion rate, a cable exposure early warning model is established by calculating the difference between two adjacent topographic data periods.

[0059] in, For cable burial depth, For flushing rate; Wind turbine foundation scour risk: Combining foundation type (monopile / jacket / gravity type) and local flow velocity (calculated through tidal flow model), CFD numerical simulation is used to correct the scour depth prediction formula.

[0060] in, The basic type coefficient is 1.2 for single pile, 0.8 for jacket, and 0.5 for gravity type. This is the actual flow rate. The critical scouring velocity. For runtime, The experience index (ranged from 0.3 to 0.7); By overlaying dynamic risk level distribution maps, cable exposure risk maps, and foundation scour risk maps, a comprehensive risk heat map is generated using GIS spatial overlay analysis. Extract the area proportion of each risk level region, the time series of typical risk events, and the risk evolution trend, and automatically generate a 3D visualization report containing spatiotemporal evolution characteristics, supporting dynamic display at multiple scales by year / quarter / month.

[0061] In practical applications, the method provided in this embodiment has been applied to regional wind farms, such as the Guishan Wind Farm after its completion (2019-2024). Figure 6 and Figure 7 As shown, its reliability has been verified. Furthermore, the technology can be combined with artificial intelligence for predictive analysis, further improving operational efficiency.

[0062] In summary, the integrated air-ground-water engineering monitoring method for offshore wind farm water and sediment operation and maintenance proposed in this invention demonstrates significant advantages and beneficial effects in terms of efficiency, economy, high-precision inversion, dynamic risk assessment, system integration, and adaptability. The implementation of this invention will strongly promote the development and application of offshore wind farm monitoring technology, providing a strong guarantee for the sustainable development of the offshore wind power industry.

[0063] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0064] The above embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of protection of the present invention. Any non-substantial changes and substitutions made by those skilled in the art based on the present invention shall fall within the scope of protection claimed by the present invention.

Claims

1. A method for monitoring and maintaining water and sediment in offshore wind farms based on integrated air-ground-water engineering, characterized in that, include: Multi-source data acquisition and preprocessing steps: Multispectral image data of the target sea area is acquired through a satellite remote sensing platform. Images with cloud coverage below a preset threshold are selected and subjected to radiometric correction and cloud removal. Simultaneously, on-site water sample collection and high-resolution UAV image acquisition were carried out in the wind farm area to determine the measured values ​​of suspended sediment concentration and ensure the spatiotemporal consistency of the images. Construct an integrated dataset that matches time and space, combining remote sensing reflectance data with measured suspended sediment concentration data; Steps for constructing a suspended sediment inversion model: Based on the correlation analysis between multispectral band combinations and measured suspended sediment concentration data, band combinations that meet the preset correlation threshold are selected. At least two types of mathematical inversion models were constructed using measured data, and the optimal model was selected as the suspended sediment concentration inversion model through cross-validation. Steps for monitoring and analyzing the difference in wind turbine wake: A multi-level annular buffer zone was established with the wind turbine foundation as the center and sector-shaped analysis units were divided. The wake influence area was identified by combining tidal flow data and the wake intensity index was calculated. The wake intensity index was introduced as a correction parameter in the suspended sediment inversion model. The differences in wake characteristics of wind turbines with different foundation structure types under high tide and low tide conditions were analyzed separately, and the correction parameters of the suspended sediment inversion model were dynamically adjusted according to the foundation structure type. Steps for water and sediment environmental risk assessment and visualization output: The suspended sediment concentration inversion model was used to perform long-term suspended sediment concentration inversion on images before and after the wind farm construction, generating a risk level distribution map of dynamic changes in water and sediment. Integrating underwater topographic monitoring data, the system assesses the risks of sediment migration, submarine cable exposure, and wind turbine foundation scour, and outputs a comprehensive risk assessment report on the water and sediment environment that includes spatiotemporal evolution characteristics.

2. The method according to claim 1, characterized in that, When acquiring multispectral image data of a target sea area through a satellite remote sensing platform, the following steps are included: Simultaneously acquire Sentinel-2 and Landsat-8 / 9 images, which must cover at least the target wind farm's sea area and a surrounding 5km buffer zone; The automatic cloud detection algorithm filters images with cloud coverage less than a preset threshold and removes invalid data with strip noise, missing rows, or abnormal radiation. Atmospheric correction was performed using a DSF algorithm improved based on the dark target method. An improved Acolite declouding algorithm is applied to generate a binary cloud mask, and neighborhood interpolation is used to repair the residual cloud shadow areas. Orthorectification was performed using the RPC model, and the image was projected onto the WGS84 coordinate system; Spatial resolution fusion was performed on Sentinel-2 and Landsat-8 / 9 images, and a daily 10m resolution reflectance dataset was generated using the STARFM temporal fusion model. Construct a data cube that combines blue, green, red, near-infrared, and short-wave infrared bands.

3. The method according to claim 2, characterized in that, The following steps are included when conducting on-site water sample collection and high-resolution UAV image acquisition: At the sampling site pre-set within the target wind farm sea area, water samples were collected underwater using a vertical water sampler. The sampling site covered a 500m radius around the wind turbine foundation and key areas of the tidal current channel. The collected water samples were sealed and stored in brown glass containers, placed in an insulated transport device, and then analyzed in the laboratory. During the laboratory analysis, vacuum filtration was performed using a glass fiber membrane, and the concentration of suspended sediment was calculated. Using a multi-rotor drone equipped with a visible-near-infrared camera, orthophotos of the area covered by the sampling were acquired within a set time period before and after sampling.

4. The method according to claim 3, characterized in that: Image data obtained through satellite remote sensing platforms and orthorectified and projected onto the WGS84 coordinate system, as well as orthorectified image data obtained by UAVs, are unified under the same geographic coordinate reference frame. Based on the specific time of on-site water sample collection, satellite remote sensing image data and UAV image data acquired within a certain time window before and after the sampling time were selected; For satellite remote sensing data: data cubes containing combinations of blue, green, red, near-infrared, and shortwave infrared bands will be constructed from Sentinel-2 and Landsat-8 / 9 images. Based on the target wind farm sea area and surrounding area, reflectance data of the corresponding area will be extracted to form a subset of satellite remote sensing reflectance data. For UAV imagery data: Band extraction is performed on high-resolution orthophotos acquired by UAVs equipped with visible light-near infrared cameras to obtain reflectivity information that corresponds to or is similar to satellite remote sensing bands, and a subset of UAV remote sensing reflectivity data is constructed. By fusing subsets of satellite remote sensing reflectance data and UAV remote sensing reflectance data, and using a pixel-based method, the data is resampled and integrated according to the spatial resolution differences of different data sources to form a unified remote sensing reflectance dataset.

5. The method according to claim 4, characterized in that: The suspended sediment concentration data obtained from laboratory analysis and calculation of each sampling site were organized according to the spatial location information of the sampling sites. Spatial coordinate information is assigned to the suspended sediment concentration data of each sampling station so that it can correspond spatially with the remote sensing reflectance data; Based on the spatial coordinates of the sampling sites, the corresponding pixels or regions are found in the unified remote sensing reflectance dataset, and the measured suspended sediment concentration data are correlated with the remote sensing reflectance data of that pixel or region. For UAV imagery data, the sampling stations are matched to the corresponding pixels; for satellite remote sensing data, the matching is performed by determining the pixel or the average value of multiple pixels where the sampling station is located, based on its spatial resolution. Establish a data fusion table or database to store the matched remote sensing reflectance data and measured suspended sediment concentration data, forming an integrated dataset with spatiotemporal matching. The dataset contains the remote sensing reflectance information and measured suspended sediment concentration value corresponding to each sampling point, as well as the corresponding spatiotemporal information.

6. The method according to claim 3, characterized in that, The specific steps for constructing the floating sediment concentration inversion model include: From the red, green, blue, near-infrared and short-wave infrared bands of the band combination data cube, all possible dual-band ratios, three-band combinations and normalized difference indices are selected as candidate feature variables. Calculate the Pearson correlation coefficient between each characteristic variable and the measured suspended sediment concentration, and select band combinations that meet the conditions as candidate variables for modeling; Variance inflation factor analysis was performed on the selected band combinations to remove multicollinear variables with VIF>10; Construct at least three mathematical models using the selected feature variables: Linear regression model: SSC = a·X + b, where X is the feature variable, and a and b are the regression coefficients; Exponential regression model: SSC = a·e (b·X) ; Log-regression model: SSB = a·ln(X) + b; Alternatively, a support vector regression model based on machine learning could be used, employing a radial basis function kernel. The actual test dataset was divided into training and test sets, and the model performance was evaluated using 5-fold cross-validation. Using the coefficient of determination, root mean square error, and mean absolute error as evaluation indicators, the model with the best overall performance is selected as the final inversion model. Validate the optimal model using an independent validation dataset. When the validation set meets the preset conditions, perform the following correction steps: Auxiliary environmental parameters are introduced as secondary feature variables; A segmented modeling strategy was adopted, and sub-models were constructed according to the concentration range of the measured suspended sediment concentration. Gaussian process regression is used to correct the model residuals to generate the final composite model.

7. The method according to claim 1, characterized in that, The specific steps for wake region identification and wake intensity calculation include: A double-layered circular buffer zone is constructed with the center of a single wind turbine as the center. The annular region is divided into N=32 sector units according to the azimuth angle; Obtain the power flow direction data at the sampling time and determine the sector unit number corresponding to the mainstream direction; With the fan-shaped unit in the mainstream direction as the center, k units are extended to both sides to form a candidate wake region; Within the candidate area, the fan-shaped unit with the highest suspended sediment concentration was selected as the core wake zone, and its two adjacent fan-shaped units were selected as the transition wake zones. Within the vertical orientation of the mainstream direction, two symmetrical sector units at a distance from the center of the wind turbine are selected as background reference areas; The wake intensity index ΔSSC is calculated using the following formula: ΔSSC = SSC w - (SSC b +a·s b ) Among them, SSC w SSC is the mean SSC value in the core wake region; b σ is the mean SSC value of the background region. b α is the standard deviation of the SSC in the background region; α is the empirical coefficient.

8. The method according to claim 7, characterized in that: The high tide and low tide conditions are divided according to the actual measured data of the tide gauge station; Under each operating condition, the following data were collected: Measured values ​​of suspended sediment concentration in the wake region of wind turbines with different foundation structure types C obs ; The wake intensity index ΔSSC corresponds to the region. For each combination of working conditions and foundation structure type, establish correction parameters. γ Dynamic mapping table: The support vector machine algorithm is used to train the working condition-structure type identification model. The input parameters are power flow direction, current velocity, water depth, and foundation type. The output is the corresponding... γ value; Embed a working condition identification module in the monitoring system to automatically match the current working condition with the basic type; Call the parameter correction library or SVM model to obtain real-time γ The value is substituted into the correction formula to update the suspended sediment inversion results.

9. The method according to any one of claims 1 to 8, characterized in that, The suspended sediment concentration was inverted using a suspended sediment concentration inversion model to retrieve long-term suspended sediment concentrations from images before and after wind farm construction. The steps included: Geometric correction and radiometric normalization were performed on multiple satellite remote sensing images before and after the wind farm construction to eliminate sensor differences and atmospheric effects. Based on the time difference threshold between the actual sampling time and the image acquisition time, and the spatial matching threshold between the spatial sampling point and the image pixel, the spatiotemporally synchronized image data and the measured data are filtered. The preprocessed images are input into the suspended sediment concentration inversion model, and the suspended sediment concentration values ​​of each period of images are calculated pixel by pixel to generate a long-term concentration dataset covering at least 1 year before the wind farm construction and at least 3 years after the construction. Trend analysis was performed on long-term time-series datasets, and the Mann-Kendall test was used to identify regions of significant concentration changes. The four concentration thresholds are set according to the "Marine Functional Zoning Water Quality Standards"; Based on the trend analysis results, the areas with excessive concentrations are divided into areas of continuous deterioration, areas of fluctuating exceedances, and areas of temporary exceedances, and a dynamic risk level distribution map including the time dimension is generated by overlaying these areas. Kriging interpolation is used to spatially interpolate the missing data to ensure the spatial continuity of the risk level map.

10. The method according to claim 9, characterized in that: A three-dimensional water and sediment environment database is constructed by integrating underwater topographic data and long-term suspended sediment concentration data obtained by a multibeam bathymetry system. Slope analysis of topographic data, combined with changes in suspended sediment concentration gradients, identifies areas of active sediment migration; Based on measured data of cable burial depth and sediment erosion rate, a cable exposure early warning model is established: in, For cable burial depth, For flushing rate; Combining the basic flow type and local flow velocity, CFD numerical simulation is used to correct the scour depth prediction formula: in, Basic type coefficients, This is the actual flow rate. The critical scouring velocity. For runtime, It is an experience index; By overlaying dynamic risk level distribution maps, cable exposure risk maps, and foundation scour risk maps, a comprehensive risk heat map is generated using GIS spatial overlay analysis. Extract the area proportion of each risk level region, the time series of typical risk events, and the risk evolution trend, and automatically generate a 3D visualization report containing spatiotemporal evolution characteristics, supporting dynamic display at multiple scales by year / quarter / month.

Images