CN-121997724-A - Marine wind power plant water Sha Yunwei monitoring method based on sky and land hydraulic engineering integration

CN121997724ACN 121997724 ACN121997724 ACN 121997724ACN-121997724-A

Abstract

The invention provides a sky-land-hydraulic-integration-based offshore wind farm water Sha Yunwei monitoring method, which comprises a multi-source data acquisition and pretreatment step, a suspended sediment inversion model construction step, a fan wake monitoring and difference analysis step and a water and sand environment risk assessment and visual output step. The method can solve the problems of low efficiency and high cost of the traditional monitoring method, realize large-scale, high-frequency and low-cost monitoring of the water and sand environment of the offshore wind farm, improve the inversion precision of suspended sediment, reduce errors through multi-source data fusion, accurately monitor the wake effect of a fan, quantify the influence of different foundation types and tide conditions on the water and sand disturbance, early warn foundation flushing risks, generate a water and sand change risk level diagram through constructing an integrated monitoring system, and provide scientific basis for the safe operation and maintenance of the wind farm.

Inventors

WANG ZHONGQUAN
WU XIN
DU LONGJIANG
LIU SHA
DENG XIAOYA
ZHU CHUNSHENG
DU TIANCANG
Deng Huasen
ZHAO ANXIN
QIN PAN

Assignees

南方海上风电联合开发有限公司

Dates

Publication Date: 20260508
Application Date: 20260109

Claims (10)

1. The marine wind power plant water Sha Yunwei monitoring method based on sky land hydraulic integration is characterized by comprising the following steps: the multi-source data acquisition and preprocessing steps are as follows: Acquiring multispectral image data of a target sea area through a satellite remote sensing platform, screening images with cloud coverage rate lower than a preset threshold value, and performing radiation correction and cloud removal processing; Synchronously carrying out field water sample acquisition and unmanned aerial vehicle high-resolution image acquisition in a wind power field area, measuring actual measurement values of suspended sediment concentration and ensuring the space-time consistency of images; Constructing a space-time matching integrated data set, and fusing remote sensing reflectivity data with actual measurement suspended sediment concentration data; the construction method of the suspended sediment inversion model comprises the following steps: Based on correlation analysis of multispectral wave band combinations and actually measured suspended sediment concentration data, screening wave band combinations meeting preset correlation thresholds; Respectively constructing at least two types of mathematical inversion models by using measured data, and selecting an optimal model as a suspended sediment concentration inversion model through cross validation; Fan wake flow monitoring and difference analysis steps: Establishing a multistage annular buffer zone with a fan foundation as a center, dividing a sector analysis unit, identifying a wake flow influence area by combining tide data, calculating wake flow intensity indexes, and introducing the wake flow intensity indexes into a suspended sediment inversion model as correction parameters; respectively analyzing wake characteristic differences of fans of different foundation structure types under the working conditions of the rising tide/falling tide, and dynamically adjusting correction parameters of a suspended sediment inversion model according to the foundation structure types; And (3) water and sand environment risk assessment and visual output: Carrying out long-time-sequence inversion on the suspended sediment concentration of the images before and after the wind field construction by using a suspended sediment concentration inversion model to generate a water-sediment dynamic change risk level distribution map; Integrating underwater topography monitoring data, evaluating sediment migration, submarine cable exposure and fan foundation scouring risk, and outputting a water and sand environment comprehensive risk evaluation report containing time-space evolution characteristics.
2. The method according to claim 1, wherein the step of acquiring multispectral image data of the target sea area by the satellite remote sensing platform comprises the steps of: Synchronously acquiring a Sentinel-2 image and a Landsat-8/9 image, wherein the images at least cover a target wind power plant sea area and a peripheral 5km buffer area; Screening images with cloud coverage rate smaller than a preset threshold value through an automatic cloud detection algorithm, and removing invalid data with stripe noise, line loss or radiation abnormality; performing atmospheric correction by adopting a DSF algorithm improved based on a dark target method; Generating a binary Yun Yanmo by applying an improved Acolite cloud removal algorithm, and repairing the residual cloud shadow area by adopting neighborhood interpolation; Adopting an RPC model to carry out orthographic correction, and projecting the image to a WGS84 coordinate system; carrying out spatial resolution fusion on Sentinel-2 and Landsat-8/9 images, and generating a day scale 10m resolution reflectivity data set by adopting a STARFM time sequence fusion model; A band combining data cube is constructed containing blue, green, red, near infrared, short wave infrared.
3. The method according to claim 2, wherein when performing in-field water sample collection and unmanned aerial vehicle high resolution image acquisition, comprising the steps of: a sampling station is preset in the sea area of a target wind power plant, a vertical water sampler is adopted to collect water samples underwater, and the sampling station covers a 500m range of the periphery of a fan foundation and a key region of a tide channel; sealing and storing the collected water sample in a brown glass container, and storing in a heat-insulating transportation device to finish laboratory analysis, wherein a glass fiber filter membrane is adopted for vacuum suction filtration and suspended sediment concentration is calculated during laboratory analysis; The multi-rotor unmanned aerial vehicle is used for carrying a visible light-near infrared camera, and the orthographic image acquisition covering the sampling area is completed within a set time before and after the sampling.
4. A method according to claim 3, characterized in that: The method comprises the steps of unifying image data which are obtained through a satellite remote sensing platform and projected to a WGS84 coordinate system through orthographic correction and orthographic image data obtained by an unmanned aerial vehicle to the same geographic coordinate reference frame; screening satellite remote sensing image data and unmanned aerial vehicle image data acquired in a certain time window before and after sampling time by taking specific time of field water sample acquisition as a reference; Integrating data cubes which are constructed from Sentinel-2 and Landsat-8/9 images and contain blue, green, red, near infrared and short wave infrared band combinations, and extracting reflectivity data of corresponding areas according to the sea area and peripheral area range of a target wind power plant to form a satellite remote sensing reflectivity data subset; Extracting a wave band of a high-resolution orthophoto image obtained by carrying a visible light-near infrared camera on the unmanned aerial vehicle to obtain reflectivity information corresponding to or similar to a satellite remote sensing wave band, and constructing an unmanned aerial vehicle remote sensing reflectivity data subset; and fusing the satellite remote sensing reflectivity data subset with the unmanned aerial vehicle remote sensing reflectivity data subset, and resampling and integrating data according to the spatial resolution difference of different data sources by adopting a pixel-based method to form a unified remote sensing reflectivity data set.
5. The method according to claim 4, wherein: the suspended sediment concentration data obtained by each sampling station through laboratory analysis and calculation are arranged according to the spatial position information of the sampling station; the method comprises the steps of giving space coordinate information to suspended sediment concentration data of each sampling station so that the suspended sediment concentration data and remote sensing reflectivity data can be spatially corresponding; Finding out a corresponding pixel or region in the unified remote sensing reflectivity data set according to the space coordinates of the sampling site, and correlating the actually measured suspended sediment concentration data with the remote sensing reflectivity data of the pixel or region; For satellite remote sensing data, according to the spatial resolution, determining the pixel or the average value of a plurality of pixels where the sampling site is located and the like for matching; and establishing a data fusion table or database, and storing the matched remote sensing reflectivity data and the actually measured suspended sediment concentration data to form a space-time matched integrated data set, wherein the data set comprises remote sensing reflectivity information and actually measured suspended sediment concentration values corresponding to each sampling point and corresponding space-time information.
6. A method according to claim 3, wherein the step of constructing an inversion model of the concentration of the floating sand comprises: Selecting all possible dual-band ratios, three-band combinations and normalized difference indexes from red, green, blue, near infrared and short wave infrared bands of the band combination data cube as candidate characteristic variables; Calculating the Pearson correlation coefficient of each characteristic variable and the actually measured suspended sediment concentration, and screening the wave band combination meeting the conditions as a modeling candidate variable; Performing variance expansion factor analysis on the screened wave band combination, and removing multiple collinearity variables of VIF > 10; and respectively constructing at least three mathematical models by using the screened characteristic variables: SSC=a.X+b, wherein X is a characteristic variable, and a and b are regression coefficients; An exponential regression model ssc=a·e (b·X) ; logistic regression model: ssb=a·ln (X) +b; or a support vector regression model based on machine learning, adopting a radial basis function; Dividing the actual measurement data set into a training set and a testing set, and evaluating the performance of the model by adopting a 5-fold cross validation method; taking the determined coefficient, the root mean square error and the average absolute error as evaluation indexes, and selecting a model with optimal comprehensive performance as a final inversion model; verifying the optimal model by using the independent verification data set, and executing the following correction steps when the verification set meets the preset condition: Introducing auxiliary environment parameters as secondary characteristic variables; Respectively constructing sub-models according to concentration intervals of actually measured suspended sediment concentration by adopting a sectional modeling strategy; And carrying out Gaussian process regression correction on the model residual error to generate a final composite model.
7. The method according to claim 1, wherein the steps of wake region identification and wake intensity calculation specifically comprise: a double-layer circular buffer zone is constructed by taking the center of a single fan as the center of a circle; Equally dividing the annular region into n=32 sector units according to azimuth angles; acquiring tide direction data at sampling time, and determining sector unit numbers corresponding to the main stream direction; taking a fan-shaped unit in the main flow direction as a center, and expanding k units to two sides to form candidate wake areas; selecting a sector unit with highest suspended sediment concentration in the candidate area as a core wake area, and taking 2 adjacent sector units as transition wake areas; in the vertical azimuth range of the main flow direction, 2 symmetrical fan-shaped units which are away from the center of the fan are selected as background reference areas; the wake intensity index Δssc is calculated as: ΔSSC = SSC w - (SSC b +α·σ b ) wherein SSC w is the average SSC value of the core wake region, SSC b is the average SSC value of the background region, sigma b is the standard deviation SSC of the background region, and alpha is an empirical coefficient.
8. The method according to claim 7, wherein: dividing the tide rise and fall working conditions according to the actual measurement data of the tide level station; under each working condition, the following data are collected respectively: The actual measurement value C obs of the suspended sediment concentration in wake areas of fans of different foundation structure types; wake intensity index Δssc of the corresponding region; for each working condition and basic structure type combination, establishing a dynamic mapping table of correction parameters gamma: Training a working condition-structure type identification model by adopting a support vector machine algorithm, wherein input parameters are a tide direction, a flow speed, a water depth and a basic type, and the input parameters are output as corresponding gamma values; embedding a working condition identification module in the monitoring system, and automatically matching the current working condition with the basic type; and calling a correction parameter library or an SVM model to obtain a real-time gamma value, and substituting the real-time gamma value into a correction formula to update the inversion result of the suspended sediment.
9. The method according to any one of claims 1 to 8, wherein the long-time-series suspended sediment concentration inversion is performed on images before and after wind farm construction by using a suspended sediment concentration inversion model, comprising the steps of: Performing geometric correction and radiation normalization processing on multi-period satellite remote sensing images before and after wind field construction, and eliminating sensor differences and atmospheric influences; screening space-time synchronous image data and actual measurement data based on a time difference threshold value between the field sampling time and the image acquisition time and a space matching threshold value between a space sampling point and an image pixel; Inputting the preprocessed images into a suspended sediment concentration inversion model, calculating suspended sediment concentration values of images in each period pixel by pixel, and generating long time sequence concentration data sets covering at least 1 year before construction and at least 3 years after construction of a wind field; trend analysis is carried out on the long time sequence data set, and a Mann-Kendall test method is adopted to identify a region with obvious concentration change; Setting a four-level concentration threshold according to ocean functional division water quality standard; Dividing a concentration exceeding area into a continuous deterioration area, a fluctuation exceeding area and a temporary exceeding area by combining a trend analysis result, and superposing to generate a dynamic risk level distribution map containing a time dimension; And (3) performing spatial interpolation on the missing data by adopting a Kriging interpolation method, and ensuring the spatial continuity of the risk level graph.
10. The method according to claim 9, wherein: Integrating underwater topography data and long-time-sequence suspended sediment concentration data acquired by a multi-beam sounding system to construct a three-dimensional water and sand environment database; Slope analysis is carried out on the topographic data, and a sediment migration active area is identified by combining the gradient change of the suspended sediment concentration; Based on the cable burial depth measured data and the sediment scouring rate, a cable exposure early warning model is established: Wherein, the Is the depth of the cable to be buried, Is the flushing rate; Combining the basic type and the local flow velocity, adopting CFD numerical simulation to correct a scouring depth prediction formula: Wherein, the As a coefficient of the basic type, For the actual flow rate, For a critical flush flow rate, In order to be able to run for a long period of time, Is an experience index; Superposing a dynamic risk level distribution diagram, a cable exposure risk diagram and a basic scouring risk diagram, and generating a comprehensive risk thermodynamic diagram by adopting GIS space superposition analysis; And extracting the area ratio of each risk level region, the time sequence of typical risk events and the risk evolution trend, and automatically generating a three-dimensional visual report containing space-time evolution characteristics, so as to support multi-scale dynamic display according to year/season/month.

Description

Marine wind power plant water Sha Yunwei monitoring method based on sky and land hydraulic engineering integration Technical Field The invention relates to the technical field of offshore wind farm environment monitoring, in particular to a sky-land-hydraulic-integration-based offshore wind farm water Sha Yunwei monitoring method, which aims to realize comprehensive, efficient and accurate monitoring of a water and sand environment of an offshore wind farm. Background Currently, marine wind power plant water and sand environment monitoring mainly depends on traditional methods and remote sensing technology. Traditional methods include in-situ sampling (water sampling using a water sampler and suspended sediment concentration analysis by laboratory gravimetric methods) and local hydrologic station monitoring. In recent years, the remote sensing technology is gradually applied to the inversion of suspended sediment in a large range, such as monitoring by using Landsat series satellite (Landsat-8/9) or Sentinel-2 satellite data, and existing offshore wind farm monitoring cases such as Shanghai Lingang, fujian Changle, jiangsu Dafeng and the like at home and abroad. The prior art has the following disadvantages: the traditional method has the defects of high field sampling cost, low frequency, limited space coverage, difficulty in realizing long-term dynamic monitoring and great influence of weather and sea conditions. The integration of the remote sensing technology is insufficient, the existing remote sensing monitoring is dependent on a single data source, and the existing remote sensing monitoring is lack of cooperation with field data and unmanned aerial vehicle monitoring, so that inversion accuracy is limited (for example, a model decision coefficient R 2 is only 0.6-0.7, and root mean square error is higher). The wind turbine wake monitoring is imperfect, the wake extraction method is simple, the tidal direction change and the wind turbine foundation type difference are not considered, disturbance evaluation is incomplete, and if the wind turbine foundation scour possibly causes a self-digging phenomenon, the risk of the wind turbine foundation scour is difficult to quantify in the prior art. The systematic risk evaluation is missing, a sky land hydraulic engineering integrated monitoring system is not formed, a water and sand change risk level diagram cannot be provided, and operation and maintenance decisions of a wind power plant are difficult to support. Disclosure of Invention The invention mainly aims to provide a sky-land-water-engineering-integrated-based marine wind power plant water Sha Yunwei monitoring method, which is based on the' sky-land-water-engineering-integrated monitoring concept, combines satellite remote sensing, field monitoring, unmanned aerial vehicle data and engineering parameters, and solves the following technical problems: 1. The method solves the problems of low efficiency and high cost of the traditional monitoring method, and realizes the monitoring of the water and sand environment of the offshore wind farm in a large scale at high frequency and low cost. 2. And the inversion precision of suspended sediment is improved, and errors are reduced through multi-source data fusion. 3. And 3, accurately monitoring the wake effect of the fan, quantifying the influence of different foundation types and tidal conditions on water and sand disturbance, and early warning the foundation scouring risk. 4. And constructing an integrated monitoring system, generating a water-sand change risk level diagram, and providing a scientific basis for the safe operation and maintenance of the wind power plant. The invention realizes the above purpose through the following technical scheme: an offshore wind farm water Sha Yunwei monitoring method based on sky and land hydraulic integration, comprising: the multi-source data acquisition and preprocessing steps are as follows: Acquiring multispectral image data of a target sea area through a satellite remote sensing platform, screening images with cloud coverage rate lower than a preset threshold value, and performing radiation correction and cloud removal processing; Synchronously carrying out field water sample acquisition and unmanned aerial vehicle high-resolution image acquisition in a wind power field area, measuring actual measurement values of suspended sediment concentration and ensuring the space-time consistency of images; Constructing a space-time matching integrated data set, and fusing remote sensing reflectivity data with actual measurement suspended sediment concentration data; the construction method of the suspended sediment inversion model comprises the following steps: Based on correlation analysis of multispectral wave band combinations and actually measured suspended sediment concentration data, screening wave band combinations meeting preset correlation thresholds; Respectively constructing at least two types of mathematical