A method and system for determining wind resistance parameters of a marine engineering project based on multi-source satellite data
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANDONG PROVINCIAL CLIMATE CENT
- Filing Date
- 2026-07-06
- Publication Date
- 2026-08-04
AI Technical Summary
[0006]本发明的目的在于，针对上述现有技术难以针对单个海上目标工程点位构建专属台风风速样本，极端风速计算结果精度不足的缺陷，提供设计一种基于多源卫星数据的海上工程台风抗风参数确定方法及系统，以解决上述技术问题
极值统计与参数计算模块，根据台风大风过程样本集的数据特征，从预设统计方法集合中选择对应方法进行拟合，计算得到目标工程点位在预设重现期下的极端风速，作为台风抗风设计参数。
Smart Images
Figure CN122508380A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of data processing and marine engineering technology, specifically relating to a method and system for determining typhoon resistance parameters for marine engineering based on multi-source satellite data. Background Technology
[0002] In existing technologies, the calculation of typhoon resistance parameters for offshore engineering typically relies on various types of sea surface wind data, such as typhoon wind fields, ship observations, and satellite remote sensing, to construct a global grid of annual maximum wind speed sequences. This is then combined with extreme value distribution models to estimate the spatial distribution of extreme wind speeds at different return periods, thus providing regional climate background information for the wind-resistant design of offshore engineering projects. However, this spatially mapping-oriented technical approach suffers from insufficient adaptability when directly serving the targeted design of specific offshore engineering sites.
[0003] In practical applications, wind-resistant design for offshore engineering focuses on the extreme wind speeds at specific locations under the influence of typhoon events. Existing technologies generate spatial distribution maps that require secondary extraction or interpolation before they can be used for engineering points. Their sample construction method, based on the annual maximum wind speed of a grid, fails to reflect the dynamic correlation between typhoon paths and specific target engineering points. For near-shore areas such as the Yellow and Bohai Seas, although this method alleviates the data sparsity problem through multi-source fusion, the generated samples are not typhoon process samples specific to that location. This results in design parameters calculated based on these samples lacking specificity and accuracy in representing the actual extreme typhoon wind risks faced by that location.
[0004] This demonstrates that existing technologies have limitations in constructing specific typhoon wind speed samples for individual offshore engineering sites, resulting in insufficient accuracy in extreme wind speed calculations. This is a shortcoming of existing technologies.
[0005] In view of this, it is very necessary to provide a method and system for determining typhoon resistance parameters for marine engineering based on multi-source satellite data to solve the above-mentioned defects in the prior art. Summary of the Invention
[0006] The purpose of this invention is to address the shortcomings of existing technologies, such as the inability to construct dedicated typhoon wind speed samples for individual offshore engineering sites and insufficient accuracy in extreme wind speed calculations. This invention provides a method and system for determining typhoon resistance parameters for offshore engineering projects based on multi-source satellite data, thereby solving the aforementioned technical problems.
[0007] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, this application provides a method for determining typhoon resistance parameters for marine engineering based on multi-source satellite data, comprising the following steps: Step S1: Obtain multi-source satellite sea surface wind inversion data and marine fixed-point observation data, perform error correction on the multi-source satellite sea surface wind inversion data to obtain corrected sea surface wind data, and generate a gridded sea surface wind dataset for the target sea area based on the corrected sea surface wind data. Step S2: Centered on the target engineering point, screen out individual typhoon cases that affect the target engineering point, and extract the wind speed data corresponding to each typhoon case from the gridded sea surface wind dataset to construct a sample set of typhoon gale processes for the target engineering point. Step S3: Based on the data characteristics of the typhoon wind process sample set, select the corresponding method from the preset statistical method set for fitting, calculate the extreme wind speed of the target engineering point under the preset return period, and use it as the typhoon wind resistance design parameter.
[0008] Preferably, in step S1, the error correction includes: A time-space matching window is defined with the geographical location of each fixed-point marine observation station as the center; Multi-source satellite sea surface wind inversion data are collected within the spatiotemporal matching window. Each set of multi-source satellite sea surface wind inversion data and the corresponding marine fixed-point observation data form a matching data pair. All matching data pairs within the target sea area are aggregated to construct a training sample set for the error correction model; The preset error correction model is trained using a training sample set, and the trained error correction model is used to correct the multi-source satellite sea surface wind inversion data to obtain the corrected sea surface wind data.
[0009] By adopting the above technical solution, by defining a time-space matching window centered on a fixed marine observation station, constructing matching data pairs and training an error correction model, it is possible to systematically correct satellite inversion data using measured data, reduce systematic bias and random errors in satellite sea surface wind inversion data, and improve the accuracy and reliability of sea surface wind data.
[0010] Preferably, in step S1, the error correction model employs the gradient boosting decision tree algorithm; the gradient boosting decision tree algorithm is an additive model constructed based on multiple CART regression trees, and its mapping function expression is:
[0011] in, To correct the sea surface wind data, input feature vector The wind speeds retrieved from the sea surface via multi-source satellites are shown in sequence. Satellite incident angle Significant wave height Sea surface temperature Marine environmental parameters ; The initial constant value is the mean of the fixed-point marine observation data in the training sample set; This is the output of the m-th regression tree; is the weight coefficient corresponding to the m-th regression tree; M is the total number of regression trees.
[0012] By adopting the above technical solution and constructing an error correction model using the gradient boosting decision tree algorithm, it is able to capture the nonlinear and complex relationship between multi-dimensional features such as satellite-retrieved wind speed, satellite incidence angle, significant wave height, sea surface temperature, and marine environmental parameters and the actual wind speed. Compared with traditional linear regression or single decision tree methods, it has higher fitting accuracy and generalization ability. At the same time, through the superposition learning of multiple regression trees, it can automatically handle the interaction between features, further improving the accuracy of the corrected sea surface wind data.
[0013] Preferably, in step S1, generating the gridded sea surface wind dataset for the target sea area specifically includes: Establish a regular latitude and longitude grid within the target sea area, dividing the target sea area into several grid units, with the center point of each grid unit serving as the grid point. A preset spatial range is defined with each grid point as the center. Corrected sea surface wind data at the same time within the spatial range are extracted, and the maximum wind speed is selected as the representative wind speed of that grid point. Traverse all grid points and time points to generate a spatiotemporally aligned gridded sea surface wind dataset.
[0014] By adopting the above technical solution, by establishing a regular latitude and longitude grid and extracting the maximum wind speed value within the spatial range with the grid point as the center, discrete satellite observation data can be transformed into spatiotemporally continuous and spatially uniform gridded data, eliminating the spatial discontinuity caused by satellite orbit gaps and data sparsity. At the same time, by selecting the maximum wind speed within the spatial range as a representative value, the extreme wind speed characteristics during the passage of typhoons are effectively preserved, avoiding the underestimation of strong wind risk due to spatial averaging.
[0015] Preferably, in step S1, if a grid point does not have valid data within a preset spatial range at the corresponding time, the value of that grid point at that time is marked as a default value.
[0016] By adopting the above technical solution, grid points without valid data are marked with default values, which can clearly identify data blank areas and avoid the use of invalid or unreliable interpolated data in subsequent statistical analysis, thus preventing the results from being distorted.
[0017] Preferably, in step S2, the screening of typhoon cases affecting the target engineering site specifically includes: Delineate the typhoon-affected area centered on the target project site; Iterate through all historical typhoon instances in the typhoon best path dataset. If the center of a typhoon falls within the typhoon's influence area at any given time, then the typhoon is determined to be a valid typhoon affecting the target engineering site and is added to the candidate typhoon list.
[0018] By adopting the above technical solution, the typhoon-affected area is delineated with the target project site as the center, and valid typhoon cases are screened based on whether the typhoon center falls into the area. This can identify typhoon events that have a real impact on the target project site and exclude storms that are far away or have no direct impact.
[0019] Preferably, step S2, which involves constructing a sample set of typhoon wind processes for the target engineering site, specifically includes: For each typhoon case in the candidate typhoon list, the complete period of impact of the typhoon on the target engineering site is determined based on the time when the typhoon enters and leaves the typhoon's influence area; In the gridded sea surface wind dataset, locate the grid point that is spatially closest to the target engineering point, extract all wind speeds of the grid point during the complete impact period, and obtain the wind speed sequence corresponding to the typhoon. By merging the wind speed sequences corresponding to all typhoons, a sample set of typhoon gale processes for the target engineering site is obtained.
[0020] By adopting the above technical solution, by extracting all wind speed data of the grid point that is spatially closest to the target project site during the entire period of typhoon impact, it is possible to reflect the actual wind speed change process of the target project site during the typhoon's passage, thus avoiding errors introduced by spatial interpolation or distant grid points.
[0021] Preferably, in step S3, selecting a corresponding method from a preset set of statistical methods for fitting specifically includes: The data characteristics of the typhoon gale process sample set are evaluated, including the number of independent typhoon events, the span of years covered by the sample, and the continuity of the annual typhoon observation sequence. Based on the data feature evaluation results, the optimal model is adaptively matched from the preset multi-class extreme value statistical models to complete the sample data fitting.
[0022] By adopting the above technical solution, the data characteristics of the sample set of typhoon wind process are evaluated, the optimal extreme value statistical model is matched, and the statistical strategy is flexibly adjusted according to the actual situation of different sea areas and different data conditions, so as to avoid fitting bias caused by using a single model.
[0023] Preferably, the preset multi-type extreme value statistical models include extreme value type I distribution, Pearson type III distribution, and Poisson-Gumbel distribution; When the number of independent typhoon events in the sample set is not less than the preset threshold and there are no gaps in the annual data of the time series, the extreme value type I distribution is selected. When the number of independent typhoon events is not less than the preset threshold, but the extreme wind speeds show a skewed distribution, the Pearson Type III distribution is selected. When the number of independent typhoon events is less than the preset threshold, or when there are gaps in annual data, the Poisson-Günbel distribution is selected.
[0024] Using the above technical solution, clear model selection rules are formulated for different data conditions of the sample set: when the sample is sufficient and there are no gaps, the extreme value type I distribution is selected to ensure statistical stability; when the wind speed is skewed, the Pearson type III distribution is selected to better characterize the tail features; when the sample is insufficient or there are gaps, the Poisson-Gumbel distribution is selected to make full use of the limited data information. This hierarchical classification strategy can obtain reliable extreme wind speed estimates under different data quality conditions.
[0025] Secondly, this application provides a system for determining typhoon resistance parameters for marine engineering based on multi-source satellite data, comprising: The wind field correction and gridding module acquires multi-source satellite sea surface wind inversion data and marine fixed-point observation data, performs error correction on the multi-source satellite sea surface wind inversion data to obtain corrected sea surface wind data, and generates a gridded sea surface wind dataset for the target sea area based on the corrected sea surface wind data. The typhoon screening and sample construction module, centered on the target engineering site, screens individual typhoon cases that affect the target engineering site, and extracts the wind speed data corresponding to each typhoon case from the gridded sea surface wind dataset to construct a sample set of typhoon gale processes for the target engineering site. The extreme value statistics and parameter calculation module selects the corresponding method from the preset statistical method set for fitting based on the data characteristics of the typhoon wind process sample set, and calculates the extreme wind speed of the target engineering point under the preset return period, which is used as the typhoon wind resistance design parameter.
[0026] The beneficial effects of this invention are as follows: it separately screens historical typhoons for designated offshore target engineering sites and constructs a dedicated typhoon wind process sample set, solving the problems of weak correlation between general samples and target engineering sites in existing technologies and the inability to reflect the dynamic impact of typhoons; it completes wind field error correction based on multi-source satellite data combined with machine learning algorithms, and uses neighborhood extreme value sampling to generate gridded wind field data, directly extracting typhoon wind speed sequences from nearby grids, overcoming the defects of sparse nearshore sea area data and inaccurate extreme wind characterization; it adaptively matches extreme value statistical models according to the number of typhoon samples, temporal completeness, and distribution characteristics, avoiding the problems of poor adaptability of single extreme value models and easy statistical distortion of small nearshore samples, thus enabling stable output of extreme wind speed parameters adapted to single-point engineering projects.
[0027] Furthermore, the design principle of this invention is reliable, the structure is simple, and it has a very wide range of application prospects.
[0028] Therefore, it is evident that the present invention has substantial features and progress compared with the prior art, and the beneficial effects of its implementation are also obvious. Attached Figure Description
[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0030] Figure 1 This is a flowchart of a method for determining typhoon resistance parameters in marine engineering based on multi-source satellite data.
[0031] Figure 2 This is a schematic diagram of a system for determining typhoon resistance parameters in marine engineering based on multi-source satellite data.
[0032] Among them, 1-wind field correction and gridding module, 2-typhoon screening and sample construction module, 3-extreme value statistics and parameter calculation module. Detailed Implementation
[0033] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The following embodiments are explanations of the present invention, but the present invention is not limited to the following implementation methods.
[0034] Example 1: This embodiment provides a method for determining typhoon resistance parameters for marine engineering based on multi-source satellite data, such as... Figure 1 As shown, it includes the following steps: Step S1: Obtain multi-source satellite sea surface wind inversion data and marine fixed-point observation data, perform error correction on the multi-source satellite sea surface wind inversion data to obtain corrected sea surface wind data, and generate a gridded sea surface wind dataset for the target sea area based on the corrected sea surface wind data. Step S2: Centered on the target engineering point, screen out individual typhoon cases that affect the target engineering point, and extract the wind speed data corresponding to each typhoon case from the gridded sea surface wind dataset to construct a sample set of typhoon gale processes for the target engineering point. Step S3: Based on the data characteristics of the typhoon wind process sample set, select the corresponding method from the preset statistical method set for fitting, and calculate the extreme wind speed at the target engineering point under the preset return period, which is used as the typhoon wind resistance design parameter. Step S1: Use high-reliability marine fixed-point observation data to perform error correction on multi-source satellite sea surface wind inversion data, laying the data foundation for generating a high-quality gridded wind field dataset.
[0035] In this embodiment of the application, the multi-source satellite sea surface wind inversion data comes from multiple satellite sensors in orbit, including QuikSCAT, MetOp series, Fengyun series and CSCAT; these satellite sensors invert the wind speed at a height of 10 meters above the sea surface through active microwave remote sensing technology and passive microwave remote sensing technology. The marine fixed-point observation data were collected from marine buoys, oil platform meteorological stations, offshore wind measurement towers, island meteorological stations, coastal meteorological stations, and shore-based laser wind measurement radars deployed in the target sea area and adjacent areas. All marine fixed-point observation data were uniformly processed into wind speeds at a height of 10 meters above the sea surface, which served as the reference true values used to correct the multi-source satellite sea surface wind inversion data.
[0036] To correct data errors, spatiotemporal matching is performed between multi-source satellite sea surface wind inversion data and fixed-point marine observation data. For each fixed-point marine observation station, a spatiotemporal matching window is defined centered on its geographical location. In this embodiment, the spatial range of the spatiotemporal matching window is [missing information]. Latitude and longitude coordinates, time range: Within this spatiotemporal matching window, multi-source satellite sea surface wind inversion data points are collected. Each set of multi-source satellite sea surface wind inversion data points and the corresponding fixed-point marine observation data at that time together form a set of matching data pairs.
[0037] All matching data pairs from multiple stations within the target sea area over a long time series are aggregated to construct a training sample set for the error correction model. In this embodiment, a gradient boosting decision tree is used to construct the error correction model. The inputs to the error correction model are multi-source satellite sea surface wind inversion wind speed, satellite incidence angle, significant wave height, sea surface temperature, and marine environmental parameters. The output of the error correction model is the corrected sea surface wind speed. The input-output relationship is expressed as follows:
[0038] in, The wind speed is retrieved from the sea surface wind using multi-source satellites; X represents optional auxiliary features, including satellite incidence angle, significant wave height, sea surface temperature, and marine environmental parameters. For gradient boosting decision tree mapping function; To correct the sea surface wind data.
[0039] The gradient boosting decision tree is an additive model built based on multiple CART regression trees, and its mapping function expression is:
[0040] Among them, the input feature vector The wind speeds retrieved from the sea surface via multi-source satellites are shown in sequence. Satellite incident angle Significant wave height Sea surface temperature Marine environmental parameters ; The initial constant value is the mean of the fixed-point marine observation data in the training sample set; This is the output of the m-th regression tree; is the weight coefficient corresponding to the m-th regression tree; M is the total number of regression trees.
[0041] During model training, each iteration uses the current residual, i.e. the negative gradient direction, as the fitting target of the new regression tree. The optimization objective is to minimize the mean square error between the corrected sea surface wind data and the wind speed observed at fixed points at sea. The regression tree structure and weight coefficients are updated step by step to obtain a stable nonlinear error correction mapping relationship. In a preferred embodiment, the total number of regression trees M is set to a range of 100-500, the maximum depth of the regression trees is set to a range of 3-8, the learning rate is set to a range of 0.01-0.1, and a subsampling ratio of 0.7-0.9 is used to suppress model overfitting.
[0042] After the model training is completed, the mean deviation and root mean square error are used as quantitative evaluation indicators to comprehensively test the overall correction effect of the error correction model. The expression for the average deviation is:
[0043] The expression for the root mean square error is:
[0044] Where N is the total number of matched data pairs; Here is the corrected wind speed corresponding to the i-th sample group; This represents the fixed-point marine observation data corresponding to the i-th sample group.
[0045] During model training, model parameters are continuously optimized to ensure that the corrected sea surface wind data and the fixed-point observation data at sea maintain consistent statistical characteristics, minimize the mean deviation and root mean square error, and ensure that the model's correction effect is stable across the entire wind speed range. After model training is completed, batch and automated systematic error correction can be performed on all historical multi-source satellite sea surface wind inversion data and future real-time multi-source satellite sea surface wind inversion data within the target sea area.
[0046] The obtained corrected sea surface wind data is transformed into a gridded sea surface wind dataset through spatial reconstruction and integration.
[0047] The corrected sea surface wind data consists of a large number of discrete sea surface wind data points with specific geographical locations and times; In this embodiment, the target sea area is the designated sea area for calculating typhoon resistance parameters; a regular latitude and longitude grid system is established within the target sea area, and in this embodiment, the grid spatial resolution is set to... This means that the spacing between adjacent grid points in both longitude and latitude directions is 0.25 degrees. The entire target sea area is divided into... There are 1 grid cell, and the center point of each grid cell is defined as a grid point. , where j is the grid point index; The expression for a single sea surface wind data point is:
[0048] in, , These are the longitude and latitude of the k-th sea surface wind data point, respectively; This represents the corrected sea surface wind speed corresponding to the kth sea surface wind data point. This represents the observation time corresponding to the sea surface wind data points.
[0049] This embodiment employs a neighborhood maximum sampling algorithm to assign wind speed values from sea surface wind data points to regular grid points; for any given grid point... The system defines a preset spatial range centered on the grid point's geographical location, extracts all discrete data at the same time within the range, and selects the maximum wind speed as the representative wind speed for that grid point.
[0050] grid points coordinates Define its neighborhood space as the center. For a rectangular region, the expression is as follows:
[0051] in, The coordinates of the point to be judged; The half-width of the space search; in this embodiment, we take... That is, a single search area is a square with a side length of 0.5 degrees of latitude and longitude.
[0052] For a given time t, the grid points The corresponding formula for calculating gridded wind speed is:
[0053] in, For grid points The gridded wind speed value at time t; This represents the corrected sea surface wind speed corresponding to the kth sea surface wind data point. The coordinates of the k-th sea surface wind data point; The observation time corresponds to the kth sea surface wind data point; For grid points The neighborhood space range; t is the target time.
[0054] This formula represents: filtering spatial range All sea surface wind data points that are within the target time t The maximum value of the corrected sea surface wind speed is taken as the current grid point. The wind speed at time t.
[0055] If a grid point has no valid discrete data within the preset spatial range at the corresponding time, then the value of that grid point at that time is marked as the default value.
[0056] Iterate through all grid points within the target sea area in sequence. By sampling and calculating the maximum value of the neighborhood at each time point t, a spatiotemporally aligned gridded sea surface wind dataset is generated. This dataset is a three-dimensional data field containing longitude, latitude, and time, with each grid cell corresponding to a specific wind speed at each time point.
[0057] After this step, the gridded sea surface wind dataset inherits the accuracy of the corrected sea surface wind data, and at the same time, it is adapted to extreme wind field analysis scenarios in the sea area by relying on grid regularization and extreme value preservation strategies.
[0058] Step S2 involves filtering and extracting all historical typhoon wind events that have affected the location from the gridded sea surface wind dataset to construct a wind speed sample set for the target engineering location at sea.
[0059] In this embodiment of the application, the target engineering location refers to the specific geographical location of offshore wind power, offshore platforms, cross-sea bridges, and other offshore engineering projects where wind resistance parameter design is carried out, and its coordinates are marked as follows: .
[0060] The optimal track dataset of tropical cyclones in the Northwest Pacific, compiled by an authoritative meteorological agency, was used as the optimal track data for typhoons. It includes the number, name, hourly center location, minimum central pressure, and maximum wind speed near the center of all tropical cyclones over the years, with a data recording interval of 6 hours.
[0061] With the target project location Delineate a circular area of typhoon impact centered on the typhoon. The mathematical expression is:
[0062] in, Here, R is the function for calculating the great circle distance between two points; R is the preset radius of influence; in this embodiment, considering engineering requirements and typhoon scale, we take... ; Iterate through all historical typhoon cases in the optimal typhoon path dataset, and for typhoon number m, check its center position at each time point. If the center position of this typhoon falls within the affected area at any time point... If the typhoon falls within the specified range, it is determined to be a valid typhoon affecting the target project site and is added to the candidate typhoon list. .
[0063] Regarding the candidate typhoon list Each typhoon case within the scope is analyzed based on its entry into and departure from the affected area. Based on the time of the typhoon, extend 12 hours before and after it to determine the complete period of impact on the target engineering site, denoted as _____. , where m is the typhoon index; In the gridded sea surface wind dataset, locate the target engineering points. The closest grid point in space Extract grid points During the period of typhoon impact The total wind speeds within the area are used to obtain a wind speed sequence:
[0064] By merging the wind speed sequences corresponding to all typhoons in the candidate typhoon list, a sample set of typhoon gale processes for the target engineering site is obtained. The expression is:
[0065] Where ∪ is the union operator; This is a list of candidate typhoons.
[0066] This sample set consists of historical typhoon wind observation data for all typhoons that have entered the influence range of the target project site. Compared with traditional annual maximum wind speed sequences that do not distinguish between weather systems, this sample set directly corresponds to typhoon disaster scenarios, with clear physical mechanisms. It highly matches the extreme typhoon wind risks actually faced by the target project site, and serves as a reliable input for subsequent extreme value statistics and calculation of typhoon wind resistance design parameters.
[0067] At this point, step S2 completes the conversion from a global general grid wind field to a single-point specific typhoon wind speed sample.
[0068] Step S3 uses the data characteristics of the typhoon wind process sample set to adaptively select the optimal extreme value statistical model to carry out fitting analysis, thereby estimating the extreme wind speed value of the target engineering point under the preset return period.
[0069] In this embodiment of the application, the typhoon gale process sample set It includes complete wind speed records corresponding to multiple historical typhoon events; the data characteristics of the typhoon gale process sample set include the total number of valid samples, the number of independent typhoon events, the distribution characteristics of peak wind speeds for each typhoon process, and the completeness of time series data; these data characteristics directly determine the fitting effect and estimation accuracy of the extreme value statistical model, so it is necessary to conduct a systematic feature evaluation of the typhoon gale process sample set.
[0070] Sample set of typhoon wind processes The main evaluation dimensions for data feature assessment include: the total number of valid wind speed records in the sample set, the number of independent typhoon events covered by the sample, the year span covered by the sample data, and the continuity of the annual typhoon observation sequence; and determining whether there are any gaps in annual data due to missing satellite data, thereby judging the data quality and statistical applicability of the sample set.
[0071] Based on the above data feature evaluation results, the optimal model is adaptively matched from the preset multi-class extreme value statistical models to complete the sample data fitting; the preset extreme value statistical models in this embodiment include extreme value type I distribution, Pearson type III distribution and Poisson-Günber distribution, which are adapted to the data features of different typhoon samples respectively. Among them, the extreme value type I distribution is suitable for scenarios with sufficient sample size, good temporal continuity, and approximately symmetrical annual extreme value distribution; the Pearson type III distribution has excellent characterization ability for hydrological and meteorological skewed data and is suitable for wind speed samples with obvious skewness characteristics; the Poisson-Günber distribution is a composite extreme value model that is suitable for scenarios where the frequency of typhoon occurrence follows a Poisson distribution and the extreme value of a single typhoon follows a Günber distribution. It is especially suitable for sparse sample scenarios with few typhoon samples, uneven annual typhoon activity frequency, and time gaps, and can maximize the mining of effective information from limited typhoon samples.
[0072] The adaptive model selection rules are set based on data characteristics, specifically as follows: When the number of independent typhoon events in the sample set is not less than the first preset threshold, and the time series is continuous and complete with no annual data gaps, the extreme value type I distribution is preferred for fitting; in this embodiment, the first preset threshold is 15 times. When the total sample size is sufficient and the number of independent typhoon events is not less than the first preset threshold, but the extreme wind speed values show a skewed distribution, the Pearson Type III distribution is selected; the criterion for judging the skewed distribution characteristics is: calculate the skewness coefficient of the sample wind speed, and if the absolute value of the skewness coefficient is not less than 0.3, it is judged as skewed. When the number of independent typhoon events is less than the first preset threshold, or when there are gaps in annual data and a complete annual extreme value sequence cannot be constructed, the Poisson-Günbel distribution is selected; the criteria for judging gaps in annual data are: there are two or more consecutive years without valid typhoon records in the sample coverage years, or the proportion of valid record years to the total number of years covered by the sample is less than 60%.
[0073] After determining the optimal extreme value statistical model, the sample set of typhoon wind processes was analyzed. Perform model parameter estimation and overall fitting; If the Type I extreme value distribution is chosen, the method of moments is used to solve for the model parameters; the cumulative distribution function of the Type I extreme value distribution is:
[0074] in, For position parameters; For scale parameters; Wind speed not greater than The cumulative probability.
[0075] Using the method of moments, position parameters and scale parameters The calculation formula is:
[0076]
[0077] in, This represents the mean wind speed of the sample. denoted as the standard deviation of the sample wind speed.
[0078] If the Pearson Type III distribution is chosen, the method of moments is used to solve for the model parameters; the probability density function of the Pearson Type III distribution is:
[0079] in, For shape parameters, For scale parameters, For position parameters, It is a gamma function; Using the method of moments, the formulas for calculating the three parameters are as follows:
[0080] in, This represents the mean wind speed of the sample. The coefficient of variation for the sample wind speed. The skewness coefficient is the sample wind speed.
[0081] If the Poisson-Günbel distribution is selected, a hierarchical modeling method is used; the number of typhoon impacts each year within the sample coverage period is statistically analyzed, and the Poisson distribution parameters are fitted. , The value is taken as the annual average number of typhoon impacts; the peak wind speed of each typhoon event is extracted to form an annual extreme value sample sequence, and the location parameters of the Gumbel distribution are fitted using the method of moments estimation. and scale parameters The calculation formula is the same as the moment estimation formula for the extreme value type I distribution; the Poisson-Gumbel distribution corresponds to the return period. The formula for calculating extreme wind speed is:
[0082] in, For the location parameters of the Gumbel distribution, The scaling parameter of the Gumbel distribution. This represents the average number of typhoons per year. For wind speed recurrence interval, To correspond to extreme wind speeds during the recurrence period.
[0083] Based on the optimal model parameters after fitting convergence, the extreme wind speed values at the target engineering location under each preset return period are calculated and denoted as follows. ,in The preset wind speed return period is used; for example, the extreme wind speed results corresponding to different commonly used return periods for engineering projects of 30 years, 50 years, and 100 years can be calculated in batches.
[0084] The extreme wind speed values for each return period obtained from the fitting calculation The system identifies and outputs the typhoon-resistant design parameters corresponding to the target offshore engineering site. These parameters can be directly applied to the structural wind resistance design, wind disaster risk assessment, and climate feasibility study of offshore engineering projects such as offshore wind power, offshore platforms, and cross-sea bridges.
[0085] Thus, step S3 completes the entire chain of intelligent calculations, from the target engineering site's specific typhoon sample set to high-precision wind-resistant design parameters.
[0086] Example 2: This embodiment provides a system for determining typhoon resistance parameters for marine engineering based on multi-source satellite data, such as... Figure 2 As shown, it includes: Wind field correction and gridding module 1 acquires multi-source satellite sea surface wind inversion data and marine fixed-point observation data, performs error correction on the multi-source satellite sea surface wind inversion data to obtain corrected sea surface wind data, and generates a gridded sea surface wind dataset for the target sea area based on the corrected sea surface wind data. Typhoon screening and sample construction module 2, with the target engineering point as the center, screens individual typhoon cases that affect the target engineering point, and extracts the wind speed data corresponding to each typhoon case from the gridded sea surface wind dataset to construct a sample set of typhoon gale processes at the target engineering point; Extreme value statistics and parameter calculation module 3, based on the data characteristics of the typhoon wind process sample set, selects the corresponding method from the preset statistical method set for fitting, calculates the extreme wind speed of the target engineering point under the preset return period, and uses it as the typhoon wind resistance design parameter.
[0087] The above-disclosed embodiments are merely preferred embodiments of the present invention, but the present invention is not limited thereto. Any non-creative variations that can be conceived by those skilled in the art, as well as any improvements and modifications made without departing from the principles of the present invention, should fall within the protection scope of the present invention.
Claims
1. A method for determining typhoon resistance parameters for marine engineering based on multi-source satellite data, characterized in that, Includes the following steps: Step S1: Obtain multi-source satellite sea surface wind inversion data and marine fixed-point observation data, perform error correction on the multi-source satellite sea surface wind inversion data to obtain corrected sea surface wind data, and generate a gridded sea surface wind dataset for the target sea area based on the corrected sea surface wind data. Step S2: Centered on the target engineering point, screen out individual typhoon cases that affect the target engineering point, and extract the wind speed data corresponding to each typhoon case from the gridded sea surface wind dataset to construct a sample set of typhoon gale processes for the target engineering point. Step S3: Based on the data characteristics of the typhoon wind process sample set, select the corresponding method from the preset statistical method set for fitting, calculate the extreme wind speed of the target engineering point under the preset return period, and use it as the typhoon wind resistance design parameter.
2. The method according to claim 1, characterized in that, In step S1, the error correction includes: A time-space matching window is defined with the geographical location of each fixed-point marine observation station as the center; Multi-source satellite sea surface wind inversion data are collected within the spatiotemporal matching window. Each set of multi-source satellite sea surface wind inversion data and the corresponding marine fixed-point observation data form a matching data pair. All matching data pairs within the target sea area are aggregated to construct a training sample set for the error correction model; The preset error correction model is trained using a training sample set, and the trained error correction model is used to correct the multi-source satellite sea surface wind inversion data to obtain the corrected sea surface wind data.
3. The method according to claim 2, characterized in that, In step S1, the error correction model employs the gradient boosting decision tree algorithm; the gradient boosting decision tree algorithm is an additive model constructed based on multiple CART regression trees, and its mapping function expression is: in, To correct the sea surface wind data, input feature vector The wind speeds retrieved from the sea surface via multi-source satellites are shown in sequence. Satellite incident angle Significant wave height Sea surface temperature Marine environmental parameters ; The initial constant value is the mean of the fixed-point marine observation data in the training sample set; This is the output of the m-th regression tree; is the weight coefficient corresponding to the m-th regression tree; M is the total number of regression trees.
4. The method according to claim 1, characterized in that, In step S1, generating a gridded sea surface wind dataset for the target sea area specifically includes: Establish a regular latitude and longitude grid within the target sea area, dividing the target sea area into several grid units, with the center point of each grid unit serving as the grid point. A preset spatial range is defined with each grid point as the center. Corrected sea surface wind data at the same time within the spatial range are extracted, and the maximum wind speed is selected as the representative wind speed of that grid point. Traverse all grid points and time points to generate a spatiotemporally aligned gridded sea surface wind dataset.
5. The method according to claim 4, characterized in that, In step S1, if a grid point does not have valid data within the preset spatial range at the corresponding time, the value of the grid point at the corresponding time is marked as the default value.
6. The method according to claim 1, characterized in that, In step S2, individual typhoon cases affecting the target project site are selected, specifically including: Delineate the typhoon-affected area centered on the target project site; Iterate through all historical typhoon instances in the typhoon best path dataset. If the center of a typhoon falls within the typhoon's influence area at any given time, the typhoon is determined to be a valid typhoon affecting the target engineering site and is added to the candidate typhoon list.
7. The method according to claim 6, characterized in that, In step S2, constructing the typhoon wind process sample set for the target engineering site specifically includes: For each typhoon case in the candidate typhoon list, the complete period of typhoon impact on the target engineering site is determined based on the time when the typhoon enters and leaves the typhoon's impact area; In the gridded sea surface wind dataset, locate the grid point that is spatially closest to the target engineering point, extract all wind speeds of the grid point during the complete impact period, and obtain the wind speed sequence corresponding to the typhoon. By merging the wind speed sequences corresponding to all typhoons, a sample set of typhoon gale processes for the target engineering site is obtained.
8. The method according to claim 1, characterized in that, In step S3, a corresponding method is selected from the preset set of statistical methods for fitting, specifically including: The data characteristics of the typhoon gale process sample set are evaluated, including the number of independent typhoon events, the span of years covered by the sample, and the continuity of the annual typhoon observation sequence. Based on the data feature evaluation results, the optimal model is adaptively matched from the preset multi-class extreme value statistical models to complete the sample data fitting.
9. The method according to claim 8, characterized in that, The preset multi-type extreme value statistical models include the extreme value type I distribution, the Pearson type III distribution, and the Poisson-Gumbel distribution; When the number of independent typhoon events in the sample set is not less than the preset threshold and there are no gaps in the annual data of the time series, the extreme value type I distribution is selected. When the number of independent typhoon events is not less than the preset threshold, but the extreme wind speeds show a skewed distribution, the Pearson Type III distribution is selected. When the number of independent typhoon events is less than the preset threshold, or when there are gaps in annual data, the Poisson-Günbel distribution is selected.
10. A system for determining typhoon resistance parameters for marine engineering based on multi-source satellite data, characterized in that, include: The wind field correction and gridding module acquires multi-source satellite sea surface wind inversion data and marine fixed-point observation data, performs error correction on the multi-source satellite sea surface wind inversion data to obtain corrected sea surface wind data, and generates a gridded sea surface wind dataset for the target sea area based on the corrected sea surface wind data. The typhoon screening and sample construction module takes the target engineering site as the center, filters individual typhoon cases that affect the target engineering site based on the best typhoon path data, and extracts the wind speed data corresponding to each typhoon case from the gridded sea surface wind dataset to construct a sample set of typhoon gale processes for the target engineering site. The extreme value statistics and parameter calculation module selects the corresponding method from the preset statistical method set for fitting based on the data characteristics of the typhoon wind process sample set, and calculates the extreme wind speed of the target engineering point under the preset return period, which is used as the typhoon wind resistance design parameter.