Offshore wind turbine foundation weight estimation method and system

By constructing a sample library mapping multidimensional engineering parameters to foundation weight and using weighted multidimensional distance interpolation, a fast and high-precision estimation of offshore wind turbine foundation weight was achieved, solving the problems of slow speed and insufficient accuracy in traditional methods and improving the economic optimization capability of offshore wind farms.

CN123116403APending Publication Date: 2026-11-17HUANENG CLEAN ENERGY RES INST +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611320161.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In the early design phase of offshore wind farms, traditional methods for estimating the weight of offshore wind turbine foundations are slow and inaccurate, leading to distorted economic evaluations and an inability to be effectively integrated into the global optimization loop. This becomes a technical obstacle to achieving optimal global economic performance for offshore wind farms.

Method used

A basic sample library containing the mapping relationship between multidimensional engineering parameters and basic weight is constructed, and neighbor sample retrieval and interpolation calculation are performed by weighted multidimensional distance to achieve fast and high-precision basic weight estimation.

Benefits of technology

The basic weight estimation time has been reduced from several days to seconds, improving estimation accuracy and efficiency, meeting engineering requirements, and can be seamlessly embedded into the global economic optimization cycle of offshore wind turbine selection and layout, solving the problems of slow speed and insufficient accuracy of traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116403A_ABST
    Figure CN123116403A_ABST
Patent Text Reader

Abstract

This application relates to the field of offshore wind power engineering technology, and provides a method and system for estimating the weight of offshore wind turbine foundations. The method includes: constructing a foundation sample library, which stores mapping data between multi-dimensional engineering parameters and foundation weights; obtaining target engineering parameters for the target turbine location; retrieving neighboring sample points and performing interpolation calculations based on the weighted multi-dimensional distance between the target engineering parameters and each sample point in the foundation sample library to obtain an estimated foundation weight for the target turbine location; and outputting the estimated foundation weight. This application prioritizes the time-consuming high-fidelity design work during the sample library construction stage, enabling extremely rapid response in online applications. Furthermore, the use of weighted multi-dimensional distance for retrieval and interpolation calculations ensures that the estimation accuracy meets engineering requirements.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of offshore wind power engineering technology, specifically to a method and system for estimating the weight of offshore wind turbine foundations. Background Technology

[0002] Currently, in the early design phase of offshore wind farms, it is necessary to conduct technical and economic comparisons of various turbine models and turbine location schemes. The offshore wind turbine foundation is a critical structure supporting the wind turbine generator, and its cost accounts for a significant proportion of the total investment cost of an offshore wind farm. Traditional foundation design processes, tailored to specific site locations and turbine loads, require comprehensive structural static and dynamic analyses, geotechnical engineering analyses, and other detailed designs, making the entire process time-consuming. This method is completely impractical when comparing dozens or even hundreds of turbine locations. The industry lacks reliable and rapid estimation tools, and design institutes often rely on empirical formulas or simple linear extrapolations for estimations during the early comparison phase. These methods result in significant errors, distorting economic evaluations and potentially misleading turbine selection and investment decisions.

[0003] Furthermore, the slow speed and insufficient accuracy of basic cost estimation prevent it from being effectively integrated into the global optimization loop that considers wake and submarine cables, becoming a technical obstacle to achieving the optimal global economic efficiency of offshore wind farms. Summary of the Invention

[0004] To address the issues of slow speed and insufficient accuracy in estimating the weight of offshore wind turbine foundations in existing technologies, this application proposes a method and system for estimating the weight of offshore wind turbine foundations. This system constructs a basic sample library containing the mapping relationship between multidimensional engineering parameters and foundation weight, and performs neighbor sample retrieval and interpolation calculation based on weighted multidimensional distance, thereby achieving rapid and high-precision estimation of foundation weight under complex marine engineering conditions.

[0005] To achieve the above objectives, this application adopts the following technical solution: A method for estimating the foundation weight of an offshore wind turbine includes: constructing a foundation sample library, wherein the foundation sample library stores mapping relationship data between multidimensional engineering parameters and foundation weight; obtaining target engineering parameters of the target turbine location; based on the weighted multidimensional distance between the target engineering parameters and each sample point in the foundation sample library, retrieving neighboring sample points and performing interpolation calculations to obtain an estimated foundation weight value for the target turbine location; and outputting the estimated foundation weight value.

[0006] The above solution pre-constructs a basic sample library covering the mapping relationship between multidimensional engineering parameters and basic weights, and uses weighted multidimensional distance that takes into account the influence of parameters to perform neighbor sample retrieval and interpolation calculation. This puts the time-consuming high-fidelity design work in the sample library construction stage, thereby achieving a second-level or even millisecond-level response in online applications, while ensuring that the estimation accuracy meets engineering requirements.

[0007] Optionally, the multidimensional engineering parameters include foundation type, water depth, geological parameters, marine hydrological parameters, and wind turbine load parameters.

[0008] The above scheme ensures the completeness and representativeness of the sample database by clearly defining the core physical dimensions that affect the base weight, thus providing a reliable data foundation for subsequent high-precision interpolation.

[0009] Optionally, the weighted multidimensional distance is calculated based on the influence weights of each dimension of the multidimensional engineering parameters on the base weight.

[0010] The above scheme avoids estimation bias caused by simple geometric distance by assigning corresponding weights to parameters of each dimension that have different effects on the basic weight, and significantly improves the estimation accuracy and robustness in complex and non-uniform design spaces.

[0011] Optionally, the basic sample library is constructed in the following way: within the design space composed of the multi-dimensional engineering parameters, multiple sample points are arranged using a grid sampling method, wherein different water depth sampling ranges and sampling intervals are set for different foundation types; a structural design process is executed for each sample point to obtain the foundation weight and structural parameters of each sample point; and the input parameters and output parameters of each sample point are associated and stored.

[0012] The above scheme, by setting differentiated sampling intervals and spacings for different foundation types, can more reasonably cover the actual stress and design spatial distribution of different foundation types, thereby improving the engineering applicability of the sample library.

[0013] Optionally, the interpolation calculation employs an inverse distance weighted average algorithm based on the weighted multidimensional distance; the inverse distance weighted average algorithm includes: calculating the weight value corresponding to each of the neighboring sample points, wherein the weight value is inversely proportional to the power of the weighted multidimensional distance; and performing a weighted average based on the base weight of each of the neighboring sample points and its corresponding weight value to obtain the estimated base weight value.

[0014] The above scheme uses an inverse distance weighted average algorithm, which makes the sample points closer to the target point contribute more to the estimation. This is consistent with the continuous and gradual change characteristics of engineering parameters in local space, and further improves the accuracy of the estimation.

[0015] Optionally, the basic sample library also stores the structural parameters of the sample points, including the main component dimensions and material grades.

[0016] The above solution, by synchronously storing structural parameters in the sample library, enables the output of weight estimates while providing a reference range for component dimensions and material selection for subsequent detailed design, thereby enhancing the engineering practical value of the solution.

[0017] Optionally, the geological parameters and the marine hydrological parameters are selected from parameter combinations of typical coastal sea areas of China, and each parameter combination includes soil parameters and hydrological parameters of the corresponding sea area.

[0018] The above scheme, by selecting typical parameter combinations for specific sea areas, enables the sample library to cover the geological and hydrological characteristics of major coastal sea areas of China, thereby enhancing the reliability of the estimation method in different sea areas.

[0019] In addition, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the above-described method for estimating the weight of offshore wind turbine foundations.

[0020] The above solution solidifies the method steps through a hardware carrier, realizing the physical operation of the estimation method and meeting the deployment requirements of different computing environments.

[0021] In addition, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for estimating the weight of offshore wind turbine foundations.

[0022] The above scheme protects the method through a software carrier, which facilitates the propagation, distribution, and flexible invocation of the estimation algorithm.

[0023] Furthermore, this application also provides an offshore wind turbine foundation weight estimation system, comprising: a construction module for constructing a foundation sample library, wherein the foundation sample library stores mapping relationship data between multidimensional engineering parameters and foundation weight; an acquisition module for acquiring target engineering parameters of a target turbine location; a retrieval module for retrieving neighboring sample points based on the weighted multidimensional distance between the target engineering parameters and each sample point in the foundation sample library and performing interpolation calculations to obtain an estimated foundation weight value for the target turbine location; and an output module for outputting the estimated foundation weight value.

[0024] The above solution provides protection from the system product side, covering the software system that integrates the estimation method, and achieving multi-carrier protection coverage.

[0025] Beneficial effects: The time for estimating the basic weight of a single data point has been reduced from several days to seconds, significantly improving efficiency and fully meeting the need for rapid comparison of numerous solutions in the early stages. The principle behind this is that the time-consuming, detailed design is moved to the offline sample library construction phase, while online estimation only requires the execution of lightweight distance calculation and interpolation algorithms.

[0026] By using a high-quality sample library and a weighted multidimensional distance interpolation algorithm that considers the influence of parameters, the weight estimation error can be stably controlled within the range allowed by the feasibility study design, which is far superior to empirical formulas. The principle is that the weighted multidimensional distance overcomes the nonlinear coupling effect between parameters of different dimensions, enabling the accurate extraction and fusion of the optimal samples in the local space.

[0027] This avoids the need for expensive, detailed design for every potential site during the scheme comparison phase, significantly saving on design consulting fees and manpower costs.

[0028] As a core enabling technology, its speed and reliability enable it to be seamlessly integrated into the global economic optimization cycle of offshore wind turbine selection and layout, solving long-standing technical bottlenecks and driving the industry toward digital and intelligent design transformation. Attached Figure Description

[0029] Figure 1 This is a schematic flowchart of a method for estimating the weight of an offshore wind turbine foundation according to an embodiment of this application; Figure 2 This is a schematic diagram of the multi-dimensional design space sample point layout and neighborhood retrieval in an embodiment of this application; Figure 3 This is a schematic diagram illustrating the principle of the multi-parameter fusion interpolation calculation process in an embodiment of this application; Figure 4 This is a schematic flowchart of another method for estimating the weight of an offshore wind turbine foundation according to an embodiment of this application; Figure 5 This is a schematic diagram of a system for estimating the weight of an offshore wind turbine foundation according to an embodiment of this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application 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 application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort should fall within the scope of protection of this application.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. In the description of this application, it should be understood that the terms “first,” “second,” etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated.

[0032] Example 1: like Figure 1As shown, this embodiment provides a method for estimating the weight of offshore wind turbine foundations. The method follows an overall framework of "offline database construction - online retrieval and interpolation," placing the time-consuming, detailed structural design offline, while the online estimation stage only performs lightweight distance calculation and interpolation algorithms, thus achieving extremely fast response times in seconds or even milliseconds. Specifically, it includes the following steps: Step S110: Construct a basic sample library, which stores data on the mapping relationship between multi-dimensional engineering parameters and basic weight.

[0033] The basic sample library is pre-built in an offline phase, which is not real-time. During this phase, detailed foundation designs covering the entire design space are performed for various typical operating conditions that offshore wind power may encounter. Detailed design typically requires the use of specialized offshore wind turbine foundation design software for a full-process structural design conforming to industry standards, including static and dynamic analysis, pile-soil interaction analysis, and fatigue analysis. This process is computationally intensive and time-consuming, often requiring several days or even weeks. By pre-executing these high-fidelity design tasks, the resulting input parameters (i.e., multi-dimensional engineering parameters) and output parameters (i.e., foundation weight) are associated and stored in a structured form, forming a basic sample library that can be queried and calculated online. The mechanism of this offline pre-calculation lies in transforming the complex physical field analysis and structural iteration process into static data query resources, completely offloading the computational burden of online estimation.

[0034] Step S120: Obtain the target engineering parameters of the target machine site.

[0035] During the online estimation phase, the system receives specific parameters of the target offshore wind turbine site to be estimated. These parameters correspond to the dimensions of multidimensional engineering parameters stored in the basic sample library, including, for example, the target foundation type, target water depth, target geological parameters, target hydrological parameters, and target wind turbine load parameters. In practical engineering applications, these target engineering parameters can be received through user input on the front-end interface or automatically imported through interfaces with other planning software. Since this step involves only data reading and transmission operations and does not involve complex numerical calculations, its execution time is extremely short.

[0036] Step S130: Based on the weighted multidimensional distance between the target engineering parameters and each sample point in the basic sample library, retrieve neighboring sample points and perform interpolation calculations to obtain the basic weight estimate of the target machine point.

[0037] After acquiring the target engineering parameters, the system quickly retrieves several sample points closest to the target location from the basic sample library. The retrieval process is based on weighted multidimensional distance, which comprehensively considers the differences in the influence of various engineering parameters on the foundation weight, assigns corresponding weights to different dimensions, and calculates the distance between the target point and each sample point in multidimensional space. After determining the neighboring sample points through retrieval, interpolation calculation is performed. Interpolation calculation can employ various mathematical methods, such as inverse distance weighted average algorithm, Kriging interpolation, or radial basis function neural network interpolation. It should be understood that any algorithm capable of sample fusion calculation based on multidimensional distance is included in this category. Since the online phase only requires multidimensional distance calculation and simple interpolation formula substitution, without the need for finite element mesh generation and equation solving, the computational scale is reduced from massive engineering numerical analysis to lightweight algebraic operations. This is the core mechanism for achieving second-level response. Through this step, the system can output a high-precision estimate of the foundation weight in a very short time.

[0038] Step S140: Output the estimated base weight.

[0039] The calculated basic weight estimate is fed back to the user or downstream application module. The output can be a numerical value directly displayed on the graphical user interface, a report file containing the estimation results, or a data stream returned to the upper-level global optimization software via an API interface. The output estimate can be directly used for subsequent cost calculations and economic evaluations. Through the above steps, this embodiment reduces the basic weight design estimation, which traditionally takes several days, to the second level, improving efficiency by several orders of magnitude. Furthermore, due to the high-quality sample library and interpolation algorithm that considers the weighting of physical influences, the estimation accuracy meets engineering requirements.

[0040] Example 2: Building upon Example 1, this example elaborates on the specific dimensions of the multidimensional engineering parameters and the construction strategy of the basic sample library. The multidimensional engineering parameters include foundation type, water depth, geological parameters, marine hydrological parameters, and wind turbine load parameters.

[0041] Specifically, the foundation type determines the overall force transmission path of the structure. For example, a monopile foundation transfers the load to deep soil through a single large-diameter steel pipe pile, while a jacket foundation distributes the load to multiple smaller-diameter pile legs through a spatial truss structure. Water depth directly determines the height of the structure and the force arm of the waves and currents it withstands. Geological parameters, including soil shear wave velocity, internal friction angle, and cohesion, determine the pile penetration depth and side friction resistance. Marine hydrological parameters, including significant wave heights and spectral peak periods (return to 50 years), directly determine the peak hydrodynamic pressure of environmental loads. Wind turbine load parameters, including ultimate bending moment at the tower base, shear force, and fatigue load spectrum, are the core input force sources for foundation design. By using these five dimensions of parameters together as dimensions of multi-dimensional engineering parameters, the core physical factors affecting foundation weight can be comprehensively covered, ensuring the completeness of the sample library within the design space.

[0042] The basic sample library is constructed in the following way: within the design space composed of multi-dimensional engineering parameters, multiple sample points are laid out using a grid sampling method, with different water depth sampling ranges and sampling intervals set for different foundation types; the structural design process is executed for each sample point to obtain the foundation weight and structural parameters of each sample point; and the input parameters and output parameters of each sample point are associated and stored.

[0043] like Figure 2 As shown, when deploying sample points within the design space, if a uniform sampling strategy is adopted, i.e., the same sampling interval is used for all foundation types within the same water depth range, it will result in sparse samples for some foundation types in specific water depth ranges. For example, monopile foundations are usually suitable for shallow water areas. If the same sampling density is maintained in deep water areas as in shallow water areas, it will not only increase unnecessary offline calculation costs, but also lead to insufficient sample density in key transition sections of shallow water areas, causing the estimation error of subsequent interpolation calculations to exceed 5%. Therefore, this embodiment sets a denser sampling interval for monopile foundations in shallow water areas, for example, deploying 19 water depth sampling points at 2m intervals in the water depth range of 15m to 51m; and sets a corresponding sampling interval for jacket foundations in deep water areas. The engineering reason for this differentiated sampling is that the weight of different foundation types varies in sensitivity and nonlinearity with water depth. The weight of monopile foundations in shallow water areas increases steeply with increasing water depth, requiring dense samples to capture local high gradient changes; while the weight change of jacket foundations in deep water areas is relatively gradual, allowing for a more relaxed sampling interval. By using differentiated sampling, the estimation accuracy of each area within the design space is ensured while controlling offline computing costs.

[0044] also, Figure 2The distribution characteristics of sample points in the basic sample library and the neighborhood retrieval process during the online estimation stage are visually illustrated using a 3D scatter plot. The vertical axis represents the foundation weight, and the two horizontal axes represent water depth and load level, respectively. Different markings distinguish sample points from different sea area groups (e.g., Jiangsu group and Southern Fujian group). It can be observed that the foundation weight increases with both water depth and load level. A pentagram marks the target working point in the center of the plot, with parameters of 32m water depth and 15MW load level. A dashed line extending from this target point connects to the four nearest neighbor sample points, with the corresponding weighted multidimensional distance values ​​marked on the dashed line, visually demonstrating the geometric relationship of the K-nearest neighbor retrieval. This visualization helps to understand the mechanism of weighted multidimensional distance in retrieving neighboring samples in a non-uniform design space.

[0045] The geological and marine hydrological parameters were selected from parameter combinations of typical coastal areas of China. Each parameter combination includes soil parameters and hydrological parameters of the corresponding sea area.

[0046] Specifically, the geological and hydrological conditions of China's coastal waters vary significantly. For example, some areas are dominated by soft clay and have long wave periods, while others are dominated by dense sand and are frequently hit by typhoons. This embodiment selects 11 representative parameter combinations from typical coastal waters of China, namely, Bohai Bay, northern Shandong Peninsula, southern Shandong Peninsula, Jiangsu, Zhejiang, northern Fujian, southern Fujian, eastern Guangdong, western Guangdong, Beibu Gulf, and Hainan. Each parameter combination binds the soil parameters and hydrological parameters of a specific sea area.

[0047] It should be understood that the 11 typical parameter combinations listed here are for illustrative purposes only and not restrictive. In other embodiments, specific parameter combinations may be added or replaced according to the actual distribution of the target wind farm, as long as they can cover the geological and hydrological characteristics of the target project area.

[0048] The basic sample library also stores the structural parameters of the sample points, including the dimensions of the main components and the material grade.

[0049] Specifically, during the offline phase of the structural design process for each sample point, not only is the total foundation weight obtained, but also the dimensions of major components (such as the pile diameter, wall thickness, and pile length of a single pile, or the cross-sectional dimensions and leg spacing of the jacket support) and material grades (such as the yield strength grade of steel) are recorded simultaneously. The purpose of storing these structural parameters in association with the foundation weight is that when the estimated foundation weight is output during the online phase, the system can simultaneously extract the structural parameters of neighboring sample points, providing a reference range for component dimensions and material selection for downstream detailed design, thereby further enhancing the practical value of the estimation method in the pre-engineering design phase.

[0050] The data in the basic sample database is stored in a structured format in a relational database. Taking a single pile foundation as an example, the core fields of the database tables are designed as shown in the table below:

[0051] In this database, `sample_id` is the primary key, `foundation_type`, `water_depth`, `geo_group_id`, `hydro_group_id`, and `load_group_id` constitute the input parameters (i.e., feature vectors), and `total_weight`, `pile_diameter`, `pile_thickness`, `pile_length`, and `material_grade` constitute the output parameters (i.e., target values). Through this structured storage, the basic sample database supports efficient query and retrieval operations, providing a data foundation for rapid neighborhood retrieval during the online estimation phase.

[0052] Example 3: Building upon Example 1, this example elaborates on the calculation logic of the weighted multidimensional distance and the specific implementation of the interpolation algorithm. The weighted multidimensional distance is calculated based on the influence weights of each dimension of the multidimensional engineering parameters on the foundation weight.

[0053] Specifically, multidimensional engineering parameters include variables of different types. For continuous variables such as water depth and wind turbine load parameters, the square of the numerical difference can be directly calculated. For categorical variables such as geological parameters and marine hydrological parameters, since they usually exist in specific parameter combinations, it is difficult to directly perform algebraic subtraction, so a differentiated processing logic is required. For example, for categorical variables, a group difference function can be introduced, where the difference value is 0 when the category of the target engineering parameter is the same as the category of the sample point, and 1 when they are different. When calculating the weighted multidimensional distance, the system assigns corresponding weight coefficients based on the degree of influence of each dimension on the foundation weight. For example, water depth and load have a significant nonlinear positive correlation on the foundation weight, so they are given higher weights; while the combination of geology and hydrology affects the pile foundation penetration depth and hydrodynamic pressure, the variation range in a specific sea area is relatively small, so it is given relatively lower weights. Through this distance measurement method based on physical influence weights, the problem of treating parameters with different dimensions and different physical meanings equally by simple geometric distance is avoided. This ensures that the retrieved neighboring sample points are truly close to the target machine point in terms of engineering physical characteristics, which is the key mechanism for improving the estimation accuracy in non-uniform design spaces.

[0054] Specifically, the formula for calculating the weighted multidimensional distance is: ; Where D is the weighted multidimensional distance between the target machine point and the sample point; As the weight of the water depth dimension, For the geological parameter dimension weights, For the weights of hydrological parameters, Weights for load parameter dimensions; For the target water depth, The sample point water depth is represented by geo_group, which represents the geological parameter group; hydro_group represents the hydrological parameter group; and load_group represents the load group. This is the group difference function, which takes the value 0 when the category of the target engineering parameter is the same as the category of the sample point, and takes the value 1 when they are different.

[0055] in, to These are weighting coefficients for water depth, geological parameters, hydrological parameters, and load parameters, respectively. Their values ​​are determined based on the degree of influence of each dimension on the foundation weight. For example, water depth and load dimensions can be assigned higher weights (e.g., 0.4), while geological and hydrological dimensions can be assigned lower weights (e.g., 0.1). This is the group difference function, which takes the value 0 when the category of the target engineering parameter is the same as the category of the sample point, and 1 when they are different. Using this formula, the system can uniformly handle the distance calculation between continuous variables and categorical variables, generating a comprehensive distance metric that reflects the differences in the physical characteristics of the engineering project.

[0056] The interpolation calculation uses an inverse distance weighted average algorithm based on weighted multidimensional distance.

[0057] The inverse distance weighted average algorithm includes: calculating the weight value corresponding to each neighboring sample point, the weight value being inversely proportional to the power of the weighted multidimensional distance; and performing a weighted average based on the basic weight of each neighboring sample point and its corresponding weight value to obtain the estimated basic weight value.

[0058] like Figure 3 As shown, assume the target machine location retrieves four neighboring sample points: sample 1, sample 2, sample 3, and sample 4. First, calculate the weight value corresponding to each neighboring sample point using the following formula: ;in, For the first The weight values ​​of each neighboring sample point; For the first The weighted multidimensional distance between each neighboring sample point and the target machine point; The parameter is an exponential function, typically set to 2. The closer a sample point is to the target location in the multidimensional engineering parameter space, the greater its basic weight contributes to the estimation result. This aligns with the characteristic that engineering parameters change continuously and gradually in the local space.

[0059] Subsequently, a weighted average is calculated based on the base weights of each neighboring sample point and their corresponding weight values ​​to obtain the estimated base weight. The calculation formula is as follows: ;in, This is the estimated base weight of the target machine location; This represents the number of neighboring sample points retrieved. For the first The system uses the inverse distance weighted average algorithm described above to fuse data from multiple neighboring samples, smooth out local sample noise, and output stable and highly accurate estimates.

[0060] The inverse distance weighted average algorithm is only a preferred implementation and not a limiting one. In other embodiments, interpolation calculations can also use Kriging interpolation or radial basis function neural network interpolation as equivalent alternatives. Specifically, Kriging interpolation, by introducing a variogram to consider the spatial correlation between sample points, can provide the optimal unbiased estimate when the sample points are unevenly distributed, and is suitable for areas with drastic changes in geological parameters; radial basis function neural network interpolation, on the other hand, uses a neural network to fit the complex nonlinear mapping relationship between the design space and the foundation weight, and is more adaptable to high-dimensional nonlinear parameters such as wind turbine loads. Different algorithms have different adaptability in non-uniform design spaces. In practical applications, they can be flexibly selected or combined according to the density and parameter distribution characteristics of the sample library to establish defense depth and meet the accuracy and speed requirements of different engineering scenarios.

[0061] Example 4: Based on the above embodiments, this embodiment takes a near-shore wind farm project in China as an example to verify the operational effectiveness and accuracy of the offshore wind turbine foundation weight estimation method in a real environment. This embodiment demonstrates the operational effectiveness of the technical solution in a real environment, establishes a causal chain of structure, mechanism, and effect through specific data, and verifies the project's usability.

[0062] Specifically, the parameters for the target turbine location are set as follows: foundation type is monopile, target water depth is 32m, target geological and marine hydrological parameters are selected from the combination of parameters in the Minnan sea area, a typical coastal area of ​​China, and the target wind turbine load parameters are the typical ultimate load and fatigue load spectrum of a 15MW unit. The system first executes the step of obtaining the target engineering parameters of the target turbine location and assembles the above parameters into a query vector.

[0063] Subsequently, based on the weighted multidimensional distance between the target engineering parameters and each sample point in the basic sample library, the system retrieves neighboring sample points and performs interpolation calculations to obtain the estimated foundation weight of the target machine site. During the retrieval process, the system calculates the weighted multidimensional distance between the target point and all monopile foundation sample points in the basic sample library. Since water depth and load have a significant non-linear positive correlation with the foundation weight, the system assigns high weights to the water depth and load dimensions when calculating the distance. For categorical variables such as geological parameters and marine hydrological parameters, the system uses a group difference function, assigning a difference value of 0 when the category of the target engineering parameter is the same as the category of the sample point, and 1 when they are different, and assigning them relatively low weights. Through this distance metric method based on physical influence weights, the system can accurately retrieve neighboring sample points that are truly close to the target machine site in terms of engineering physical characteristics. In this embodiment, the system retrieves K=4 nearest neighbor sample points, namely: Sample 1 (single pile, 30m water depth, Minnan geological and hydrological combination, 15MW load, foundation weight 1420t), Sample 2 (single pile, 34m water depth, Minnan geological and hydrological combination, 15MW load, foundation weight 1480t), Sample 3 (single pile, 30m water depth, Minnan geological and hydrological combination, 10MW load, foundation weight 1380t) and Sample 4 (single pile, 34m water depth, Minnan geological and hydrological combination, 20MW load, foundation weight 1580t).

[0064] After retrieving neighboring sample points, the system performs interpolation calculations. In this embodiment, the interpolation calculation uses an inverse distance weighted average algorithm based on weighted multidimensional distance. Subsequently, the system performs a weighted average based on the base weight of each neighboring sample point and its corresponding weight value to obtain an estimated base weight value.

[0065] Substituting the above sample data and distances into the calculations, the system ultimately outputs an estimated foundation weight of 1450 tons for the target machine site. Simultaneously, the system can also extract structural parameters from the four adjacent sample points, providing a reference range for downstream detailed design with pile diameters between 2.6m and 3.0m and pile lengths between 75m and 85m, further enhancing the practical value of the estimation method in pre-engineering design.

[0066] Furthermore, the system can also output a confidence interval for the estimated foundation weight based on the dispersion of the weighted multidimensional distance between neighboring sample points and the target machine point. For example, when the distance distribution of neighboring sample points is relatively concentrated, the confidence level of the estimation result is higher and the confidence interval is narrower; conversely, when the distance distribution of neighboring sample points is relatively dispersed, the confidence interval is correspondingly wider. In this embodiment, the system outputs a confidence interval of ±3% (approximately 1407 tons to 1493 tons) for the estimated value of 1450 tons, providing an uncertainty reference for engineering decisions. To verify the accuracy of the estimation method of this application, leave-one-out cross-validation is used. A portion of the samples are randomly hidden from the constructed basic sample library as validation points, and the remaining samples are used as the basic sample library to estimate these validation points. For example, 50 samples are randomly hidden from 500 samples of a single pile foundation as validation points, and the remaining 450 samples are used as the basic sample library to estimate these 50 points. The verification results show that, compared with the actual refined design weight, the estimated weight of 50 verification points has an average absolute error controlled at 3.8%, a maximum error of 4.9%, and the error of more than 95% of the points is controlled within 5%, meeting the engineering accuracy requirements of the feasibility study stage. In terms of efficiency, traditional refined design requires 1 to 2 weeks per machine point, while the estimation method of this application requires only 0.1 seconds per point, improving efficiency by several orders of magnitude. This proves that by pre-constructing a basic sample library covering the mapping relationship between multidimensional engineering parameters and basic weight, and using weighted multidimensional distance considering the influence of parameters for neighbor sample retrieval and interpolation calculation, this application can achieve a second-level or even millisecond-level ultra-fast response while ensuring acceptable engineering accuracy, solving the problems of low accuracy and slow speed in rapid estimation of basic weight under complex marine engineering conditions.

[0067] As an example, combined Figure 4 As shown, taking a single pile foundation as an example, the entire process from sample library construction to online estimation is demonstrated.

[0068] Offline sample library construction phase: A1, Determine the design space and dimensions.

[0069] Identify five key design variables that affect the weight of a monopile foundation and construct a five-dimensional design space: Foundation type: In this embodiment, it is fixed as "single pile".

[0070] Water depth: The design range is 15 meters to 51 meters.

[0071] Geological parameter groups: Soil parameters from 11 typical coastal areas of China are selected. Each group of parameters includes surface soil type, shear wave velocity profile, internal friction angle, cohesion, etc. For example, "G_Jiangsu" represents a combination of parameters for typical soft clay and silt soil in the Jiangsu sea area.

[0072] Marine hydrological parameter sets: Corresponding to geological regions, these include significant wave heights (once every 50 years), spectral peak periods, and ocean current velocities. For example, "H_Jiangsu" represents typical wave conditions in the Jiangsu sea area.

[0073] Wind turbine load parameter groups: corresponding to the ultimate load and fatigue load spectrum of different capacity models. This embodiment selects three groups: L_10MW, L_15MW, and L_20MW. Each group includes the ultimate bending moment, shear force, torque, and fatigue load spectrum at the tower base.

[0074] A2, Set up sample points.

[0075] The system employs a full-factor gridded sampling strategy to select representative operating points: Water depth: 19 water depth points (15, 17, 19, …, 51) in intervals of 2 meters, ranging from 15m to 51m.

[0076] Geological and Hydrological Group: 11 typical parameters were selected (Central Bohai Bay, Northern Shandong Peninsula, Southern Shandong Peninsula, Jiangsu, Zhejiang, Northern Fujian, Southern Fujian, Eastern Guangdong, Western Guangdong, Beibu Gulf, and Hainan).

[0077] Load groups: Select 3 groups (10MW, 15MW, 20MW).

[0078] Theoretically, the total number of sample points for a single pile foundation is: 19 (water depth) × 11 (geological and hydrological groups) × 3 (load groups) = 627. In actual construction, obviously unreasonable combinations (such as extremely shallow water depth matching excessive loads) can be eliminated based on engineering experience, and the initial plan is to construct approximately 500 high-quality sample points.

[0079] A3, meticulous design.

[0080] For each sample point, a full-process design conforming to industry standards (such as DNVGL-ST-0126) is carried out using professional integrated offshore wind turbine foundation design software (such as commercial software SACS). Input: Accurately input water depth, geological parameters, wave parameters, and loads.

[0081] Design process: Static and dynamic analysis, pile-soil interaction analysis, fatigue analysis, scour and siltation analysis, etc. are carried out, and safe and economical foundation dimensions are determined through iterative optimization.

[0082] Output records: The key output parameters for each sample point should include at least: Total weight of the foundation: in tons, including the weight of all materials such as steel for the steel pipe pile, grouting materials, and anti-corrosion coating.

[0083] Main dimensions: pile diameter, pipe wall thickness, pile length.

[0084] Material grade: such as Q355C.

[0085] The complete input parameters (feature vector) and output parameters (target value) of each sample point are associated and stored.

[0086] A4, Build a sample library.

[0087] Store all sample data in a structured format in a relational database (such as MySQL) or a high-performance matrix file.

[0088] Online rapid estimation stage: Suppose we need to estimate the foundation weight of a target wind turbine site in Zhangzhou, Fujian Province. The parameters are: single pile foundation, target water depth = 32m, target geological parameters belong to the "Southern Fujian" group (G_Southern Fujian), target hydrological parameters belong to the "Southern Fujian" group (H_Southern Fujian), and the target wind turbine is a 15MW unit (L_15MW).

[0089] B1 receives target operating condition parameters.

[0090] The target parameters are configured as a query vector: V_target = ["single pile", 32.0, "G_Minnan", "H_Minnan", "L_15MW"].

[0091] B2, retrieve neighboring samples.

[0092] Distance metric: Weighted Euclidean distance is used. Due to the different dimensions and importance of each parameter, a weight is assigned to each dimension. For example, water depth and load have the greatest impact on weight and are assigned higher weights (e.g., 0.4); the geological and hydrological groups are categorical variables and are assigned medium weights (e.g., 0.1) after coding.

[0093] Retrieval process: Calculate the weighted distance between V_target and all single-pile sample points in the sample library.

[0094] Result: Quickly find the K sample points that are closest to the target point.

[0095] B3, multi-parameter fusion interpolation calculation.

[0096] The inverse distance weighted interpolation (IDW) method was used for estimation.

[0097] Substituting the above sample data and distance into the calculation, the final estimated weight is approximately 1450 tons.

[0098] B4 outputs the estimation result.

[0099] Output result: Base weight estimate: 850 tons.

[0100] Main size reference range: Based on nearby samples, the recommended pile diameter range is 2.6m-3.0m and the pile length range is 75m-85m.

[0101] Confidence assessment: It can output a confidence interval based on the dispersion of the sample point distance, such as ±3% (i.e., 825-875 tons).

[0102] Example 5: Based on the above embodiments, such as Figure 5 As shown, this embodiment provides an offshore wind turbine foundation weight estimation system from the system product side. The system includes a construction module, an acquisition module, a retrieval module, and an output module.

[0103] The building module is used to build a basic sample library, which stores data on the mapping relationship between multi-dimensional engineering parameters and basic weights.

[0104] Specifically, the build module is deployed on an offline computing server, responsible for executing the time-consuming, detailed structural design process and generating a structured basic sample library. In a microservice architecture, the build module can be encapsulated as an independent offline design microservice, internally integrating a professional offshore wind turbine foundation design software interface. The build module persistently stores the generated multi-dimensional engineering parameters and foundation weight mapping data in a relational database or high-performance matrix file, serving as the data source for subsequent online estimation. By physically or logically decoupling the build module from the online estimation module, updates and maintenance of the offline high-fidelity design do not interfere with the stability of the online estimation service.

[0105] The acquisition module is used to acquire the target engineering parameters of the target machine location.

[0106] Specifically, the acquisition module, acting as the system's front-end access layer, is responsible for receiving target location parameters from external input. In practical engineering applications, the acquisition module can provide a RESTful API interface to receive JSON-formatted data streams sent by the upper-layer global optimization software or front-end planning platform. After receiving the target engineering parameters, the acquisition module performs format validation and vectorization processing, assembling them into query vectors that conform to the dimensional requirements of the basic sample database, providing standardized input data for subsequent retrieval calculations.

[0107] The retrieval module is used to retrieve neighboring sample points and perform interpolation calculations based on the weighted multidimensional distance between the target engineering parameters and each sample point in the basic sample library, so as to obtain the basic weight estimate of the target machine point.

[0108] Specifically, the retrieval module is the core computing engine that enables the system to achieve ultra-fast online response. The retrieval module reads data from the basic sample library maintained by the construction module and receives query vectors from the acquisition module. Internally, the retrieval module integrates weighted multidimensional distance calculation logic and interpolation algorithms. First, it comprehensively evaluates the influence weights of each dimension parameter on the base weight to calculate the multidimensional distance. Then, it quickly retrieves the K nearest neighbor sample points and finally performs inverse distance weighted average interpolation calculations. Because the retrieval module only performs lightweight algebraic operations, it does not require finite element mesh generation or equation solving, thus enabling it to output high-precision base weight estimates within milliseconds or even seconds. In a microservice architecture, the retrieval module can be deployed on computing nodes with high concurrency processing capabilities to handle concurrent estimation requests from a large number of machine points in the global optimization loop.

[0109] The output module is used to output the basic weight estimate.

[0110] Specifically, the output module is responsible for feeding back the basic weight estimate calculated by the retrieval module to the user or downstream application module. The output module can encapsulate the estimation results into standardized data messages and return them to the upper-level global optimization software via an API interface for subsequent cost calculations and economic evaluations. Simultaneously, the output module can also generate a visual report displaying the estimated values ​​and structural parameter reference ranges extracted from neighboring sample points on a graphical user interface. Through the coordinated operation and data flow of these four modules, the system in this embodiment achieves a complete closed loop from offline database construction to online estimation, providing a fast and reliable basic weight estimation tool for the early design phase of offshore wind farms.

[0111] Example 6: Based on the above embodiments, this embodiment provides an electronic device and a computer-readable storage medium from the hardware and software carrier side, and solidifies the steps of the above-mentioned offshore wind turbine foundation weight estimation method through hardware execution logic to achieve object compliance protection.

[0112] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory. When the processor executes the computer program, it implements the above-mentioned method for estimating the weight of offshore wind turbine foundations.

[0113] Specifically, the electronic device can be a standalone physical server, a cloud cluster server, or a computing terminal integrated into an offshore wind farm engineering design workstation. The memory stores the data of the basic sample library and the computer program. It can use high-speed random access memory for data caching during online estimation, and non-volatile storage media such as solid-state drives for persistent storage of the offline-built sample library. The processor, as the hardware computing core, triggers the following hardware execution logic when reading and executing the computer program from memory: first, it receives the target engineering parameters of the target machine point through the bus interface; then, it calls the internally integrated arithmetic logic unit to execute the weighted multidimensional distance calculation instructions, quickly retrieving neighboring sample points; finally, it executes the algebraic operation instructions for interpolation calculation and feeds back the basic weight estimate through the output interface. Since the above steps mainly involve data reading, algebraic operations, and writing at the hardware level, without involving complex iterative solutions to finite element equations, the processor can complete large-scale concurrent computing tasks in a very short time, ensuring a second-level or even millisecond-level ultra-fast response from the hardware execution mechanism.

[0114] This embodiment also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described method for estimating the weight of offshore wind turbine foundations.

[0115] Specifically, computer-readable storage media can include, but is not limited to, various physical media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, magnetic disks, or optical disks. When the computer program is loaded into the processor of an electronic device and runs, it can transform the aforementioned offline library construction and online interpolation estimation logic into a machine-executable sequence of instructions. For example, the computer program includes construction instructions to control the processor to read sample point data within a design space composed of multi-dimensional engineering parameters; it also includes retrieval instructions to control the processor to calculate weighted multi-dimensional distances based on target engineering parameters and filter neighboring sample points; and it includes interpolation instructions to control the processor to execute an inverse distance weighted average algorithm and output an estimated value. By encapsulating the algorithm logic into a computer program and storing it in a tangible physical medium, the estimation method can be distributed, deployed, and flexibly invoked in the form of a software product, facilitating its application in different design institutions or computing platforms. It should be understood that the above specific descriptions of electronic devices and storage media are illustrative only and not restrictive. Any physical device or storage medium capable of implementing the above method steps through hardware execution logic should fall within the scope of protection of this application.

[0116] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for estimating the weight of offshore wind turbine foundations, characterized in that, include: Construct a basic sample library, which stores data on the mapping relationship between multi-dimensional engineering parameters and basic weight; Obtain the target engineering parameters of the target machine location; Based on the weighted multidimensional distance between the target engineering parameters and each sample point in the basic sample library, neighboring sample points are retrieved and interpolation calculations are performed to obtain the basic weight estimate of the target machine point. Output the estimated base weight.

2. The method according to claim 1, characterized in that, The multidimensional engineering parameters include foundation type, water depth, geological parameters, marine hydrological parameters, and wind turbine load parameters.

3. The method according to claim 1, characterized in that, The weighted multidimensional distance is calculated based on the influence weights of each dimension of the multidimensional engineering parameters on the base weight.

4. The method according to claim 2, characterized in that, The basic sample library is constructed in the following way: Within the design space defined by the multidimensional engineering parameters, multiple sample points are deployed using a gridded sampling method, with different water depth sampling ranges and sampling intervals set for different foundation types. For each of the sample points, a structural design process is executed to obtain the basic weight and structural parameters of each sample point; The input parameters and output parameters of each sample point are associated and stored.

5. The method according to claim 3, characterized in that, The interpolation calculation employs an inverse distance-weighted average algorithm based on the weighted multidimensional distance; The inverse distance weighted average algorithm includes: Calculate the weight value corresponding to each of the neighboring sample points, wherein the weight value is inversely proportional to the power of the weighted multidimensional distance; The estimated base weight is obtained by weighting the base weight of each neighboring sample point and its corresponding weight value.

6. The method according to claim 1, characterized in that, The basic sample library also stores the structural parameters of the sample points, including the dimensions of the main components and the material grade.

7. The method according to claim 2, characterized in that, The geological parameters and the marine hydrological parameters are selected from parameter combinations of typical coastal sea areas of China. Each parameter combination includes soil parameters and hydrological parameters of the corresponding sea area.

8. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored on the memory, wherein the processor executes the computer program to implement the method of any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

10. A system for estimating the weight of offshore wind turbine foundations, characterized in that, include: The construction module is used to build a basic sample library, which stores data on the mapping relationship between multi-dimensional engineering parameters and basic weights; The acquisition module is used to acquire the target engineering parameters of the target machine location; The retrieval module is used to retrieve neighboring sample points and perform interpolation calculations based on the weighted multidimensional distance between the target engineering parameters and each sample point in the basic sample library, so as to obtain the basic weight estimate of the target machine point. The output module is used to output the estimated base weight value.