WO2023045244A1

WO2023045244A1 - Offshore wind turbine support structure optimization design method and system based on proxy model

Info

Publication number: WO2023045244A1
Application number: PCT/CN2022/077804
Authority: WO
Inventors: 周昳鸣; 李卫东; 刘鑫; 闫姝; 张波; 陈新明; 唐巍; 郭雨桐; 孙曼洁; 吕晓静
Original assignee: 中国华能集团清洁能源技术研究院有限公司; 华能海上风电科学技术研究有限公司
Priority date: 2021-09-26
Filing date: 2022-02-25
Publication date: 2023-03-30
Also published as: CN115391926A

Abstract

Provided are an offshore wind turbine support structure optimization design method and system based on a proxy model. The method comprises: establishing, on the basis of a Kriging proxy model, a relationship between design parameters of support structures of offshore wind turbine generators having different capacities and optimally designed structural performance responses; and meanwhile, performing optimization design on a tower and a single-pile foundation to find optimal design of the lightest mass of the whole support structure. The method can avoid the problem that only the lightest design of a tower or a single pile is found due to the fact that an offshore wind turbine support structure falls within a locally optimal solution during optimization design.

Description

Method and system for optimal design of offshore wind turbine support structure based on proxy model

This application claims the priority of the domestic application submitted to the China Patent Office on September 26, 2021 with the application number 202111130009.0 and the title of the invention "Method and System for Optimal Design of Offshore Wind Turbine Support Structure Based on Proxy Model", the entire content of which is incorporated by reference incorporated in this application.

technical field

The invention belongs to the technical field of support structure design for offshore wind power generating sets, and in particular relates to an optimization design method and system for support structures of offshore wind turbines based on an agent model.

Background technique

Benefiting from the technological progress and scale expansion of wind power, the price of wind turbines, the investment cost of wind power development and the cost of operation and maintenance are showing a downward trend. From the perspective of wind turbine prices, the support structure of offshore wind turbines includes two parts: towers and foundations: the cost of wind turbine towers accounts for about 8% of the investment cost of offshore wind power projects, and the foundation of offshore wind turbines mainly includes single piles, jackets, high pile caps, etc. Different foundation forms generally account for about 14% of the investment cost of offshore wind power projects, that is, the cost of the overall support structure accounts for about 22% of the total construction cost. Therefore, reducing the support structure cost of offshore wind power can effectively reduce the levelized cost of electricity of offshore wind power.

At present, domestic offshore wind power projects usually adopt a step-by-step iterative design method when bidding. Generally, the wind turbine manufacturer provides the tower design and guarantees the tower engineering quantity. During the bid evaluation process, the tower weight will be ranked and scored; details after bidding During the design stage, the wind turbine manufacturer and the design institute will optimize the design of the tower and foundation respectively. Under this process, the wind turbine manufacturer will give the lightest local optimal design scheme of the tower as much as possible, but the lightest design of the tower is often not the global optimal design scheme of the lightest overall support structure. Existing studies often only optimize the design of a part of the overall support structure (tower or foundation), so the results obtained have certain limitations.

There are three parts involved in designing the support structure of offshore wind turbines: load calculation, tower design and foundation design.

1) Load calculation

The supporting structure of offshore wind power is subject to the combined effects of various environmental loads such as wind, waves, and currents. Most fan manufacturers in the industry use GH-Bladed for integrated modeling and load calculation.

Integrated modeling includes environmental condition input and overall support structure model building. Among them, environmental conditions include wind resource parameters, ocean hydrological parameters, engineering geological parameters and other special working conditions (sea ice, earthquake, typhoon, etc.); collectively referred to as infrastructure).

The anisotropic effect of wind and waves is considered in the load calculation. According to IEC specifications, various working conditions such as normal power generation, emergency shutdown, start-up, normal shutdown, idling, and maintenance need to be considered. According to the joint distribution of wind and waves, it can be divided into more than 20,000 working conditions.

2) Tower design

In tower design, it is necessary to check the ultimate strength, buckling strength and fatigue strength of the tower main body and local structures. The ultimate strength check includes the check of the tower shell, the tower flange, the door opening and the sea cable hole, the anchor bolt cage and other local structures; the buckling strength check includes the check of the tower shell, the door opening and the sea cable hole and other structures ; Fatigue strength check includes the check of tower shell welds, flange connection bolts, door frames and sea cable holes, top flanges, anchor bolt cages and other structures.

3) Basic design

In the design of the main body of the foundation structure, it mainly includes the analysis of the strength and bearing capacity under extreme sea conditions, the analysis of normal service conditions, the analysis of ship collisions, and the analysis of earthquake conditions. In the load combination, the extreme combination of waves, ocean currents and wind turbine operating loads under the most unfavorable water level that may occur is considered. Fatigue strength analysis uses S-N curve and Miner linear cumulative damage theory for fatigue calculation. The cumulative damage degree of each pipe joint under fatigue load is calculated separately, and the fatigue design safety of the structure is evaluated by using the cumulative damage degree.

At present, the domestic wind power industry mostly adopts the step-by-step iterative design method. Figure 1 shows a schematic diagram of the overall support structure of the offshore monopile foundation. As shown in Figure 1, the overall support structure is divided by the design interface, the tower above the interface, and the foundation structure below the interface. Figure 2 shows the flow of the step-by-step iterative design method. First, the design institute provides the environmental input of the project; the fan manufacturer provides the initial configuration of the tower and foundation according to the environmental input, and performs overall modeling and load calculation. After obtaining the optimal tower, the load at the design interface, The tower type and frequency requirements are submitted to the design institute; then the design institute checks and optimizes the basic structure under the premise of given load and tower type, and meets the frequency requirements given by the fan manufacturer; finally, the fan manufacturer obtains After optimizing the basic structure, judge whether it is converged. If it is satisfied, the iteration will end. If it is not satisfied, it will remodel and perform load calculation. There are two types of convergence criteria here: one is the design criteria for checking the tower and foundation according to the code; the other is whether the difference in quality and frequency between the optimized design obtained in this round and the previous round is within 1%.

Here, it needs to be explained that when the current design method of offshore wind turbine support structure is adopted in China, after the preliminary configuration (the diameter of the tower and monopile foundation) is determined, it generally takes 2-4 rounds of iterations to converge. Both require load calculations, tower and foundation design optimization. Further optimization of the tower and monopile diameters to find the design with the lightest overall support structure would be computationally time consuming and affect the project schedule. Therefore, in order to provide the construction drawings of the tower and monopile foundation as soon as possible in actual engineering projects, there is often not enough time for optimization, and in this process, the design and optimization of the tower and foundation are carried out sequentially, are two independent design domains, and the goal is to find the optimal design in their respective design domains. Therefore, in domestic actual projects, the final design is often the lightest local optimal design of the tower, rather than the global optimal design of the lightest overall support structure.

If a sensitivity-based optimization algorithm is to be used to solve the optimal design of the offshore wind turbine support structure, the sensitivity of the structural performance to the design variables can only be obtained through the finite difference method. The workload is very large and cannot meet the requirements of the actual construction period. Therefore, one way to overcome this difficulty is to reduce the number of required computer simulation structural analysis and improve optimization efficiency; another way has also received extensive attention. This approach is to use commercial structural analysis software as a black box, use the black box to design and calculate the performance response of complex structures for a batch of samples, and construct a proxy model of the structural analysis model on this basis.

According to the different agent models constructed, the main agent models include polynomial response surface method (Response Surface Method, RSM), radial basis function (Radial Basis Function, RBF), support vector regression method (Support Vector Regression, SVR), Kriging wait. In the present invention, we propose to use the Kriging proxy model, and it should be noted that other proxy model methods can also be used to achieve the purpose of optimization.

The Kriging surrogate model can give the predicted value and variance of sample points, which is suitable for highly nonlinear problems, and can describe the approximate original model smoothly with fewer sample points. However, since the Kriging method is an interpolation method, it is sensitive to numerical noise.

The Kriging model can be written as:

y(x)=F(β,x)+z(x)=f ^T β+z(x) (1)

Among them, β is the regression coefficient; f(x) is an approximate function, expressed as a 0-order, first-order or second-order polynomial of x; z(x) is a randomly distributed error, which has the following statistical properties:

E[z(x)]=0

Where x _i , x _j are any two sample points in the sample set, R(θ, _xi ,x _j ) is a correlation function with parameter θ, which can be given as:

where n _v is the number of design variables,

are the kth component of sample points x _i and x _j respectively. Among them, R _k (θ _k ,d _k ) has many forms, including:

Linear function R _k (θ _k ,d _k )=max{0,1-θ _k d _k }

quadratic polynomial function

Exponential function R _k (θ _k ,d _k )＝exp(-θ _k d _k )

Gaussian function

Among them, the calculation effect is better, and the widely used correlation function is the Gaussian correlation function.

It is worth noting that we can construct a response surface with the smallest error by choosing different regression functions and correlation functions, which will be further discussed in Chapter 4.

Suppose there are ns sample points, given the sample set S=[x ₁ ,x ₂ ,...,x _ns ] and its response set Y=[y ₁ ,y ₂ ,...,y _ns ], for the model (1), the predicted response value for any point x _new to be measured is:

The forecast error is:

in

In order to ensure the unbiasedness of the simulation process, the mean value of the error should be zero, that is:

Available:

F ^T cf = 0 (7)

The prediction variance of formula (1-4) is:

in:

This formula expresses the spatial correlation between the new sample point x _new and each sample point. At this time, the difference coefficient c can be determined by minimizing the forecast variance of the forecast value, and the optimization model is:

By solving it can be obtained:

Substituting c obtained from the above formula into formulas (4) and (8), the predicted value and predicted value variance of the point x _new to be measured are obtained:

by regression problem

The generalized least squares estimate of can be obtained:

β ^* ＝(F ^T R ^-1 F) ^-1 F ^T R ^-1 Y (13)

Substituting formula (13) into (12), and considering the margin expression Rγ ^* =Y-Fβ ^* , we can get:

Since the matrix F, R and vector Y are calculated by a given sample S and have nothing to do with x _new , so β and γ have nothing to do with x _new , then in formula (14) there are only vectors f(x _new ) and r( x _new ) is related to x _new . That is to say, for any point x _new to be measured, only f(x _new ) and r(x _new ) are required to obtain the predicted response value of this point.

Substituting formula (1-11) into formula (1-18), the prediction variance of the predicted value is:

in:

So far, equations (14) and (15) can be used to calculate the predicted value of any point and the predicted variance of the predicted value. There are two parameters in these two formulas

and θ. These two parameters can be calculated by maximizing the likelihood estimate of the response value. z(x) obeys a normal distribution, then y(x) should also obey a normal distribution, and its likelihood function is:

Taking the logarithm of the above formula and removing the constant term, we can get:

Match the above formula to

Taking the partial derivatives and setting them to zero gives:

Substituting (17) into (16) and ignoring the constant term, we can get:

The θ value can be obtained by maximizing (17), and the maximization problem can be solved:

In summary, the construction of the Kriging surrogate model is transformed into a nonlinear unconstrained optimization problem.

If a sensitivity-based optimization algorithm is to be used to solve the optimal design of the offshore wind turbine support structure, the sensitivity of the structural performance to the design variables can only be obtained through the finite difference method. The workload is very large and cannot meet the requirements of the actual construction period.

Contents of the invention

In order to solve the problems existing in the prior art, the present invention provides a method and system for optimal design of support structures of offshore wind turbines based on surrogate models, and at the same time optimizes the design of towers and foundation structures, and optimizes offshore wind turbines from the perspective of finding the global optimal design. The support structure is designed to reduce the LCOE and design cycle of offshore wind power.

In order to achieve the above object, the technical solution adopted in the present invention is: a method and system for optimal design of support structure of offshore wind turbine based on proxy model, comprising the following steps:

Based on determining the design variables of the wind turbine support structure, a sampling method is used to generate an initial sample set in the design space;

Perform numerical simulation on each sample point in the initial sample set to obtain corresponding structural performance response results;

Selecting a proxy model, based on the initial sample set and the structural performance response results, using the corresponding fitting or interpolation method to establish a mapping relationship between the initial sample set and the structural performance response results;

Use the proxy model for structural optimization design to obtain the preliminary design results of the tower and foundation structure. When performing structural optimization design, check and optimize the design of the tower and foundation structure at the same time after load calculation;

Use numerical simulation to simulate the obtained preliminary design results, confirm its feasibility and optimality, and check whether the convergence criterion is met; if the convergence criterion is not met, select the appropriate point addition criterion to add sample points, update the proxy model, and use the update The final proxy model is optimized until the convergence criterion is met, and the optimization iteration ends.

The design variables are tower bottom diameter and single pile diameter; the structural performance response is tower mass, single pile mass and overall support structure frequency.

The initial sample set was generated using a central composite trial approach.

Structural performance responses are tower mass, single pile mass, and overall support structure frequency.

The Kriging agent model is adopted.

The convergence criterion is:

Δ _weight /Weigh _n-1 ≤1%

Δ _frequency /Frequency _n-1 ≤1%

Among them, Δ _weight /Weigh _n-1 ≤ 1% means whether the difference between the optimally designed tower mass and single pile mass obtained in this round and the previous round is within 1%; Δ _frequency /Frequency _n-1 ≤ 1% means that the current round Whether the frequency difference of the overall support structure obtained from the optimized design obtained in the last round is within 1%.

The optimized column formula in the optimization process using the surrogate model is:

find: Diameter, wall thickness, weld height of tower bottom and monopile foundation

minimum: overall support structure quality

subject to:

SRFULS_tower≥1 ①

SRFFLS_tower≥1 ②

UCULS_foundation≤1 ③

DamageFLS_foundation≤1 ④

Design compressive bearing capacity of single pile foundation ≤ allowable compressive bearing capacity of single pile foundation ⑤

Design deformation value of single pile foundation ≤ allowable deformation value of single pile foundation ⑥

Among them: SRFULS_tower represents the safety margin of the tower structure in the limit state; SRFFLS_tower represents the safety margin of the tower structure in the fatigue state; UCULS_foundation represents the unit index of the single pile foundation structure in the limit state; DamageFLS_foundation represents the single pile foundation structure of fatigue damage.

① The lightest design of the tower is not the lightest design of the overall support structure;

② The lightest design of the overall support structure corresponds to a larger tower diameter: 6.0m for 3-5MW units, 7.0m for 6-8MW units, and the diameter of the corresponding single pile is slightly larger than the diameter of the tower bottom by 0.5m- Within the range of 1.0m, that is, the diameter of a single pile for a 3-5MW unit is 6.5-7.0m, and the diameter of a single pile for a 6-8MW unit is 7.5-8.0m.

Another aspect of the present invention is to provide an offshore wind turbine support structure optimization design system based on an agent model, including an initial sample set generation module, a structural performance response acquisition module, a mapping relationship construction module, an optimization module, and an iterative verification module;

The initial sample set generation module is based on the design variables of the fan support structure, and adopts a sampling method to generate an initial sample set in the design space;

The structural performance response acquisition module is used to numerically simulate each sample point in the initial sample set to obtain corresponding structural performance response results;

The mapping relationship building module is used to select a proxy model, and based on the initial sample set and the structural performance response results, use the corresponding fitting or interpolation method to establish a mapping relationship between the initial sample set and the structural performance response results;

The optimization module uses the proxy model for structural optimization design, and obtains the preliminary design results of the tower and foundation structure. When performing structural optimization design, the tower and foundation structure are checked and optimized at the same time after the load calculation;

The iterative verification module uses numerical simulation to simulate the obtained preliminary design results, confirms its feasibility and optimality, and verifies whether the convergence criterion is met; if the convergence criterion is not satisfied, select the appropriate point addition criterion to add sample points and update the agent Model, using the updated surrogate model to optimize the design until the convergence criterion is satisfied, then the optimization iteration ends.

The present invention also provides a computer device, including a processor and a memory, the memory is used to store a computer executable program, the processor reads and executes the computer executable program from the memory, and the present invention can be realized when the processor executes the computer executable program The proxy model-based optimization design method for offshore wind turbine support structures.

Compared with the prior art, the present invention has at least the following beneficial effects:

Adopting the method of the present invention can avoid falling into a local optimal solution when optimizing the design of the support structure of the offshore wind turbine, resulting in finding only the lightest design of the tower or single pile; using the method and process used in the present invention, after the load calculation Checking and optimizing the design of the tower and foundation structure can find the global optimal design, which makes the structure more integrated, the stiffness distribution is more uniform, and the force is more reasonable. Finally, the overall support structure has the lightest weight and reduces the support of offshore wind turbines. The purpose of structural cost: optimize the design based on the surrogate model to reduce the number of required computer simulation structural analysis and improve the efficiency of optimization.

Description of drawings

Exemplary embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, and the above and other features and advantages of the present invention will become clearer. In the accompanying drawings:

Figure 1 is a schematic diagram of the offshore support structure.

Fig. 2 is the step-by-step iterative design process of offshore wind turbine support structure.

Figure 3 is the process flow of the proxy model optimization algorithm.

Figure 4 is the process flow of the overall optimization design method for the support structure of offshore wind turbines.

Figure 5 shows the initial sample points of tower bottom diameter and single pile diameter of 3-5MW offshore wind turbines.

Figure 6 shows the initial sample points of tower bottom diameter and single pile diameter of 6-8MW offshore wind turbines.

Figure 7 is a surrogate model for the mass of a 3-5MW offshore wind turbine tower.

Figure 8 is a quality proxy model of the overall support structure of 3-5MW offshore wind turbines.

Figure 9 is a surrogate model for the mass of a 6-8MW offshore wind turbine tower.

Figure 10 is a quality proxy model of the overall support structure of 6-8MW offshore wind turbines.

Detailed ways

The present invention will be further described in detail below with reference to FIGS. 3-10 and specific embodiments.

Referring to Figure 3, the general flowchart of the agent model optimization algorithm includes the following steps:

Step 1. Obtain the design features of the offshore support structure and screen the design variables through engineering project experience. In the present invention, the design variables are the diameter of the tower bottom and the diameter of the single pile;

Step 2, adopt sampling method to generate initial sample set in the design space, the present invention adopts central compound test method;

Step 3: Perform numerical simulation analysis on each sample point in the initial sample set to obtain the corresponding structural performance response. In the present invention, the structural performance response is the mass of the tower, the mass of the single pile and the frequency of the overall support structure.

Step 4: Select a surrogate model (suggested but not limited to Kriging surrogate model), and use the corresponding appropriate fitting or interpolation method to establish the mapping relationship between them according to the input and output in step 2 and step 3.

Step 5, use the proxy model to carry out structural optimization design, and obtain the preliminary design results of the tower and foundation structure. When performing structural optimization design, check and optimize the design of the tower and foundation structure at the same time after load calculation;

Step 6. Use numerical simulation to conduct simulation analysis on the preliminary design results obtained in step 5, confirm its feasibility and optimality, and check whether the convergence criterion is satisfied; if the convergence criterion is not satisfied, select the appropriate point addition criterion to add sample points, Update the proxy model and go back to step 5. If the convergence criterion is met, the optimization iteration ends; the convergence criterion here refers to whether the difference between the response value of the optimized design obtained in the current round and the last round, as well as the physical quantities such as the mass of the tower, the mass of the single pile, and the frequency of the overall support structure are within 1%.

The method for numerically simulating the sample points in step 2 above is a proxy model-based overall optimization design method for offshore wind turbine support structures proposed by the present invention; refer to the process flow of the overall optimization design method for support structures shown in Figure 4, offshore The biggest difference between the overall optimal design method of the wind power support structure and the design method of the step-by-step iterative method is that after the load calculation, the tower and the foundation structure are checked and optimized at the same time, and the lightest overall support structure is found in the entire design domain. Excellent design. After obtaining the load, the wind turbine manufacturer and the design institute will optimize the design of the tower and the foundation structure at the same time, and design from the perspective of finding the global optimal design. At present, the procurement cost of materials for offshore support structures basically only considers the quality factor. The quality of the overall support structure can be reduced through the overall optimization design method of the support structure, so as to achieve the purpose of reducing the LCOE of offshore wind power. The optimized columns are as follows:

minimum: overall support structure quality

subject to:

SRFULS_tower≥1 ①

SRFFLS_tower≥1 ②

UCULS_foundation≤1 ③

DamageFLS_foundation≤1 ④

The optimal support structure that can be obtained through the proxy model-based optimization design method for offshore wind turbine support structures proposed by the present invention:

② The lightest design of the overall support structure corresponds to a larger tower diameter (3-5MW unit is 6.0m, 6-8MW unit is 7.0m), and the corresponding single pile diameter is slightly larger than the diameter of the tower bottom by 0.5m Within the range of -1.0m, that is, 6.5-7.0m for 3-5MW units, and 7.5-8.0m for 6-8MW units.

Take the optimal design of the supporting structure of 3-5MW and 6-8MW offshore wind turbines as an example:

Step 1, select the tower bottom diameter and single pile diameter as design variables.

Step 2, select the central composite test method to generate initial sample points (Figure 5 is the initial sample point for the tower bottom diameter and single pile diameter of 3-5MW offshore wind turbines; Figure 6 is the tower bottom diameter and single pile diameter for 6-8MW offshore wind turbines pile diameter initial sample points).

In step 3, the overall optimization of the support structure of the offshore wind turbine is used to obtain the corresponding structural performance response (tower mass, single pile mass, frequency of the overall support structure, etc.).

Step 4, select the Kriging proxy model, use the corresponding appropriate fitting or interpolation method according to the input and output in step 2), 3) to establish a proxy model of the mapping relationship between them (as shown in Figure 7 and Figure 8 It is the 3-5MW offshore wind turbine tower quality proxy model and the overall support structure quality proxy model; Figure 9 and Figure 10 are the 6-8MW offshore wind turbine tower bottom diameter and the initial sample point of the single pile diameter).

Step 5, use the proxy model instead of the original model for structural optimization design.

Step 6: Check the optimal design obtained in Step 5 using the original model to confirm its feasibility and optimality, and check the convergence criteria. If it is not satisfied, select the appropriate point addition criterion to add sample points, update the proxy model, and return to step 5. If satisfied, the optimization iteration ends.

The convergence criterion here refers to whether the difference between the response values (tower mass, single pile mass, overall support structure frequency, etc.) of the optimal design obtained in this round and the previous round is within 1%.

The optimized columns are as follows:

find: tower bottom diameter, single pile diameter

minimum: overall support structure quality

subject to:

SRFULS_tower≥1

SRFFLS_tower≥1

UCULS_foundation≤1

DamageFLS_foundation≤1

The design compressive bearing capacity of the single pile foundation ≤ the allowable compressive bearing capacity of the single pile foundation.

Design deformation value of single pile foundation ≤ allowable deformation value of single pile foundation.

Table 1 Calculation results of 3-5MW offshore wind turbine support structure

Table 2 Calculation results of 6-8MW offshore wind turbine support structure

Table 1 and Table 2 respectively show the calculation results of the support structure of 3-5MW and 6-8MW offshore wind turbines based on the optimization algorithm of the proxy model. Among them, for 3-5MW units, the lightest tower design is D tower = 5.5m, D single pile = 6.0m; the lightest design of the overall support structure is D tower = 6.0m, D pile = 6.5m. For 6-8MW units, the lightest design of the tower is D tower = 6.5m, D single pile = 8.5m; the lightest design of the overall support structure is D tower = 7.0m, D pile = 7.5m.

The optimal support structure obtainable through this example is characterized by:

1) The lightest design of the tower is not the lightest design of the overall support structure;

2) The lightest design of the overall support structure corresponds to a larger tower diameter (3-5MW unit is 6.0m, 6-8MW unit is 7.0m), and the corresponding single pile diameter is 0.5m larger than the tower bottom diameter , that is, 3-5MW units are 6.5m, and 6-8MW units are 7.5m.

Proxy model-based optimization design system for offshore wind turbine support structure, including initial sample set generation module, structural performance response acquisition module, mapping relationship building module, optimization module, and iterative verification module;

The iterative verification module uses numerical simulation to simulate the obtained preliminary design results, confirms its feasibility and optimality, and verifies whether the convergence criterion is met; if the convergence criterion is not satisfied, select the appropriate point addition criterion to add sample points and update the agent Model, using the updated surrogate model to optimize the design until the convergence criterion is met, then the optimization iteration ends.

The present invention also provides a computer device, including a processor and a memory, the memory is used to store a computer executable program, the processor reads and executes the executable program from the memory, and when the processor executes the computer executable program, the invention can be realized The optimization design method of offshore wind turbine support structure based on surrogate model is described.

The computer device may be a laptop, tablet, desktop or workstation.

The processor can be a central processing unit (CPU), digital signal processor (DSP), application specific integrated circuit (ASIC), or off-the-shelf programmable gate array (FPGA).

For the memory of the present invention, it can be an internal storage unit of a notebook computer, a tablet computer, a desktop computer, a mobile phone or a workstation, such as a memory, a hard disk; an external storage unit can also be used, such as a mobile hard disk, a flash memory card.

Claims

A method for optimal design of offshore wind turbine support structure based on proxy model, characterized in that it includes the following steps:

Based on determining the design variables of the wind turbine support structure, a sampling method is used to generate an initial sample set in the design space;

Perform numerical simulation on each sample point in the initial sample set to obtain corresponding structural performance response results;

Selecting a proxy model, based on the initial sample set and the structural performance response results, using the corresponding fitting or interpolation method to establish a mapping relationship between the initial sample set and the structural performance response results;

Use the proxy model for structural optimization design, and obtain the preliminary design results of the tower and foundation structure. When performing structural optimization design, check and optimize the design of the tower and foundation structure at the same time after load calculation;

Use numerical simulation to simulate the obtained preliminary design results, confirm its feasibility and optimality, and check whether the convergence criterion is met; if the convergence criterion is not met, select the appropriate point addition criterion to add sample points, update the proxy model, and use the update The final proxy model is optimized until the convergence criterion is met, and the optimization iteration ends.
The method for optimal design of support structures for offshore wind turbines based on proxy models according to claim 1, wherein the design variables are the diameter of the tower bottom and the diameter of the single pile; the structural performance response is the mass of the tower, the mass of the single pile, and the overall support structure frequency.
The method for optimal design of support structures of offshore wind turbines based on surrogate models according to claim 1, wherein the initial sample set is generated by using a central composite test method.
The proxy model-based optimization design method for offshore wind turbine support structures according to claim 1, wherein the structural performance response is tower mass, single pile mass, and overall support structure frequency.
The proxy model-based optimization design method for offshore wind turbine support structures according to claim 1, characterized in that a Kriging proxy model is used.
The proxy model-based optimization design method for offshore wind turbine support structures according to claim 1, wherein the convergence criterion is:

Δ weight /Weigh n-1 ≤1%

Δ frequency /Frequency n-1 ≤1%

Among them, Δ weight /Weigh n-1 ≤ 1% means whether the difference between the optimally designed tower mass and single pile mass obtained in this round and the previous round is within 1%; Δ frequency /Frequency n-1 ≤ 1% means that the current round Whether the frequency difference of the overall support structure obtained from the optimized design obtained in the last round is within 1%.
According to claim 1, the method for optimal design of support structures for offshore wind turbines based on proxy models is characterized in that the optimized column formula in the optimization process using the proxy model is:

find: Diameter, wall thickness, weld height of tower bottom and monopile foundation

minimum: overall support structure quality

subject to:

SRFULS_tower≥1 ①

SRFFLS_tower≥1 ②

UCULS_foundation≤1 ③

DamageFLS_foundation≤1 ④

Design compressive bearing capacity of single pile foundation ≤ allowable compressive bearing capacity of single pile foundation ⑤

Design deformation value of single pile foundation ≤ allowable deformation value of single pile foundation ⑥

Among them: SRFULS_tower represents the safety margin of the tower structure in the limit state; SRFFLS_tower represents the safety margin of the tower structure in the fatigue state; UCULS_foundation represents the unit index of the single pile foundation structure in the limit state; DamageFLS_foundation represents the single pile foundation structure of fatigue damage.
The method for optimal design of offshore wind turbine support structure based on proxy model according to claim 1, characterized in that,

① The lightest design of the tower is not the lightest design of the overall support structure;

② The lightest design of the overall support structure corresponds to a larger tower diameter: 6.0m for 3-5MW units, 7.0m for 6-8MW units, and the diameter of the corresponding single pile is slightly larger than the diameter of the tower bottom by 0.5m- Within the range of 1.0m, that is, the diameter of a single pile for a 3-5MW unit is 6.5-7.0m, and the diameter of a single pile for a 6-8MW unit is 7.5-8.0m.
The offshore wind turbine support structure optimization design system based on the proxy model is characterized in that it includes an initial sample set generation module, a structural performance response acquisition module, a mapping relationship construction module, an optimization module, and an iterative verification module;

The initial sample set generation module is based on the design variables of the fan support structure, and adopts a sampling method to generate an initial sample set in the design space;

The structural performance response acquisition module is used to numerically simulate each sample point in the initial sample set to obtain corresponding structural performance response results;

The mapping relationship building module is used to select a proxy model, and based on the initial sample set and the structural performance response results, use the corresponding fitting or interpolation method to establish a mapping relationship between the initial sample set and the structural performance response results;

The optimization module uses the proxy model for structural optimization design, and obtains the preliminary design results of the tower and foundation structure. When performing structural optimization design, the tower and foundation structure are checked and optimized at the same time after the load calculation;

The iterative verification module uses numerical simulation to simulate the obtained preliminary design results, confirms its feasibility and optimality, and verifies whether the convergence criterion is met; if the convergence criterion is not satisfied, select the appropriate point addition criterion to add sample points and update the agent Model, using the updated surrogate model to optimize the design until the convergence criterion is met, then the optimization iteration ends.
A computer device, characterized by comprising a processor and a memory, the memory is used to store a computer executable program, the processor reads and executes the computer executable program from the memory, and the claim can be realized when the processor executes the computer executable program The proxy model-based optimization design method for offshore wind turbine support structures described in any one of 1 to 8.