Offshore floating type fan mooring system optimization method based on proxy model

Abstract

The invention relates to the technical field of offshore wind power equipment, and discloses an offshore floating type wind turbine mooring system optimization method based on a proxy model, and the method comprises the following steps: S1, determining key design parameters; s2, sampling to generate design points; s3, establishing a total index function of multi-target coupling; s4, generating a sample database through numerical simulation; s5, constructing an agent model by using a Kriging method; s6, calling the proxy model to search an optimal solution by adopting a differential evolution algorithm; s7, performing numerical simulation verification on the optimal solution and judging convergence; and S8, if not, updating the database and returning to S5 for iteration, and if so, outputting the optimal configuration. According to the method, the problems that a traditional optimization method is high in simulation calculation cost and too long in consumed time are solved, searching is carried out through the proxy model instead of high-precision simulation, the verification iteration step is combined, the optimization efficiency is greatly improved while the result accuracy is guaranteed, and the multi-target mooring scheme with the optimal comprehensive performance can be efficiently obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of offshore wind power equipment, in particular to a mooring system optimization method for offshore floating wind turbine based on a proxy model. BACKGROUND

[0002] With the growth of global energy demand and the gradual development of onshore wind energy resources, offshore wind power has become an important direction of renewable energy development. When the wind farm develops to the deep sea, the floating foundation platform becomes the key structure to support the wind turbine generator unit. As the core component of anchoring the floating platform in a fixed position, the design of the mooring system directly determines the safety, stability and economy of the entire floating wind turbine system.

[0003] When designing and optimizing the mooring system of offshore floating wind turbine, it is usually necessary to use numerical simulation tools to evaluate the dynamic response of the system under different design parameters. However, these high-precision time-domain coupled dynamic simulations often consume a large amount of computing resources and time, and a complete simulation calculation may take several hours or even longer. A complete optimization process requires the evaluation of hundreds or thousands of design schemes in the parameter space. If the optimization algorithm is directly coupled with high-precision simulation tools, the huge time cost is unacceptable in engineering practice.

[0004] In addition, the design of the mooring system is a complex multi-objective optimization problem. The designer needs to reduce the engineering cost as much as possible while ensuring that the performance indicators such as mooring cable tension, platform movement amplitude and anchoring system safety meet the specification requirements. There is often a conflict between these performance indicators, and traditional optimization methods often have difficulty in finding a global optimal balance point between multiple conflicting objectives, and are easily trapped in local optimal solutions, resulting in a final design scheme that is not the best in comprehensive performance. SUMMARY

[0005] In view of the shortcomings of the prior art, the present application provides a mooring system optimization method for offshore floating wind turbine based on a proxy model, which solves the problems of high computational cost and long time consumption caused by excessive reliance on high-precision numerical simulation in traditional optimization methods, and effectively balances the performance indicators in the complex multi-objective optimization problem of the mooring system and avoids falling into local optimal solutions.

[0006] To achieve the above purpose, the present application realizes the following technical solutions:

[0007] A mooring system optimization method for offshore floating wind turbine based on a proxy model, comprising the following steps:

[0008] S1: Determine the key design parameters of the offshore floating wind turbine mooring system, and set the parameter value range;

[0009] S2: sampling in the parameter space determined in S1 to generate a batch of design points;

[0010] S3: establishing a total index coupling motion, tension and cost as an objective function of optimization;

[0011] S4: calling a numerical simulation tool to perform dynamic response simulation on each design point generated in S2, and evaluating the objective function defined in S3 based on the simulation results to generate a sample database;

[0012] S5: using the Kriging method to model the sample database generated in S4 to construct a surrogate model;

[0013] S6: using the differential evolution algorithm and calling the surrogate model constructed in S5 to evaluate the design scheme and search for the optimal parameter combination;

[0014] S7: simulating the optimal parameter combination searched in S6 by the numerical simulation tool to calculate the real fitness value, and comparing the real fitness value with the fitness value predicted by the surrogate model to determine whether the convergence requirement is met;

[0015] S8: if the convergence requirement is not met in S7, updating the sample database in S4 and returning to S5 for iteration; if the convergence requirement is met, outputting the current optimal mooring configuration.

[0016] Preferably, the key design parameters in S1 include mooring radius, mooring cable length, ballast block density and ballast block position.

[0017] Preferably, the objective function in S3 is obtained by multiplying the dimensionless index of mooring line tension, the dimensionless index of anchor point tension, the dimensionless index of motion constraint and the dimensionless index of cost by respective weight coefficients and then summing them up, and the sum of all weight coefficients is 1.

[0018] Preferably, the calculation of the motion constraint dimensionless index comprises the following sub-steps:

[0019] S3.1: performing fast Fourier transform on the time series of six degrees of freedom motion of the floating wind turbine to obtain a motion spectrum;

[0020] S3.2: selecting spectral peaks greater than five percent of the maximum peak value from the motion spectrum;

[0021] S3.3: summing the peak values of the selected spectral peaks to obtain representative values of six degrees of freedom, i.e. surge, sway, heave, roll, pitch and yaw;

[0022] S3.4: calculating the motion constraint dimensionless index based on the representative values of the six degrees of freedom and their respective reference values.

[0023] Preferably, the cost dimensionless index is calculated in the following way: the difference between the current mooring line mass and the reference mooring line mass is divided by twice the reference mooring line mass to obtain a first term; the difference between the current ballast mass and the reference ballast mass is divided by twice the reference ballast mass to obtain a second term; and the cost dimensionless index is the sum of the first term and the second term.

[0024] Preferably, the numerical simulation tool in S4 simulates using the following model:

[0025] The blade element momentum method is used to simulate the aerodynamic force on the wind turbine blade.

[0026] The potential flow theory and the Morison equation are used to calculate the hydrodynamic load on the floating wind turbine.

[0027] The lumped mass method is used to calculate the displacement and tension of the mooring line.

[0028] Preferably, the Kriging method in S5 uses a model predicted value represented by an objective function, which is a regression term composed of a basis function vector and a regression coefficient vector, plus a stationary Gaussian process with a mean of zero.

[0029] Preferably, the differential evolution algorithm in S6 specifically includes: setting the population size to 10d to 20d, the mutation rate to 0.5 to 0.9, and the crossover rate to 0.6 to 0.95, where d is the dimension of the continuous variables in S1.

[0030] Preferably, the convergence requirement in S7 is that the absolute error between the true fitness value and the fitness value predicted by the surrogate model is less than 5%.

[0031] Preferably, the mooring radius in the key design parameters in S1 is a plurality of candidate values; the optimization process of S1 to S8 is performed once for each candidate mooring radius to obtain the optimal mooring configuration; and the method is further limited in that the mooring scheme with the best overall performance is selected from all the obtained optimal mooring configurations as the global optimal solution.

[0032] The present application provides a method for optimizing the mooring system of a floating wind turbine at sea based on a surrogate model.

[0033] 1. The present application significantly improves the optimization efficiency of the mooring system. By constructing the Kriging surrogate model in step S5 and calling the surrogate model in the optimization search of step S6 to replace the high computational cost numerical simulation (such as OpenFAST) in step S4 for a large number of fitness evaluations, the number of simulation tool calls is greatly reduced, the design and optimization cycle is shortened, and the computational burden in the optimization process is reduced; at the same time, the differential evolution algorithm is introduced to find the optimal parameter combination with global search ability, and the optimization quality is improved.

[0034] 2. The present application ensures the accuracy and reliability of the optimization results. By setting the convergence judgment of step S7 and the iteration update mechanism of step S8, the present method performs real numerical simulation verification on the optimal solution searched in step S6 based on the surrogate model. If the error does not meet the requirements (such as the absolute error is greater than 5%), the sample database is updated and the model is retrained in S5 to ensure the local precision of the surrogate model in the optimal solution area and avoid the deviation of the optimization results from the real physical response.

[0035] 3. The present application realizes the global comprehensive optimization of the mooring system multi-objective. By establishing a total index coupling motion, tension and cost in step S3, the complex multi-objective optimization problem is converted into a single objective problem; and combined with the global search ability of the differential evolution algorithm (DE) in step S6, the safety, cost and platform motion performance and other multiple conflicting design indicators can be effectively balanced, the global optimal parameter combination is searched, and the local optimal solution is avoided. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The optimization framework diagram of the present application;

[0037] Figure 2 The target function prediction distribution diagram of the surrogate model of the present application in the parameter space;

[0038] Figure 3 The prediction accuracy diagram of the target function of the surrogate model of the present application;

[0039] Figure 4 The optimization result diagram of the present application;

[0040] Figure 5 The mooring tension comparison diagram before and after optimization of the present application. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the specification of the present application.

[0042] Please refer to the drawings in the specification of the present application Figure 1 - the drawings in the specification of the present application Figure 5The embodiment of the present application provides a mooring system optimization method for offshore floating wind turbine based on a proxy model, comprising the following steps:

[0043] S1: determining key design parameters of the mooring system of the offshore floating wind turbine, and setting a parameter value range;

[0044] S2: sampling in the parameter space determined in S1 to generate a batch of design points;

[0045] S3: establishing a total index coupling motion, tension and cost as an objective function of optimization;

[0046] S4: calling a numerical simulation tool to perform dynamic response simulation on each design point generated in S2, and evaluating the objective function defined in S3 based on the simulation result to generate a sample database;

[0047] S5: modeling the sample database generated in S4 using the Kriging method to construct a proxy model;

[0048] S6: adopting a differential evolution algorithm and calling the proxy model constructed in S5 to evaluate a design scheme and search for an optimal parameter combination;

[0049] S7: performing numerical simulation on the optimal parameter combination searched in S6 by using the numerical simulation tool, calculating a real fitness value, comparing the real fitness value with a fitness value predicted by the proxy model, and judging whether a convergence requirement is met;

[0050] S8: if the convergence requirement is not met in S7, updating the sample database in S4 and returning to S5 for iteration; if the convergence requirement is met, outputting a current optimal mooring configuration.

[0051] Specifically, the key design parameters in S1 include a mooring radius, a mooring cable length, a ballast block density and a ballast block position.

[0052] The determination of the parameter value range can refer to related research and industry specifications. For example, the ratio of the mooring radius to the water depth can be set to be between 5 and 10; the mooring cable length in the optimization is set to be the mooring radius r to 1.03r. In combination with commonly used counterweight materials and related research, the linear density of the ballast block is 10,000 kg / m to 40,000 kg / m. In order to reduce the vertical tension component of the mooring line lower end and avoid collision between the counterweight and the seabed, the position of the counterweight (i.e. the ballast block) is set to be 10 meters to 40 meters away from the floating wind turbine floating body.

[0053] In the S2 step, the Sobol method is used to sample in the entire parameter space determined in S1 to generate a batch of design points. If the dimension of the continuous variable in S1 is d, the recommended sample number (i.e. the number of design points) is 10d to 20d.

[0054] In the S3 step, a total index coupling motion, tension, cost and other performance indexes is established as an objective function of optimization. The objective function in S3 is obtained by multiplying the dimensionless indexes of mooring line tension, anchor point tension, motion constraint and cost by their respective weight coefficients and then summing them up, and the sum of all weight coefficients is 1.

[0055] In one embodiment, the expression of the objective function F is:

[0056] F = λ l R lt + λ a R at + λ m R m + λ c R c

[0057] Wherein, F is the objective function for quantifying the overall performance of the mooring system. R lt , R at , R m and R c are four dimensionless indexes respectively evaluating the relative deviation of a mooring scheme in four aspects of mooring line tension, anchor point tension, motion constraint and cost. λ l , λ a , λ m , λ c are the corresponding weight coefficients, and the sum of the four coefficients is 1.

[0058] The expression of the dimensionless index of mooring line tension R lt is:

[0059]

[0060] Wherein, T DEL,i represents the short-term damage equivalent load of the i-th mooring line in a mooring scheme; T DEL,ri represents the corresponding same quantity in a reference mooring scheme. η1 and η2 are weight coefficients, and satisfy 2η1+η2=1.

[0061] The expression of the dimensionless index of anchor point tension R at is:

[0062]

[0063] Wherein, T aveAi represents the average value of the total tension of the i-th anchor point; T ave-zAi represents the average value of the vertical uplift tension of the i-th anchor point (i=1, 2, 3); T max-zA represents the maximum value of the vertical uplift tension of the i-th anchor point; T DELAirepresents the damage equivalent load of the ith anchor point; all symbols with subscript r represent the corresponding values of the reference scheme. θ1, θ2, θ3, θ4 are weight coefficients, in one specific embodiment, θ3 = 0.25, θ4 = 0.25.

[0064] The calculation of the motion constraint dimensionless index includes the following sub-steps:

[0065] S3.1: Perform fast Fourier transform on the time series of the six degrees of freedom of the floating wind turbine to obtain the motion spectrum;

[0066] S3.2: Select the spectral peaks greater than 5% of the maximum peak value from the motion spectrum;

[0067] S3.3: Sum the peak values of the selected spectral peaks to obtain the representative values of the six degrees of freedom, i.e. surge, sway, heave, roll, pitch, and yaw, respectively;

[0068] S3.4: Calculate the motion constraint dimensionless index based on the representative values of the six degrees of freedom and their respective reference values.

[0069] The motion constraint dimensionless index R m is expressed as:

[0070]

[0071] where x and θ represent the representative values of translational and rotational motion, respectively, and the subscripts surge, heave, pitch, sway, roll, and yaw are used to indicate the specific motion degree of freedom. μ1 and μ2 are weight coefficients, and satisfy 3μ1 + 3μ2 = 1. The specific calculation process of the representative values (x or θ) is as follows: perform fast Fourier transform on the time series of the motion of a given degree of freedom to obtain the motion spectrum, select the spectral peaks greater than 5% of the maximum peak value, and sum the peak values of the selected spectral peaks.

[0072] The calculation method of the cost dimensionless index is as follows: divide the difference between the current mooring line mass and the reference mooring line mass by twice the reference mooring line mass to obtain the first term; divide the difference between the current counterweight mass and the reference counterweight mass by twice the reference counterweight mass to obtain the second term; and the cost dimensionless index is the sum of the first and second terms.

[0073] The cost dimensionless index R c is expressed as:

[0074]

[0075] where M Land M CM M and M represent the mass of the mooring line and the counterweight, respectively; M L,r and M CM,r M and M represent the mass of the mooring line and the counterweight, respectively; M

[0076] In the S4 step, different wind and wave conditions are set based on a specific sea area, and a numerical simulation tool is called to simulate the dynamic response of each sample point generated in S2. In an embodiment, the numerical simulation tool is preferably OpenFAST. The simulation duration is set to 4000s, and the first 400s is used for the wind turbine to reach steady state.

[0077] The numerical simulation tool in S4 uses the following models:

[0078] The blade element momentum method is used to simulate the aerodynamic force on the wind turbine blades;

[0079] The potential flow theory and the Morison equation are used to calculate the hydrodynamic load on the floating wind turbine;

[0080] The lumped mass method is used to calculate the displacement and tension of the mooring line.

[0081] In this embodiment, the OpenFAST tool integrates the above models, which simulate the aerodynamic force by the blade element momentum method, calculate the hydrodynamic load by the potential flow theory and the Morison equation, and calculate the mooring line dynamic response by the lumped mass method. S4 outputs the parameters for evaluating the objective function defined in S3 based on the simulation results, and stores the samples and corresponding index values in the database to generate a high-quality sample database.

[0082] In the S5 step, the kriging method is used to model the sample database generated in S4.

[0083] In S5, the model predicted value represented by the objective function of the kriging method is a regression term composed of a basis function vector and a regression coefficient vector, and is added to a stationary Gaussian process with a mean of zero.

[0084] The kriging method fits the relationship between the input parameters of S1 and the objective function (response) of S3. After constructing the model, the model accuracy is evaluated by comparing the predicted value and the true value at a specific sample, for example, by quantitative means such as calculating the determination coefficient (R2) and the root mean square error (MSE) based on the predicted value and the true value at the validation sample, and setting a qualified threshold, such as R2≥0.9, MSE≥0.8, to ensure that it has good prediction ability.

[0085] In the S6 step, based on the surrogate model constructed in S5, the differential evolution algorithm (DE) is used to search for the optimal parameter combination.

[0086] The differential evolution algorithm in S6 specifically includes: setting the population size to 10d-20d, the mutation rate to 0.5-0.9, and the crossover rate to 0.6-0.95, where d is the dimension of the continuous variable in S1.

[0087] The algorithm calls the proxy model constructed in S5 to evaluate the fitness (i.e., the S3 objective function value) to search for the optimal parameter combination that meets multiple performance indicators.

[0088] In the S7 step, the optimal parameter combination searched in S6 is simulated by OpenFAST, the real fitness value is calculated, and the size of the real fitness value is compared with the fitness value predicted by the proxy model in S5.

[0089] The convergence requirement in S7 is that the absolute error between the real fitness value and the fitness value predicted by the proxy model is less than 5%.

[0090] In the S8 step, it is judged whether the result of S7 meets the convergence requirement. If it does not meet (i.e., the absolute error is greater than or equal to 5%), the parameter combination simulated in S7 and its real fitness value are updated to the sample database in S4, and step S5 is returned for iteration to retrain the proxy model; if it meets (i.e., the absolute error is less than 5%), the optimal mooring configuration under the current radius is output.

[0091] The mooring radius in the key design parameters in S1 is a plurality of candidate values; the optimization process of S1-S8 is performed once for each candidate mooring radius to obtain the respective optimal mooring configuration; the method further limits: from all the obtained optimal mooring configurations, the mooring scheme with the best overall performance is selected as the global optimal solution.

[0092] In a preferred embodiment, to screen the global optimal solution, the complete optimization process of S1-S8 can be performed repeatedly for the plurality of candidate mooring radii (e.g., 500m, 600m, 700m, etc.) set in S1. After obtaining the optimal mooring configurations corresponding to all the candidate mooring radii, the S3 objective function F values of these optimal configurations are compared, and the mooring scheme with the best overall performance (i.e., the smallest F value) is selected as the global optimal solution output by the method.

[0093] The present application will be further described below in conjunction with specific implementation cases to further illustrate the feasibility and progressiveness of the present application.

[0094] The data set used in the examples of the present application is selected from measured and derived wave data collected from ocean wave measuring buoys anchored in the Gulf of Maine, USA. The data is from the Queensland Government data, mainly including wave data from 2006 to 2018, containing multiple wave field characteristics, which can be seen in Table 1, and the data can be downloaded from the National Ocean Big Data Platform. The optimization results of the test optimization framework in the sea area for the OC45MW semi-submersible wind turbine mooring system.

[0095] The dynamic responses of the optimized floating wind turbine, such as its six-degree-of-freedom rigid body motion and mooring line tension, are studied and evaluated. According to industry standards, limit state (ULS) and fatigue state (FLS) evaluations are performed to verify that the optimized wind turbine meets design or safety requirements. Subsequently, under representative load conditions, the dynamic responses of the wind turbine with the optimized mooring configuration are compared with those of a reference wind turbine at a water depth of 200 meters, and it is found that the overall performance of the optimized design is not inferior to that of the reference wind turbine.

[0096] Table 1. Environmental load data

[0097]

[0098] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and variations can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. An optimization method for offshore floating wind turbine mooring systems based on a surrogate model, characterized in that, Includes the following steps: S1: Determine the key design parameters of the offshore floating wind turbine mooring system and set the parameter value range; S2: Sampling is performed in the parameter space determined by S1 to generate a batch of design points; S3: Establish a comprehensive index that couples motion, tension, and cost as the objective function for optimization; S4: Call the numerical simulation tool to perform dynamic response simulation on each design point generated in S2, and evaluate the objective function defined in S3 based on the simulation results to generate a sample database; S5: Use the Kriging method to model the sample database generated in S4 and build a surrogate model; S6: Employ the differential evolution algorithm and call the surrogate model built in S5 to evaluate the design scheme and search for the optimal parameter combination; S7: Simulate the optimal parameter combination found in S6 using the numerical simulation tool, calculate the actual fitness value, and compare the actual fitness value with the fitness value predicted by the surrogate model to determine whether the convergence requirement is met. S8: If the convergence requirement is not met in S7, update the sample database in S4 and return to S5 for iteration; if the convergence requirement is met, output the current optimal mooring configuration.

2. The optimization method for offshore floating wind turbine mooring system based on surrogate model according to claim 1, characterized in that, The key design parameters in S1 include mooring radius, mooring cable length, ballast block density, and ballast block position.

3. The optimization method for offshore floating wind turbine mooring system based on surrogate model according to claim 1, characterized in that, The objective function in S3 is obtained by multiplying the dimensionless index of mooring line tension, dimensionless index of anchor point tension, dimensionless index of motion constraint, and dimensionless index of cost by their respective weight coefficients and then summing them up, with the sum of all weight coefficients being 1.

4. The optimization method for offshore floating wind turbine mooring system based on surrogate model according to claim 3, characterized in that, The calculation of the dimensionless index of the motion constraint includes the following sub-steps: S3.1: Perform a fast Fourier transform on the six-degree-of-freedom motion time series of the floating wind turbine to obtain the motion spectrum; S3.2: Select spectral peaks from the motion spectrum that are greater than five percent of the maximum peak value; S3.3: Sum the peak values ​​of the selected spectral peaks to obtain representative values ​​for the six degrees of freedom: sway, heave, roll, pitch, and yaw. S3.4: Calculate the dimensionless index of the motion constraint based on the representative values ​​of the six degrees of freedom and their respective reference values.

5. The optimization method for offshore floating wind turbine mooring system based on surrogate model according to claim 3, characterized in that, The dimensionless cost index is calculated as follows: the difference between the current mooring cable mass and the reference mooring cable mass is divided by twice the reference mooring cable mass to obtain the first term; the difference between the current counterweight mass and the reference counterweight mass is divided by twice the reference counterweight mass to obtain the second term; the dimensionless cost index is the sum of the first term and the second term.

6. The optimization method for offshore floating wind turbine mooring system based on surrogate model according to claim 1, characterized in that, The numerical simulation tool in S4 uses the following model: The aerodynamic forces on wind turbine blades were simulated using the blade element-momentum method. Hydrodynamic loads on floating wind turbines are calculated using potential flow theory and Morrison's equations. The displacement and tension of the mooring line are calculated using the lumped mass method.

7. The optimization method for offshore floating wind turbine mooring system based on surrogate model according to claim 1, characterized in that, The Kriging method in S5 uses a target function whose model prediction is composed of a regression term consisting of a basis function vector and a regression coefficient vector, which is then added to a stationary Gaussian process with a mean of zero.

8. The optimization method for offshore floating wind turbine mooring system based on surrogate model according to claim 1, characterized in that, The differential evolution algorithm in S6 specifically includes: setting the population size to 10d to 20d, the mutation rate to 0.5 to 0.9, and the crossover rate to 0.6 to 0.95, where d is the dimension of the continuous variable in S1.

9. The optimization method for offshore floating wind turbine mooring system based on surrogate model according to claim 1, characterized in that, The convergence requirement in S7 is that the absolute error between the actual fitness value and the fitness value predicted by the surrogate model is less than five percent.

10. The optimization method for offshore floating wind turbine mooring system based on surrogate model according to claim 1, characterized in that, The mooring radius in the key design parameter S1 is a plurality of candidate values; the optimization process from S1 to S8 is executed once for each candidate mooring radius to obtain the optimal mooring configuration for each; the method is further defined as follows: from all the obtained optimal mooring configurations, the mooring scheme with the best overall performance is selected as the global optimal solution.

Images