A wind turbine blade limit load analysis method, system, medium, and apparatus

Abstract

The application provides a wind turbine blade limit load analysis method, system, medium and equipment, and belongs to the field of wind turbine blade load analysis. The method comprises the following steps: building a wind turbine model to simulate limit working condition load; determining the limit characteristic load, azimuth angle and pitch angle of each section of the blade according to the simulation data; determining the section oscillation moment based on the azimuth angle and the pitch angle, and taking the section oscillation moment as the gravity load of the blade section; judging whether the gravity load component is dominant, and if not, determining the limit load according to the limit characteristic load, and if yes, determining the first blade section limit load based on the gravity load and other types of load; reducing the gravity acceleration by using the safety factor of the gravity load, determining the optimal reduction coefficient by using the adaptive constraint genetic algorithm based on the Kriging proxy model, and determining the blade section limit load based on the optimal reduction coefficient. The application realizes the lightweight design of the blade under the premise of ensuring the safety margin.

Description

Technical Field

[0001] This invention belongs to the field of wind turbine blade load analysis technology, and particularly relates to a method, system, medium and equipment for ultimate load analysis of wind turbine blades. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] As the installed capacity of wind power expands year by year, the single-unit capacity of wind turbines continues to increase. As the core load-bearing component of the unit, the accuracy of the load calculation of the blades directly determines the level of blade lightweighting and the overall cost of the turbine. Wind turbine blades are subjected to the combined effects of gravity and inertial loads, aerodynamic loads, control operation loads, and other loads (such as wake loads and ice loads). The ultimate load is the core input basis for blade strength verification and ply structure design. The rationality of the load calculation scheme directly affects the reliability of the unit's operation and its market competitiveness.

[0004] Traditional blade ultimate load calculations include post-processing methods. These methods statistically analyze load data obtained through load simulation software and, based on design requirements in the IEC 61400-1 standard, compensate for the statistically obtained ultimate characteristic loads of each blade section using a uniform safety factor to ensure safe design, considering load uncertainties. However, this traditional approach struggles to comprehensively and accurately reflect the load characteristics under actual operating conditions, especially under gravity-dominated conditions. It fails to consider the basis and methods for determining the gravity load safety factor as specified in the standard, potentially leading to insufficient design margins or over-design. This can result in blade strength failing to meet requirements or increased blade cost and weight, ultimately affecting the reliability of the wind turbine. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, the present invention provides a method, system, medium and equipment for analyzing the ultimate load of wind turbine blades, which can achieve efficient and accurate analysis of the ultimate load of blades considering the safety factor of gravity load, avoid the safety factor value being too conservative, and realize the lightweight design of blades.

[0006] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions: The first aspect of this invention provides a method for analyzing the ultimate load of wind turbine blades; A method for analyzing the ultimate load of wind turbine blades includes: Build a wind turbine model to simulate the load under extreme operating conditions of the wind turbine; Based on the load data obtained from the simulation, the ultimate characteristic load of each section of the blade and the blade azimuth and pitch angle at the moment when the ultimate characteristic load occurs are determined; based on the azimuth and pitch angle, the swaying moment of each section of the unit in a static state is determined as the gravity load of the blade section. Determine whether the gravity load of the blade section dominates the ultimate characteristic load of the blade section. If not, determine the ultimate load of the blade section based on the ultimate characteristic load of each blade section. If so, determine the swaying moment of each section by applying the corresponding safety factor for the gravity load and other types of loads, and record it as the ultimate load of the first blade section. Gravitational acceleration is reduced by a safety factor for gravity load, and the optimal reduction coefficient is determined by an adaptive constrained genetic algorithm based on the Kriging surrogate model. The ultimate load of the blade section is then determined based on the optimal reduction coefficient.

[0007] As a further technical solution, based on the load data obtained from simulation, the ultimate characteristic load of each section of the blade and the blade azimuth angle and pitch angle at the time of occurrence of the ultimate characteristic load are determined, including: According to the characteristic load value rules specified in the design standard, the blade section oscillation moment under each extreme working condition is statistically processed to obtain the characteristic loads under different working conditions at each section. The maximum value of the characteristic loads at each section is taken as the ultimate characteristic load of the blade section, and the blade azimuth angle and pitch angle corresponding to the ultimate characteristic load are obtained simultaneously from the simulation results.

[0008] As a further technical solution, the ultimate characteristic load of each section of the blade is subtracted from the gravity load component, and the gravity load component is compared with the subtraction result. If the gravity load component is greater than the subtraction result, it indicates that the gravity load dominates the characteristic load; otherwise, it indicates that the gravity load does not dominate the characteristic load.

[0009] As a further technical solution, when the gravity load component does not dominate, the ultimate characteristic load of each section of the blade is uniformly amplified using the first safety factor, and the amplified load value is used as the ultimate load of the blade section. When the gravity load component is dominant, the gravity load of the blade section is amplified using a second safety factor, and the other types of loads other than the gravity load in the ultimate characteristic load of the blade section are amplified using a first safety factor. The two are superimposed to obtain the oscillation moment of each section, which is used as the ultimate load of the first blade section.

[0010] As a further technical solution, the adaptive constrained genetic algorithm based on the Kriging surrogate model specifically includes: Initial sample points are generated within the reduction coefficient range using Latin hypercube sampling. High-fidelity load simulation is performed on each sample point to construct an initial sample library. Kriging surrogate models are constructed based on the sample library, including objective function surrogate models and constraint function surrogate models; The optimal candidate point is obtained by iteratively optimizing the surrogate model using an adaptive constrained genetic algorithm. New sample points are selected using the constraint expectation improvement criterion, high-fidelity simulation verification is performed, the surrogate model is updated, and the process is iterated until convergence.

[0011] As a further technical solution, the prediction formula of the Kriging proxy model is as follows:

[0012] in, This represents the Kriging model prediction at point x; This represents the reduction factor for the gravitational acceleration to be predicted; This is a vector of regression coefficients; The vector of regression basis functions; This is the correlation vector, reflecting the spatial correlation between the point to be predicted and each sample point; This is the correlation matrix; For the response vector; This is the regression basis function matrix. The prediction variance is:

[0013] in, Mean square error; For process variance; For auxiliary vectors; This is the Fisher information matrix for estimating the regression coefficients.

[0014] The relevant kernel uses a Gaussian kernel:

[0015] in, For sample points , The correlation coefficient between them; These are hyperparameters in the relevant function.

[0016] As a further technical solution, the calculation formula for the constraint expectation improvement criterion is as follows:

[0017] in, To constrain the expected improvement, we express the expected improvement that can be obtained by performing a high-fidelity simulation at the design variables, taking into account the constraints. This represents the expected improvement relative to the current optimal solution that can be obtained by performing high-fidelity simulation at the design variables. Output the mean of the predicted constraint values ​​for the surrogate model of the constraint function; The standard deviation of the predictions for the surrogate model of the constraint function; The formula for improving the unconstrained expectation index (EI) is:

[0018] in, This represents the expected improvement relative to the current optimal solution that can be obtained by performing high-fidelity simulation at the design variables. This is the minimum value of the ultimate load on the second blade section; The standard deviation of the predictions of the surrogate model for the objective function; It follows a standard normal cumulative distribution; It represents the standard normal probability density.

[0019] The second aspect of the present invention provides a system for analyzing the ultimate load of wind turbine blades.

[0020] A wind turbine blade ultimate load analysis system, comprising: The operating condition simulation module is configured to: build a wind turbine model and simulate the load under extreme operating conditions of the wind turbine. The gravity load extraction module is configured to: determine the ultimate characteristic load of each section of the blade and the blade azimuth and pitch angle at the moment of occurrence of the ultimate characteristic load based on the load data obtained from the simulation; and determine the oscillation moment of each section of the unit in a static state based on the azimuth and pitch angle, as the gravity load of the blade section. The dominant judgment and first load calculation module is configured to: determine whether the section gravity load is dominant in the ultimate characteristic load of the blade section; if not, determine the ultimate load of the blade section according to the ultimate characteristic load of each section of the blade; if so, determine the oscillation moment of each section by applying the corresponding safety factor to the gravity load and other types of loads, and record it as the first blade section ultimate load. The surrogate model optimization module is configured to: reduce gravitational acceleration using a safety factor for the gravity load, determine the optimal reduction coefficient using an adaptive constrained genetic algorithm based on the Kriging surrogate model, and determine the ultimate load of the blade section based on the optimal reduction coefficient.

[0021] A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in the method for analyzing the ultimate load of a wind turbine blade as described in the first aspect of the present invention.

[0022] A fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the wind turbine blade ultimate load analysis method as described in the first aspect of the present invention.

[0023] The above one or more technical solutions have the following beneficial effects: (1) This invention uses the blade azimuth angle and pitch angle at the moment of occurrence of the ultimate characteristic load to reverse simulate the pure gravity load of the unit in a static state, and realizes the accurate separation of the gravity component and other types of loads in the ultimate characteristic load of the blade section. On this basis, the direction comparison and dual proportion judgment criteria are introduced to overcome the technical defects of the traditional method that cannot quantify the proportion of gravity load and blindly adopt a uniform safety factor, and provide an accurate decision basis for the scientific application of differentiated safety factors.

[0024] (2) Under the condition where gravity load is dominant, this invention abandons the conservative approach of applying the same safety factor to all loads in the traditional method. Instead, it uses different safety factors for gravity load and other loads according to the IEC 61400-1:2019 standard, and achieves the physical embedding of the partial factors through the equivalent method of reducing gravity acceleration. This ensures that the safety margin is not lower than the standard requirements, and avoids the over-design caused by uniform scaling. Under the premise of meeting the blade strength requirements, it effectively reduces the blade weight and the overall cost of the machine.

[0025] (3) To address the technical challenge of high-fidelity load simulation computation during the optimization of the gravitational acceleration reduction coefficient, this invention proposes an adaptive constrained genetic algorithm (KAC-GA) based on the Kriging surrogate model. This algorithm constructs an initial sample library through Latin hypercube sampling, approximates the true response using the Kriging surrogate model, and adaptively adds points using the Constraint Expectation Improvement (CEI) criterion. It only requires high-fidelity simulation to perform on a small number of the most valuable candidate points to converge to the optimal solution. Compared with the standard genetic algorithm, it reduces the number of simulation calls, making the optimization time, which was originally unacceptable in engineering practice, feasible.

[0026] (4) Based on the optimization output of the surrogate model, this invention further constructs a closed loop for comparison and verification between the ultimate load of the second blade section and the ultimate load of the first blade section. By dynamically adjusting the reduction factor until the second load is greater than the first load, it ensures that the final output ultimate load of the blade section meets the design requirements of the IEC standard while retaining an additional safety margin. This dual guarantee mechanism effectively avoids the safety risks that may be caused by the prediction error of the surrogate model and significantly improves the reliability and engineering confidence of the blade design.

[0027] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0028] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0029] Figure 1 This is a flowchart of the method in the first embodiment.

[0030] Figure 2 This is a system structure diagram of the second embodiment. Detailed Implementation

[0031] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, 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 invention pertains.

[0032] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0033] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0034] Example 1 This embodiment discloses a method for analyzing the ultimate load of wind turbine blades. The ultimate characteristic load and corresponding blade attitude are obtained through simulation. The pure gravity load is solved separately and the load dominance type is determined. The gravity-dominant working condition is calculated by reducing the equivalent partial safety factor of gravity acceleration. After iterative verification, the compliant load is output, realizing the lightweight and precise design of the blade.

[0035] like Figure 1 As shown, a method for analyzing the ultimate load of wind turbine blades includes: Step S1: Build a wind turbine model to simulate the load under extreme operating conditions of the wind turbine.

[0036] A wind turbine model was built using load simulation software to simulate the loads under various normal design limit load conditions of the turbine.

[0037] In this embodiment, a simulation model can be built using Bladed software to simulate the various normal design limit load conditions specified in the wind turbine design requirements. By simulating various conditions, load data under each condition can be obtained, such as the bending moment in the direction of yaw at the blade root and other locations, which can be used for subsequent calculation of the ultimate load of the blade section.

[0038] Step S2: Based on the load data obtained from the simulation, determine the ultimate characteristic load of each section of the blade and the blade azimuth angle and pitch angle at the time when the ultimate characteristic load occurs; based on the azimuth angle and pitch angle, determine the oscillation moment of each section of the unit in a static state, as the gravity load of the blade section.

[0039] Step S21, according to the IEC 61400-1:2019 Wind Turbine Generator Design Requirements, for load conditions with specific deterministic wind field events, the characteristic load is the calculated worst-case instantaneous value. If multiple simulations are performed at a given wind speed (represented by the rotor azimuth angle), the characteristic load is the average of the calculated worst-case instantaneous values ​​at each azimuth angle. If the inflow condition of the load condition is turbulent wind, the characteristic load is the average of the worst-case values ​​calculated under different 10-minute random turbulent wind conditions at the same wind speed. Under DLC2.1, DLC2.2, and DLC5.1 conditions, the characteristic load is the average of half of the maximum load.

[0040] In this embodiment, the characteristic loads of the shimmy moment of each section of the three blades are first classified and statistically analyzed according to standard requirements for each ultimate load condition. Then, the characteristic loads of the three blades are sorted in descending order, and the largest moment is selected and recorded as the ultimate characteristic load of the blade section. Meanwhile, in the simulation results of the ultimate characteristic load at the blade root position, the azimuth angle of each blade when the ultimate load occurs was found. and pitch angle .

[0041] Step S22: Simulate the blade load when the unit is in a stopped and stationary state. In the simulation software, set the inflow velocity to 0 and set the azimuth angle of each blade in the stationary state to [value missing]. The blade pitch angle is The oscillation moment of each section of the blade at this time was calculated. Under static conditions, since the incoming air velocity is 0, the blade is not affected by external aerodynamic loads, nor by inertial loads or operating loads. Therefore, the load on the blade section under static conditions is the gravity load generated by the blade itself.

[0042] Step S3: Determine whether the gravity load of the blade section dominates the ultimate characteristic load of the blade section. If not, determine the ultimate load of the blade section according to the ultimate characteristic load of each section. If so, determine the swaying moment of each section by using the corresponding safety factor for the gravity load and other types of loads, and record it as the ultimate load of the first blade section.

[0043] Step S31, ultimate characteristic load of blade section It includes gravity loads and other types of loads, which can be obtained through the steps above. Gravity load component The corresponding time is the time when the ultimate characteristic load at the leaf root position occurs. If this condition is met, it means that gravity load dominates the characteristic load at this time.

[0044] The blade section gravity load component obtained in step S22 Compared with the ultimate characteristic load of the blade section obtained in step S21 Perform orientation comparison. If... and If the directions are the same, the gravity load is determined to be an unfavorable load under this extreme condition; if the directions are opposite, the gravity load cancels out the ultimate load, and based on conservative design principles, the gravity load is not treated separately. For cases where the gravity load is not an unfavorable load, the ultimate characteristic loads of each section of the blade are... The first safety factor is adopted uniformly. The load value is magnified and used as the ultimate load of the wind turbine blade section under this condition. ,Right now:

[0045] Step S32: When the gravity load is an unfavorable load, further determine whether the gravity load dominates among the characteristic loads. The determination of dominance uses a dual condition: First, determine whether the absolute value of the gravity load component is greater than the absolute value of the ultimate characteristic load, i.e. Is this condition met? If so, it indicates that the gravity load component has exceeded the total characteristic load. In this case, according to the IEC standard's provisions on the safety factor when gravity load is dominant, the ultimate characteristic load of each blade section is directly output and multiplied by the safety factor corresponding to the gravity load dominant condition. This serves as the ultimate load on the blade section of the wind turbine under this condition.

[0046] If not satisfied Then further judgment Does this condition hold? The left side of the inequality represents the absolute value of the gravity load component, and the right side represents the absolute value of the non-gravity load component (i.e., the difference between the total characteristic load and the gravity load component). If this condition holds, it means that the gravity load component accounts for more than 50% of the characteristic load, and it can be determined that the gravity load is dominant in the characteristic load; otherwise, it can be determined that the gravity load is not dominant.

[0047] The logical basis for judging the dominant position mentioned above is that the higher the proportion of gravity load, the more significant the reduction effect of a safety factor value below 1.35 on the ultimate characteristic load. Setting a 50% proportion threshold as the judgment criterion is a more conservative approach within the standard framework, and can also be used as an internal control standard for enterprises to achieve a balance between safety and economy.

[0048] Step S33: If, based on the above judgment, gravity load dominates the characteristic load, then according to the calculation formula for the safety factor of gravity load in IEC 61400-1:2019, the gravity load component and other types of load components are treated differently using corresponding safety factors. Specifically, for normal design load conditions, the safety factor corresponding to gravity load is... Take 1.1, and the safety factor for other types of loads. Take 1.35. Multiply the two types of loads by their respective safety factors and then sum them to obtain the swaying moment at each section, which is denoted as the ultimate load of the first blade section. :

[0049] Step S4: The gravitational acceleration is reduced by the safety factor of the gravity load, the optimal reduction coefficient is determined by the adaptive constrained genetic algorithm based on the Kriging surrogate model, and the ultimate load of the blade section is determined based on the optimal reduction coefficient.

[0050] Because the extreme conditions are re-simulated using reduced gravitational acceleration, each simulation requires extensive time-domain load calculations and post-processing for the DLC conditions, resulting in a lengthy evaluation time. If standard genetic algorithms are used directly for iterative optimization, hundreds of simulation evaluations are needed, making the computational cost unacceptable in engineering practice. Therefore, this embodiment proposes an adaptive constrained genetic algorithm (KAC-GA) based on the Kriging surrogate model, which significantly reduces the number of high-fidelity simulation calls while maintaining optimization accuracy.

[0051] Step S41, specifically, firstly, the optimization problem of the reduction coefficient is initially set as a single-objective constrained optimization problem, and the optimization variable is set as the gravitational acceleration reduction coefficient. Its value space is The optimization objective is to minimize the ultimate load on the second blade section, i.e. By substituting into the reduction function The result is obtained by multiplying the simulation by a safety factor of 1.35. The constraint condition is that the ultimate load on the second blade section must be greater than the ultimate load on the first blade section to ensure a safety margin. .

[0052] Step S42, followed by Latin hypercube sampling in Multiple initial sample points are generated within the interval, denoted as... In a one-dimensional design space, 8-15 sample points can be selected, which captures the basic trend of the objective function while keeping the initial simulation cost acceptable.

[0053] For each sample point Perform a complete high-fidelity load simulation process, setting the reduced gravitational acceleration as... Time-domain simulations of all normal design ultimate load conditions were completed, and the corresponding results were obtained through post-processing and statistical analysis. Simultaneously, it is directly calculated from step S32. Calculate the constraint function value The obtained , , Store the training dataset to build the initial sample library.

[0054] Step S43: Construct the Kriging model based on the sample library, as shown below:

[0055] in, This represents the Kriging model prediction at point x; This represents the reduction factor for the gravitational acceleration to be predicted; This is a vector of regression coefficients; The vector of regression basis functions; This is the correlation vector, reflecting the spatial correlation between the point to be predicted and each sample point; This is the correlation matrix; For the response vector; This is the regression basis function matrix. The prediction variance is:

[0056] in, Mean square error; For process variance; For auxiliary vectors; This is the Fisher information matrix for estimating the regression coefficients.

[0057] The relevant kernel uses a Gaussian kernel:

[0058] in, For sample points , The correlation coefficient between them; These are hyperparameters in the relevant function.

[0059] In this embodiment, the reduction factor is used. To share input, the second limit load and constraint margin were used as training labels, and their hyperparameters were independently estimated using the same Kriging modeling process, thus constructing the objective function surrogate model in parallel. and constraint function surrogate model The objective function surrogate model outputs the predicted mean of the second limit load. ,variance The constrained function surrogate model outputs the mean of the predicted constraint values. ,variance .

[0060] Step S44, Proxy model of objective function and constraint function surrogate model An iterative genetic population is established using a generation-constrained genetic algorithm, with a population size of 50-100 and an individual reduction factor. .

[0061] Then, the fitness function is constructed, with the original objective being... ,constraint Fitness is constructed using an external penalty function:

[0062] in, The fitness value is the reduction factor; The coefficient is a penalty factor that significantly increases fitness when constraints are not met, eliminating infeasible individuals.

[0063] Next, GA genetic operations are performed, using tournament selection to retain individuals with low fitness; real-valued single-point crossover and Gaussian small-perturbation mutation are performed to ensure that the variables are within the normal range; finally, through internal iteration, the optimal candidate point at the surrogate model level is output. .

[0064] Step S45: Relying solely on the surrogate model's optimal point results in prediction errors. Therefore, the constrained expectation CEI criterion is further used to quantify the added benefit for the optimal candidate point. Using the CEI criterion as a benchmark, a comprehensive evaluation is conducted from two dimensions: "potential for improvement of the objective" and "probability of feasibility of constraints". Based on the validation value of each point globally, the point with the highest CEI is selected as a new sample point. This achieves a balance between developing the optimal and exploring uncertainty, ensuring that the most valuable points are verified in each iteration and avoiding erroneous convergence due to surrogate model errors.

[0065] The formula for unconstrained expected improvement of EI is:

[0066] in, This represents the expected improvement relative to the current optimal solution that can be obtained by performing high-fidelity simulation at the design variables. This is the minimum value of the ultimate load on the second blade section; The standard deviation of the predictions of the surrogate model for the objective function; It follows a standard normal cumulative distribution; It represents the standard normal probability density.

[0067] CEI (Constraint Expected Improvement) with introduced constraints: Only calculate the constraint satisfaction Expected improvements in the region:

[0068] in, To constrain the expected improvement, we express the expected improvement that can be obtained by performing a high-fidelity simulation at the design variables, taking into account the constraints. This represents the expected improvement relative to the current optimal solution that can be obtained by performing high-fidelity simulation at the design variables. Output the mean of the predicted constraint values ​​for the surrogate model of the constraint function; This is the prediction standard deviation of the surrogate model for the constraint function.

[0069] Solve within the interval The reduction factor corresponding to the maximum value Add new sample points; if the maximum CEI is less than the convergence threshold, it means the surrogate model has sufficiently approximated the true response, and no more samples are needed, thus entering the convergence output stage; otherwise, add new samples. It is then fed into a high-fidelity simulation.

[0070] Step S46, for the selected Perform high-fidelity load simulation of the complete wind turbine under extreme operating conditions to calculate the actual load. and actual constraint values .

[0071] Will , , Add the sample library, and using the expanded full sample library, adaptively reconstruct a new surrogate model to eliminate the prediction bias of the old model. Based on the updated surrogate model, perform constrained GA optimization again, iterating cyclically. When the iteration meets the termination condition or reaches the maximum number of iterations, output the optimal reduction coefficient. The optimal reduction factor Substitute the values ​​and output the optimal ultimate load of the second blade section as the final design load.

[0072] Example 2 This embodiment discloses a wind turbine blade ultimate load analysis system; like Figure 2 As shown, a wind turbine blade ultimate load analysis system includes: The operating condition simulation module is configured to: build a wind turbine model and simulate the load under extreme operating conditions of the wind turbine. The gravity load extraction module is configured to: determine the ultimate characteristic load of each section of the blade and the blade azimuth and pitch angle at the moment of occurrence of the ultimate characteristic load based on the load data obtained from the simulation; and determine the oscillation moment of each section of the unit in a static state based on the azimuth and pitch angle, as the gravity load of the blade section. The dominant judgment and first load calculation module is configured to: determine whether the section gravity load is dominant in the ultimate characteristic load of the blade section; if not, determine the ultimate load of the blade section according to the ultimate characteristic load of each section of the blade; if so, determine the oscillation moment of each section by applying the corresponding safety factor to the gravity load and other types of loads, and record it as the first blade section ultimate load. The surrogate model optimization module is configured to: reduce gravitational acceleration using a safety factor for the gravity load, determine the optimal reduction coefficient using an adaptive constrained genetic algorithm based on the Kriging surrogate model, and determine the ultimate load of the blade section based on the optimal reduction coefficient.

[0073] Example 3 The purpose of this embodiment is to provide a computer-readable storage medium.

[0074] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the wind turbine blade ultimate load analysis method as described in Example 1.

[0075] Example 4 The purpose of this embodiment is to provide an electronic device.

[0076] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the wind turbine blade ultimate load analysis method as described in Example 1.

[0077] The steps and methods involved in the apparatuses of Embodiments 2, 3, and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0078] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0079] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for analyzing the ultimate load of wind turbine blades, characterized in that, include: Build a wind turbine model to simulate the load under extreme operating conditions of the wind turbine; Based on the load data obtained from the simulation, the ultimate characteristic load of each section of the blade and the blade azimuth and pitch angle at the moment when the ultimate characteristic load occurs are determined; based on the azimuth and pitch angle, the swaying moment of each section of the unit in a static state is determined as the gravity load of the blade section. To determine whether the gravity load dominates the ultimate characteristic load of the blade section, specifically: subtract the ultimate characteristic load of each blade section from the gravity load component, and compare the gravity load component with the subtraction result. If the gravity load component is greater than the subtraction result, it indicates that the gravity load dominates the characteristic load. At this time, the corresponding safety factor is used to determine the oscillation moment of each section for the gravity load and other types of loads, which is recorded as the ultimate load of the first blade section. Conversely, if the gravity load does not dominate the characteristic load, then the ultimate load of the blade section is determined based on the ultimate characteristic load of each section of the blade. Gravitational acceleration is reduced by using a safety factor for gravity load, and the optimal reduction coefficient is determined by an adaptive constrained genetic algorithm based on the Kriging surrogate model. The ultimate load of the blade section is then determined based on the optimal reduction coefficient. The adaptive constrained genetic algorithm based on the Kriging surrogate model specifically includes: The optimization problem of the reduction coefficient is initially set as a single-objective constrained optimization problem. The optimization variable is set as the gravitational acceleration reduction coefficient. The optimization objective is to minimize the ultimate load of the second blade section. The ultimate load of the second blade section is obtained by substituting it into the reduction function, resimulating it, and then multiplying it by the corresponding safety factor. The constraint condition is that the ultimate load of the second blade section is greater than the ultimate load of the first blade section to ensure a safety margin. Latin hypercube sampling is used to generate initial sample points within the reduction coefficient range. High-fidelity load simulation is performed on each sample point to construct an initial sample library. Kriging surrogate models are constructed based on the sample library, including objective function surrogate models and constraint function surrogate models; The optimal candidate point is obtained by iteratively optimizing the surrogate model using an adaptive constrained genetic algorithm. New sample points are selected using the constraint expectation improvement criterion, high-fidelity simulation verification is performed, the surrogate model is updated, and the process is iterated until convergence.

2. The method for analyzing the ultimate load of wind turbine blades as described in claim 1, characterized in that, Based on the load data obtained from the simulation, the ultimate characteristic loads of each blade section and the blade azimuth and pitch angles at the time of occurrence of the ultimate characteristic loads are determined, including: According to the characteristic load value rules specified in the design standard, the blade section oscillation moment under each extreme working condition is statistically processed to obtain the characteristic loads under different working conditions at each section. The maximum value of the characteristic loads at each section is taken as the ultimate characteristic load of the blade section, and the blade azimuth angle and pitch angle corresponding to the ultimate characteristic load are obtained simultaneously from the simulation results.

3. The method for analyzing the ultimate load of wind turbine blades as described in claim 1, characterized in that, When the gravity load component is not dominant, the ultimate characteristic load of each section of the blade is uniformly amplified using the first safety factor, and the amplified load value is used as the ultimate load of the blade section. When the gravity load component is dominant, the gravity load of the blade section is amplified using a second safety factor, and the other types of loads other than the gravity load in the ultimate characteristic load of the blade section are amplified using a first safety factor. The two are superimposed to obtain the oscillation moment of each section, which is used as the ultimate load of the first blade section.

4. The method for analyzing the ultimate load of wind turbine blades as described in claim 1, characterized in that, The prediction formula of the Kriging proxy model is: in, This represents the Kriging model prediction at point x; This represents the reduction factor for the gravitational acceleration to be predicted; This is a vector of regression coefficients; The vector of regression basis functions; This is the correlation vector, reflecting the spatial correlation between the point to be predicted and each sample point; This is the correlation matrix; For the response vector; The regression basis function matrix; The prediction variance is: in, Mean squared error; For process variance; For auxiliary vectors; Fisher information matrix for estimating regression coefficients; The relevant kernel uses a Gaussian kernel: in, For sample points , The correlation coefficient between them; These are hyperparameters in the relevant function.

5. The method for analyzing the ultimate load of wind turbine blades as described in claim 1, characterized in that, The formula for calculating the constraint expectation improvement criterion is as follows: in, To constrain the expected improvement, we express the expected improvement that can be obtained by performing a high-fidelity simulation at the design variables, taking into account the constraints. This represents the expected improvement relative to the current optimal solution that can be obtained by performing high-fidelity simulation at the design variables. Output the mean of the predicted constraint values ​​for the surrogate model of the constraint function; The standard deviation of the predictions for the surrogate model of the constraint function; The formula for improving the unconstrained expectation index (EI) is: in, This represents the expected improvement relative to the current optimal solution that can be obtained by performing high-fidelity simulation at the design variables. This is the minimum value of the ultimate load on the second blade section; The standard deviation of the predictions of the surrogate model for the objective function; It follows a standard normal cumulative distribution; It represents the standard normal probability density.

6. A wind turbine blade ultimate load analysis system, characterized in that, include: The operating condition simulation module is configured to: build a wind turbine model and simulate the load under extreme operating conditions of the wind turbine. The gravity load extraction module is configured to: determine the ultimate characteristic load of each section of the blade and the blade azimuth and pitch angle at the moment of occurrence of the ultimate characteristic load based on the load data obtained from the simulation; and determine the oscillation moment of each section of the unit in a static state based on the azimuth and pitch angle, as the gravity load of the blade section. The dominant judgment and first load calculation module is configured to: determine whether the gravity load of the blade section dominates the ultimate characteristic load; specifically: subtract the ultimate characteristic load of each blade section from the gravity load component, compare the gravity load component with the subtraction result; if the gravity load component is greater than the subtraction result, it indicates that the gravity load dominates the characteristic load, and at this time, the corresponding safety factor is used to determine the oscillation moment of each section for the gravity load and other types of loads, which is recorded as the first blade section ultimate load; otherwise, it indicates that the gravity load does not dominate the characteristic load, and the ultimate load of the blade section is determined according to the ultimate characteristic load of each blade section. The surrogate model optimization module is configured to: reduce the gravitational acceleration using a safety factor for the gravity load, determine the optimal reduction coefficient using an adaptive constrained genetic algorithm based on the Kriging surrogate model, and determine the ultimate load of the blade section based on the optimal reduction coefficient; The adaptive constrained genetic algorithm based on the Kriging surrogate model specifically includes: The optimization problem of the reduction coefficient is initially set as a single-objective constrained optimization problem. The optimization variable is set as the gravitational acceleration reduction coefficient. The optimization objective is to minimize the ultimate load of the second blade section. The ultimate load of the second blade section is obtained by substituting it into the reduction function, resimulating it, and then multiplying it by the corresponding safety factor. The constraint condition is that the ultimate load of the second blade section is greater than the ultimate load of the first blade section to ensure a safety margin. Latin hypercube sampling is used to generate initial sample points within the reduction coefficient range. High-fidelity load simulation is performed on each sample point to construct an initial sample library. Kriging surrogate models are constructed based on the sample library, including objective function surrogate models and constraint function surrogate models; The optimal candidate point is obtained by iteratively optimizing the surrogate model using an adaptive constrained genetic algorithm. New sample points are selected using the constraint expectation improvement criterion, high-fidelity simulation verification is performed, the surrogate model is updated, and the process is iterated until convergence.

7. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the method for analyzing the ultimate load of wind turbine blades as described in any one of claims 1-5.

8. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the wind turbine blade ultimate load analysis method as described in any one of claims 1-5.

Images