A safety assessment method for a wind turbine

Abstract

The application discloses a kind of safety evaluation methods of wind turbine generator set, it is related to wind power generator technical field, including the monitoring data of generator blade, the monitoring data includes wind speed, turbulence intensity and temperature, the edge distribution of the monitoring data is coupled by mixed Copula function to the fitting of the monitoring data;According to the shape parameter of generator set, establish three-dimensional model, carry out a variety of scene simulation by analysis software, generate the maximum bending moment of wing root under different scenes, obtain the mapping relationship of scene-bending moment;Based on the mapping relationship of scene-bending moment.The method can more accurately capture the influence of environmental changes on blade bending moment by combining real-time monitoring data such as wind speed, turbulence intensity, temperature and aerodynamic principles, thereby significantly improving the accuracy of safety evaluation.

Description

Technical Field

[0001] This invention relates to the field of wind turbine technology, and in particular to a method for safety assessment of wind turbine generator sets. Background Technology

[0002] Wind power has become an important part of global energy production. As the core equipment for wind energy conversion, the safety and reliability of wind turbine generators directly affect the overall efficiency and economy of wind power systems. Therefore, assessing the safety of wind turbine generators, especially the generator blades, is particularly important.

[0003] The safety assessment of wind turbine generators primarily relies on the analysis of blade stress and fatigue. Traditional assessment methods are typically based on structural mechanics models, combined with common load models and environmental conditions. However, these traditional methods often exhibit significant uncertainties when facing complex and variable environmental conditions, especially the impact of factors such as wind speed, turbulence intensity, and temperature on blade performance. Particularly in nonlinear environments where wind speed, turbulence intensity, and temperature interact, the predictive accuracy of traditional models is low. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for safety assessment of wind turbine generator sets.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a safety assessment method for wind turbine generator sets, comprising: collecting monitoring data of the generator blades, the monitoring data including wind speed, turbulence intensity, and temperature; fitting the monitoring data; coupling the edge distribution of the monitoring data through a hybrid Copula function; establishing a three-dimensional model based on the shape parameters of the generator set; performing various scenario simulations using analysis software to generate the maximum bending moment at the blade root under different scenarios, obtaining the scenario-bending moment mapping relationship; based on the scenario-bending moment mapping relationship, combining the spatiotemporal correlation features of the monitoring data extracted from historical monitoring data, and analyzing historical data... The influence of temperature changes on generator blades was investigated to determine correction factors. A Gaussian process regression model integrating aerodynamic principles was constructed and trained. Historical monitoring data was input into the trained Gaussian process regression model to obtain several bending moment prediction results. Samples close to the failure threshold were selected, and multiple risk levels were set. Markov chain migration sampling was performed on samples at each risk level to dynamically adjust the failure threshold. The failure probabilities between levels were calculated and integrated. A safety threshold was set based on the failure probability. Real-time monitoring data was input into the trained Gaussian process regression model to calculate the current blade root bending moment value. The current blade root bending moment value was compared with the safety threshold to determine the current safety status of the generator set.

[0007] As a preferred embodiment of the safety assessment method for wind turbine generators described in this invention, the wind speed is fitted using a two-parameter Weibull distribution, the turbulence intensity is fitted using a truncated log-normal distribution, the temperature is fitted using a hybrid distribution model, and a joint probability model for non-stationary environments is established.

[0008] As a preferred embodiment of the safety assessment method for wind turbine generator sets described in this invention, the spatiotemporal correlation characteristics of the monitoring data extracted from historical monitoring data include the following steps: calculating the rate of change of wind speed, turbulence intensity, and temperature over time to obtain the time gradients of wind speed, turbulence intensity, and temperature; calculating the covariance between wind speed, turbulence intensity, and temperature to obtain the covariance matrix; and integrating the time gradients of wind speed, turbulence intensity, and temperature with each element in the covariance matrix to form spatiotemporal correlation characteristics.

[0009] As a preferred embodiment of the safety assessment method for wind turbine generator sets described in this invention, the determination of the correction factor includes the following steps: defining a model relating temperature and the elastic modulus of generator blade material; measuring and recording the elastic modulus of the generator blades at different temperatures to form a temperature-elastic modulus mapping relationship; and calculating the correction factor through a linear regression equation.

[0010] As a preferred embodiment of the safety assessment method for wind turbine generator sets described in this invention, the fused aerodynamic principle refers to defining a relationship model between bending moment and angle of attack based on the airfoil aerodynamic equation, defining that the partial derivative of bending moment with respect to angle of attack should satisfy aerodynamic theory, and using it as a physical regularization term for training the Gaussian process regression model.

[0011] As a preferred embodiment of the safety assessment method for wind turbine generators described in this invention, the Markov chain migration sampling refers to performing directional migration sampling through a gradient-sensitive strategy, and the dynamic adjustment of the failure threshold refers to readjusting the threshold based on the failure ratio of the samples after Markov chain migration.

[0012] As a preferred embodiment of the safety assessment method for wind turbine generator sets described in this invention, the kernel function of the Gaussian process regression model includes a radial basis function kernel function, a spatiotemporal correlation kernel function, and a temperature correction kernel function.

[0013] As a preferred embodiment of the safety assessment method for wind turbine generator sets described in this invention, after collecting the monitoring data, a preprocessing operation is first performed.

[0014] As a preferred embodiment of the safety assessment method for wind turbine generator sets described in this invention, the shape parameters include the blade length, width, thickness, material, curvature, blade torsion angle, and tower height.

[0015] As a preferred embodiment of the safety assessment method for wind turbine generator sets described in this invention, the calculation and integration of failure probabilities between levels refers to first calculating the failure probability of each level, and then summing the failure probabilities of each level based on exponential decay weights.

[0016] The beneficial effects of this invention are as follows: by combining real-time monitoring data such as wind speed, turbulence intensity, and temperature with aerodynamic principles, the influence of environmental changes on blade bending moment can be captured more accurately, thereby significantly improving the accuracy of safety assessment; by introducing real-time monitoring data and combining it with the Markov chain migration sampling method, the failure threshold can be dynamically adjusted, and the safety status of wind turbine generators can be assessed in real time based on actual environmental changes. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a safety assessment method for wind turbine generators.

[0019] Figure 2 This is a schematic diagram of the multimodal data fusion process for the safety assessment method of wind turbine generators.

[0020] Figure 3 A schematic diagram of the joint distribution modeling of environmental parameters for the safety assessment method of wind turbine generators.

[0021] Figure 4 A schematic diagram illustrating the construction of a Gaussian process regression model for safety assessment of wind turbine generators. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Example 1, referring to Figures 1-4 This is the first embodiment of the present invention, which provides a safety assessment method for wind turbine generator sets, including the following steps:

[0026] S1. Collect wind speed, turbulence intensity, and temperature data at positions 0.3R, 0.7R, and 0.9R at the leading edge of the blade, where R represents the blade radius. Calculate the average value of the collected wind speed, turbulence intensity, and temperature data at these positions. For example, average wind speed = (wind speed at 0.3R + wind speed at 0.7R + wind speed at 0.9R) / 3. After collecting the wind speed, turbulence intensity, and temperature data, synchronize the data in time. Then, use a Kalman filter algorithm to smooth the wind speed, turbulence intensity, and temperature data, removing noise introduced by sensor errors or external factors. Use the Z-Score method to detect outliers and remove data that does not conform to the normal range of variation. If data is missing, use linear interpolation or spline interpolation algorithms to supplement it.

[0027] The wind speed is fitted using a two-parameter Weibull distribution, and is expressed as follows:

[0028]

[0029] In the formula, u is the wind speed, λ is the scale parameter, and l is the shape parameter. Parameters λ and k need to be estimated through data fitting. Parameters λ and k are estimated by maximizing the log-likelihood function of the Weibull distribution, which is expressed as:

[0030]

[0031] In the formula, u i Let λ be the i-th wind speed data point and n be the number of data points. Preferably, the optimal values ​​of λ and k can be obtained through optimization algorithms.

[0032] We assume that the turbulence intensity follows a truncated log-normal distribution, and the probability density function (PDF) of the truncated log-normal distribution is:

[0033]

[0034] In the formula, y is the turbulence intensity, and μ and σ are the mean and standard deviation of the log-normal distribution.

[0035] Temperature may exhibit a multimodal distribution under different environmental conditions; therefore, a Gaussian mixture model (GMM) is used to fit the temperature data. The probability density function (PDF) of the Gaussian mixture distribution is:

[0036]

[0037] In the formula, π1 and π2 are the weights of the mixed components, μ1 and μ2 are the means of the two Gaussian distributions, and σ1 and σ2 are the standard deviations. For example, the temperature fitting steps are as follows: acquire temperature data and check the multimodal characteristics of the data; estimate the parameters π1, π2, μ1, μ2, σ1, and σ2 of the mixed distribution using the expectation-maximization (EM) algorithm; iterate repeatedly using the EM algorithm until the log-likelihood function converges.

[0038] After obtaining the marginal distributions of each monitored variable, the dependencies between these variables are modeled using a Copula function. This method chooses a hybrid Copula function to capture the nonlinear correlation between wind speed, turbulence intensity, and temperature. The hybrid Copula function has the following form:

[0039] C(u,v,q)=Φ ρ (Φ -1 (u),Φ -1 (v),Φ -1 (w))

[0040] In the formula, Φ ρ It is Gaussian Copula, Φ -1 () is the inverse CDF of the standard normal distribution, and u, v, w are the CDFs of the edge distributions of wind speed, turbulence intensity, and temperature, respectively.

[0041] By combining the marginal distributions of wind speed, turbulence intensity, and temperature using the Copula function, a joint probability distribution under non-stationary conditions can be obtained:

[0042] F(U,I t ,T)=C(f U (u),f Y (y),f T (μ1))

[0043] In the formula, f U (u), f Y (y) and f T (μ1) are the edge distribution functions for wind speed, turbulence intensity, and temperature, respectively.

[0044] The above steps establish a joint probability model for nonstationary environments based on hybrid Copula, which can effectively describe the interdependence between wind speed, turbulence intensity, and temperature.

[0045] S2. Establish a three-dimensional model based on the shape parameters of the generator set, perform various scenario simulations using analysis software, generate the maximum bending moment at the flange root under different scenarios, and obtain the mapping relationship between scenario and bending moment.

[0046] Specifically, the shape parameters of the wind turbine generator are acquired and determined. These parameters include length, width, thickness, material, curvature, blade torsion angle, and tower height. A 3D model of the wind turbine generator is created based on these shape parameters using design software (such as SolidWorks, AutoCAD, ANSYS, etc.). Then, CAE software (such as ANSYS, Simulink, etc.) is used to simulate different operating conditions of the wind turbine generator, generating the maximum bending moment at the blade root under different scenarios. The simulation results (maximum bending moment values ​​at the blade root for each wind speed, turbulence intensity, temperature, etc. scenario) are mapped to the corresponding scenario conditions (wind speed, turbulence intensity, temperature) to form a dataset. Regression analysis or interpolation methods can be used to construct the mapping relationship between scenarios and bending moments.

[0047] S3. Based on the scenario-bending moment mapping relationship, combined with the spatiotemporal correlation characteristics of the monitoring data extracted from historical monitoring data, and by analyzing the influence of historical temperature changes on generator blades to determine the correction factor, a Gaussian process regression model integrating aerodynamic principles is constructed and trained.

[0048] Specifically, the rate of change of wind speed, turbulence intensity, and temperature over time is calculated to obtain the time gradients of wind speed, turbulence intensity, and temperature.

[0049] The wind speed time gradient is expressed as:

[0050]

[0051] In the formula, U(t) is the wind speed at time t, and Δt is the time interval.

[0052] Turbulence intensity time gradient:

[0053]

[0054] In the formula, I t (t) represents the turbulence intensity at time t.

[0055] Temperature time gradient:

[0056]

[0057] In the formula, T(t) is the temperature value at time t.

[0058] The covariance between wind speed, turbulence intensity, and temperature is calculated to obtain the covariance matrix, which is expressed as follows:

[0059]

[0060] In the formula, Cov(X,Y) represents the covariance between variables X and Y, which measures the linear relationship between the two variables.

[0061] Integrating the temporal gradients of wind speed, turbulence intensity, and temperature with the elements of the covariance matrix, we form a spatiotemporal correlation feature, represented as:

[0062]

[0063] The determination of the correction factor includes the following steps.

[0064] Define a model relating temperature and the elastic modulus of generator blade material, expressed as follows:

[0065] E(T)=E0·(1-θ T ·(TT ref ))

[0066] In the formula, E0 is the elastic modulus at the reference temperature, and θ T T is the temperature coefficient. ref The standard reference temperature for material performance testing is usually 20℃, where T is the current temperature.

[0067] The elastic modulus of the generator blades was measured and recorded at different temperatures to establish a temperature-elastic modulus mapping relationship.

[0068] The correction factor is calculated using the linear regression equation and is expressed as follows:

[0069]

[0070] In the formula, The average experimental temperature. The measured mean elastic modulus is given by m, where m is the number of experimental samples, and T is the mean elastic modulus. i Let E be the temperature of the i-th sample. i Let be the elastic modulus of the i-th sample.

[0071] Using the correction factor as a quantitative indicator of the effect of temperature on material properties, a temperature correction kernel function is constructed.

[0072]

[0073] In the formula, T ′ T and T are two temperature sample points that are input.

[0074] The integrated aerodynamic principle includes the following steps.

[0075] Based on the airfoil aerodynamic equations, the relationship between bending moment and angle of attack is defined as follows:

[0076]

[0077] In the formula, ρ is the air density (obtained from environmental sensors), c is the blade chord length (geometric parameters of the three-dimensional model), U is the wind speed (input from monitoring data), and C... L (α) is a function of the lift coefficient as a function of the angle of attack α. This function is usually calibrated by wind tunnel experiments and is existing technology, so it is not limited here.

[0078] The partial derivative of the bending moment with respect to the angle of attack is defined to satisfy aerodynamic theory, expressed as follows:

[0079]

[0080] The partial derivative of bending moment with respect to angle of attack is used as the physical regularization term for training the Gaussian process regression model.

[0081] The kernel functions of the Gaussian process regression model include the radial basis function, the spatiotemporal correlation kernel function, and the temperature correction kernel function, expressed as follows:

[0082] k(x,x′)=k RBF (x,x′)+k ST (x,x′)+k T (T,T′)

[0083]

[0084] In the formula, k T (T,T′) is the temperature correction kernel function, k RBF (x,x′) is the radial basis kernel function, k ST (x,x′) is the spatiotemporal correlation kernel function, γ is the spatial decay coefficient (controlling the rate of decay of the correlation between geographical locations), and x space This refers to the geographical coordinates (latitude and longitude or relative position code) of the wind turbine.

[0085] Using historical monitoring data and the mapping relationship between scene and bending moment, a Gaussian process regression model is trained by maximizing the log-likelihood function.

[0086] S4. Input historical monitoring data into the trained Gaussian process regression model to obtain several bending moment prediction results. Select samples that meet the pre-set requirements and set multiple risk levels. Perform Markov chain migration sampling in the samples of each risk level, dynamically adjust the failure threshold, calculate the failure probability between levels, and integrate them.

[0087] Specifically, the failure threshold M t h r Defined as:

[0088] Mt h r =μ sim +3σ sim

[0089] In the formula, μ sim σ represents the average failure bending moment obtained from the three-dimensional simulation. sim This is expressed as the standard deviation of the simulation results.

[0090] Meeting the pre-defined requirements refers to samples near the failure threshold. The conditions can be set as follows:

[0091] S={M i |M thr -ΔM≤M i ≤M thr +ΔM}

[0092] In the formula, M i Let be the predicted flange root bending moment for the i-th sample (unit: kN·m), and ΔM be the buffer width, typically set to 0.2σ. sim .

[0093] Dividing risk levels refers to setting multiple risk ranges based on failure thresholds. In this embodiment, they are divided into four levels: safe zone, warning zone, risk zone, and high-risk zone.

[0094]

[0095] The Markov chain transfer sampling refers to performing targeted transfer sampling using a gradient-sensitive strategy, denoted as,

[0096]

[0097] In the formula, ∈ represents the migration step size, and ∈ = 0.05σ. sim , The gradient of the bending moment output by the model with respect to the input features is used to indicate the direction of the most sensitive change.

[0098] The dynamically adjusted failure threshold refers to readjusting the threshold based on the failure rate of samples after Markov chain migration, expressed as:

[0099]

[0100] In the formula, P fail The percentage of samples exceeding the limit after migration is η, and η is the adjustment rate used to prevent oscillations, η = 0.01σ. sim β is the preset failure tolerance, N fail N represents the number of samples exceeding the limit after migration. talal This represents the total number of samples after the migration.

[0101] The calculation and integration of failure probabilities between levels refers to first calculating the failure probability of each level, and then summing the failure probabilities of each level based on exponential decay weights.

[0102] The failure probability for each level is calculated as follows:

[0103]

[0104] In the formula, Ⅱ() represents the indicator function, and N k P is the number of samples at level k. k Let be the failure probability of the kth level.

[0105] Using exponentially decaying weights to integrate risks at each level, the risk representation is as follows:

[0106]

[0107] In the formula, γ is the attenuation coefficient used to control the weight distribution, |k-2| represents the degree of deviation from the safe zone, k is the current risk level, j is the index variable for traversing all levels, and w k As the weight, P total This represents the overall failure probability after integration.

[0108] S5. Based on the failure probability, a safety threshold is set. Real-time monitoring data is input into the trained Gaussian process regression model to calculate the current flange root bending moment value. The current flange root bending moment value is compared with the safety threshold to determine the current safety status of the generator set.

[0109] Specifically, first set a safety factor based on the overall failure probability.

[0110]

[0111] The safety threshold is then calculated using the following formula:

[0112] M safe =M th -k·σ th

[0113] In the formula, σ th This represents the standard deviation of the threshold adjustment process.

[0114] In summary, this invention combines real-time monitoring data such as wind speed, turbulence intensity, and temperature with aerodynamic principles to more accurately capture the impact of environmental changes on blade bending moments, thereby improving the accuracy of safety assessments. Furthermore, by extracting the spatiotemporal correlation features of historical monitoring data and calculating the rates of change of wind speed, turbulence intensity, and temperature over time, as well as their covariance matrices, more representative spatiotemporal features are extracted, enhancing the model's adaptability to environmental changes and improving its generalization ability. Finally, by introducing a physical regularization term based on aerodynamic theory into the Gaussian process regression model, combining aerodynamic principles with machine learning models enhances the model's physical rationality and interpretability, resulting in highly accurate predictions.

[0115] This embodiment also provides a computer device applicable to the safety assessment method for wind turbine generator sets, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the safety assessment method for wind turbine generator sets as proposed in the above embodiment.

[0116] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0117] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the safety assessment method for wind turbine generators as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0118] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A safety assessment method for wind turbine generator sets, characterized in that: include, Collect monitoring data of generator blades, including wind speed, turbulence intensity and temperature, fit the monitoring data, and couple the edge distribution of the monitoring data by a hybrid Copula function; A three-dimensional model is established based on the shape parameters of the generator set. Various scenarios are simulated using analysis software to generate the maximum bending moment at the flange root under different scenarios, and the mapping relationship between scenario and bending moment is obtained. Based on the aforementioned scenario-bending moment mapping relationship, and combined with the spatiotemporal correlation features of the monitoring data extracted from historical monitoring data, the specific steps are as follows: Calculate the rate of change of wind speed, turbulence intensity, and temperature over time to obtain the time gradients of wind speed, turbulence intensity, and temperature; Calculate the covariance between wind speed, turbulence intensity, and temperature to obtain the covariance matrix; By integrating the temporal gradients of wind speed, turbulence intensity, and temperature with the elements in the covariance matrix, spatiotemporal correlation features are formed. By analyzing the impact of historical temperature changes on generator blades, correction factors were determined, and a Gaussian process regression model integrating aerodynamic principles was constructed and trained. The integrated aerodynamic principle refers to defining the relationship model between bending moment and angle of attack based on the airfoil aerodynamic equation. The partial derivative of bending moment with respect to angle of attack is defined to satisfy aerodynamic theory and is used as the physical regularization term for training the Gaussian process regression model. The kernel functions of the Gaussian process regression model include the radial basis function, the spatiotemporal correlation kernel function, and the temperature correction kernel function; Historical monitoring data is input into the trained Gaussian process regression model to obtain several bending moment prediction results. Samples that meet the pre-set requirements are selected, and multiple risk levels are set. Markov chain migration sampling is performed on samples of each risk level to dynamically adjust the failure threshold, calculate the failure probability between levels, and integrate them. Based on the failure probability, a safety threshold is set. Real-time monitoring data is input into the trained Gaussian process regression model to calculate the current flange root bending moment value. The current flange root bending moment value is compared with the safety threshold to determine the current safety status of the generator set.

2. The safety assessment method for wind turbine generator sets as described in claim 1, characterized in that: Wind speed was fitted using a two-parameter Weibull distribution, turbulence intensity was fitted using a truncated log-normal distribution, and temperature was fitted using a mixed distribution model to establish a joint probability model for non-stationary environments.

3. The safety assessment method for wind turbine generator sets as described in claim 1, characterized in that: The determination of the correction factor includes the following steps. Define a model relating temperature and the elastic modulus of generator blade materials; The elastic modulus of the generator blades was measured and recorded at different temperatures to establish a temperature-elastic modulus mapping relationship. The correction factor is calculated using a linear regression equation.

4. The safety assessment method for wind turbine generator sets as described in claim 1, characterized in that: The Markov chain migration sampling refers to performing directional migration sampling through a gradient-sensitive strategy, and the dynamically adjusted failure threshold refers to readjusting the threshold based on the failure ratio of samples after Markov chain migration.

5. The safety assessment method for wind turbine generator sets as described in claim 1, characterized in that: After collecting the monitoring data, preprocessing operations are performed first.

6. The safety assessment method for wind turbine generator sets as described in claim 1, characterized in that: The shape parameters include the blade's length, width, thickness, material, curvature, blade twist angle, and tower height.

7. The safety assessment method for wind turbine generator sets as described in claim 4, characterized in that: The calculation and integration of failure probabilities between levels refers to first calculating the failure probability of each level, and then summing the failure probabilities of each level based on exponential decay weights.

Images