A wind turbine blade life intelligent monitoring method and system

Abstract

The application discloses a wind turbine blade life intelligent monitoring method and system, and relates to the technical field of wind power, which comprises the following steps: simulating based on a wind turbine blade structure model to obtain simulation data of a wind turbine blade surface fatigue crack; performing abnormal point elimination and degradation quantity extraction on the simulation data, and fitting a fitting curve by using the obtained degradation quantity difference; judging whether the fitting curve has an inflection point, if yes, using a two-stage random failure threshold Wiener degradation model to predict the life information of the wind turbine blade, and if not, using a first-stage random failure threshold Wiener degradation model to predict the life information of the wind turbine blade. Through the steps of abnormal point elimination and degradation quantity extraction, the degradation quantity difference is obtained, then linear fitting is performed based on the degradation quantity difference, the inflection point is judged, and the life of the wind turbine blade is predicted in two stages based on the inflection point, so that the prediction accuracy of the life of the wind turbine blade is improved.

Description

Technical Field

[0001] This invention relates to the field of wind power technology, and in particular to a method and system for intelligent monitoring of wind turbine blade life. Background Technology

[0002] In wind power systems, wind turbine blades are one of the key components for converting wind energy into mechanical energy. However, due to long-term exposure to wind impacts and environmental influences, the materials and structure of the blades gradually suffer damage and fatigue, thus reducing their performance and lifespan. Therefore, accurately monitoring and assessing the lifespan of wind turbine blades is crucial. Traditional methods for monitoring the lifespan of wind turbine blades mainly rely on manual inspections and offline tests, which suffer from high costs, poor timeliness, and discontinuous data acquisition. In recent years, with the development of intelligent monitoring technology, intelligent monitoring methods based on data analysis and model building have gradually become a research hotspot. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent monitoring method and system for wind turbine blade life, thereby improving the accuracy of wind turbine blade life prediction.

[0004] To achieve the above objectives, the present invention provides the following solution:

[0005] A method for intelligent monitoring of wind turbine blade life, the method comprising:

[0006] Simulation data of fatigue cracks on the surface of wind turbine blades were obtained by performing simulation based on the structural model of wind turbine blades.

[0007] Outlier removal is performed on the simulation data. Degradation is extracted from the simulation data after outlier removal. The difference in degradation of the simulation data after different simulation cycles is taken as the degradation difference.

[0008] Linear fitting is performed based on the difference in degradation amount to obtain the fitted curve;

[0009] Determine whether the fitted curve has an inflection point. If an inflection point exists, use a two-stage random failure threshold Wiener degradation model to predict the lifespan information of the wind turbine blade. The two-stage random failure threshold Wiener degradation model includes a first-stage random failure threshold Wiener degradation model and a second-stage random failure threshold Wiener degradation model. If no inflection point exists, use the first-stage random failure threshold Wiener degradation model to predict the lifespan information of the wind turbine blade.

[0010] Optionally, the two-stage random failure threshold Wiener degradation model is constructed based on the random failure threshold, the degradation amount of the simulation data in the first stage, the degradation amount of the simulation data in the second stage, the drift coefficient in the first stage, the drift coefficient in the second stage, the diffusion coefficient in the first stage, and the diffusion coefficient in the second stage.

[0011] The two-stage random failure threshold Wiener degradation model is as follows:

[0012] When 0 < t ≤ t k hour,

[0013] ;

[0014] When t k When < t,

[0015]

[0016] In the formula, q α (t) represents the time during which the wind turbine continues to operate normally with a probability of 1-α, R -1 (*) represents the inverse function of the reliability R(*) of the wind turbine blade, α is a parameter between 0 and 1, t is the number of simulation cycles; D is the random failure threshold, a is the degradation of the simulation data when t = 0, and a1 is the degradation of the simulation data when t = t k The degradation of the simulation data at time t, μ1 and μ2 are the drift coefficients of the first and second stages respectively, σ1 and σ2 are the diffusion coefficients of the first and second stages respectively, and t is the simulation cycle number. k It is the number of simulation iterations for the inflection point k;

[0017] The reliability function for wind turbine blades is as follows:

[0018]

[0019] Where R1(*) and R2(*) are the reliability functions for the first and second stages, respectively.

[0020] Optionally, the process of determining the inflection point k includes:

[0021] The difference in degradation is substituted into the formula for the minimum AICc criterion to find the minimum AICc value. The formula for the minimum AICc criterion is as follows:

[0022] AlCc(k i )=-2In(L)+2k i +(2k i (k i +1)) / (nk i -1);

[0023] In the formula, k i is the number of parameters in the two-stage Wiener model, n is the size of the two-stage Wiener model sample, and In(L) is the maximum likelihood function value of the two-stage Wiener model;

[0024] in:

[0025]

[0026] In the formula, Δx i Let be the difference in degradation between the i-th and i-1-th simulation cycles; k is the number of the last degradation in the first stage, i.e., the inflection point; m is the number of the last degradation in the second stage; where μ1 and μ2 are the drift coefficients of the first and second stages, respectively, σ1 and σ2 are the diffusion coefficients of the first and second stages, respectively, and Δt is the value of the degradation difference between the i-th and i-1-th simulation cycles; Δt is the value of the degradation difference between the i-th and i-1-th simulation cycles; σ1 and σ2 are the drift coefficients of the first and second stages, respectively; ... i Let be the difference between the number of simulation cycles in the i-th and i-1-th iterations. This represents the average value of the degradation difference in the simulation data extracted in the first stage. This represents the average value of the difference in degradation amount of the simulation data extracted in the second stage.

[0027] The minimum point of AICc is determined based on the minimum AICc criterion, and the minimum point of AICc is the inflection point k;

[0028]

[0029] Simulation loop count t based on inflection point k k The simulation degradation is divided into two stages: the first stage is 0 < t ≤ t k The second stage is t k ≤t.

[0030] Optionally, the process of determining the drift coefficient of the first stage, the drift coefficient of the second stage, the diffusion coefficient of the first stage, and the diffusion coefficient of the second stage includes:

[0031] Take the partial derivatives of the maximum likelihood function value In(L) of the two-stage Wiener model, and let... It can be obtained The maximum likelihood estimates of the parameters are as follows:

[0032]

[0033] Substituting k and m into the above equation, we obtain the drift coefficient μ1 of the first stage, the drift coefficient μ2 of the second stage, the diffusion coefficient σ1 of the first stage, and the diffusion coefficient σ2 of the second stage.

[0034] Optionally, the process for determining the random failure threshold D includes:

[0035]

[0036] In the formula, θ1 and θ2 are preset parameters that follow a normal distribution.

[0037] Optionally, before obtaining the fitted curve by performing linear fitting based on the degradation difference, the method further includes: removing outliers in the degradation difference using an adaptive function.

[0038] Optionally, outliers in the degradation difference can be removed using an adaptive function, specifically including:

[0039] The standard value of the degradation difference is obtained based on the extracted simulation data and the measured data of fatigue cracks on the surface of wind turbine blades.

[0040] The degradation difference below the standard value is set as an outlier. An adaptive function is used to remove outliers, and the adaptive parameters in the adaptive function are updated based on the extracted simulation data and measured data.

[0041] Optionally, after obtaining the lifespan information of the wind turbine blades, the method also includes: issuing an early warning and performing maintenance on the wind turbine blades when their lifespan exceeds the warning threshold.

[0042] A smart monitoring system for the lifespan of wind turbine blades, the system comprising:

[0043] The simulation module is used to perform simulations based on the wind turbine blade structure model to obtain simulation data of fatigue cracks on the surface of the wind turbine blade.

[0044] The data processing module is used to remove outliers from the simulation data, extract the degradation amount from the simulation data after outlier removal, and use the difference in degradation amount of the simulation data after different simulation cycles as the degradation amount difference.

[0045] The fitting module is used to perform linear fitting based on the difference in degradation amount to obtain a fitted curve;

[0046] The prediction module is used to determine whether the fitted curve has an inflection point. If an inflection point exists, the two-stage random failure threshold Wiener degradation model is used to predict the lifespan information of the wind turbine blade. The two-stage random failure threshold Wiener degradation model includes a first-stage random failure threshold Wiener degradation model and a second-stage random failure threshold Wiener degradation model. If no inflection point exists, the first-stage random failure threshold Wiener degradation model is used to predict the lifespan information of the wind turbine blade.

[0047] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0048] This invention improves the accuracy of wind turbine blade life prediction by removing outliers and extracting degradation amounts, then performing linear fitting based on the degradation difference and determining the inflection point, and predicting the life of wind turbine blades in two stages based on the inflection point. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. 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.

[0050] Figure 1 This is a first flowchart of the intelligent monitoring method for wind turbine blade life provided in an embodiment of the present invention;

[0051] Figure 2 This is a second flowchart of the intelligent monitoring method for wind turbine blade life provided in an embodiment of the present invention;

[0052] Figure 3 A diagram of the intelligent diagnostic system for wind turbine blades provided in an embodiment of the present invention;

[0053] Figure 4 A reliability comparison chart of wind turbine blades under different simulation cycle numbers provided in this embodiment of the invention;

[0054] Figure 5 This is a structural diagram of a wind turbine blade life intelligent monitoring system module provided in an embodiment of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] The purpose of this invention is to provide an intelligent monitoring method and system for wind turbine blade life. By eliminating abnormal points and extracting degradation amount, the degradation amount difference is obtained. Then, based on the degradation amount difference, linear fitting is performed and the inflection point is determined. Based on the inflection point, the life of wind turbine blades is predicted in two stages, thereby improving the prediction accuracy of wind turbine blade life.

[0057] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] Example 1

[0059] like Figure 1-3 As shown, the present invention provides an intelligent monitoring method for the lifespan of wind turbine blades, the method comprising:

[0060] Step 101: Simulation is performed based on the wind turbine blade structure model to obtain simulation data of fatigue cracks on the surface of the wind turbine blade.

[0061] Specifically, the wind turbine structural model is imported into simulation software; the simulation software is ANSYS software; the wind turbine structural model is a SolidWorks 3D model.

[0062] Configure simulation parameters for the simulation software; specifically, configuring simulation parameters includes:

[0063] Set the geometry, material properties, loading conditions, boundary conditions, simulation time, mesh density, and solver for the actual wind turbine blade.

[0064] The degradation model selected in this embodiment is the blade surface fatigue crack propagation model.

[0065] After running the simulation, simulation data of fatigue cracks on the surface of wind turbine blades were obtained at different simulation cycle counts as the number of simulation cycles increased. The simulation data included at least one of blade condition and performance data.

[0066] Step 102: Remove outliers from the simulation data, extract the degradation amount from the simulation data after outlier removal, and use the difference in degradation amount of the simulation data after different simulation cycles as the degradation amount difference.

[0067] Specifically, the simulation data obtained in step 101, as well as the measured data of fatigue cracks on the surface of the wind turbine blades, are collected. The simulation data and measured data specifically include parameters such as operating time, rotational speed, temperature, and crack length.

[0068] The collected simulation data and measured data are preprocessed, including: data denoising, handling missing values, and outlier removal; in this embodiment, outlier removal is used.

[0069] Feature extraction is performed on the preprocessed data. The extracted features include: maximum rotational speed, vibration amplitude, degradation amount, etc.; in this embodiment, the degradation amount is extracted.

[0070] A leaf degradation database was established using the extracted features (degradation amount).

[0071] In this embodiment, the data sources should be extensive, covering different working conditions and operating situations to obtain comprehensive and accurate data; noise reduction can reduce the impact of noise on the data and improve data quality; interpolation or padding methods can be used to fill in missing data and ensure data integrity; the purpose of feature extraction is to capture key information in the blade degradation process and provide effective input for subsequent life prediction modeling.

[0072] Step 103: Perform linear fitting based on the difference in degradation amount to obtain the fitted curve;

[0073] Step 104: Determine whether the fitted curve has an inflection point. If an inflection point exists, use the two-stage random failure threshold Wiener degradation model to predict the lifespan information of the wind turbine blade. The two-stage random failure threshold Wiener degradation model includes a first-stage random failure threshold Wiener degradation model and a second-stage random failure threshold Wiener degradation model. If no inflection point exists, use the first-stage random failure threshold Wiener degradation model to predict the lifespan information of the wind turbine blade.

[0074] As an optional implementation, after obtaining the lifespan information of the wind turbine blades in step 104, the method further includes: issuing an early warning and performing maintenance on the wind turbine blades when their lifespan exceeds a warning threshold. The warning threshold is set based on historical wind turbine blade lifespan information.

[0075] As an optional implementation, before performing linear fitting based on the degradation difference in step 103 to obtain the fitted curve, the method further includes: removing outliers from the degradation difference using an adaptive function; specifically, removing outliers from the degradation difference using an adaptive function includes:

[0076] The standard value of the degradation difference is obtained based on the extracted simulation data and the measured data of fatigue cracks on the surface of wind turbine blades.

[0077] The degradation difference below the standard value is set as an outlier. An adaptive function is used to remove outliers, and the adaptive parameters in the adaptive function are updated based on the extracted simulation data and measured data.

[0078] In this study, the mean squared error (MSE) in linear regression is used as the main body of the loss function to construct an adaptive function, thereby obtaining the standard value of the degradation difference.

[0079] The loss function formula is:

[0080]

[0081] Where N represents the number of samples, y i This represents the measured data. This represents simulation data.

[0082] As an optional implementation, the two-stage random failure threshold Wiener degradation model is constructed based on the random failure threshold, the degradation amount of the simulation data in the first stage, the degradation amount of the simulation data in the second stage, the drift coefficient in the first stage, the drift coefficient in the second stage, the diffusion coefficient in the first stage, and the diffusion coefficient in the second stage.

[0083] The two-stage random failure threshold Wiener degradation model is as follows:

[0084] When 0 < t ≤ t k hour,

[0085]

[0086] When t k When < t,

[0087]

[0088] In the formula, q α (t) represents the time during which the wind turbine continues to operate normally with a probability of 1-α, R -1 (*) represents the inverse function of the reliability R(*) of the wind turbine blade, α is a parameter between 0 and 1, t is the number of simulation cycles; D is the random failure threshold, a is the degradation of the simulation data when t = 0, and a1 is the degradation of the simulation data when t = t k The degradation of the simulation data at time t, μ1 and μ2 are the drift coefficients of the first and second stages respectively, σ1 and σ2 are the diffusion coefficients of the first and second stages respectively, and t is the simulation cycle number. k It is the number of simulation cycles for the inflection point k.

[0089] The reliability function for wind turbine blades is as follows:

[0090]

[0091] Where R1(*) and R2(*) are the reliability functions for the first and second stages, respectively.

[0092] The relationship between the reliability and lifespan of wind turbine blades is as follows:

[0093]

[0094] q α (t)=R -1 ((1-α)R(t))-t.

[0095] As an optional implementation, the process of determining the inflection point k includes:

[0096] The difference in degradation is substituted into the formula for the minimum AICc criterion to find the minimum AICc value. The formula for the minimum AICc criterion is as follows:

[0097] AlCc(k i )=-2In(L)+2k i +(2k i (k i +1)) / (nk i -1).

[0098] In the formula, k i denoted as the number of parameters in the two-stage Wiener model, n is the sample size of the two-stage Wiener model, and In(L) is the maximum likelihood function value of the two-stage Wiener model.

[0099] in:

[0100]

[0101] In the formula, Δx i Let be the difference in degradation between the i-th and (i-1)-th simulation cycles; k be the number of the last degradation in the first stage, i.e., the inflection point; m be the number of the last degradation in the second stage; where μ1 and μ2 are the drift coefficients of the first and second stages, respectively, σ1 and σ2 are the diffusion coefficients of the first and second stages, respectively, and Δt is the value of the degradation difference between the i-th and (i-1)-th simulation cycles; Δt is the value of the degradation difference between the i-th and (i-1)-th simulation cycles; σ1 and σ2 are the diffusion coefficients of the first and second stages, respectively; Δt is the value of the degradation difference between the i-th and (i-1)-th simulation cycles ...σ1 and σ2 are the drift coefficients of the first and second stages, respectively; Δt is the value of i Let be the difference between the number of simulation cycles in the i-th and (i-1)-th iterations. This represents the average value of the degradation difference in the simulation data extracted in the first stage. This represents the average value of the difference in degradation amount of the simulation data extracted in the second stage.

[0102] Substituting the maximum likelihood function value In(L) of the two-stage Wiener model into the definition of the AICc criterion, we get:

[0103]

[0104] in,

[0105] The minimum point of AICc is determined based on the minimum AICc criterion, and the minimum point of AICc is the inflection point k.

[0106]

[0107] Simulation loop count t based on inflection point k k The simulation degradation is divided into two stages: the first stage is 0 < t ≤ t kThe second stage is t k ≤t.

[0108] As an optional implementation, the process of determining the drift coefficient of the first stage, the drift coefficient of the second stage, the diffusion coefficient of the first stage, and the diffusion coefficient of the second stage includes:

[0109] Take the partial derivatives of the maximum likelihood function value In(L) of the two-stage Wiener model, and let... It can be obtained The maximum likelihood estimates of the parameters are as follows:

[0110]

[0111] Substituting k and m into the above equation, we obtain the drift coefficient μ1 of the first stage, the drift coefficient μ2 of the second stage, the diffusion coefficient σ1 of the first stage, and the diffusion coefficient σ2 of the second stage.

[0112] As an optional implementation, the process of determining the random failure threshold D includes:

[0113]

[0114] In the formula, θ1 and θ2 are preset parameters that follow a normal distribution.

[0115] In this embodiment Figure 4 The invention presents life prediction curves for wind turbine blades with and without a random failure threshold, where M0 represents a two-stage method including the random failure threshold, and M1 represents a two-stage method without the random failure threshold. This invention addresses the complexity of the wind turbine blade degradation process, the uncertainty of the failure threshold, and the performance differences within the same batch of products. Therefore, it avoids using complex high-order nonlinear degradation models, performing reliability analysis and wind turbine blade life prediction while ensuring simulation accuracy, thus improving the accuracy and computational efficiency of wind turbine blade life prediction.

[0116] Example 2

[0117] like Figure 5 As shown in the figure, this embodiment provides an intelligent monitoring system for the lifespan of wind turbine blades, the system comprising:

[0118] The simulation module is used to perform simulations based on the wind turbine blade structure model to obtain simulation data of fatigue cracks on the surface of the wind turbine blade.

[0119] The data processing module is used to remove outliers from the simulation data, extract the degradation amount from the simulation data after outlier removal, and use the difference in degradation amount of the simulation data after different simulation cycles as the degradation amount difference value.

[0120] The fitting module is used to perform linear fitting based on the difference in degradation amount to obtain a fitted curve;

[0121] The prediction module is used to determine whether the fitted curve has an inflection point. If an inflection point exists, the lifespan information of the wind turbine blade is predicted using a two-stage random failure threshold Wiener degradation model. The two-stage random failure threshold Wiener degradation model includes a first-stage random failure threshold Wiener degradation model and a second-stage random failure threshold Wiener degradation model. If no inflection point exists, the lifespan information of the wind turbine blade is predicted using the first-stage random failure threshold Wiener degradation model.

[0122] This invention improves the accuracy of wind turbine blade life prediction by removing outliers and extracting degradation amounts to obtain degradation difference values. Then, it performs linear fitting based on these degradation difference values ​​to determine inflection points and predicts the lifespan of wind turbine blades in two stages based on these inflection points. First, this invention constructs a wind turbine blade network, acquires measured and online degradation data, and performs linear fitting using MATLAB or Origin software. Next, it evaluates the fitting results and selects degradation models for different stages. Adaptive parameter updates are performed on the degradation difference values, and the AICc criterion is used to find change points and perform interval estimation. The failure thresholds for different stages are obtained using the least squares method to establish a two-stage reliability function considering the randomness of the failure thresholds. Based on the relationship between reliability and lifespan, a lifespan prediction model is constructed. This invention avoids the use of complex nonlinear degradation models, performing reliability analysis and product lifespan prediction while ensuring simulation accuracy, effectively improving the accuracy and computational efficiency of wind turbine blade lifespan prediction.

[0123] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0124] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for intelligent monitoring of wind turbine blade life, characterized in that, The method includes: Simulation data of fatigue cracks on the surface of wind turbine blades were obtained by performing simulation based on the structural model of wind turbine blades. Outlier removal is performed on the simulation data. Degradation is extracted from the simulation data after outlier removal. The difference in degradation of the simulation data after different simulation cycles is taken as the degradation difference. Linear fitting is performed based on the difference in degradation amount to obtain the fitted curve; Determine whether the fitted curve has an inflection point. If an inflection point exists, use a two-stage random failure threshold Wiener degradation model to predict the lifespan information of the wind turbine blade. The two-stage random failure threshold Wiener degradation model includes a first-stage random failure threshold Wiener degradation model and a second-stage random failure threshold Wiener degradation model. If no inflection point exists, use the first-stage random failure threshold Wiener degradation model to predict the lifespan information of the wind turbine blade. The two-stage random failure threshold Wiener degradation model is constructed based on the random failure threshold, the degradation amount of the simulation data in the first stage, the degradation amount of the simulation data in the second stage, the drift coefficient in the first stage, the drift coefficient in the second stage, the diffusion coefficient in the first stage, and the diffusion coefficient in the second stage. The two-stage random failure threshold Wiener degradation model is as follows: when hour, ； when hour, ； In the formula, This represents the time during which the wind turbine continues to operate normally with a probability of 1-α. Indicates the reliability of wind turbine blades inverse function, The cumulative distribution function represents the standard normal distribution, where α is a parameter between 0 and 1, and t is the number of simulation cycles; This is the random failure threshold. This represents the degradation of the simulation data at t=0. For when t=t k The amount of degradation of the simulation data at that time, These are the drift coefficients for the first and second stages, respectively. These are the diffusion coefficients for the first and second stages, respectively. k is the number of the last degradation in the first stage, i.e., the inflection point, and t... k It is the number of simulation cycles for the inflection point k. This is the reliability function for the first stage; The reliability function for wind turbine blades is as follows: ； in, This is the reliability function for the second stage.

2. The intelligent monitoring method for wind turbine blade life according to claim 1, characterized in that, The process of determining the inflection point k includes: The difference in degradation is substituted into the formula for the minimum AICc criterion to find the minimum AICc value. The formula for the minimum AICc criterion is as follows: ； In the formula, represents the number of parameters in the two-stage Wiener model, and n represents the sample size of the two-stage Wiener model. The maximum likelihood function value of the two-stage Wiener model; in: ； In the formula, Δx i Let be the difference in degradation between the i-th and i-1-th simulation cycles; m is the number of the last degradation in the second stage; where, These are the drift coefficients for the first and second stages, respectively. These are the diffusion coefficients for the first and second stages, respectively, Δt. i This is the difference between the number of simulation cycles in the i-th and i-1-th iterations; The minimum point of AICc is determined based on the minimum AICc criterion, and the minimum point of AICc is the inflection point k; ； Simulation loop count t based on inflection point k k The simulation degradation is divided into two stages: the first stage is 0 < t ≤ t k The second stage is t k ≤t.

3. The intelligent monitoring method for wind turbine blade life according to claim 2, characterized in that, The process of determining the drift coefficient of the first stage, the drift coefficient of the second stage, the diffusion coefficient of the first stage, and the diffusion coefficient of the second stage includes: Maximum likelihood function value of the two-stage Wiener model Take the partial derivatives respectively, and let You can get The maximum likelihood estimates of the parameters are as follows: ； Substituting k and m into the above equation, we obtain the estimated drift coefficient for the first stage. The estimated drift coefficients for the second stage The estimated diffusion coefficient for the first stage Second-stage diffusion coefficient estimates .

4. The intelligent monitoring method for wind turbine blade life according to claim 1, characterized in that, The process of determining the random failure threshold D includes: ； In the formula, , θ1 and θ2 are preset parameters that follow a normal distribution.

5. The intelligent monitoring method for wind turbine blade life according to claim 1, characterized in that, Before obtaining the fitted curve by performing linear fitting based on the degradation difference, the process also includes: removing outliers in the degradation difference using an adaptive function.

6. The intelligent monitoring method for wind turbine blade life according to claim 1, characterized in that, Outliers in the degradation difference are removed using an adaptive function, specifically including: The standard value of the degradation difference is obtained based on the extracted simulation data and the measured data of fatigue cracks on the surface of wind turbine blades. The degradation difference below the standard value is set as an outlier. An adaptive function is used to remove outliers, and the adaptive parameters in the adaptive function are updated based on the extracted simulation data and measured data.

7. The intelligent monitoring method for wind turbine blade life according to claim 1, characterized in that, After obtaining the lifespan information of the wind turbine blades, the system also includes: issuing warnings and performing maintenance on the wind turbine blades when their lifespan exceeds the warning threshold.

8. A smart monitoring system for the lifespan of wind turbine blades, characterized in that, The system includes: The simulation module is used to perform simulations based on the wind turbine blade structure model to obtain simulation data of fatigue cracks on the surface of the wind turbine blade. The data processing module is used to remove outliers from the simulation data, extract the degradation amount from the simulation data after outlier removal, and use the difference in degradation amount of the simulation data after different simulation cycles as the degradation amount difference. The fitting module is used to perform linear fitting based on the difference in degradation amount to obtain a fitted curve; The prediction module is used to determine whether the fitted curve has an inflection point. If an inflection point exists, the two-stage random failure threshold Wiener degradation model is used to predict the lifespan information of the wind turbine blade. The two-stage random failure threshold Wiener degradation model includes a first-stage random failure threshold Wiener degradation model and a second-stage random failure threshold Wiener degradation model. If no inflection point exists, the first-stage random failure threshold Wiener degradation model is used to predict the lifespan information of the wind turbine blade. The two-stage random failure threshold Wiener degradation model is constructed based on the random failure threshold, the degradation amount of the simulation data in the first stage, the degradation amount of the simulation data in the second stage, the drift coefficient in the first stage, the drift coefficient in the second stage, the diffusion coefficient in the first stage, and the diffusion coefficient in the second stage. The two-stage random failure threshold Wiener degradation model is as follows: when hour, ； when hour, ； In the formula, This represents the time during which the wind turbine continues to operate normally with a probability of 1-α. Indicates the reliability of wind turbine blades inverse function, The cumulative distribution function represents the standard normal distribution, where α is a parameter between 0 and 1, and t is the number of simulation cycles; This is the random failure threshold. This represents the degradation of the simulation data at t=0. For when t=t k The amount of degradation of the simulation data at that time, These are the drift coefficients for the first and second stages, respectively. These are the diffusion coefficients for the first and second stages, respectively. k is the number of the last degradation in the first stage, i.e., the inflection point, and t... k It is the number of simulation cycles for the inflection point k. This is the reliability function for the first stage; The reliability function for wind turbine blades is as follows: ； in, This is the reliability function for the second stage.

Images