A fan blade life prediction and fault early warning method based on digital twinning

Abstract

The application relates to the technical field of fan blade operation monitoring, and discloses a fan blade life prediction and fault early warning method based on digital twinning; an optical fiber grating sensor is pre-buried in a blade cap fiber layer, an acceleration sensor is arranged at a blade root, and multi-source data are collected in combination with wind speed, rotating speed and pitch angle sensors; a blade finite element digital twinning model is established, stress field dynamic assimilation is realized in combination with aerodynamic-structure coupling simulation and extended Kalman filtering; fatigue residual life is calculated based on rain flow counting, S-N curve and Miner theory, crack propagation life is evaluated by adopting a Paris model, and the minimum value of the two is taken as comprehensive life; a dynamic safety threshold considering wind speed fluctuation, environmental temperature and operation time is further constructed, and early warning is triggered when the comprehensive life is less than or equal to the threshold. The application can realize accurate prediction and real-time early warning of blade life, and solves the problems of insufficient monitoring coverage, incomplete life evaluation and fixed threshold.

Description

Technical Field

[0001] This invention relates to the field of wind turbine blade operation monitoring technology, and more specifically, to a method for predicting wind turbine blade life and providing early warning of faults based on digital twins. Background Technology

[0002] Wind power, as a crucial component of clean energy, has experienced rapid global development in recent years, with a continuously increasing application of large-scale wind turbines. As a key load-bearing and energy conversion component of wind turbines, the blades' service condition directly impacts the unit's power output and operational safety. However, due to long-term exposure to a complex operating environment, blades are susceptible to multiple factors such as alternating wind loads, rain, snow, freezing, temperature variations, and dust erosion, leading to gradual accumulation of material fatigue damage and crack propagation, potentially resulting in failure or fracture. Blade failure not only causes unit downtime and significant economic losses but may also endanger the safety of surrounding personnel and equipment.

[0003] Traditional blade life prediction methods are mostly based on Miner's linear cumulative damage theory. This involves calculating the cumulative damage factor and estimating the life by counting rainflows on the stress time history and combining this with the material's S-N curve. While these methods are computationally simple, they have two drawbacks: first, they insufficiently consider the nonlinear characteristics of material fatigue behavior and the multi-scale damage mechanism under complex loads, often leading to significant life prediction errors; second, they rely on stress assumptions under design conditions, making it difficult to reflect the impact of actual wind fluctuations and changes in operating conditions, potentially resulting in a significant discrepancy between the predicted and actual lifespans.

[0004] In recent years, some scholars have attempted to use sensor monitoring and data-driven models to assess the condition of blades. For example, strain gauges, accelerometers, and other sensors are installed on the blade surface or root, and damage identification is achieved by combining signal processing and machine learning methods. However, due to limitations in sensor placement, the acquired signals often fail to cover the entire blade structure, and the data-driven models lack physical constraints, making them prone to overfitting and insufficient interpretability, resulting in low reliability of the prediction results.

[0005] With the development of digital twin technology, researchers have gradually introduced it into the wind power field. Digital twins can achieve virtual reproduction and prediction of blade operating status through finite element modeling, aerodynamic-structural coupling simulation, and fusion of measured data. However, existing research still has the following shortcomings: First, digital twin models often remain at the static simulation level, lacking dynamic assimilation with real-time monitoring data, leading to deviations between model predictions and actual operating conditions; second, most studies only consider fatigue cumulative lifespan, neglecting key failure mechanisms such as crack initiation and propagation, resulting in insufficient comprehensive lifespan assessment; third, traditional lifespan prediction methods generally use fixed thresholds as early warning criteria, failing to dynamically correct based on operating conditions such as wind speed fluctuations, ambient temperature, and cumulative operating time, easily causing false alarms or missed alarms. In summary, existing technologies cannot simultaneously meet the requirements of high-precision lifespan prediction and real-time reliable early warning.

[0006] Therefore, we urgently need to design a method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins to solve these problems. Summary of the Invention

[0007] The purpose of this invention is to solve the technical problems mentioned in the background section and to provide a method for predicting the lifespan and predicting the faults of wind turbine blades based on digital twins.

[0008] The above-mentioned objective of the present invention is achieved through the following technical solution:

[0009] A method for predicting the lifespan and providing early warning of faults in wind turbine blades based on digital twins includes the following steps:

[0010] S1) Collect wind turbine operating parameters, including wind speed obtained by the laser anemometer on the top of the nacelle, rotational speed obtained by the main shaft encoder, pitch angle obtained by the pitch angle sensor, strain signal obtained by the fiber optic grating strain sensor embedded in the fiber layer of the blade truss cap, and vibration signal obtained by the accelerometer arranged on the inner side of the blade root web, to form multi-source operating data.

[0011] S2) Based on the geometric and material parameters of the blade, a finite element model is established, and combined with the operating data, aerodynamic-structural coupling simulation is performed to obtain the digital twin simulation stress field of the blade.

[0012] S3) Using the extended Kalman filter method, the simulated stress field is fused and corrected with the observed stress obtained by inversion from strain signal and vibration signal to obtain the corrected real-time stress field;

[0013] S4) Perform rainflow counting on the time history data of the real-time stress field to obtain the stress cycle amplitude, average cycle stress, and number of cycles;

[0014] S5) Calculate the cumulative damage factor of the blade based on the fatigue life curve of the material and the linear cumulative damage theory;

[0015] S6) Based on the cumulative damage factor and the damage evolution rate, predict the remaining fatigue life of the blade.

[0016] S7) Based on the crack propagation model and combined with the real-time stress field, calculate the crack propagation life. The crack propagation model considers the initial crack length, critical crack length, stress intensity factor, crack propagation threshold, material fracture toughness, and geometric correction coefficient.

[0017] S8) The combined fatigue life and crack propagation life are used to obtain the overall life of the blade;

[0018] S9) Construct a dynamic safety threshold, which is determined based on wind speed fluctuation, ambient temperature, and cumulative operating time; when the overall lifespan is less than or equal to the dynamic safety threshold, output a fault warning signal.

[0019] As a preferred technical solution of the present invention, the fiber optic strain sensor is pre-embedded in the fiber layer of the upper and lower stringers during the blade manufacturing stage, and is arranged at positions of 0.25 times, 0.50 times and 0.75 times the spanwise length of the blade, and is arranged parallel to the main force direction of the blade; the accelerometer is arranged on the inner side of the blade root web, fixed by structural adhesive, and led to the blade root signal interface through a shielded cable.

[0020] As a preferred technical solution of the present invention, the digital twin simulation stress field establishes a blade structure model through the finite element method, and loads aerodynamic loads determined by wind speed, chord length, aerodynamic coefficient and air density into the model to obtain the stress distribution of the blade under different working conditions.

[0021] As a preferred technical solution of the present invention, the real-time stress field is obtained by extended Kalman filtering. The filtering process comprehensively considers the observation matrix, prediction covariance and observation noise covariance to achieve dynamic fusion of digital twin simulation results and measured stress signals.

[0022] As a preferred technical solution of the present invention, rainflow counting is performed on the time history data of the real-time stress field to obtain stress cycle parameters, and the corresponding cycle life is determined by combining the fatigue life curve of the material. Then, the cumulative damage factor of the blade during operation is calculated based on the linear cumulative damage theory.

[0023] As a preferred technical solution of the present invention, the fatigue remaining life is calculated by extrapolating the cumulative damage factor and combining it with the damage evolution rate, which is applicable when the cumulative damage factor is less than a critical value.

[0024] As a preferred technical solution of the present invention, the crack propagation life is calculated based on a segmented crack propagation model. The model considers the stress intensity factor amplitude, crack propagation threshold, material fracture toughness, safety factor and geometric correction factor, and calculates the life by integrating the crack propagation process from the initial length to the critical length.

[0025] As a preferred technical solution of the present invention, the comprehensive life is the minimum value of fatigue remaining life and crack propagation life, which is used to reflect the true remaining life of the blade.

[0026] As a preferred embodiment of the present invention, the dynamic security threshold is determined as follows:

[0027] Wind speed fluctuations are divided into different levels, and different levels correspond to different lifespan correction coefficients.

[0028] Ambient temperature is used as a correction factor for material performance degradation;

[0029] The cumulative running time is used as a weighting factor for fatigue accumulation;

[0030] A dynamic safety threshold is obtained through weighted calculation. When the overall lifespan is less than or equal to the threshold, the system triggers a fault warning signal.

[0031] This invention also proposes a digital twin wind turbine blade life prediction and fault early warning system, comprising:

[0032] The data acquisition module is used to collect wind speed, rotational speed, blade pitch angle, strain, and vibration signals.

[0033] The digital twin modeling module is used to build finite element models and generate simulated stress fields for blades;

[0034] The data assimilation module is used to perform extended Kalman filtering and output the corrected real-time stress field;

[0035] The fatigue life prediction module is used to perform rainflow counting, fatigue life curve and cumulative damage calculation to obtain fatigue life;

[0036] The crack propagation prediction module is used to perform crack propagation model calculations to obtain the crack propagation life.

[0037] The comprehensive assessment and early warning module is used to calculate the overall blade life and compare it with the dynamic safety threshold, and output an early warning signal.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] First, this invention constructs a multi-source monitoring system covering the entire operating condition of the wind turbine by pre-embedding fiber grating sensors in the fiber layer of the blade truss and arranging accelerometers on the inner side of the blade root web, while combining them with a laser anemometer, a main shaft encoder, and a pitch angle sensor. Compared with traditional methods that rely on a single strain gauge or local monitoring points, this arrangement can achieve real-time monitoring of the blade's operating status throughout its entire life cycle without affecting the blade's aerodynamic performance and structural integrity, effectively solving the problems of insufficient sensor coverage and incomplete monitoring data in existing technologies.

[0040] Second, this invention introduces a digital twin model and an extended Kalman filter algorithm to achieve dynamic fusion of simulated stress fields and measured signals. The simulated stress distribution of the blade is obtained through finite element modeling and aerodynamic-structural coupling, and assimilated and corrected using data collected by sensors, enabling the twin to reflect the actual stress state of the blade in real time. Based on this, rainflow counting, S-N curves, and Miner's cumulative damage theory are combined to predict the remaining fatigue life. Simultaneously, the Paris model is introduced to calculate crack propagation life, and the two are integrated to obtain a more realistic comprehensive life. This method overcomes the shortcomings of traditional Miner models, which are singular and neglect crack propagation mechanisms, thus improving the accuracy and physical rationality of life prediction.

[0041] Third, based on lifespan prediction, this invention proposes a method for constructing dynamic safety thresholds. It incorporates key operating conditions such as wind speed fluctuations, ambient temperature, and cumulative operating time into the threshold correction, enabling the early warning criteria to dynamically adjust with changes in the operating environment. Compared to the fixed threshold method used in existing technologies, this invention effectively reduces false alarms and missed alarms, achieving more sensitive and reliable fault early warning. Therefore, this invention not only improves the accuracy of wind turbine blade lifespan prediction but also enhances the adaptability and engineering application value of the early warning system, thus comprehensively solving the problems of large prediction deviations, lack of comprehensive evaluation, and unreasonable thresholds in existing technologies. Attached Figure Description

[0042] 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.

[0043] Figure 1 This is a schematic diagram of the overall process of the method of the present invention;

[0044] Figure 2 This is a schematic diagram of the wind turbine blade sensor system layout;

[0045] Figure 3 This is a schematic diagram of the data assimilation process;

[0046] Figure 4 This is a schematic diagram of the lifespan prediction and determination process;

[0047] Figure 5 This is a schematic diagram illustrating the construction of dynamic security thresholds;

[0048] Figure 6 This is a schematic diagram of the system implementation framework of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the following references are made. Figures 1-6 The present invention will be further described in detail below with reference to the embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0050] Example 1: This example of the present invention proposes a method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins, which can monitor the operating status of wind turbine blades in real time, predict their lifespan, and provide early warning of faults under complex operating conditions.

[0051] During wind turbine operation, it is first necessary to collect key operating parameters from multiple sources, specifically wind speed. The data is obtained from a laser anemometer installed on the top of the nacelle. This anemometer can perform non-contact measurements of the incoming airflow field, avoiding the need to lay out additional aerodynamic measuring points on the blade surface, thereby ensuring the smoothness of the blade surface and aerodynamic efficiency.

[0052] Spindle speed The data is collected in real time by an encoder at the end of the spindle. This encoder is fixed to the spindle by mechanical coupling and does not affect the rotation of the spindle.

[0053] Pitch angle The pitch angle is collected by a pitch angle sensor installed at the blade root. This sensor is linked to the hydraulic pitch system and can accurately reflect the real-time pitch angle of the blade.

[0054] For blade strain signals The data is collected using fiber optic grating (FBG) strain sensors. The FBG sensors are pre-embedded inside the fiber layers of the upper and lower stringers during the blade manufacturing stage. Typical placement positions are 0.25R, 0.50R, and 0.75R along the blade span (R is the blade radius). One or two sensing fibers are laid at each position, arranged parallel to the main force direction of the blade, in order to accurately capture the bending moment and tensile and compressive strain of the blade. Since the diameter of the FBG is less than 0.2mm, it is extremely lightweight and will not damage the structural integrity of the blade or change the aerodynamic shape of the blade after being embedded in the composite material layer.

[0055] The sensing fiber is led through the internal web of the blade to the blade root and then transmitted to the signal acquisition unit via an electric slip ring.

[0056] For blade vibration acceleration A MEMS-type accelerometer is installed on the inner side of the blade root section web. This location is less affected by centrifugal loads, and the installation of the sensor will not cause significant disturbance to the dynamic balance of the blade. The sensor is fixed to the inner wall with structural adhesive and led to the blade root signal interface through a shielded cable. Through the above arrangement, multi-source operational monitoring data of wind speed, rotational speed, blade pitch angle, strain and acceleration are generated.

[0057] Based on the collected data, a digital twin of the blade was established. This twin was formed by coupling finite element modeling with aerodynamic load calculation. The blade's geometric and material parameters were input into the finite element model, and combined with the aerodynamic loads, the simulated stress field of the digital twin was obtained. Digital twin simulation of stress field Calculate using the following formula:

[0058] ;

[0059] in, , for exhibition location aerodynamic normal force; air density; Wind speed; The length of the chord; Normal force coefficient; This is the equivalent area of ​​force application; The simulated stress field is obtained using a digital twin model. This method can reflect the mechanical distribution of the blade under ideal operating conditions. However, due to wind turbulence, load fluctuations, and material nonlinearities in actual operation, the simulated stress field deviates from the measured state.

[0060] Therefore, this invention introduces an extended Kalman filter algorithm for data assimilation. The data assimilation process uses an extended Kalman filter, and its state update formula is as follows:

[0061] ;

[0062] in, This is the corrected real-time stress; For simulating stress; To obtain the strain-stress relationship matrix from , The measured stress obtained; The observation matrix; The Kalman gain is calculated using the following formula:

[0063] ;

[0064] in, To predict the covariance matrix, To observe the noise covariance, this process enables dynamic coupling correction between the twin and the measured data, thereby improving the accuracy of stress prediction.

[0065] Obtain the corrected real-time stress field Then, rainflow counting was performed on the stress time history to obtain the stress cycle amplitude. Mean stress and number of cycles .

[0066] Cycle life was then determined based on the SN curve. Its mathematical form is:

[0067] ;

[0068] Or through the form of mean stress correction:

[0069] ;

[0070] in, , , These are material constants; This refers to the stress amplitude; This is the fatigue strength coefficient; It is the mean stress constant; For the first Average stress over one stress cycle; This is the average stress correction factor; This refers to cycle life.

[0071] Cumulative damage factors The calculation is as follows:

[0072] ;

[0073] in, Stress level The number of loops below; This corresponds to the number of fatigue life cycles. This represents the total number of stress cycle intervals.

[0074] Based on damage factors It can predict the remaining fatigue life:

[0075] , ;

[0076] in, For damage evolution rate, For fatigue remaining life, Based on damage factors.

[0077] Furthermore, to consider the crack propagation behavior after crack initiation, the crack propagation life is calculated based on the piecewise Paris model, and the crack propagation rate formula is as follows:

[0078] ;

[0079] in, The length of the crack; This represents the crack propagation rate. These are material constants; The stress intensity factor amplitude; This is the crack propagation threshold. For the fracture toughness of the material; For safety factor; This is the geometric correction factor.

[0080] Therefore, crack life The calculation is as follows:

[0081] ;

[0082] in, The initial crack length is... This represents the critical crack length.

[0083] Ultimately, the overall lifespan is obtained by minimizing the remaining fatigue life and the crack life:

[0084] ;

[0085] The minimum value between fatigue life and crack life is taken as the true remaining life of the blade.

[0086] To achieve early warning, this invention proposes a dynamic threshold, the mathematical form of which can be expressed as follows: ,in, This refers to the amplitude of wind speed fluctuations. For ambient temperature, The ambient temperature is obtained by a temperature sensor located outside the cabin to accumulate operating time.

[0087] In its implementation, the dynamic threshold is constructed by considering not only wind speed, temperature, and running time, but also by introducing a grading and correction mechanism:

[0088] Wind speed fluctuations are classified into three levels: small fluctuations, medium fluctuations, and large fluctuations. Different levels correspond to different lifespan reduction factors.

[0089] Ambient temperature acts as a correction factor for material performance degradation. As the temperature increases, the fatigue strength decreases, and the threshold value decreases accordingly.

[0090] The cumulative running time serves as a weighting factor for fatigue accumulation, and the threshold level gradually decreases as the running time increases.

[0091] The final dynamic safety threshold is obtained by weighting the above factors, thus more accurately reflecting the blade's safety boundary under complex operating conditions. When the overall lifespan... Less than the threshold When this happens, the system issues a fault warning signal, prompting maintenance personnel to take timely measures, such as load reduction or shutdown for maintenance.

[0092] In constructing the dynamic threshold, this invention not only considers wind speed fluctuations, ambient temperature, and cumulative operating time, but also classifies and corrects each factor. Specifically:

[0093] Wind speed fluctuations are divided into three levels: small fluctuations, medium fluctuations, and large fluctuations, corresponding to different life reduction factors, which are used to reflect the impact of wind load fluctuations on blade life.

[0094] Ambient temperature is used as a correction factor for material performance degradation. When the temperature increases, the fatigue strength of the material decreases, and the life prediction results are adjusted accordingly.

[0095] Cumulative running time serves as a weighting factor for fatigue accumulation; the longer the running time, the lower the safety threshold, reflecting the gradual accumulation of fatigue effects.

[0096] The final dynamic security threshold is obtained by weighting the above correction factors.

[0097] In this way, the dynamic threshold can more realistically reflect the safety boundary of the blade in a complex operating environment. When the overall lifespan is lower than the threshold, the system outputs a fault warning signal.

[0098] The method of this invention can be implemented through a system combining hardware and software. This system includes the following modules:

[0099] Data acquisition module: The laser anemometer, spindle encoder, pitch angle sensor, fiber optic strain sensor and accelerometer installed on the wind turbine blades and nacelle are used to collect wind speed, rotational speed, pitch angle, strain and vibration signals in real time and transmit them to the data processing unit.

[0100] Digital twin modeling module: Based on finite element modeling technology, a digital twin model of the blade is constructed, and loads determined by aerodynamic parameters are applied to the model to generate a simulated stress field.

[0101] Data assimilation module: The extended Kalman filter algorithm is run in the central processing unit to fuse the simulated stress field with the measured stress signal to obtain the corrected real-time stress field.

[0102] Fatigue life prediction module: It performs rain flow counting on the real-time stress field, combines the material fatigue life curve and cumulative damage theory to calculate the fatigue damage factor and estimate the remaining fatigue life.

[0103] Crack propagation prediction module: Utilizes a segmented crack propagation model to calculate the blade's propagation life from the initial crack length to the critical crack length.

[0104] Comprehensive assessment and early warning module: It integrates fatigue life and crack life to obtain comprehensive life, and compares it with dynamic safety threshold. When the comprehensive life is less than or equal to the threshold, it outputs an early warning signal.

[0105] The system can run on the wind turbine monitoring and data acquisition (SCADA) platform, enabling real-time monitoring of blade status, life prediction, and intelligent fault early warning.

[0106] Example 2: In this example, an onshore wind turbine with a rated power of 3MW is used as an example. The lifespan analysis of the blade is performed using the wind turbine blade life prediction and fault early warning method based on digital twin proposed in this invention.

[0107] In this unit, multiple sensors are first deployed to achieve real-time acquisition of operating parameters: a laser anemometer on the top of the nacelle is used to obtain the incoming wind speed, an encoder at the end of the main shaft is used to obtain the rotational speed, and an angle sensor on the pitch mechanism is used to obtain the pitch angle. During the blade manufacturing stage, fiber optic grating (FBG) strain sensors are pre-embedded in the fiber layers of the upper and lower truss caps, positioned at 0.25R, 0.50R, and 0.75R of the blade spanwise length. The optical fibers are laid parallel to the main stress direction, led through the web to the blade root, and transmitted to the acquisition unit via slip rings. At the same time, a MEMS-type accelerometer is installed on the inner side of the blade root web and connected to the acquisition system via a shielded cable. All sensors sample synchronously at a frequency of 100Hz to ensure the temporal consistency of wind speed, rotational speed, pitch angle, strain, and vibration signals.

[0108] Based on this, a finite element model of the blade was established as a digital twin. Aerodynamic loads were calculated using air density, incoming wind speed, chord length, and aerodynamic coefficients, and then applied to the finite element model to obtain the simulated stress field on the blade surface. At an incoming wind speed of 11 m / s, the simulation results showed that the stress amplitude of the bent main fiber layer at 0.75R of the blade was approximately 55 MPa. Considering the presence of turbulence in actual wind speeds and the potential deviation between simulated and measured stresses, an extended Kalman filter was used to fuse the simulated stress with the observed stress obtained from FBG strain and acceleration inversion. At a certain moment, the observed stress was approximately 58 MPa, the simulated value was 55 MPa, the prediction covariance was set to 100, the observation noise covariance was set to 400, and the Kalman gain calculation results were as follows:

[0109] ;

[0110] The corrected real-time stress is as follows:

[0111] ;

[0112] It is closer to the actual measured state.

[0113] Rainflow counting was performed on the corrected stress time history to obtain the cyclic distribution of different amplitudes: approximately 20 MPa at around 90,000 cycles, approximately 35 MPa at 7,000 cycles, and approximately 55.6 MPa at 800 cycles. These values ​​were then substituted into the SN curve of the blade material (assuming parameters). , ),have:

[0114] ;

[0115] The corresponding calculation result is: the cycle life at 20MPa is approximately 35MPa 55.6MPa Then, based on Miner's linear cumulative damage theory:

[0116] ;

[0117] Assuming the blade has been in operation for many years, the cumulative damage is... Then, according to the extrapolation formula:

[0118] ;

[0119] The remaining fatigue life was approximately 4.74 years.

[0120] The influence of crack propagation is also considered. Let the initial crack length be... Critical crack length The material Paris constant is taken as , Geometric correction factor This represents a stress amplitude of 55.6 MPa. According to the Paris model integral formula:

[0121] ;

[0122] Substituting the values, the crack life for propagation from 2 mm to 50 mm is approximately 2.66 × 10⁻⁶. 7 This cycle. If the average daily cycle number of leaves is approximately 1.7 × 10⁻⁶. 4 The corresponding crack propagation life is:

[0123] ;

[0124] The minimum value between the remaining fatigue life and the crack propagation life is ultimately taken as the overall blade life.

[0125] ;

[0126] To further improve prediction reliability, this invention introduces a dynamic safety threshold. This threshold, which comprehensively considers wind speed fluctuations, ambient temperature, and cumulative operating time, is calculated using the following formula:

[0127] ;

[0128] In this embodiment, the wind speed fluctuation amplitude is 3 m / s, corresponding to the coefficient. The ambient temperature is 35°C, and the corresponding coefficient is... It has been running for a total of 9 years, and the corresponding coefficient is... benchmark threshold .therefore:

[0129] ;

[0130] Since the overall lifespan of 4.29 years exceeds the threshold of 3.85 years, the system does not trigger an early warning, but the margin is only: The warning indicates that the blades are approaching the risk boundary, and monitoring should be strengthened and maintenance should be planned in advance.

[0131] Furthermore, the method described in this embodiment is implemented through a system combining hardware and software. The system includes: a data acquisition module for acquiring wind speed, rotational speed, blade pitch angle, strain, and vibration signals; a digital twin modeling module for establishing a finite element model and generating a simulated stress field for the blade; a data assimilation module for performing extended Kalman filtering and outputting a corrected real-time stress field; a fatigue remaining life prediction module for performing rainflow counting, fatigue life curve calculation, and cumulative damage calculation to obtain the fatigue remaining life; a crack propagation prediction module for performing crack propagation model calculation to obtain the crack propagation life; and a comprehensive evaluation and early warning module for calculating the blade's comprehensive life, comparing it with a dynamic safety threshold, and outputting an early warning signal. This system can be integrated into the SCADA platform of a wind turbine to achieve real-time and engineering applications of blade life prediction and fault early warning.

[0132] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting the lifespan and providing early warning of faults in wind turbine blades based on digital twins, characterized in that, Includes the following steps: S1) Collect wind turbine operating parameters, including wind speed obtained by the laser anemometer on the top of the nacelle, rotational speed obtained by the main shaft encoder, pitch angle obtained by the pitch angle sensor, strain signal obtained by the fiber optic grating strain sensor embedded in the fiber layer of the blade truss cap, and vibration signal obtained by the accelerometer arranged on the inner side of the blade root web, to form multi-source operating data. S2) Based on the geometric and material parameters of the blade, a finite element model is established, and combined with the operating data, aerodynamic-structural coupling simulation is performed to obtain the digital twin simulation stress field of the blade. S3) Using the extended Kalman filter method, the simulated stress field is fused and corrected with the observed stress obtained by inversion from strain signal and vibration signal to obtain the corrected real-time stress field; S4) Perform rainflow counting on the time history data of the real-time stress field to obtain the stress cycle amplitude, average cycle stress, and number of cycles; S5) Calculate the cumulative damage factor of the blade based on the fatigue life curve of the material and the linear cumulative damage theory; S6) Based on the cumulative damage factor and the damage evolution rate, predict the remaining fatigue life of the blade. S7) Based on the crack propagation model and combined with the real-time stress field, calculate the crack propagation life. The crack propagation model considers the initial crack length, critical crack length, stress intensity factor, crack propagation threshold, material fracture toughness, and geometric correction coefficient. S8) The combined fatigue life and crack propagation life are used to obtain the overall life of the blade; S9) Construct a dynamic safety threshold, which is determined based on wind speed fluctuation, ambient temperature, and cumulative operating time; when the overall lifespan is less than or equal to the dynamic safety threshold, output a fault warning signal.

2. The method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins according to claim 1, characterized in that, The fiber optic strain sensor is pre-embedded in the fiber layer of the upper and lower stringers during the blade manufacturing stage, and is arranged at positions of 0.25 times, 0.50 times, and 0.75 times the spanwise length of the blade, and is arranged parallel to the main force direction of the blade; the accelerometer is arranged on the inner side of the blade root web, fixed by structural adhesive, and led to the blade root signal interface through a shielded cable.

3. The method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins according to claim 1, characterized in that, The digital twin simulation stress field establishes a blade structure model using the finite element method, and loads aerodynamic loads determined by wind speed, chord length, aerodynamic coefficient, and air density into the model to obtain the stress distribution of the blade under different operating conditions.

4. The method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins according to claim 1, characterized in that, The real-time stress field is obtained through extended Kalman filtering. The filtering process comprehensively considers the observation matrix, prediction covariance, and observation noise covariance to achieve dynamic fusion of digital twin simulation results and measured stress signals.

5. The method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins according to claim 1, characterized in that, Rainflow counting is performed on the time history data of the real-time stress field to obtain stress cycle parameters. The corresponding cycle life is determined by combining the fatigue life curve of the material. Then, the cumulative damage factor of the blade during operation is calculated based on the linear cumulative damage theory.

6. The method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins according to claim 5, characterized in that, The remaining fatigue life is calculated by extrapolating the cumulative damage factor and combining it with the damage evolution rate. This method is applicable when the cumulative damage factor is less than a critical value.

7. The method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins according to claim 1, characterized in that, The crack propagation life is calculated based on a segmented crack propagation model, which considers the stress intensity factor amplitude, crack propagation threshold, material fracture toughness, safety factor, and geometric correction factor, and calculates the life by integrating the crack propagation process from the initial length to the critical length.

8. A method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins, as described in claim 6 or 7, characterized in that... The combined life is the minimum of the fatigue remaining life and the crack propagation life, used to reflect the true remaining life of the blade.

9. The method for predicting the lifespan and providing early warning of faults of wind turbine blades based on digital twins according to claim 1, characterized in that, The dynamic security threshold is determined as follows: Wind speed fluctuations are divided into different levels, and different levels correspond to different lifespan correction coefficients. Ambient temperature is used as a correction factor for material performance degradation; The cumulative running time is used as a weighting factor for fatigue accumulation; A dynamic safety threshold is obtained through weighted calculation. When the overall lifespan is less than or equal to the threshold, the system triggers a fault warning signal.

10. A wind turbine blade life prediction and fault early warning system based on the method of any one of claims 1-9, characterized in that, include: The data acquisition module is used to collect wind speed, rotational speed, blade pitch angle, strain, and vibration signals. The digital twin modeling module is used to build finite element models and generate simulated stress fields for blades; The data assimilation module is used to perform extended Kalman filtering and output the corrected real-time stress field; The fatigue life prediction module is used to perform rainflow counting, fatigue life curve and cumulative damage calculation to obtain fatigue life; The crack propagation prediction module is used to perform crack propagation model calculations to obtain the crack propagation life. The comprehensive assessment and early warning module is used to calculate the overall blade life and compare it with the dynamic safety threshold, and output an early warning signal.

Images