A wind turbine blade fault diagnosis method based on fiber grating strain gauge

Abstract

The application provides a wind turbine blade fault diagnosis method based on a fiber grating strain gauge, and belongs to the technical field of wind turbine blade fault diagnosis. The fiber grating strain gauge is arranged in the orthogonal layout of a blade root, strain data are collected, after effective working condition screening and azimuth angle alignment, the double-dimension bending moments of waving and oscillation of a blade are calculated through decoupling, the precision diagnosis of a blade stiffness abnormal fault and the positioning of an abnormal blade are completed through the comparison of the homologous load dispersion of three blades.

Description

Technical Field

[0001] This invention belongs to the field of wind turbine blade fault diagnosis technology, and particularly relates to a wind turbine blade fault diagnosis method based on fiber optic strain gauges. Background Technology

[0002] With the rapid development of the wind power industry, the operation and maintenance costs and operational reliability of in-service wind turbines have become core concerns for the industry. As the core component for capturing wind energy, wind turbine blades are subjected to harsh environments such as complex alternating loads, extreme temperature differences, wind and sand erosion, and rain, snow, and icing for extended periods. These conditions make them highly susceptible to surface cracks, substrate fatigue damage, and uneven icing, which in turn lead to a decrease in blade stiffness, causing impeller imbalance and excessive loads. In severe cases, this can result in major safety accidents such as blade breakage and turbine overturning, causing direct economic losses of over ten million yuan per unit.

[0003] Currently, the mainstream health monitoring solutions for wind turbine blades mainly include four categories: vibration monitoring, image recognition monitoring, acoustic emission monitoring, and strain monitoring. All of these have core shortcomings that cannot be resolved. (1) Vibration monitoring scheme: It is susceptible to strong background noise from the rotation of the wind turbine rotor and tower vibration. The characteristic signals of small stiffness changes caused by early blade cracks and slight icing are difficult to extract. It can only identify serious faults, cannot achieve early warning, and cannot accurately locate abnormal blades. (2) Image recognition monitoring scheme: It is greatly affected by severe weather such as rain, snow, fog, and night. It can only identify visible cracks on the surface of the blade, but cannot detect damage to the internal matrix of the blade or hidden cracks at the leading edge, and cannot quantify the degree of change in blade stiffness. (3) Acoustic emission monitoring scheme: It has extremely high requirements for sensor installation location and environmental signal-to-noise ratio. Under the complex working conditions of in-service wind turbines, the signal attenuation is severe and there is a lot of interference noise. It has poor engineering applicability, high installation and maintenance costs, and is difficult to promote and apply on a large scale. (4) Conventional strain monitoring schemes: resistance strain gauges are mostly used, which are easily affected by electromagnetic interference and temperature drift, resulting in poor long-term operational stability. Existing fiber optic strain monitoring schemes mostly compare single-blade strain data with historical health benchmarks, which are greatly affected by fluctuations in operating conditions, have a high false alarm rate, and are difficult to achieve accurate early fault identification.

[0004] In existing technologies, a technical approach has been proposed to calculate blade axial torque based on the axial force of pitch bolts and monitor impeller imbalance faults by detecting the dispersion of the axial torque of the three blades. This approach solves the problem of poor anti-interference capability in traditional vibration monitoring and lays the foundation for the application of same-source load comparison in the field of wind turbine fault monitoring. The core implementation process is as follows: several bolts on the root flange of a single wind turbine blade are replaced with smart bolts, and bolt axial force data is collected; the axial torque of the single blade is calculated based on the bolt axial force data, and impeller azimuth angle and operating condition data are collected simultaneously; with one rotation of the impeller as the cycle, the azimuth angles of the three blades are aligned, and the dispersion of the axial torque of the three blades is calculated; based on the dispersion threshold, it is determined whether there is an impeller imbalance fault, thus realizing impeller imbalance monitoring.

[0005] However, the existing technology still has the following defects: (1) It can only monitor the overall imbalance fault of the impeller, but cannot locate the specific abnormal blade, and cannot identify the stiffness abnormal faults such as cracks and uneven icing of the blade itself; (2) It is based on the axial force of the pitch bolt for monitoring. The bolt axial force is not sensitive to the early small stiffness changes of the blade. It can only identify serious impeller imbalance faults and cannot realize early warning of blade faults; (3) It can only monitor the axial torque in a single dimension. It cannot decouple the load changes in both directions of blade flapping and oscillation. It cannot fully capture stiffness abnormal faults of different positions and types of blades, and is prone to missed detection; (4) It cannot simultaneously take into account the identification of early small stiffness changes and the elimination of operating condition interference. It cannot achieve accurate and stable monitoring of the health status of wind turbine blades under complex operating conditions. Summary of the Invention

[0006] This invention addresses the practical needs of wind turbine blade maintenance by providing a method for wind turbine blade fault diagnosis based on fiber Bragg grating strain gauges. This method is primarily used for: (1) To solve the problems of poor anti-interference ability, inability to achieve early warning of blade failure, and poor engineering applicability of existing conventional monitoring schemes, and to achieve ultra-early identification of micron-level cracks and slight stiffness changes caused by uneven icing on blades; (2) Solve the problem that the closest existing technology can only monitor the overall imbalance of the impeller, cannot locate abnormal blades, and cannot identify abnormal blade stiffness faults, and realize the accurate location of abnormal blades and the initial judgment of fault location; (3) To solve the core pain points of existing fiber optic grating monitoring schemes, which are greatly affected by operating condition fluctuations and have a high false alarm rate, by comparing the same load on the three blades, the interference of operating condition fluctuations such as wind speed, rotation speed, and pitch angle on the monitoring results is completely eliminated, and the false alarm rate is reduced. (4) To solve the problem of missed detection in the existing single-dimensional monitoring scheme, by decoupling the bending moment of flapping and oscillation and linking the dispersion analysis, the stiffness abnormality faults of different positions and types of blades can be fully captured, and the fault identification rate can be greatly improved. (5) To solve the problems of poor environmental adaptability and insufficient long-term operational stability of existing monitoring schemes, fiber optic grating sensing technology is adopted, which is completely immune to electromagnetic interference, lightning strikes and temperature drift, and can operate stably for a long time in onshore wind farms and offshore high humidity and high salt spray environments. (6) It solves the problems of existing monitoring schemes requiring major modifications to the wind turbine structure, high installation and maintenance costs, and difficulty in large-scale promotion. It only requires attaching strain gauges to the blade root section to achieve monitoring, without modifying the original structure of the wind turbine, and is suitable for both newly built and in-service wind turbine retrofits.

[0007] The first aspect of this invention discloses a method for fault diagnosis of wind turbine blades based on fiber optic strain gauges, the method comprising: S1. Perform data acquisition and preprocessing; Fiber Bragg grating strain gauges are orthogonally attached to the blade root sections of three wind turbine blades. Fiber Bragg grating strain data of the three blades, rotor azimuth data of the wind turbine, and operating condition data are collected simultaneously. The strain data are preprocessed to obtain corrected standard strain data.

[0008] S2. Perform effective working condition screening and azimuth alignment; Based on the operating condition data, the effective operating range is selected. Taking one rotation of the impeller as a cycle, the standard strain data of the three blades passing through the same target azimuth angle range in a single cycle is extracted to complete the spatial alignment of the load data of the three blades. S3. Perform two-dimensional bending moment decoupling calculation; Based on the standard strain data within the target azimuth range, the flapping moment and oscillation moment of the three blades are calculated respectively. S4. Perform two-dimensional dispersion calculation; Based on the flapping moment of the three blades under the same target azimuth angle, calculate the flapping side dispersion; based on the oscillation moment of the three blades under the same target azimuth angle, calculate the oscillation side dispersion. S5. Perform initial fault diagnosis; Based on the preset health dispersion threshold, the dispersion of the waving side and the oscillation side are initially judged. If there is an abnormality, the fault location and diagnosis process is initiated.

[0009] S6. Locate and diagnose the faults in the abnormal blades; Calculate the relative deviations of flapping moment and sway moment of a single blade and locate abnormal blades; combine the abnormal characteristics of two-dimensional dispersion to determine that the abnormal blades have stiffness reduction faults caused by icing / cracks / fatigue damage, and complete the fault diagnosis of wind turbine blades.

[0010] According to the method of the first aspect of the present invention, in S1: Fiber grating strain gauges are located on the windward, leeward, leading, and trailing edges of the blade root section, with adjacent strain gauges spaced 90° apart circumferentially and orthogonally distributed. The installation positions, quantities, and specifications of the strain gauges on the three blades are completely identical, and their circumferential numbers correspond one-to-one. The time synchronization error of strain data, impeller azimuth angle data, and operating condition data does not exceed 1ms. The synchronization accuracy is used to maintain the spatiotemporal correspondence between the azimuth angle and strain data and reduce the load calculation deviation introduced by the synchronization error. Preprocessing includes moving average filtering for noise reduction, 3 The criteria include removing outliers, temperature compensation correction, noise elimination, and data jumps and temperature drift interference on strain data. Operating data includes real-time wind speed, impeller speed, and pitch angle of the three blades.

[0011] According to the method of the first aspect of the present invention, in S2, the effective running interval screening condition is: The wind speed is within the stable operating range between the wind turbine's cut-in wind speed and the rated wind speed, with wind speed fluctuations ≤ ±1m / s; The impeller speed is stable, with speed fluctuation ≤ ±5% of the rated speed; The pitch angles of the three blades are synchronized and consistent, with an angle deviation of ≤ ±0.5°.

[0012] According to the method of the first aspect of the present invention, in S2, when performing azimuth alignment: the strain data of the three blades all come from the same spatial orientation and the same working environment, eliminating the interference of working condition fluctuations on load comparison, and providing an unbiased reference for the common source comparison of the three blades.

[0013] According to the method of the first aspect of the present invention, in S3: The formula for calculating the swing moment is: in, For the bending moment of the blade flapping, The elastic modulus of the blade root section is given by [value]. Let be the section modulus of the blade root section about the neutral axis of the flapping motion. This represents the standard strain value of the fiber optic strain gauge on the windward side. The standard strain value of the fiber optic strain gauge on the leeward side; The formula for calculating the sway moment is: in, For the blade oscillation bending moment, Let be the section modulus of the blade root section about the neutral axis of the oscillation. This is the standard strain value of the leading-edge fiber optic strain gauge. This is the standard strain value for the trailing edge fiber optic strain gauge.

[0014] According to the method of the first aspect of the present invention, in S4: The formula for calculating the dispersion on the waving side is: in, This is the arithmetic mean of the flapping moments of the three blades under the same target azimuth angle. The overall standard deviation of the flapping moment of the three blades; The formula for calculating the dispersion on the oscillation side is: in, This represents the arithmetic mean of the swaying bending moments of the three blades under the same target azimuth angle. This represents the overall standard deviation of the oscillation bending moment of the three blades.

[0015] According to the method of the first aspect of the present invention, in S5, the initial fault determination specifically includes: If the dispersion of the waving side And the dispersion on the oscillation side The three leaflets were determined to be in good health. like or If the system is found to be abnormal, the fault location and diagnosis process will begin. in, The preset health dispersion threshold is obtained by calibrating the full-condition data of the wind turbine blade under full health conditions.

[0016] According to the method of the first aspect of the present invention, in S6, locating the abnormal blade specifically includes: Calculate the relative deviation of the flapping moment and the relative deviation of the oscillation moment of the three blades; The relative deviation of the single-blade flapping moment is: in, =1,2,3 correspond to three blades. For the first The flapping moment of the blades; The relative deviation of the bending moment of a single blade is: in, For the first The swaying bending moment of the blades; If a certain leaf or Exceeding the preset single-blade deviation threshold If the deviation is greater than the relative deviation of the other two leaves, the leaf is determined to be an abnormal leaf.

[0017] According to the method of the first aspect of the present invention, in S6, the fault diagnosis includes determining the fault type and fault location, specifically including: If the dispersion of the flapping side and the oscillating side exceeds the standard simultaneously, and the relative deviation of the two-dimensional bending moment of the abnormal blade exceeds the threshold, it is determined that the abnormal blade has a stiffness reduction fault caused by uneven icing / overall fatigue damage. If the dispersion only exceeds the standard on the flapping side, and the abnormal blade only has a flapping bending moment relative deviation exceeding the threshold, the fault is determined to be located on the windward / leeward side of the blade. If only the dispersion on the oscillation side exceeds the standard, and the relative deviation of the oscillation bending moment of the abnormal blade exceeds the threshold, the fault is determined to be located at the leading / trailing edge of the blade.

[0018] A second aspect of this invention discloses a wind turbine blade fault diagnosis system based on fiber Bragg grating strain gauges, the system comprising: Synchronous acquisition unit; configured as: Data acquisition and preprocessing were performed; fiber optic strain gauges were orthogonally attached to the blade root sections of the three blades of the wind turbine, and fiber optic strain data of the three blades, rotor azimuth data of the wind turbine, and operating condition data were collected simultaneously; the strain data were preprocessed to obtain corrected standard strain data.

[0019] The data processing unit is configured as follows: Perform effective operating condition screening and azimuth alignment; based on the operating condition data, screen the effective operating range, take one revolution of the impeller as a cycle, extract the standard strain data of the three blades passing through the same target azimuth interval in a single cycle, and complete the spatial alignment of the load data of the three blades. Perform two-dimensional decoupled bending moment calculations; based on standard strain data within the target azimuth angle range, calculate the flapping moment and oscillation moment of the three blades respectively; Perform two-dimensional dispersion calculation; calculate the dispersion on the flapping side based on the flapping moment of the three blades under the same target azimuth angle; calculate the dispersion on the oscillation side based on the oscillation moment of the three blades under the same target azimuth angle. Fault diagnosis unit; configured as: Perform initial fault assessment; based on the preset health dispersion threshold, perform initial assessment of the dispersion on the waving side and the dispersion on the swing side. If there is an abnormality, proceed to the fault location and diagnosis process. Perform fault location and diagnosis of abnormal blades; calculate the relative deviation of flapping moment and swaying moment of a single blade and locate the abnormal blade; combine the abnormal characteristics of two-dimensional dispersion to determine that the abnormal blade has a stiffness reduction fault caused by icing / cracks / fatigue damage, and complete the fault diagnosis of wind turbine blades. The early warning output unit is configured as follows: Output fault warning information, abnormal blade location results, preliminary fault location judgment results and operation and maintenance suggestions, and connect to the wind farm operation and maintenance platform.

[0020] Compared with the prior art, the present invention has the following beneficial effects: (1) Strong anti-interference ability and complete elimination of the influence of operating condition fluctuations: By comparing the same source loads of the three blades at the same spatial azimuth angle, no matter how the wind speed, rotation speed and pitch angle change, the three blades are always in the same operating condition, which completely eliminates the interference of operating condition fluctuations on the monitoring results. The on-site engineering verification has reduced the false alarm rate of faults. (2) Extremely strong early fault identification capability: It can capture micron-level early cracks in blades and small stiffness changes caused by uneven icing less than 0.5mm. Compared with traditional vibration monitoring schemes, the fault identification time is 6-12 months earlier, and more than 12 months earlier than image recognition schemes, which can effectively avoid major safety accidents caused by the expansion of faults. (3) High diagnostic accuracy, can directly locate abnormal blades and fault locations: Through dual-dimensional discreteness linkage analysis, it can accurately distinguish stiffness abnormal faults at different locations of the blades, directly lock the specific abnormal blades, without the need to stop the machine for disassembly and inspection, and greatly reduce the unplanned downtime of wind turbine units; (4) Strong environmental adaptability and good long-term operational stability: The fiber optic strain gauge is completely immune to electromagnetic interference, lightning strikes and temperature drift. It can operate stably for a long time in extreme environments of -40℃ to 85℃, and is suitable for various onshore and offshore wind turbines. It is especially suitable for harsh offshore environments with high humidity and high salt spray. (5) The project is highly practical and has low modification costs: only four fiber optic strain gauges need to be pasted on the blade root section of the three blades. No modification is required to the wind turbine blades or hub structure. It is suitable for both newly built wind turbines and the modification of existing wind turbines. It is easy to install and maintain and can be widely promoted and applied. (6) Significantly reduce operation and maintenance costs: It can realize early warning of blade failures, avoid blade replacement and unit overhaul caused by the expansion of failures. According to calculations, the annual downtime loss and operation and maintenance costs of a single unit can be reduced by more than RMB 150,000, which has significant economic benefits. (7) Comprehensive fault coverage with no risk of missed detection: Through the decoupling and dispersion analysis of the two-dimensional bending moment of flapping and oscillation, it can comprehensively cover various faults on the windward side, leeward side, leading edge and trailing edge of the blade. Compared with the single-dimensional monitoring scheme, the fault identification rate is increased by more than 40%, which completely solves the problem of missed detection in single-dimensional monitoring. Attached Figure Description

[0021] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 This is a flowchart of a wind turbine blade fault diagnosis method based on fiber optic strain gauges according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the orthogonal distribution of fiber optic strain gauges at the blade root section of a wind turbine blade according to an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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.

[0024] The abbreviations and key terms involved in the embodiments of this invention are defined as follows.

[0025] Fiber Bragg grating strain gauge (FBG strain gauge): A strain sensor with fiber Bragg grating as the core sensing element. It utilizes the characteristic that the Bragg wavelength of the fiber Bragg grating changes with strain and temperature to achieve high-precision acquisition of strain data. It has the characteristics of anti-electromagnetic interference, anti-lightning strike, small temperature drift and strong long-term stability, and is the core sensing element of this solution.

[0026] Wagging moment: The bending moment perpendicular to the plane of rotation of a wind turbine blade is generated by wind energy load during its rotation. It is a core parameter reflecting the load state and structural stiffness of the windward / leeward side of the blade.

[0027] Swaying moment: The bending moment parallel to the plane of rotation of a wind turbine blade generated by wind energy load during rotation is a core parameter reflecting the loading state of the leading / trailing edge of the blade and the structural stiffness.

[0028] Blade root section: The largest section at the root of the wind turbine blade where it connects to the pitch bearing is the key section for load transfer on the blade and the installation location of the strain sensor in this scheme. The strain data of this section can directly reflect the overall load and stiffness state of the blade.

[0029] Dispersion: In this scheme, it specifically refers to the degree of dispersion of the bending moment load parameters of the three blades under the same operating conditions and the same spatial azimuth angle. It is quantified by statistical indicators such as the coefficient of variation, and is used to characterize the consistency of the loads on the three blades. It is the core indicator for initial fault diagnosis.

[0030] Coefficient of variation (CV): A standardized metric used in probability and statistics to measure the dispersion of data. It is the ratio of the population standard deviation to the arithmetic mean. It can eliminate the influence of data magnitude and dimensions on the judgment of dispersion and is the core quantification method of dispersion in this scheme.

[0031] Comparison of loads from the same source: The core technical idea of ​​this solution refers to comparing load data of the three blades under completely consistent operating conditions with the same rotation period and spatial azimuth. This completely eliminates the interference of operating condition fluctuations on load consistency judgment and provides an unbiased benchmark for fault diagnosis. This approach does not rely on any external benchmarks or historical data, fundamentally solving the problem of interference from operating condition fluctuations on monitoring results.

[0032] SCADA system: Supervisory Control And Data Acquisition, which in this solution refers to the wind turbine's built-in operation monitoring system, can provide real-time operating condition data such as wind speed, speed, and pitch angle.

[0033] Fiber Bragg grating demodulator: A signal demodulation device used in conjunction with fiber Bragg grating strain gauges. It can acquire wavelength change data of fiber Bragg gratings in real time and convert it into corresponding strain values. It is the core device for data acquisition in this solution.

[0034] Stiffness reduction: The reduction in structural bending stiffness (including flapping stiffness and yaw stiffness) of wind turbine blades due to problems such as surface cracks, matrix fatigue damage, and uneven icing is the core monitoring target of this scheme and a key early characteristic of blade failure.

[0035] This invention collects strain data using fiber optic strain gauges orthogonally arranged at the blade roots. After effective working condition screening and azimuth alignment, it decouples and calculates the two-dimensional bending moments of blade flapping and oscillation. By comparing the dispersion of the same load on the three blades, it completes the accurate diagnosis of blade stiffness abnormality faults and the location of abnormal blades.

[0036] The first aspect of this invention discloses a method for fault diagnosis of wind turbine blades based on fiber optic strain gauges (the method flow is as follows). Figure 1 (As shown).

[0037] S1: Data Acquisition and Preprocessing.

[0038] Four fiber Bragg grating strain gauges were orthogonally attached to the blade root section of three wind turbine blades. The fiber Bragg grating strain data of the three blades, the rotor azimuth angle data of the wind turbine, and the operating condition data were collected simultaneously. The strain data were preprocessed to obtain the corrected standard strain data.

[0039] Fiber Bragg grating strain gauge installation rules: (e.g.) Figure 2 As shown, four fiber optic strain gauges are located on the windward, leeward, leading, and trailing edges of the blade root section, respectively. Adjacent strain gauges are circumferentially spaced at 90° and orthogonally distributed. The installation positions, quantities, and specifications of the strain gauges on the three blades are completely identical, and their circumferential numbers correspond one-to-one.

[0040] Synchronization requirement: The time synchronization error of strain data, impeller azimuth angle data, and operating condition data is ≤1ms. This synchronization accuracy can ensure the spatiotemporal correspondence between azimuth angle and strain data and avoid load calculation deviations caused by synchronization errors.

[0041] Preprocessing operations: including moving average filtering for noise reduction, 3 The criteria remove outliers and perform temperature compensation corrections to eliminate interference from noise, data jumps, and temperature drift on strain data.

[0042] Operating data includes real-time wind speed, impeller speed, and pitch angle of the three blades.

[0043] S2: Effective operating condition screening and azimuth alignment.

[0044] Based on the operating condition data, the effective operating range is selected. Taking one rotation of the impeller as a cycle, the standard strain data of the three blades passing through the same target azimuth angle range in a single cycle is extracted to complete the spatial alignment of the load data of the three blades.

[0045] Valid operating range filtering criteria (must be met simultaneously): The wind speed is within the stable operating range between the wind turbine's cut-in wind speed and the rated wind speed, with wind speed fluctuations ≤ ±1m / s; The impeller speed is stable, with speed fluctuation ≤ ±5% of the rated speed; The pitch angles of the three blades are synchronized and consistent, with an angle deviation of ≤ ±0.5°.

[0046] Azimuth alignment logic: ensures that the strain data of the three blades all come from the same spatial orientation and the same working environment, completely eliminates the interference of working condition fluctuations on load comparison, and provides an unbiased reference for the common source comparison of the three blades.

[0047] S3: Two-dimensional bending moment decoupling calculation.

[0048] Based on the standard strain data within the target azimuth angle range, the flapping moment and oscillation moment of the three blades are calculated respectively.

[0049] Swing moment calculation formula: ;in, For the bending moment of the blade flapping, The elastic modulus of the blade root section is given by [value]. Let be the section modulus of the blade root section about the neutral axis of the flapping motion. This represents the standard strain value of the fiber optic strain gauge on the windward side. This represents the standard strain value of the fiber optic strain gauge on the leeward side.

[0050] Formula for calculating the moment of oscillation: ;in, For the blade oscillation bending moment, Let be the section modulus of the blade root section about the neutral axis of the oscillation. This is the standard strain value of the leading-edge fiber optic strain gauge. This is the standard strain value for the trailing edge fiber optic strain gauge.

[0051] S4: Two-dimensional discreteness calculation.

[0052] The flapping moment of the three blades at the same target azimuth angle is used to calculate the dispersion on the flapping side; the oscillation moment of the three blades at the same target azimuth angle is used to calculate the dispersion on the oscillation side. This scheme uses the coefficient of variation to standardize the dispersion.

[0053] Swing-side dispersion: ;in, This is the arithmetic mean of the flapping moments of the three blades under the same target azimuth angle. The overall standard deviation of the flapping moment of the three blades; Swing side dispersion: ;in, This represents the arithmetic mean of the swaying bending moments of the three blades under the same target azimuth angle. This represents the overall standard deviation of the oscillation bending moment of the three blades.

[0054] S5: Initial fault diagnosis.

[0055] Based on the preset health dispersion threshold, the dispersion of the waving side and the oscillation side are initially judged. If there is an abnormality, the fault location and diagnosis process is initiated.

[0056] Initial fault diagnosis rule: If the dispersion on the waving side... And the dispersion on the oscillation side The three leaflets were determined to be in a healthy state; if or If the system is found to be abnormal, the fault location and diagnosis process will begin. Threshold explanation: The preset health dispersion threshold is obtained by calibrating the full-condition data of the wind turbine blade under full health conditions.

[0057] S6: Abnormal blade location and fault diagnosis.

[0058] The relative deviations of flapping moment and sway moment of a single blade are calculated to identify abnormal blades. By combining the abnormal characteristics of two-dimensional dispersion, it is determined that the abnormal blade has stiffness reduction faults caused by icing / cracks / fatigue damage, thus completing the health monitoring and fault diagnosis of wind turbine blades.

[0059] Abnormal blade location includes: Calculate the relative deviations of the flapping moment and the oscillation moment of the three blades: Relative deviation of single-blade flapping moment: ;in, =1,2,3 correspond to three blades. For the first The flapping moment of the blades; Relative deviation of single-blade oscillation bending moment: ;in, For the first The swaying bending moment of the blades; If a certain leaf or Exceeding the preset single-blade deviation threshold The deviation of this leaf is significantly greater than the relative deviation of the other two leaves, thus the leaf is determined to be an abnormal leaf.

[0060] The rules for determining fault type and fault location include: If the dispersion of the flapping side and the oscillating side exceeds the standard simultaneously, and the relative deviation of the two-dimensional bending moment of the abnormal blade exceeds the threshold, it is determined that the abnormal blade has a stiffness reduction fault caused by uneven icing / overall fatigue damage. If the dispersion only exceeds the standard on the flapping side, and the abnormal blade only has a flapping bending moment relative deviation exceeding the threshold, the fault is determined to be located on the windward / leeward side of the blade. If only the dispersion on the oscillation side exceeds the standard, and the relative deviation of the oscillation bending moment of the abnormal blade exceeds the threshold, the fault is determined to be located at the leading / trailing edge of the blade.

[0061] The second aspect of this invention discloses a wind turbine blade fault diagnosis system based on fiber Bragg grating strain gauges. Used to execute the aforementioned fault diagnosis method, the system comprises four core functional units, all interconnected to form a complete monitoring-diagnosis-early warning closed loop, as detailed below: Synchronous acquisition unit: includes fiber optic strain gauge group, fiber optic demodulator, azimuth acquisition module, operating condition acquisition module, and time synchronization module. Its core function is to synchronously acquire blade strain data, impeller azimuth data, and operating condition data, ensuring that the time synchronization accuracy of all data meets the requirements. Data processing unit: Communicates with the synchronous acquisition unit. Its core functions are to perform data preprocessing, effective working condition screening, two-dimensional bending moment calculation, and two-dimensional dispersion calculation, providing standardized calculation data for fault diagnosis. Fault diagnosis unit: It communicates with the data processing unit and its core functions are to perform initial fault judgment, abnormal blade location, fault type and location determination, and complete the entire fault diagnosis logic. Early warning output unit: It communicates with the fault diagnosis unit. Its core functions are to output fault early warning information, abnormal blade location results, preliminary fault location judgment results and operation and maintenance suggestions, and connect to the wind farm operation and maintenance platform.

[0062] Compared with the prior art, the present invention has the following beneficial effects: (1) Strong anti-interference ability and complete elimination of the influence of operating condition fluctuations: By comparing the same source loads of the three blades at the same spatial azimuth angle, no matter how the wind speed, rotation speed and pitch angle change, the three blades are always in the same operating condition, which completely eliminates the interference of operating condition fluctuations on the monitoring results. The on-site engineering verification has reduced the false alarm rate of faults. (2) Extremely strong early fault identification capability: It can capture micron-level early cracks in blades and small stiffness changes caused by uneven icing less than 0.5mm. Compared with traditional vibration monitoring schemes, the fault identification time is 6-12 months earlier, and more than 12 months earlier than image recognition schemes, which can effectively avoid major safety accidents caused by the expansion of faults. (3) High diagnostic accuracy, can directly locate abnormal blades and fault locations: Through dual-dimensional discreteness linkage analysis, it can accurately distinguish stiffness abnormal faults at different locations of the blades, directly lock the specific abnormal blades, without the need to stop the machine for disassembly and inspection, and greatly reduce the unplanned downtime of wind turbine units; (4) Strong environmental adaptability and good long-term operational stability: The fiber optic strain gauge is completely immune to electromagnetic interference, lightning strikes and temperature drift. It can operate stably for a long time in extreme environments of -40℃ to 85℃, and is suitable for various onshore and offshore wind turbines. It is especially suitable for harsh offshore environments with high humidity and high salt spray. (5) The project is highly practical and has low modification costs: only four fiber optic strain gauges need to be pasted on the blade root section of the three blades. No modification is required to the wind turbine blades or hub structure. It is suitable for both newly built wind turbines and the modification of existing wind turbines. It is easy to install and maintain and can be widely promoted and applied. (6) Significantly reduce operation and maintenance costs: It can realize early warning of blade failures, avoid blade replacement and unit overhaul caused by the expansion of failures. According to calculations, the annual downtime loss and operation and maintenance costs of a single unit can be reduced by more than RMB 150,000, which has significant economic benefits. (7) Comprehensive fault coverage with no risk of missed detection: Through the decoupling and dispersion analysis of the two-dimensional bending moment of flapping and oscillation, it can comprehensively cover various faults on the windward side, leeward side, leading edge and trailing edge of the blade. Compared with the single-dimensional monitoring scheme, the fault identification rate is increased by more than 40%, which completely solves the problem of missed detection in single-dimensional monitoring.

[0063] The following are alternative solutions for achieving the core objective of this invention, all of which can achieve the same technical effect.

[0064] Replacement of strain gauge quantity: The number of fiber optic strain gauges on the root section of a single blade can be selected as 8, 12, etc. Regardless of the number of strain gauges, a symmetrical orthogonal layout in four directions (windward side, leeward side, leading edge, and trailing edge) must be maintained to ensure that the physical basis of bending moment decoupling remains unchanged. For example, two strain gauges can be attached to each of the windward side, leeward side, leading edge, and trailing edge of the blade root section to form redundancy and improve the reliability of data acquisition.

[0065] Alternative to discrete quantification: In addition to the coefficient of variation, statistical indicators such as range, standard deviation, and variance can be used for discrete quantification. It is only necessary to achieve a quantitative assessment of the degree of dispersion of the load on the three blades.

[0066] Alternative target azimuth angle: The target azimuth angle can be any fixed azimuth angle such as 90°, 180°, 270°, etc. It is only necessary to ensure that the data is aligned under the same azimuth angle within the same rotation cycle of the three blades, without being limited to 0° azimuth angle; multiple target azimuth angles can also be selected for multiple sets of data comparison to further improve the stability of diagnostic results.

[0067] Alternative threshold calibration methods: Health dispersion threshold, single blade deviation threshold, etc., can be calibrated using health big data of the same type of unit, or the industry-standard wind turbine blade load design threshold can be used, without being limited to the specific values ​​in the examples.

[0068] Alternatives to data preprocessing methods: In addition to moving average filtering, Kalman filtering, wavelet denoising, and other methods can be used for data denoising, which only requires eliminating noise interference in the data; temperature compensation can also be achieved by adding an independent fiber optic temperature sensor, without being limited to the differential temperature compensation method in the embodiment.

[0069] Replacement for compatible units: This solution is compatible with horizontal axis wind turbines of various power ratings of 1.5MW, 3MW, 5MW, 8MW and above. It is only necessary to adjust the installation position of the strain gauges and the calibration parameters of the bending moment calculation formula according to the blade root cross-section size and material parameters.

[0070] Alternatives to fault determination logic: Classification algorithms such as Support Vector Machine (SVM), Random Forest, or Deep Neural Network (DNN) can be used, with the flapping moment and oscillation moment of the three blades at the same azimuth angle as input features, and the degree of dispersion deviation and blade deviation as classification labels, to achieve intelligent identification of fault types.

[0071] Please note that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. The above embodiments only illustrate several implementations of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the appended claims.

Claims

1. A method for fault diagnosis of wind turbine blades based on fiber optic strain gauges, characterized in that, The method includes: S1. Perform data acquisition and preprocessing; Fiber Bragg grating strain gauges are orthogonally attached to the blade root sections of three wind turbine blades. Fiber Bragg grating strain data of the three blades, rotor azimuth data of the wind turbine, and operating condition data are collected simultaneously. The strain data are preprocessed to obtain corrected standard strain data. S2. Perform effective working condition screening and azimuth alignment; Based on the operating condition data, the effective operating range is selected. Taking one rotation of the impeller as a cycle, the standard strain data of the three blades passing through the same target azimuth angle range in a single cycle is extracted to complete the spatial alignment of the load data of the three blades. S3. Perform two-dimensional bending moment decoupling calculation; Based on the standard strain data within the target azimuth range, the flapping moment and oscillation moment of the three blades are calculated respectively. S4. Perform two-dimensional dispersion calculation; Based on the flapping moment of the three blades under the same target azimuth angle, calculate the flapping side dispersion; based on the oscillation moment of the three blades under the same target azimuth angle, calculate the oscillation side dispersion. S5. Perform initial fault diagnosis; Based on the preset health dispersion threshold, the dispersion of the waving side and the oscillation side are initially judged. If there is an abnormality, the fault location and diagnosis process is initiated. S6. Locate and diagnose the faults in the abnormal blades; Calculate the relative deviations of flapping moment and sway moment of a single blade and locate abnormal blades; combine the abnormal characteristics of two-dimensional dispersion to determine that the abnormal blades have stiffness reduction faults caused by icing / cracks / fatigue damage, and complete the fault diagnosis of wind turbine blades.

2. The method for fault diagnosis of wind turbine blades based on fiber optic strain gauges according to claim 1, characterized in that, In S1: Fiber grating strain gauges are located on the windward, leeward, leading, and trailing edges of the blade root section, with adjacent strain gauges spaced 90° apart circumferentially and orthogonally distributed. The installation positions, quantities, and specifications of the strain gauges on the three blades are completely identical, and their circumferential numbers correspond one-to-one. The time synchronization error of strain data, impeller azimuth angle data, and operating condition data does not exceed 1ms. The synchronization accuracy is used to maintain the spatiotemporal correspondence between the azimuth angle and strain data and reduce the load calculation deviation introduced by the synchronization error. Preprocessing includes moving average filtering for noise reduction, 3 The criteria include removing outliers, temperature compensation correction, noise elimination, and data jumps and temperature drift interference on strain data. Operating data includes real-time wind speed, impeller speed, and pitch angle of the three blades.

3. The method for fault diagnosis of wind turbine blades based on fiber optic strain gauges according to claim 2, characterized in that, In S2, the effective running interval filtering criteria are: The wind speed is within the stable operating range between the wind turbine's cut-in wind speed and the rated wind speed, with wind speed fluctuations ≤ ±1m / s; The impeller speed is stable, with speed fluctuation ≤ ±5% of the rated speed; The pitch angles of the three blades are synchronized and consistent, with an angle deviation of ≤ ±0.5°.

4. The method for fault diagnosis of wind turbine blades based on fiber optic strain gauges according to claim 3, characterized in that, In S2, when performing azimuth alignment: the strain data of the three blades all come from the same spatial orientation and the same working environment, eliminating the interference of working condition fluctuations on load comparison and providing an unbiased reference for the common source comparison of the three blades.

5. The method for fault diagnosis of wind turbine blades based on fiber optic strain gauges according to claim 4, characterized in that, In S3: The formula for calculating the swing moment is: in, For the bending moment of the blade flapping, The elastic modulus of the blade root section is given by [value]. Let be the section modulus of the blade root section about the neutral axis of the flapping motion. This represents the standard strain value of the fiber optic strain gauge on the windward side. The standard strain value of the fiber optic strain gauge on the leeward side; The formula for calculating the sway moment is: in, For the blade oscillation bending moment, Let be the section modulus of the blade root section about the neutral axis of the oscillation. This is the standard strain value of the leading-edge fiber optic strain gauge. This is the standard strain value for the trailing edge fiber optic strain gauge.

6. The method for fault diagnosis of wind turbine blades based on fiber optic strain gauges according to claim 5, characterized in that, In S4: The formula for calculating the dispersion on the waving side is: in, This is the arithmetic mean of the flapping moments of the three blades under the same target azimuth angle. The overall standard deviation of the flapping moment of the three blades; The formula for calculating the dispersion on the oscillation side is: in, This represents the arithmetic mean of the swaying bending moments of the three blades under the same target azimuth angle. This represents the overall standard deviation of the oscillation bending moment of the three blades.

7. The method for fault diagnosis of wind turbine blades based on fiber optic strain gauges according to claim 6, characterized in that, In S5, the initial fault diagnosis specifically includes: If the dispersion of the waving side And the dispersion on the oscillation side The three leaflets were determined to be in good health. like or If the system is found to be abnormal, the fault location and diagnosis process will begin. in, The preset health dispersion threshold is obtained by calibrating the full-condition data of the wind turbine blade under full health conditions.

8. The method for fault diagnosis of wind turbine blades based on fiber optic strain gauges according to claim 7, characterized in that, In S6, locating abnormal blades specifically includes: Calculate the relative deviation of the flapping moment and the relative deviation of the oscillation moment of the three blades; The relative deviation of the single-blade flapping moment is: in, =1,2,3 correspond to three blades. For the first The flapping moment of the blades; The relative deviation of the bending moment of a single blade is: in, For the first The swaying bending moment of the blades; If a certain leaf or Exceeding the preset single-blade deviation threshold If the deviation is greater than the relative deviation of the other two leaves, the leaf is determined to be an abnormal leaf.

9. A method for fault diagnosis of wind turbine blades based on fiber optic strain gauges according to claim 8, characterized in that, In S6, fault diagnosis includes fault type and fault location determination, specifically including: If the dispersion of the flapping side and the oscillating side exceeds the standard simultaneously, and the relative deviation of the two-dimensional bending moment of the abnormal blade exceeds the threshold, it is determined that the abnormal blade has a stiffness reduction fault caused by uneven icing / overall fatigue damage. If the dispersion only exceeds the standard on the flapping side, and the abnormal blade only has a flapping bending moment relative deviation exceeding the threshold, the fault is determined to be located on the windward / leeward side of the blade. If only the dispersion on the oscillation side exceeds the standard, and the abnormal blade only has a relative deviation in oscillation bending moment exceeding the threshold, the fault is determined to be located at the leading / trailing edge of the blade.

10. A wind turbine blade fault diagnosis system based on fiber optic strain gauges, characterized in that, The system includes: Synchronous acquisition unit; configured as: Data acquisition and preprocessing were performed; fiber optic strain gauges were orthogonally attached to the blade root sections of the three blades of the wind turbine, and fiber optic strain data of the three blades, rotor azimuth data of the wind turbine, and operating condition data were collected simultaneously; the strain data were preprocessed to obtain corrected standard strain data. The data processing unit is configured as follows: Perform effective operating condition screening and azimuth alignment; based on the operating condition data, screen the effective operating range, take one revolution of the impeller as a cycle, extract the standard strain data of the three blades passing through the same target azimuth interval in a single cycle, and complete the spatial alignment of the load data of the three blades. Perform two-dimensional decoupled bending moment calculations; based on standard strain data within the target azimuth angle range, calculate the flapping moment and oscillation moment of the three blades respectively; Perform two-dimensional dispersion calculation; calculate the dispersion on the flapping side based on the flapping moment of the three blades under the same target azimuth angle; calculate the dispersion on the oscillation side based on the oscillation moment of the three blades under the same target azimuth angle. Fault diagnosis unit; configured as: Perform initial fault assessment; based on the preset health dispersion threshold, perform initial assessment of the dispersion on the waving side and the oscillation side. If any abnormality is found, proceed to the fault location and diagnosis process. Perform fault location and diagnosis of abnormal blades; calculate the relative deviation of flapping moment and swaying moment of a single blade and locate the abnormal blade; combine the abnormal characteristics of two-dimensional dispersion to determine that the abnormal blade has a stiffness reduction fault caused by icing / cracks / fatigue damage, and complete the fault diagnosis of wind turbine blades. The early warning output unit is configured as follows: Output fault warning information, abnormal blade location results, preliminary fault location judgment results and operation and maintenance suggestions, and connect to the wind farm operation and maintenance platform.

Images