A multi-degree of freedom vibration detection system and method for wind turbine blades

Abstract

The application discloses a multi-degree-of-freedom vibration detection system and method for a wind turbine blade, relates to the field of wind power generation equipment monitoring, and comprises the following steps: collecting wind turbine blade operation vibration data, processing the vibration data according to operation stages, extracting main frequency information to determine a vibration state, constructing a vibration characteristic space to establish a mapping relationship among multi-degree-of-freedom vibration components, and calculating a vibration coupling degree; when the vibration coupling degree exceeds a preset threshold value, decoupling the vibration components into independent vibration sources, tracing evolution rules of the vibration sources to determine a dominant vibration source; establishing a vibration propagation chain combination to construct a correlation mapping network, calculating a blade group vibration risk index based on a propagation characteristic of the dominant vibration source, and realizing early warning of blade group vibration. The method can effectively detect abnormal vibration states of the wind turbine blade, discover potential faults in time, and improve operation safety and reliability of the wind turbine.

Description

Technical Field

[0001] This invention relates to the field of wind power equipment monitoring technology, specifically to a multi-degree-of-freedom vibration detection system and method for wind turbine blades. Background Technology

[0002] With the rapid development of wind power generation technology, large wind turbine generators have been widely used globally. As a key component of wind turbine generators, the operating condition of wind turbine blades directly affects the power generation efficiency and service life of the generator. During actual operation, wind turbine blades are subjected to various factors such as complex airflow, aerodynamic loads, and structural loads, resulting in vibrations. These vibrations are characterized by multiple degrees of freedom, nonlinearity, and coupling, posing significant challenges to vibration monitoring and fault diagnosis of wind turbine blades.

[0003] Vibration detection of wind turbine blades mainly employs single-point or multi-point independent measurement methods, which are insufficient to accurately reflect the coupling characteristics of multi-degree-of-freedom vibration of the blades. Existing vibration detection methods often focus only on single characteristics such as vibration amplitude and frequency, lacking in-depth analysis of the vibration source propagation law and vibration coupling mechanism, resulting in the inability to detect and warn of potential vibration risks in a timely manner.

[0004] While some sensor network-based vibration monitoring methods have been proposed in the prior art, these methods suffer from difficulties in signal decoupling and inaccurate vibration source identification when processing multi-degree-of-freedom vibration signals. Furthermore, the lack of an effective vibration risk assessment model makes it difficult to achieve accurate early warning of the vibration status of wind turbine blades. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-degree-of-freedom vibration detection system and method for wind turbine blades, which can accurately detect multi-degree-of-freedom vibration of wind turbine blades, effectively identify vibration sources, and provide timely early warning of vibration risks, thereby improving the accuracy and reliability of wind turbine blade vibration monitoring and ensuring the safe and stable operation of wind turbine generator sets.

[0006] In a first aspect, an embodiment of the present invention provides a method for detecting multi-degree-of-freedom vibration of wind turbine blades, comprising the following steps:

[0007] Collect vibration data of wind turbine blades during operation;

[0008] The vibration data is segmented according to the operating stage of the wind turbine blade, the main frequency information of each operating stage is extracted, and the vibration state of the wind turbine blade is determined based on the main frequency information.

[0009] A vibration feature space is constructed based on the vibration state. A mapping relationship between the multi-degree-of-freedom vibration components of the wind turbine blade is established in the vibration feature space. The vibration coupling degree of the wind turbine blade is calculated based on the mapping relationship.

[0010] When the vibration coupling degree exceeds the preset coupling degree threshold, the multi-degree-of-freedom vibration components are decoupled into independent vibration sources, the evolution of the independent vibration sources over time is tracked, and the dominant vibration source of abnormal vibration of the wind turbine blade is determined.

[0011] By establishing the vibration propagation chain of the dominant vibration source, the vibration propagation chain is combined to construct an association mapping network. Based on the propagation characteristics of the dominant vibration source in the association mapping network, the vibration risk index of the blade group is calculated, thereby realizing early warning of blade group vibration.

[0012] Furthermore, the vibration data is segmented according to the operating stages of the wind turbine blades, and the dominant frequency information of each operating stage is extracted. Based on the dominant frequency information, the vibration state of the wind turbine blades is determined, including:

[0013] The vibration signal is subjected to adaptive threshold denoising to obtain the preprocessed vibration signal;

[0014] Time-frequency analysis was performed on the preprocessed vibration data to obtain instantaneous frequency characteristics;

[0015] The frequency similarity between adjacent time windows is calculated based on the instantaneous frequency characteristics. When the frequency similarity is less than a preset similarity threshold, it is determined as a segmentation point of the operation phase, and the vibration data is divided into multiple operation phases.

[0016] Spectral analysis is performed on the vibration data for each operating stage to identify the dominant frequency information in the vibration data, construct the trend of the dominant frequency information over time, and determine the vibration state transition characteristics of the wind turbine blades based on the trend.

[0017] Based on the vibration state transition characteristics, a state discrimination index is constructed, and the correspondence between the state discrimination index and the vibration state is established to determine the vibration state of the wind turbine blade.

[0018] Furthermore, a mapping relationship is established between the multi-degree-of-freedom vibration components of the wind turbine blade in the vibration characteristic space, and the vibration coupling degree of the wind turbine blade is calculated based on the mapping relationship, including:

[0019] The vibration signal is decomposed in the vibration feature space to obtain vibration components with multiple degrees of freedom;

[0020] Calculate the area of ​​the overlapping region of vibration components of each degree of freedom in the vibration feature space, and extract the phase information and amplitude information of the vibration components within the overlapping region;

[0021] The phase difference matrix of the vibration component is calculated based on the phase information, and the amplitude ratio matrix of the vibration component is calculated based on the amplitude information. The combined eigenvalue of the phase difference matrix and the amplitude ratio matrix is ​​used as the vibration coupling degree of the wind turbine blade.

[0022] Furthermore, calculating the area of ​​the overlapping region of the vibration components of each degree of freedom in the vibration characteristic space includes:

[0023] Perform Hilbert transform on the vibration components of each degree of freedom to obtain analytical signals, calculate the instantaneous frequency and instantaneous amplitude of the vibration components based on the analytical signals, and construct an energy density distribution function that includes time and frequency dimensions;

[0024] A basic grid is constructed in the time-frequency plane, and the rate of change of the energy density distribution function at adjacent grid points is calculated. When the rate of change exceeds a preset rate of change threshold, the number of grid points in the current region is increased to obtain an adaptive grid.

[0025] For any two vibration components, the energy density distribution function is used to calculate the energy density value at each grid point of the adaptive grid, and the minimum of the two energy density values ​​is determined as the overlapping energy density of that grid point.

[0026] Based on the overlapping energy density, the adaptive mesh is numerically integrated in different regions, and the area of ​​the overlapping region between vibration components is calculated based on the integration results.

[0027] Furthermore, the multi-degree-of-freedom vibration components are decoupled into independent vibration sources, and the evolution of these independent vibration sources over time is tracked to determine the dominant vibration sources of abnormal vibration in wind turbine blades, including:

[0028] Construct a vibration mode matrix containing vibration components of each degree of freedom;

[0029] The vibration mode matrix is ​​decomposed into singular value decomposition to obtain eigenvectors. Based on the orthogonality of the eigenvectors, the multi-degree-of-freedom vibration components are decoupled into independent vibration sources.

[0030] Calculate the vibration energy sequence of an independent vibration source, establish a time-series correlation matrix based on the vibration energy sequence, and track the evolution law of the independent vibration source;

[0031] Based on the evolution law of independent vibration sources, identify groups of independent vibration sources with energy transfer relationships, construct vibration propagation chains of independent vibration source groups, calculate the triggering delay time and energy attenuation ratio between vibration sources in the vibration propagation chain, and determine the source vibration.

[0032] Establish the triggering intensity matrix and energy contribution matrix of the source vibration, calculate the triggering intensity, and take the product of the energy proportion of the source vibration and the energy attenuation ratio in the energy contribution matrix as the energy contribution.

[0033] The dominant vibration source of abnormal vibration of wind turbine blades is determined based on the combined score of the triggering intensity and energy contribution.

[0034] Furthermore, the vibration energy sequence of the independent vibration source is calculated, and a time-series correlation matrix is ​​established based on the vibration energy sequence to track the evolution law of the independent vibration source, including:

[0035] Extract the energy time series of independent vibration sources;

[0036] The energy time series is segmented and accumulated to obtain an energy accumulation curve;

[0037] Calculate the location of the abrupt change point between the energy accumulation curves of adjacent independent vibration sources, and determine the triggering time of energy transfer based on the location of the abrupt change point;

[0038] Based on the triggering time, the independent vibration sources are arranged in the order of energy transfer to construct the transmission sequence of the vibration sources;

[0039] In the transmission sequence, the acceleration and decay intervals of energy transmission are identified, and the evolution law of the vibration source is determined based on the distribution of the acceleration and decay intervals.

[0040] Furthermore, by establishing the vibration propagation chain of the dominant vibration source, a correlation mapping network is constructed by combining the vibration propagation chains. Based on the propagation characteristics of the dominant vibration source in the correlation mapping network, the vibration risk index of the blade assembly is calculated, thereby achieving early warning of blade assembly vibration, including:

[0041] Collect vibration signals from the dominant vibration source;

[0042] Time-frequency analysis is performed on the vibration signal to obtain the amplitude spectrum and phase spectrum;

[0043] The amplitude ratio of adjacent measuring points on each blade is calculated based on the amplitude spectrum, and the phase difference of adjacent measuring points on each blade is calculated based on the phase spectrum. The amplitude ratio is used as the energy attenuation coefficient, and the phase difference is used as the transmission delay time.

[0044] A vibration propagation chain is constructed based on the energy attenuation coefficient and the transmission delay time. The cross-correlation degree between the vibration propagation chains is calculated. The vibration propagation chains are classified according to the cross-correlation degree, and an association mapping network containing the propagation hierarchy relationship is constructed.

[0045] Extract the propagation time and propagation intensity of each propagation level in the association mapping network, and superimpose the propagation intensity at the intersection node of the propagation level;

[0046] The vibration risk index of the blade assembly is calculated based on the transmission time and transmission intensity, and the vibration warning level is determined according to the vibration risk index.

[0047] Secondly, embodiments of the present invention provide a multi-degree-of-freedom vibration detection system for wind turbine blades, used to implement the method described in any of the foregoing claims, the system comprising:

[0048] The vibration data acquisition module is used to collect vibration data of wind turbine blades during operation.

[0049] The operation phase analysis module is used to segment the vibration data according to the operation phase of the wind turbine blade, extract the main frequency information of each operation phase, and determine the vibration state of the wind turbine blade based on the main frequency information.

[0050] The coupling degree calculation module is used to construct a vibration feature space based on the vibration state, establish a mapping relationship between the multi-degree-of-freedom vibration components of the wind turbine blade in the vibration feature space, and calculate the vibration coupling degree of the wind turbine blade according to the mapping relationship.

[0051] The vibration source identification module is used to decouple the multi-degree-of-freedom vibration components into independent vibration sources when the vibration coupling degree exceeds a preset coupling degree threshold, track the evolution of the independent vibration sources over time, and determine the dominant vibration source of abnormal vibration of the wind turbine blades.

[0052] The vibration early warning module is used to establish a vibration propagation chain of the dominant vibration source, combine the vibration propagation chains to construct an association mapping network, and calculate the blade group vibration risk index based on the propagation characteristics of the dominant vibration source in the association mapping network, so as to realize early warning of blade group vibration.

[0053] Thirdly, one technical solution provided in this embodiment of the invention is an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0054] Fourthly, one technical solution provided in this embodiment of the invention is a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the steps in any of the aforementioned methods.

[0055] By collecting and segmenting vibration data from wind turbine blades, and combining this with analysis of dominant frequency information, the vibration state of the blades can be accurately identified. By constructing a vibration characteristic space and establishing mapping relationships between multi-degree-of-freedom vibration components, vibration coupling can be effectively assessed, achieving precise decoupling of complex vibrations. By tracing the evolution of independent vibration sources, the main causes of abnormal blade vibrations can be identified in a timely manner. By establishing vibration propagation chains and correlation mapping networks, the propagation characteristics of vibration within the blade assembly can be accurately grasped, enabling quantitative assessment and early warning of vibration risks. This method overcomes the shortcomings of existing technologies in processing multi-degree-of-freedom vibration signals, improves the accuracy and reliability of vibration monitoring, and provides effective technical support for fault diagnosis and preventative maintenance of wind turbine blades, possessing significant engineering application value.

[0056] The above description of the invention is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art 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.

[0058] Figure 1 A flowchart of a method for detecting multi-degree-of-freedom vibration of wind turbine blades provided in an embodiment of the present invention;

[0059] Figure 2 This is a flowchart illustrating the calculation of the vibration coupling degree of a wind turbine blade according to an embodiment of the present invention.

[0060] Figure 3 This is a schematic diagram of a multi-degree-of-freedom vibration detection system for wind turbine blades, provided as an embodiment of the present invention. Detailed Implementation

[0061] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. In the following description relating to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements.

[0062] The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of this specification. It should be noted that the steps of the corresponding methods are not necessarily performed in the order shown and described in this specification in other embodiments. In some other embodiments, the methods may include more or fewer steps than described in this specification. Furthermore, a single step described in this specification may be broken down into multiple steps in other embodiments; and multiple steps described in this specification may be combined into a single step in other embodiments.

[0063] like Figure 1 As shown, Figure 1 A flowchart of a method for detecting multi-degree-of-freedom vibration of wind turbine blades provided by an embodiment of the present invention, the method comprising the following steps:

[0064] Collect vibration data of wind turbine blades during operation;

[0065] The vibration data is segmented according to the operating stage of the wind turbine blade, the main frequency information of each operating stage is extracted, and the vibration state of the wind turbine blade is determined based on the main frequency information.

[0066] A vibration feature space is constructed based on the vibration state. A mapping relationship between the multi-degree-of-freedom vibration components of the wind turbine blade is established in the vibration feature space. The vibration coupling degree of the wind turbine blade is calculated based on the mapping relationship.

[0067] When the vibration coupling degree exceeds the preset coupling degree threshold, the multi-degree-of-freedom vibration components are decoupled into independent vibration sources, the evolution of the independent vibration sources over time is tracked, and the dominant vibration source of abnormal vibration of the wind turbine blade is determined.

[0068] By establishing the vibration propagation chain of the dominant vibration source, the vibration propagation chain is combined to construct an association mapping network. Based on the propagation characteristics of the dominant vibration source in the association mapping network, the vibration risk index of the blade group is calculated, thereby realizing early warning of blade group vibration.

[0069] In one optional embodiment, the vibration data is segmented according to the operating stages of the wind turbine blades, and the dominant frequency information of each operating stage is extracted. The vibration state of the wind turbine blades is determined based on the dominant frequency information, including:

[0070] The vibration signal is subjected to adaptive threshold denoising to obtain the preprocessed vibration signal;

[0071] Time-frequency analysis was performed on the preprocessed vibration data to obtain instantaneous frequency characteristics;

[0072] The frequency similarity between adjacent time windows is calculated based on the instantaneous frequency characteristics. When the frequency similarity is less than a preset similarity threshold, it is determined as a segmentation point of the operation phase, and the vibration data is divided into multiple operation phases.

[0073] Spectral analysis is performed on the vibration data for each operating stage to identify the dominant frequency information in the vibration data, construct the trend of the dominant frequency information over time, and determine the vibration state transition characteristics of the wind turbine blades based on the trend.

[0074] Vibration state transition characteristics are used to construct state discrimination indexes, establish the correspondence between the state discrimination indexes and vibration states, and determine the vibration state of wind turbine blades.

[0075] This embodiment first performs adaptive threshold denoising on the vibration signal to obtain a preprocessed vibration signal. Specifically, the original vibration signal is processed using an adaptive threshold wavelet transform denoising method. This method automatically adjusts the threshold value based on the local characteristics of the signal, effectively removing noise components from the wind turbine blade vibration signal. In practical applications, the vibration signal is decomposed into four levels using the db4 wavelet basis function, with different thresholds applied to the wavelet coefficients at each level. The threshold calculation is determined based on the variance of the wavelet coefficients and the number of samples; a larger threshold is used for high-frequency detail coefficients, and a smaller threshold is used for low-frequency detail coefficients. For example, for the collected wind turbine blade vibration signal, 0.85 times the standard deviation is applied as the threshold for the first level of detail coefficients, and 0.25 times the standard deviation is applied as the threshold for the fourth level of detail coefficients, achieving adaptive noise suppression while preserving the main features of the signal.

[0076] Time-frequency analysis was performed on the preprocessed vibration data to obtain instantaneous frequency characteristics. The Hilbert-Huang transform was used for time-frequency analysis of the vibration signal. This method first decomposes the signal into multiple intrinsic mode functions (EMFs) through empirical mode decomposition, and then performs a Hilbert transform on each EMF to obtain instantaneous frequency information. For wind turbine blade vibration signals, the signal is typically decomposed into 5-8 EMFs, each representing a natural vibration mode. The instantaneous frequency of each EMF is calculated using the Hilbert transform to construct a time-spectrum, reflecting the frequency variation characteristics of the vibration signal over time. In practical applications, a Hanning window with 50% overlap is used for sliding window processing. The window length is 512 sampling points, and the sliding step size is 256 sampling points, which can effectively capture the frequency characteristic changes of wind turbine blades under different operating conditions.

[0077] The frequency similarity between adjacent time windows is calculated based on instantaneous frequency characteristics. When the frequency similarity is less than a preset similarity threshold, it is determined as a segmentation point of the operating phase, thus dividing the vibration data into multiple operating phases. Frequency similarity is measured by calculating the cosine similarity of the dominant frequency components of adjacent time windows. In wind turbine blade vibration detection, the similarity threshold is typically set to 0.75. When the frequency similarity between adjacent windows is lower than 0.75, it indicates a significant change in the operating state of the wind turbine blade, which is then marked as a segmentation point of an operating phase. In this way, the vibration data of wind turbine blades can be divided into multiple operating phases, such as the startup phase, stable operation phase, speed change phase, and shutdown phase. In practice, a typical wind turbine blade operating cycle may contain 3-5 different operating phases, each with unique vibration characteristics.

[0078] Spectral analysis was performed on vibration data from each operational phase to identify the dominant frequency information and construct a trend of its variation over time. Based on this trend, the vibration state transition characteristics of the wind turbine blades were determined. Within each operational phase, the spectrum of the vibration signal was calculated using Fast Fourier Transform (FFT), and the three frequency components with the largest amplitudes were extracted as the dominant frequency information. For each operational phase, statistical characteristics such as the mean, standard deviation, rate of change, and energy distribution ratio of the dominant frequency value were calculated to construct a time-series curve of the dominant frequency information. During stable operation, the dominant frequency value of a normally operating wind turbine blade typically fluctuates within a range of no more than 5%, and the dominant frequency energy accounts for more than 60% of the total energy. However, under abnormal conditions, the fluctuation amplitude of the dominant frequency value increases, reaching more than 15%, and the dominant frequency energy ratio decreases to below 40%, accompanied by an increase in subharmonic component energy. By analyzing these characteristics, the transition process of the wind turbine blade from a normal state to an abnormal state can be identified.

[0079] Based on the characteristics of vibration state transition, a state discrimination index is constructed, and a correspondence between the state discrimination index and the vibration state is established to determine the vibration state of the wind turbine blade. The state discrimination index includes a dominant frequency stability index, a frequency distribution index, and an energy concentration index. The dominant frequency stability index is calculated by dividing the standard deviation of the dominant frequency value over an operating phase by the mean; under normal conditions, this value is less than 0.08. The frequency distribution index reflects the dispersion of the spectral peaks and is obtained by calculating the variance of the interval between the first five peak frequencies; under normal conditions, this value is greater than 2.5. The energy concentration index represents the proportion of dominant frequency energy to total energy; under normal conditions, this value is greater than 0.6. These three indices are weighted and fused to construct a comprehensive evaluation index. According to historical data, a comprehensive evaluation index less than 0.3 is considered a normal state, greater than 0.7 is considered a severely abnormal state, and between 0.3 and 0.7 is considered a slightly abnormal state.

[0080] This method achieves accurate judgment of vibration status by segmenting wind turbine blade vibration data according to operational stages and extracting dominant frequency information. Compared with traditional methods, this method can adaptively identify different operational stages, reduce environmental interference, and improve the accuracy of status judgment. Automatic segmentation through frequency similarity calculation avoids the subjectivity of manual stage division and improves analysis efficiency. The status discrimination model built based on dominant frequency information has strong adaptability, can distinguish between normal fluctuations and abnormal changes, provides a basis for decision-making in preventive maintenance of wind turbine blades, extends equipment lifespan, reduces downtime losses due to failures, and improves the overall reliability and economic benefits of wind power generation systems.

[0081] like Figure 2 The diagram illustrates the flow of the wind turbine blade vibration coupling degree calculation method in this embodiment.

[0082] In one optional embodiment, a mapping relationship is established between the multi-degree-of-freedom vibration components of the wind turbine blade in the vibration characteristic space, and the vibration coupling degree of the wind turbine blade is calculated based on the mapping relationship, including:

[0083] The vibration signal is decomposed in the vibration feature space to obtain vibration components with multiple degrees of freedom;

[0084] Calculate the area of ​​the overlapping region of vibration components of each degree of freedom in the vibration feature space, and extract the phase information and amplitude information of the vibration components within the overlapping region;

[0085] The phase difference matrix of the vibration component is calculated based on the phase information, and the amplitude ratio matrix of the vibration component is calculated based on the amplitude information. The combined eigenvalue of the phase difference matrix and the amplitude ratio matrix is ​​used as the vibration coupling degree of the wind turbine blade.

[0086] This implementation first decomposes the vibration signal into vibration components with multiple degrees of freedom in the vibration feature space. The vibration feature space is a high-dimensional data representation containing three basic dimensions: time, frequency, and energy. The decomposition process employs a multi-component decomposition technique, which combines the advantages of wavelet transform and empirical mode decomposition, effectively handling nonlinear and non-stationary vibration signals. Specifically, the original vibration signal is first preprocessed, including trend term removal and noise suppression. Trend term removal uses a high-pass filter with a cutoff frequency of 0.1 Hz; noise suppression employs adaptive threshold wavelet denoising, using the db4 wavelet basis function for a 4-level decomposition, and applying soft thresholding to the wavelet coefficients. The preprocessed vibration signal is then decomposed into vibration components with three main degrees of freedom: flapping direction, oscillation direction, and torsional direction, using orthogonal projection. Orthogonal projection utilizes the geometric characteristics of the wind turbine blades and the sensor placement to establish a coordinate transformation matrix, converting the vibration signal in the measurement coordinate system into vibration components in the blade's physical coordinate system. For a typical wind turbine blade, the vibration frequency in the flapping direction is mainly concentrated in the range of 0.8-1.5 Hz, the vibration frequency in the swaying direction is mainly concentrated in the range of 1.2-2.0 Hz, and the vibration frequency in the torsional direction is mainly concentrated in the range of 2.5-4.0 Hz. By performing time-frequency analysis on the vibration components of each degree of freedom after decomposition, a vibration characteristic space representation including three dimensions of time, frequency, and vibration amplitude is constructed.

[0087] The overlapping area of ​​vibration components of each degree of freedom in the vibration feature space is calculated, and the phase and amplitude information of the vibration components within the overlapping area is extracted. The overlapping area refers to the region where the energy distributions of vibration components of different degrees of freedom intersect on the time-frequency plane, representing the degree of coupling of vibration energy between different degrees of freedom. The area of ​​the overlapping area is calculated using the intersection operation of the time-frequency energy distribution function. In specific implementation, a continuous wavelet transform is first performed on the vibration components of each degree of freedom to obtain the time-frequency energy distribution map. The continuous wavelet transform uses the Morlet wavelet, with the scale parameter set in the range of 1-64, corresponding to a frequency range of 0.5-20 Hz, and a time resolution of 0.05 seconds. A grid is defined on the time-frequency plane, with the grid cell size being 0.1 seconds in the time dimension and 0.2 Hz in the frequency dimension.

[0088] For each grid cell, the normalized energy density value of each degree of freedom vibration component is calculated. Regions with energy density values ​​greater than a preset threshold of 0.1 are marked as active regions for that degree of freedom. When active regions of two or more degrees of freedom overlap in the same grid cell, that grid cell is marked as an overlapping region. The area of ​​the overlapping region is equal to the sum of the areas of all overlapping grid cells. Within the overlapping region, the instantaneous phase and instantaneous amplitude information of each vibration component are extracted using a Hilbert transform. The instantaneous phase is represented by angles, ranging from 0 to 360 degrees; the instantaneous amplitude is normalized to a maximum amplitude of 1.

[0089] The phase difference matrix of the vibration components is calculated based on phase information, and the amplitude ratio matrix of the vibration components is calculated based on amplitude information. The combined eigenvalues ​​of the phase difference matrix and the amplitude ratio matrix are used as the vibration coupling degree of the wind turbine blade. The phase difference matrix is ​​a symmetric matrix, where each element represents the absolute value of the average phase difference between vibration components of different degrees of freedom. The phase difference is calculated using a statistical averaging method of instantaneous phases within the overlapping region. For each pair of vibration components, the absolute value of their instantaneous phase difference is calculated, and then the average value is calculated over the entire overlapping region. For the vibration of a three-degree-of-freedom wind turbine blade, the phase difference matrix is ​​a 3×3 symmetric matrix with diagonal elements of 0. The amplitude ratio matrix is ​​also a symmetric matrix, where each element represents the average amplitude ratio between vibration components of different degrees of freedom. The amplitude ratio is calculated using the root mean square ratio of instantaneous amplitudes within the overlapping region. For each pair of vibration components, the ratio of their root mean square instantaneous amplitudes is calculated, and then the average value is calculated over the entire overlapping region. For a normally operating wind turbine blade, the average phase difference between the flapping direction and the oscillation direction is about 80-100 degrees, the average phase difference between the flapping direction and the torsional direction is about 110-130 degrees, and the average phase difference between the oscillation direction and the torsional direction is about 140-160 degrees. In terms of amplitude ratio, the amplitude ratio between the flapping direction and the oscillation direction is about 1.3-1.7, the amplitude ratio between the flapping direction and the torsional direction is about 2.0-2.5, and the amplitude ratio between the oscillation direction and the torsional direction is about 1.5-2.0.

[0090] Vibration coupling degree is calculated based on the combined eigenvalues ​​of the phase difference matrix and the amplitude ratio matrix. First, the phase difference matrix is ​​decomposed using eigenvalue decomposition to extract the maximum eigenvalue; similarly, the amplitude ratio matrix is ​​decomposed using eigenvalue decomposition to extract the maximum eigenvalue. The maximum eigenvalue of the phase difference matrix reflects the degree of phase coupling between vibration components; a smaller value indicates stronger phase coupling. The maximum eigenvalue of the amplitude ratio matrix reflects the degree of amplitude imbalance between vibration components; a larger value indicates a greater amplitude difference. The vibration coupling degree is calculated by combining these two eigenvalues: the product of the reciprocal of the maximum eigenvalue of the phase difference matrix and the maximum eigenvalue of the amplitude ratio matrix, multiplied by an adjustment coefficient of 0.01. The vibration coupling degree typically ranges from 0 to 1; a larger value indicates stronger coupling between vibration components. The vibration coupling degree of a normally operating wind turbine blade is typically in the range of 0.1-0.3. When the vibration coupling degree reaches 0.4-0.6, it indicates potential early structural damage to the blade. When the vibration coupling degree exceeds 0.7, it indicates serious structural problems with the blade, requiring immediate inspection and maintenance.

[0091] This method achieves accurate assessment of the structural health of wind turbine blades by analyzing the coupling characteristics of multi-degree-of-freedom vibration signals. Compared with traditional single-degree-of-freedom vibration monitoring, this method can capture the interaction between different degrees of freedom, providing more comprehensive vibration characteristic information. Through time-frequency domain analysis and vibration coupling degree calculation, it can detect blade structural anomalies at an early stage, such as material fatigue, loose connections, and crack propagation. This technology effectively improves the accuracy of fault diagnosis for wind turbine blades, reduces the false alarm rate, provides strong support for preventive maintenance of wind power generation equipment, extends equipment service life, and improves power generation efficiency and economic benefits.

[0092] In an optional embodiment, calculating the area of ​​the overlapping region of the vibration components of each degree of freedom in the vibration characteristic space includes:

[0093] Perform Hilbert transform on the vibration components of each degree of freedom to obtain analytical signals, calculate the instantaneous frequency and instantaneous amplitude of the vibration components based on the analytical signals, and construct an energy density distribution function that includes time and frequency dimensions;

[0094] A basic grid is constructed in the time-frequency plane, and the rate of change of the energy density distribution function at adjacent grid points is calculated. When the rate of change exceeds a preset rate of change threshold, the number of grid points in the current region is increased to obtain an adaptive grid.

[0095] For any two vibration components, the energy density distribution function is used to calculate the energy density value at each grid point of the adaptive grid, and the minimum of the two energy density values ​​is determined as the overlapping energy density of that grid point.

[0096] Based on the overlapping energy density, the adaptive mesh is numerically integrated in different regions, and the area of ​​the overlapping region between vibration components is calculated based on the integration results.

[0097] In this embodiment, the collected vibration signals of each degree of freedom of the wind turbine blade are first preprocessed, including filtering, noise reduction, and normalization. Taking the three main degrees of freedom of the wind turbine blade (flapping direction, swaying direction, and torsional direction) as an example, Hilbert transforms are performed on each vibration component of the preprocessed degree of freedom. The Hilbert transform is achieved by convolving the original signal with a Hilbert kernel function, resulting in orthogonal components of the original signal. The original signal is then combined with its orthogonal components to form a complex analytic signal. Taking the flapping direction vibration signal of the wind turbine blade collected at a sampling frequency of 1024Hz as an example, the analytic signal obtained after the Hilbert transform is the real part of the original signal and the imaginary part is the Hilbert transform result of the original signal.

[0098] The instantaneous frequency and instantaneous amplitude of the vibration components are calculated based on the analytic signal. The instantaneous amplitude is obtained by calculating the magnitude of the analytic signal, representing the magnitude of the signal amplitude at each moment. The instantaneous frequency is obtained by calculating the time derivative of the phase angle of the analytic signal, representing the vibration frequency of the signal at each moment. Under normal operating conditions of wind turbine blades, the instantaneous amplitude in the flapping direction typically fluctuates within the range of 0.5 to 2 mm, and the instantaneous frequency is mainly concentrated in the range of 1 to 3 Hz; the instantaneous amplitude in the oscillation direction fluctuates within the range of 0.3 to 1.5 mm, and the instantaneous frequency is mainly concentrated in the range of 2 to 4 Hz; the instantaneous amplitude in the torsional direction is smaller, within the range of 0.1 to 0.8 mm, and the instantaneous frequency is higher, mainly concentrated in the range of 6 to 10 Hz. Based on the instantaneous amplitude and instantaneous frequency, an energy density distribution function containing time and frequency dimensions is constructed. This function represents the distribution of vibration signal energy in the time-frequency plane. The calculation method is to map the square of the instantaneous amplitude onto the time-frequency plane, where the time resolution is set to 0.1 s and the frequency resolution is set to 0.1 Hz, forming the energy density distribution in the time-frequency plane.

[0099] A basic mesh is constructed in the time-frequency plane to achieve a fine representation of the energy density distribution. The basic mesh is initially set as a uniform mesh with a time interval of 0.5 s and a frequency interval of 0.5 Hz. The rate of change of the energy density distribution function at adjacent mesh points is calculated. The rate of change is defined as the difference in energy density values ​​between two adjacent mesh points divided by the distance between the two points. When the rate of change exceeds a preset threshold, it indicates that the energy density in that region is changing drastically, requiring a finer mesh for accurate description. The preset threshold is determined based on the energy change characteristics under normal operating conditions and is typically set to 20% of the average energy density. In regions with drastic energy density changes, the original mesh is subdivided, with the time and frequency intervals halved, forming a locally refined mesh. After multiple iterations of subdivision, an adaptive mesh is finally obtained. This mesh has a high density in regions with drastic energy density changes and a low density in regions with gradual energy density changes, achieving efficient utilization of computational resources. In wind turbine blade vibration analysis, the mesh density is typically higher in regions with abrupt frequency changes (such as near the resonance point) and abrupt amplitude changes (such as the impact response stage).

[0100] For any two vibration components' energy density distribution functions, the energy density value is calculated at each grid point of the adaptive grid to determine the overlapping energy density. Taking the vibration components in the waving and swinging directions as an example, the energy density values ​​E1 and E2 in the two directions are calculated at each grid point of the adaptive grid. The minimum of the two energy density values ​​is determined as the overlapping energy density at that grid point. The reason for choosing the minimum value as the overlapping energy density is that both vibration components must have energy at that point to form coupling, and the coupling strength is limited by the weaker one.

[0101] Numerical integration is performed on the adaptive mesh based on overlapping energy density to calculate the area of ​​overlapping regions between vibration components. The numerical integration employs the trapezoidal integration method, dividing the adaptive mesh into multiple sub-regions and applying the trapezoidal formula to calculate the integral value within each sub-region. Specifically, for each mesh cell, the average of the overlapping energy density values ​​at its four vertices is multiplied by the area of ​​the mesh cell to obtain the integral value for that cell. The integral values ​​of all mesh cells are summed to obtain the total overlapping region area. In the vibration analysis of wind turbine blades, the overlapping regions of the three pairs of vibration components—flapping-swaying, flapping-torsion, and swaying-torsion—are calculated separately to comprehensively evaluate the vibration coupling state of the blade.

[0102] The calculated overlapping area serves as a quantitative indicator of the vibration coupling degree of wind turbine blades, used to assess the blade's health status. A threshold for the overlapping area is set, and the range of the overlapping area under normal operating conditions is determined based on historical data and expert experience. When the calculated overlapping area exceeds the threshold, an anomaly alarm is triggered. The threshold setting can employ a multi-level strategy: a slight exceedance triggers a warning, while a significant exceedance triggers an alarm. In the wind turbine blade operation monitoring system, the overlapping area is combined with other vibration characteristic indicators to construct a comprehensive evaluation model, enabling accurate diagnosis of the blade's health status.

[0103] This method achieves accurate analysis of multi-degree-of-freedom vibration of wind turbine blades through Hilbert transform and adaptive mesh technology, enabling precise assessment of the coupling relationships between vibration components. Compared to traditional fixed mesh methods, adaptive mesh technology significantly reduces computational complexity and improves analysis efficiency while maintaining computational accuracy. The instantaneous characteristics obtained by Hilbert transform can capture the dynamic changes of vibration signals, offering higher time-frequency resolution than traditional Fourier analysis. This technology has high detection sensitivity for structural anomalies in wind turbine blades, enabling the identification of potential problems at an early stage. This provides a basis for preventative maintenance decisions, extends equipment lifespan, reduces maintenance costs, and improves the overall operational efficiency and economic benefits of wind farms.

[0104] In one optional embodiment, the multi-degree-of-freedom vibration components are decoupled into independent vibration sources, and the evolution of these independent vibration sources over time is tracked to determine the dominant vibration source of the abnormal vibration of the wind turbine blades, including:

[0105] Construct a vibration mode matrix containing vibration components of each degree of freedom;

[0106] The vibration mode matrix is ​​decomposed into singular value decomposition to obtain eigenvectors. Based on the orthogonality of the eigenvectors, the multi-degree-of-freedom vibration components are decoupled into independent vibration sources.

[0107] Calculate the vibration energy sequence of an independent vibration source, establish a time-series correlation matrix based on the vibration energy sequence, and track the evolution law of the independent vibration source;

[0108] Based on the evolution law of independent vibration sources, identify groups of independent vibration sources with energy transfer relationships, construct vibration propagation chains of independent vibration source groups, calculate the triggering delay time and energy attenuation ratio between vibration sources in the vibration propagation chain, and determine the source vibration.

[0109] Establish the triggering intensity matrix and energy contribution matrix of the source vibration, calculate the triggering intensity, and take the product of the energy proportion of the source vibration and the energy attenuation ratio in the energy contribution matrix as the energy contribution.

[0110] The dominant vibration source of abnormal vibration of wind turbine blades is determined based on the combined score of the triggering intensity and energy contribution.

[0111] For example, a vibration mode matrix containing vibration components of each degree of freedom is first constructed. The vibration mode matrix consists of vibration data collected from multiple measuring points on the wind turbine blade, with each measuring point corresponding to vibration components in three directions (flapping, swaying, and torsional). Vibration data acquisition is accomplished using accelerometers positioned at key locations on the blade, with a sampling frequency of 1024 Hz and a data length of 60 s. Sensors are typically placed at the root, 1 / 4 length, 1 / 2 length, 3 / 4 length, and tip of the wind turbine blade. The number of rows in the constructed vibration mode matrix equals the number of sensors multiplied by the number of directions, and the number of columns equals the number of sampling points. For example, with 5 measuring points, each with vibration components in 3 directions, a vibration mode matrix of 15 rows and 61,440 columns is formed.

[0112] Singular value decomposition (SVD) is performed on the vibration mode matrix to isolate independent vibration sources. SVD decomposes the vibration mode matrix into the product of three matrices: a left singular matrix, a singular value diagonal matrix, and the transpose of the right singular matrix. The column vectors of the left singular matrix form eigenvectors, representing spatial modes; the diagonal elements of the singular value diagonal matrix reflect the energy magnitude of each mode; and the row vectors of the right singular matrix represent temporal modes. The eigenvectors are orthogonal, enabling the decoupling of multi-degree-of-freedom vibration components into independent vibration sources. In wind turbine blade vibration analysis, the eigenvectors corresponding to the first 5 to 8 largest singular values ​​are typically selected as the main vibration modes, which usually explain more than 90% of the vibration energy. For example, under normal operating conditions, the first vibration source of a wind turbine blade corresponds to the fundamental frequency vibration, with a characteristic frequency of approximately 1.2 Hz and an energy share of 45%; the second vibration source corresponds to second-order vibration, with a characteristic frequency of approximately 3.6 Hz and an energy share of 25%; and the third vibration source corresponds to torsional vibration, with a characteristic frequency of approximately 7.2 Hz and an energy share of 15%.

[0113] The vibration energy sequence of independent vibration sources is calculated to track their evolution. The vibration energy sequence is obtained by calculating the energy changes of each independent vibration source within a short time window. The time series is divided into several time windows of 2 seconds each with a 50% overlap. The energy value of each independent vibration source within each window is calculated, forming an energy time series. A time-series correlation matrix is ​​established based on the vibration energy sequence. This matrix describes the correlation of the energies of each vibration source under different time windows. The time-series correlation matrix is ​​constructed by calculating the Pearson correlation coefficient between the energy sequences. The matrix element values ​​range from -1 to 1, with values ​​closer to 1 indicating stronger correlations. By analyzing the characteristic patterns of the time-series correlation matrix, the evolution of independent vibration sources over time can be tracked. Under normal operating conditions, the energy sequences of each vibration source exhibit periodic changes and stable correlations; while under abnormal conditions, the energy sequences show abrupt changes, and the correlations change significantly.

[0114] Based on the evolution patterns of independent vibration sources, groups of independent vibration sources with energy transfer relationships are identified. These relationships are determined through cross-correlation analysis, calculating the cross-correlation function between the energy sequences of different vibration sources to determine the time delay between energy peaks. If the energy sequences of two vibration sources exhibit a significant correlation and a stable time delay relationship, an energy transfer relationship is considered to exist between them. Based on the identified energy transfer relationships, a vibration propagation chain for the independent vibration source group is constructed. The vibration propagation chain is a directed graph structure, where nodes represent independent vibration sources and edges represent energy transfer paths.

[0115] Calculate the trigger delay time and energy attenuation ratio between vibration sources in the vibration propagation chain. The trigger delay time is the time difference between the occurrence of the energy peak, and the energy attenuation ratio is the ratio of the peak energy of the subsequent vibration source to the peak energy of the preceding vibration source. For example, when a wind turbine blade vibrates abnormally, the trigger delay time between the fourth vibration source and the second vibration source is 0.3s, and the energy attenuation ratio is 0.72, indicating that the fourth vibration source is triggered by the second vibration source, and that approximately 28% of the energy is lost during the transmission process.

[0116] The source vibration is determined based on the vibration propagation chain. A source vibration is an independent vibration source that is not triggered by other vibration sources but can trigger them. In the vibration propagation chain, the node with an in-degree of 0 is the source vibration. By analyzing the vibration propagation chains in multiple abnormal vibration events, the frequency of each vibration source acting as a source is statistically analyzed, and the vibration source with the highest frequency is identified as the primary source vibration. A triggering intensity matrix and an energy contribution matrix for the source vibration are established. The triggering intensity matrix describes the probability of the source vibration triggering other vibration sources; the matrix element values ​​are the proportion of the number of times the source vibration triggers a specific vibration source out of the total number of triggers. The energy contribution matrix describes the contribution of the source vibration to the overall vibration energy; the matrix element values ​​are the proportion of the source vibration's energy in the total energy. The triggering intensity is calculated by the weighted average of the elements in the triggering intensity matrix, with the weights set as the relative importance of the triggered vibration source. The energy contribution is the product of the source vibration's energy proportion and its energy attenuation ratio in the energy contribution matrix. For example, the triggering intensity of the second vibration source was 0.65, and its energy contribution was 0.38. It had the highest overall score and was identified as the dominant vibration source.

[0117] The dominant vibration source of abnormal vibration in wind turbine blades is determined based on a combined score of trigger intensity and energy contribution. The combined score is calculated using a weighted average, with trigger intensity having a weight of 0.6 and energy contribution a weight of 0.4. The vibration source with the highest combined score is identified as the dominant vibration source. After identifying the dominant vibration source, the physical mechanism of the abnormal vibration and possible causes of failure can be inferred by combining the characteristic frequency and spatial distribution of the vibration source. If the characteristic frequency of the dominant vibration source is close to the blade's natural frequency and is mainly distributed at a specific location on the blade, it may indicate structural damage at that location; if the characteristic frequency is related to the rotor rotation frequency, it may indicate blade imbalance or aerodynamic problems.

[0118] This method achieves source identification and anomaly diagnosis of multi-degree-of-freedom vibrations in wind turbine blades through vibration modal decomposition and propagation chain analysis. Compared with traditional vibration analysis methods, this method not only focuses on the spectral characteristics of vibration signals but also considers the temporal correlation and energy transfer relationships between vibration sources, enabling more accurate identification of anomaly sources in complex vibration systems. By decoupling coupled vibrations into independent vibration sources through singular value decomposition, it solves the challenge of multi-degree-of-freedom vibration signal analysis. Vibration propagation chain analysis reveals the transmission path of vibration energy, providing a deeper physical explanation for fault diagnosis. This method has high computational efficiency and adaptability, capable of handling vibration anomaly detection under different operating conditions, effectively improving the accuracy and reliability of wind turbine blade vibration monitoring. It provides a powerful tool for preventive maintenance and fault diagnosis of wind power equipment, reduces maintenance costs, improves equipment availability, and extends the service life of wind turbine blades.

[0119] In one optional embodiment, calculating the vibration energy sequence of an independent vibration source, establishing a time-series correlation matrix based on the vibration energy sequence, and tracing the evolution of the independent vibration source includes:

[0120] Extract the energy time series of independent vibration sources;

[0121] The energy time series is segmented and accumulated to obtain an energy accumulation curve;

[0122] Calculate the location of the abrupt change point between the energy accumulation curves of adjacent independent vibration sources, and determine the triggering time of energy transfer based on the location of the abrupt change point;

[0123] Based on the triggering time, the independent vibration sources are arranged in the order of energy transfer to construct the transmission sequence of the vibration sources;

[0124] In the transmission sequence, the acceleration and decay intervals of energy transmission are identified, and the evolution law of the vibration source is determined based on the distribution of the acceleration and decay intervals.

[0125] In this embodiment, the energy time series of independent vibration sources is first extracted. An array of accelerometers is arranged on the wind turbine blades to collect vibration data in the flapping, swaying, and torsional directions. The sampling frequency is set to 1024 Hz, and the sampling duration is 10 minutes. The collected vibration data is organized into a matrix, with rows representing measurement points at different locations and directions, and columns representing the time series. Singular value decomposition is performed on this matrix to extract the main eigenvectors as independent vibration sources. Typically, the top 5 to 8 eigenvectors with the largest eigenvalues ​​are selected, as these eigenvectors can explain more than 90% of the energy of the original signal. The instantaneous energy of each independent vibration source is calculated, which is the square of the vibration signal amplitude, forming an energy time series.

[0126] The energy time series of independent vibration sources is accumulated in segments to obtain energy accumulation curves. Segmented accumulation aims to eliminate random fluctuations in the energy series and highlight the main trend of energy change. A sliding window technique is used, with a window length of 2 seconds and a sliding step of 0.5 seconds. Within each window, the cumulative energy values ​​are calculated to obtain the accumulated energy for that window. The accumulated energies of all windows are arranged in chronological order to form the energy accumulation curve. The energy accumulation curve reflects the accumulation process of vibration energy over time, and the slope of the curve represents the energy growth rate. Under normal operating conditions, the energy accumulation curve of wind turbine blades usually shows a stable linear growth trend; however, under abnormal operating conditions, the energy accumulation curve will show abrupt slope changes, indicating a sudden change in vibration energy. For example, the energy accumulation curve of the second independent vibration source shows a sudden slope change at 78 seconds, with the energy growth rate increasing from 0.05 J / s to 0.32 J / s, indicating a sudden energy surge at that moment.

[0127] The location of abrupt change points between the energy accumulation curves of adjacent independent vibration sources is calculated to determine the triggering time of energy transfer. The location of the abrupt change point is identified by calculating the change in the first derivative (slope) of the energy accumulation curve. Specifically, the moving average slope of the energy accumulation curve is calculated with a window length of 1 second; the standard deviation of the slope sequence is calculated; points where the slope change exceeds three times the standard deviation are identified as potential abrupt change points; the abrupt change points of adjacent independent vibration sources are compared over time. If the time interval between two abrupt change points is less than a preset threshold (usually 0.5 seconds), an energy transfer relationship is considered to exist. The order of the abrupt change points determines the direction of energy transfer; the vibration source with the first abrupt change is the triggering source, and the vibration source with the second abrupt change is the triggered source. The triggering time is defined as the moment when the triggered source experiences an abrupt change. For example, if the second independent vibration source experiences an energy abrupt change at 78 seconds and the third independent vibration source experiences an energy abrupt change at 78.3 seconds, the time interval between them is 0.3 seconds, which is less than the preset threshold of 0.5 seconds. Therefore, it is determined that the second vibration source triggered the third vibration source, and the triggering time is 78.3 seconds.

[0128] Based on the triggering time, independent vibration sources are sorted according to the order of energy transfer to construct a transmission sequence. The transmission sequence is a directed linked list structure, where nodes represent independent vibration sources, edges represent energy transfer relationships, and the direction of the edges indicates the direction of energy transfer. The construction method is as follows: all independent vibration sources are sorted according to the time of their first energy abrupt change; the triggering relationship between adjacent vibration sources is identified; and directed connections are established to form a transmission chain. If multiple vibration sources are triggered simultaneously (with a time interval less than the resolution threshold, typically 0.1 s), they are arranged side-by-side in the transmission sequence. The identified transmission sequence is: second vibration source → third vibration source → first vibration source → fourth vibration source. This indicates that the abnormal vibration initially appears at the second vibration source (in the oscillation direction) and then sequentially triggers other vibration sources. This energy transfer path is of great significance for understanding the propagation mechanism of vibration anomalies.

[0129] In the energy transfer sequence, the acceleration and decay intervals are identified to determine the evolution pattern of the vibration source. An acceleration interval is defined as the interval where the peak energy of a subsequent vibration source is greater than that of an preceding vibration source during energy transfer; a decay interval is defined as the interval where the peak energy of a subsequent vibration source is less than that of an preceding vibration source. The calculation method is as follows: compare the peak energy values ​​of adjacent vibration sources in the transfer sequence; calculate the energy ratio (peak energy of the subsequent source / peak energy of the preceding source); intervals with an energy ratio greater than 1 are considered acceleration intervals, and intervals with an energy ratio less than 1 are considered decay intervals. In the vibration analysis of wind turbine blades, an acceleration interval indicates that the vibration is amplified during transmission, potentially indicating resonance or a positive feedback mechanism; a decay interval indicates that the vibration is attenuated during transmission, potentially indicating damping or a negative feedback mechanism. Based on the distribution of acceleration and decay intervals, the evolution pattern of the vibration source can be determined, and it can be judged whether the vibration will continue to amplify, leading to serious consequences, or whether it will naturally decay and return to normal. For example, the energy ratio from the second to the third vibration source is 1.35, which falls within the acceleration range; the energy ratio from the third to the first vibration source is 0.82, which falls within the decay range; and the energy ratio from the first to the fourth vibration source is 0.51, which also falls within the decay range. This evolutionary pattern indicates that abnormal vibrations are amplified in the initial stage but gradually decay thereafter, with an overall convergent trend that does not lead to destructive consequences.

[0130] Based on the evolution patterns of the vibration source, the stability and safety of wind turbine blade vibration can be further evaluated. If the acceleration range dominates in the transmission sequence, it indicates that the vibration may continue to amplify, and the system is at risk of instability; if the decay range dominates, it indicates that the vibration will eventually decay, and the system has self-stability. By monitoring the evolution patterns of the vibration source over a long period, an early warning model for the vibration state of wind turbine blades can be established. When a transmission sequence dominated by the acceleration range is detected, a timely warning can be issued to prevent vibration amplification from causing equipment damage.

[0131] This method identifies the evolution of multi-degree-of-freedom vibrations in wind turbine blades by analyzing the energy transfer relationships of independent vibration sources. The introduction of energy accumulation curves eliminates random fluctuations in the energy sequence, highlights the main trends in energy change, and improves the reliability of abrupt change detection. The transfer sequence constructed based on trigger moments visually demonstrates the propagation path of vibration energy, providing a powerful tool for identifying vibration sources. The analysis of acceleration and decay intervals assesses the stability of the vibration system, providing a scientific basis for predicting vibration development trends. This method has significant application value for structural health monitoring and fault early warning of wind turbine blades, enabling the early detection of potential vibration anomalies, preventing serious failures, improving the safety and reliability of wind power generation equipment, extending equipment lifespan, and reducing maintenance costs.

[0132] In one optional embodiment, by establishing a vibration propagation chain of the dominant vibration source, combining the vibration propagation chains to construct an association mapping network, and calculating the blade assembly vibration risk index based on the propagation characteristics of the dominant vibration source in the association mapping network, early warning of blade assembly vibration is achieved, including:

[0133] Collect vibration signals from the dominant vibration source;

[0134] Time-frequency analysis is performed on the vibration signal to obtain the amplitude spectrum and phase spectrum;

[0135] The amplitude ratio of adjacent measuring points on each blade is calculated based on the amplitude spectrum, and the phase difference of adjacent measuring points on each blade is calculated based on the phase spectrum. The amplitude ratio is used as the energy attenuation coefficient, and the phase difference is used as the transmission delay time.

[0136] A vibration propagation chain is constructed based on the energy attenuation coefficient and the transmission delay time. The cross-correlation degree between the vibration propagation chains is calculated. The vibration propagation chains are classified according to the cross-correlation degree, and an association mapping network containing the propagation hierarchy relationship is constructed.

[0137] Extract the propagation time and propagation intensity of each propagation level in the association mapping network, and superimpose the propagation intensity at the intersection node of the propagation level;

[0138] The vibration risk index of the blade assembly is calculated based on the transmission time and transmission intensity, and the vibration warning level is determined according to the vibration risk index.

[0139] The dominant vibration source was identified through preceding vibration mode decomposition and source identification, typically manifesting as the mode with the strongest energy or triggering other vibrations. For the identified dominant vibration source, high-precision accelerometers were deployed at key locations on the wind turbine blades to collect vibration signals. The sensor placement followed structural dynamics principles, with a set of triaxial accelerometers placed at the blade root, 1 / 4 of the blade length, 1 / 2 of the blade length, 3 / 4 of the blade length, and the blade tip, respectively collecting vibration data in the flapping, oscillating, and torsional directions. The sampling frequency was set to 1024Hz to ensure the capture of high-frequency vibration components; the data acquisition duration was 10 minutes, covering different operating states of the wind turbine. The collected raw vibration signals underwent preprocessing, including removing DC components, filtering high-frequency noise, and correcting outliers, to obtain a vibration time series suitable for further analysis.

[0140] Time-frequency analysis was performed on the preprocessed vibration signal to obtain the amplitude and phase spectra. The time-frequency analysis employed the short-time Fourier transform method, dividing the long-time series into multiple short-time windows. A Fourier transform was performed on each window to obtain the time-varying spectral characteristics. The short-time window length was set to 2 seconds, the window overlap rate to 50%, and the Hanning window was selected as the window function to reduce spectral leakage. The amplitude and phase spectra were extracted from the Fourier transform results of each short-time window. The amplitude spectrum represents the energy magnitude of each frequency component, and the phase spectrum represents the phase angle of each frequency component. Under normal operating conditions of wind turbine blades, the amplitude spectrum typically exhibits a significant peak at specific frequency points, corresponding to the blade's natural frequency and rotational speed-dependent frequency; the phase spectrum shows a certain regularity in variation between measurement points.

[0141] The amplitude ratio of adjacent measuring points on each blade is calculated based on the amplitude spectrum and used as an energy attenuation coefficient. For a specific frequency point, the ratio of the amplitude values ​​of adjacent measuring points is calculated, such as A2 / A1 representing the amplitude ratio of measuring point 2 relative to measuring point 1. The amplitude ratio reflects the attenuation or amplification of vibration energy during propagation. Under normal conditions, the amplitude ratio is usually less than 1, indicating that the vibration energy attenuates with the propagation distance; under abnormal conditions, the amplitude ratio at certain frequency points may be greater than 1, indicating the existence of a local amplification effect, which may be related to structural damage or resonance. The phase difference between adjacent measuring points on each blade is calculated based on the phase spectrum and used as the propagation delay time. The phase difference is equal to the difference in phase angle between adjacent measuring points, representing the time delay of the vibration wave during propagation. The phase difference can be converted into an actual time delay by dividing the phase difference by the angular frequency. In the vibration analysis of wind turbine blades, the propagation delay time reflects the propagation speed of the vibration wave and is of great significance for assessing structural integrity.

[0142] Vibration propagation chains are constructed based on energy attenuation coefficients and propagation delay times. A vibration propagation chain is a directed graph structure describing the propagation path of vibration within the blade. Nodes represent measurement point locations, and edges represent propagation paths. Edge attributes include energy attenuation coefficients and propagation delay times. The construction method is as follows: Measurement points are sorted according to their radial position within the blade; the energy attenuation coefficient and propagation delay time between adjacent measurement points are calculated; directed connections are established to form propagation chains. For multi-blade wind turbine systems, one propagation chain is constructed for each blade. The cross-correlation between vibration propagation chains is calculated to assess the similarity of vibration propagation characteristics among different blades. The cross-correlation is obtained by calculating the correlation coefficient between the amplitude and phase of corresponding nodes on the propagation chain. Under normal conditions, the vibration propagation characteristics of each blade are similar, resulting in high cross-correlation; under abnormal conditions, the propagation characteristics of a faulty blade differ significantly from other blades, leading to decreased cross-correlation. Vibration propagation chains are graded based on cross-correlation, constructing an association mapping network containing propagation hierarchy relationships. The grading method is as follows: a cross-correlation threshold is set (usually 0.85); propagation chains with cross-correlation higher than the threshold are grouped into the same level; hierarchical relationships are established between different levels. The correlation mapping network is a multi-level network structure that represents the propagation relationship of vibration between different blades and structural components.

[0143] The propagation time and intensity of each propagation level in the correlation mapping network are extracted. Propagation time refers to the cumulative time required for vibration to propagate from the source point to a specific node, calculated by summing the propagation delay times of each segment along the propagation path. Propagation intensity refers to the relative magnitude of vibration energy transmitted to a specific node, calculated by multiplying the energy attenuation coefficient of each segment along the propagation path. The propagation intensities are superimposed at the intersection nodes of the propagation levels to reflect the cumulative effect of multi-path vibration convergence. Intersection nodes are typically common connection points of different propagation chains, such as the hub center or nacelle connection. The superposition method involves a weighted sum of the propagation intensities of each propagation path, with weights set as the reliability or importance of each path. In the wind turbine blade vibration analysis, the hub center is the intersection point of three blade propagation chains. The vibration intensities transmitted from each blade to the hub are 0.32, 0.35, and 0.28, respectively. The superimposed propagation intensity is 0.95, indicating that the vibration is significantly amplified at the hub.

[0144] The vibration risk index of the blade assembly is calculated based on transmission time and intensity, and the vibration warning level is determined. The vibration risk index is a composite index that comprehensively considers transmission time and intensity. It is calculated by dividing the transmission intensity by the square root of the transmission time, reflecting the energy transfer efficiency per unit time. A higher risk index indicates faster vibration propagation speed and smaller energy attenuation, resulting in stronger system instability. Based on historical data and expert experience, a threshold range for the risk index is set, dividing it into four warning levels: normal, attention, warning, and alarm.

[0145] This method achieves risk assessment and early warning of multi-degree-of-freedom vibration by analyzing the propagation characteristics of the dominant vibration source in wind turbine blades. The introduction of amplitude ratio and phase difference values ​​enables quantitative expression of changes in vibration propagation characteristics, improving the sensitivity of fault identification. The construction of the correlation mapping network reveals the propagation path and hierarchical relationship of vibration in complex structures, providing a new perspective for understanding system dynamics. Based on the risk index calculation method using propagation time and intensity, the scientific quantification of vibration risk is achieved, providing an objective basis for early warning decisions.

[0146] like Figure 3 As shown, Figure 3 This is a schematic diagram of a multi-degree-of-freedom vibration detection system for wind turbine blades provided in an embodiment of the present invention. The system includes:

[0147] The vibration data acquisition module is used to collect vibration data of wind turbine blades during operation.

[0148] The operation phase analysis module is used to segment the vibration data according to the operation phase of the wind turbine blade, extract the main frequency information of each operation phase, and determine the vibration state of the wind turbine blade based on the main frequency information.

[0149] The coupling degree calculation module is used to construct a vibration feature space based on the vibration state, establish a mapping relationship between the multi-degree-of-freedom vibration components of the wind turbine blade in the vibration feature space, and calculate the vibration coupling degree of the wind turbine blade according to the mapping relationship.

[0150] The vibration source identification module is used to decouple the multi-degree-of-freedom vibration components into independent vibration sources when the vibration coupling degree exceeds a preset coupling degree threshold, track the evolution of the independent vibration sources over time, and determine the dominant vibration source of abnormal vibration of the wind turbine blades.

[0151] The vibration early warning module is used to establish a vibration propagation chain of the dominant vibration source, combine the vibration propagation chains to construct an association mapping network, and calculate the blade group vibration risk index based on the propagation characteristics of the dominant vibration source in the association mapping network, so as to realize early warning of blade group vibration.

[0152] The specific embodiments described above are preferred embodiments of the present invention and are not intended to limit the specific scope of the present invention. The scope of the present invention includes, but is not limited to, these specific embodiments. All equivalent changes made in accordance with the shape and structure of the present invention are within the protection scope of the present invention.

Claims

1. A method for multi-degree of freedom vibration detection of a wind turbine blade, characterized in that, The method comprises the following steps: Collect vibration data of the wind turbine blade during operation; Segment the vibration data according to the operation stage of the wind turbine blade, extract the main frequency information of each operation stage, and determine the vibration state of the wind turbine blade according to the main frequency information; Based on the vibration state, a vibration characteristic space is constructed, a mapping relationship between the multi-degree-of-freedom vibration components of the wind turbine blade is established in the vibration characteristic space, and the vibration coupling degree of the wind turbine blade is calculated according to the mapping relationship; When the vibration coupling degree exceeds a preset coupling degree threshold, the multi-degree-of-freedom vibration components are decoupled into independent vibration sources, the evolution law of the independent vibration sources with time is tracked, and the dominant vibration source of the abnormal vibration of the wind turbine blade is determined; By establishing the vibration propagation chain of the dominant vibration source, a correlation mapping network is constructed by combining the vibration propagation chain, a blade group vibration risk index is calculated based on the propagation characteristics of the dominant vibration source in the correlation mapping network, and the vibration of the blade group is warned.

2. The method of claim 1, wherein, Segmenting the vibration data according to the operation stage of the wind turbine blade, extracting the main frequency information of each operation stage, and determining the vibration state of the wind turbine blade according to the main frequency information comprises: Adaptive threshold denoising processing is performed on the vibration signal to obtain a preprocessed vibration signal; Performing time-frequency analysis on the preprocessed vibration data to obtain instantaneous frequency characteristics; Based on the instantaneous frequency characteristics, the frequency similarity of adjacent time windows is calculated, and when the frequency similarity is less than a preset similarity threshold, the segmentation point of the operation stage is determined, and the vibration data is divided into multiple operation stages; Spectrum analysis is performed on the vibration data of each operation stage to identify the main frequency information in the vibration data, the change trend of the main frequency information with time is constructed, and the vibration state transition characteristics of the wind turbine blade are determined according to the change trend; Based on the vibration state transition characteristics, a state discrimination index is constructed, a corresponding relationship between the state discrimination index and the vibration state is established, and the vibration state of the wind turbine blade is determined.

3. The method of claim 1, wherein, In the vibration characteristic space, the mapping relationship between the multi-degree-of-freedom vibration components of the wind turbine blade is established, and the vibration coupling degree of the wind turbine blade is calculated according to the mapping relationship, which comprises: The vibration signal is decomposed in the vibration characteristic space to obtain vibration components of multiple degrees of freedom; The overlapping area of the vibration components of each degree of freedom in the vibration characteristic space is calculated, and the phase information and amplitude information of the vibration components in the overlapping area are extracted; According to the phase information, a phase difference matrix of the vibration components is calculated, and according to the amplitude information, an amplitude ratio matrix of the vibration components is calculated, and the combined eigenvalue of the phase difference matrix and the amplitude ratio matrix is taken as the vibration coupling degree of the wind turbine blade.

4. The method of claim 3, wherein, The overlapping area of the vibration components of each degree of freedom in the vibration characteristic space comprises: Performing Hilbert transform on the vibration components of each degree of freedom to obtain an analytic signal, calculating the instantaneous frequency and instantaneous amplitude of the vibration components based on the analytic signal, and constructing an energy density distribution function containing time and frequency dimensions; An adaptive grid is obtained by constructing a basic grid in a time-frequency plane, calculating a rate of change of the energy density distribution function at adjacent grid points, and increasing the number of grid points in a current region when the rate of change exceeds a preset rate of change threshold; An energy density value is calculated at each grid point of the adaptive grid for the energy density distribution function of any two vibration components, and a minimum value of the two energy density values is determined as an overlap energy density of the grid point; Based on the overlap energy density, a numerical integration is performed on the adaptive grid in a region-by-region manner, and an overlap area between the vibration components is calculated according to an integration result.

5. The method of claim 1, wherein, The multiple-degree-of-freedom vibration components are decoupled into independent vibration sources, an evolution law of the independent vibration sources is tracked, and a dominant vibration source of the abnormal vibration of the wind turbine blade is determined, including: A vibration modal matrix including the vibration components of each degree of freedom is constructed; A singular value decomposition is performed on the vibration modal matrix to obtain an eigenvector, and the multiple-degree-of-freedom vibration components are decoupled into independent vibration sources based on the orthogonality of the eigenvector; A vibration energy sequence of the independent vibration sources is calculated, a time series correlation matrix is established according to the vibration energy sequence, and the evolution law of the independent vibration sources is tracked; Based on the evolution law of the independent vibration sources, an independent vibration source group having an energy transmission relationship is identified, a vibration propagation chain of the independent vibration source group is constructed, a trigger delay time and an energy attenuation ratio between the vibration sources in the vibration propagation chain are calculated, and a source vibration is determined; A trigger action intensity matrix and an energy contribution matrix of the source vibration are established, the energy contribution of the source vibration is calculated as a product of an energy proportion of the source vibration in the energy contribution matrix and the energy attenuation ratio, and the energy contribution is calculated as a product of the trigger action intensity and the energy proportion of the source vibration in the energy contribution matrix; Based on a combination score of the trigger action intensity and the energy contribution, a dominant vibration source of the abnormal vibration of the wind turbine blade is determined.

6. The method of claim 5, wherein, The vibration energy sequence of the independent vibration sources is calculated, the time series correlation matrix is established according to the vibration energy sequence, and the evolution law of the independent vibration sources is tracked, including: An energy time sequence of the independent vibration sources is extracted; The energy time sequence is segmented and accumulated to obtain an energy accumulation curve; A mutation point position between the energy accumulation curves of adjacent independent vibration sources is calculated, and a trigger time of energy transmission is determined according to the mutation point position; The independent vibration sources are sorted according to an energy transmission order based on the trigger time, and a transmission sequence of the vibration sources is constructed; An acceleration interval and a decay interval of the energy transmission are identified in the transmission sequence, and the evolution law of the vibration sources is determined according to the distribution of the acceleration interval and the decay interval.

7. The method of claim 1, wherein, A vibration propagation chain of the dominant vibration source is established, a correlation mapping network is constructed by combining the vibration propagation chains, a blade group vibration risk index is calculated based on a propagation characteristic of the dominant vibration source in the correlation mapping network, and a warning for the blade group vibration is realized, including: A vibration signal of the dominant vibration source is collected; A time-frequency analysis is performed on the vibration signal to obtain an amplitude spectrum and a phase spectrum; An amplitude ratio of adjacent measuring points on each blade is calculated according to the amplitude spectrum, a phase difference value of the adjacent measuring points on each blade is calculated according to the phase spectrum, the amplitude ratio is taken as an energy attenuation coefficient, and the phase difference value is taken as a transmission delay time. According to the energy attenuation coefficient and the transmission delay time, a vibration propagation chain is constructed, the cross-correlation between the vibration propagation chains is calculated, the vibration propagation chains are ranked according to the cross-correlation, and a correlation mapping network containing a propagation hierarchical relationship is constructed; The transmission time and transmission intensity of each propagation hierarchy in the correlation mapping network are extracted, and the transmission intensities are superimposed at the intersection nodes of the propagation hierarchies; Based on the transmission time and transmission intensity, a vibration risk index of the blade group is calculated, and a vibration warning level is determined according to the vibration risk index.

8. A multi-degree of freedom vibration detection system for a wind turbine blade for implementing the method of any of claims 1-7, characterized in that, The system comprises: a vibration data acquisition module for acquiring vibration data of the wind turbine blade during operation; an operation stage analysis module for segmenting and processing the vibration data according to the operation stage of the wind turbine blade, extracting the main frequency information of each operation stage, and determining the vibration state of the wind turbine blade according to the main frequency information; a coupling degree calculation module for constructing a vibration characteristic space based on the vibration state, establishing a mapping relationship between the multi-degree-of-freedom vibration components of the wind turbine blade in the vibration characteristic space, and calculating the vibration coupling degree of the wind turbine blade according to the mapping relationship; a vibration source identification module for decoupling the multi-degree-of-freedom vibration components into independent vibration sources when the vibration coupling degree exceeds a preset coupling degree threshold, tracking the evolution law of the independent vibration sources over time, and determining the dominant vibration source of the abnormal vibration of the wind turbine blade; a vibration warning module for combining and constructing a correlation mapping network by establishing a vibration propagation chain of the dominant vibration source, calculating a blade group vibration risk index based on the propagation characteristics of the dominant vibration source in the correlation mapping network, and realizing the early warning of the vibration of the blade group.

9. An electronic device, comprising: It comprises: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to realize the steps in the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer program instructions, and the computer program instructions realize the steps in the method of any one of claims 1 to 7 when executed by the processor.

Images