Wind turbine blade anomaly identification method and device, and storage medium

Abstract

A method, apparatus, and storage medium for identifying blade anomalies in wind turbine generators are disclosed. The blade anomaly identification method includes: identifying the actual value of the first-order natural frequency of each blade in each of multiple wind turbine generators in a wind farm; determining the correction value of the first-order natural frequency of each blade in each wind turbine generator; calculating the first-order natural frequency over-limit factor of each blade in each wind turbine generator, and identifying blades with the first-order natural frequency over-limit factor greater than a preset threshold as candidate abnormal blades; calculating the local entropy of different frequency bands of each blade in each wind turbine generator, determining the distance factor of each wind turbine generator, and identifying faulty units among the multiple wind turbine generators; and determining whether there are abnormal blades in the multiple wind turbine generators based on the first-order natural frequency over-limit factor, distance factor, blade length, and local entropy of different frequency bands of each blade in each wind turbine generator.

Description

Technical Field

[0001] This disclosure generally relates to the field of wind power generation technology, and more specifically, to a method, device, and storage medium for identifying blade anomalies in wind turbine generators. Background Technology

[0002] As a key component for energy conversion in wind turbine generators, blades account for over 20% of the overall cost in turbine manufacturing. Due to the long-term exposure to alternating loads and harsh operating environments, blades are susceptible to problems such as icing, cracking, and surface peeling. To ensure the safe and stable operation of wind turbine generators and reduce the probability of property damage and personal injury accidents caused by blade damage, various blade condition monitoring technologies have emerged. Existing blade condition monitoring technologies include acoustic emission, infrared thermal imaging, ultrasonic detection, vibration analysis, and SCADA data analysis. However, the first few technologies suffer from high costs, difficulty in installation, and susceptibility to environmental influences. While SCADA data analysis eliminates the need for additional detection equipment, it lacks direct monitoring points on the blades, and its accuracy requires extensive support from environmental and operational parameters, resulting in a complex monitoring method with poor generalization capabilities.

[0003] On the other hand, vibration condition monitoring technology is the most widely used technology in fault diagnosis of rotating machinery and equipment. Vibration-based blade condition monitoring is currently the most widely used commercial application model in China, with high customer acceptance.

[0004] In terms of identifying blade anomalies (e.g., structural damage), current technologies are mainly divided into three categories: The first category is to identify the natural frequency of the blade by combining multi-source data such as SCADA data with environmental parameters; the second category is to establish an accurate fault sample library based on massive sample data and use deep learning models to construct blade damage identification models; the third category is to establish wind turbine blade models through finite element simulation and vibration response experiments, substitute real-time data into the model for deviation comparison, and issue alarms in a timely manner.

[0005] The above methods firstly have relatively strict requirements on the richness of the sample library and the stability of the data source. Secondly, in the face of complex wind field operating environments, the applicability of finite element models and deep learning models built on local data is difficult to guarantee. Finally, the early structural damage of long and flexible blades has a very limited impact on the first-order natural frequency of the blades, and the application effect of various natural frequency identification methods on long and flexible blades is difficult to evaluate. Summary of the Invention

[0006] Therefore, embodiments of this disclosure provide a method, apparatus, and storage medium for identifying blade anomalies in wind turbine generator sets, which can effectively solve the problems of low accuracy and poor robustness in current blade anomaly identification.

[0007] In one general aspect, a method for identifying blade anomalies in wind turbine generators is provided. The method includes: identifying the actual values ​​of the first-order natural frequencies of each blade in each of multiple wind turbine generators in a wind farm, based on low-frequency vibration data of each blade in each wind turbine generator, wherein the multiple wind turbine generators have blades of the same type; determining a correction value for the first-order natural frequency of each blade in each wind turbine generator based on the actual first-order natural frequencies of each blade; and calculating the values ​​of the correction values ​​and actual first-order natural frequencies of each blade in each wind turbine generator. The system calculates the first-order natural frequency exceedance factor of each blade and identifies blades with first-order natural frequency exceedance factors greater than a preset threshold as candidate abnormal blades. Based on the high-frequency vibration data of each blade of each wind turbine generator set, it calculates the local entropy of different frequency bands of each blade of each wind turbine generator set, determines the distance factor of each wind turbine generator set based on the calculated local entropy of different frequency bands, and identifies faulty units among the multiple wind turbine generator sets based on the distance factor of each wind turbine generator set. Based on the first-order natural frequency exceedance factor, distance factor, blade length, and local entropy of different frequency bands of each blade of each wind turbine generator set, it determines whether there are abnormal blades among the multiple wind turbine generator sets.

[0008] Optionally, the blade anomaly identification method further includes: acquiring vibration data of each blade of each of the multiple wind turbine generator sets; preprocessing the acquired vibration data to remove abnormal data and vibration data of the wind turbine generator set in the shutdown state; and performing low-pass filtering and high-pass filtering on the preprocessed vibration data to obtain low-frequency vibration data and high-frequency vibration data of each blade of each wind turbine generator set.

[0009] Optionally, the step of identifying the actual value of the first natural frequency of each blade of each wind turbine includes: for any blade, based on the low-frequency vibration data of the blade and the harmonic data of a predetermined multiple of the rotor frequency data of the wind turbine in which the blade is located, determining the actual value of the first natural frequency of the blade within a preset range centered on the theoretical value of the first natural frequency of the blade.

[0010] Optionally, the step of determining the correction value of the first natural frequency of the blades of each wind turbine generator set includes: determining a number of wind turbine generator sets whose data volume meets preset requirements from the plurality of wind turbine generator sets; determining multiple statistical values ​​of the actual value of the first natural frequency of the blades of each of the plurality of wind turbine generator sets; determining the theoretical value of the first natural frequency of the blades as the correction value of the first natural frequency of the blades of each wind turbine generator set in response to the average variance of the multiple statistical values ​​being greater than or equal to a first predetermined threshold; and determining the correction value of the first natural frequency of the blades of each wind turbine generator set based on the multiple statistical values ​​of the actual value of the first natural frequency of the blades of each wind turbine generator set in response to the average variance being less than the first predetermined threshold.

[0011] Optionally, the step of determining the correction value of the first-order natural frequency of each wind turbine blade based on multiple statistical values ​​of the actual value of the first-order natural frequency of each blade includes: calculating the average value of the actual values ​​of the first-order natural frequencies of all blades in all wind turbines for which the correction value of the first-order natural frequency of the blades has not yet been determined; determining the maximum value and the mean value of the average deviations between any two blades of the current wind turbine; and, in response to the maximum value of the average deviations being greater than or equal to a second predetermined threshold, determining the deviation between two preset quantiles of the actual value of the first-order natural frequency of each blade of the current wind turbine. The maximum value among the actual values ​​of the first natural frequencies of each blade of the current wind turbine generator set, the maximum value among the standard deviations of the actual values ​​of the first natural frequencies of each blade of the current wind turbine generator set, and the average of the minimum and second minimum values ​​among the average deviations of the average values ​​between any two blades of the current wind turbine generator set are used to determine the theoretical value of the first natural frequency of the blade as the correction value of the first natural frequency of the blade of the current wind turbine generator set, or the reference value is determined as the correction value of the first natural frequency of the blade of the current wind turbine generator set. The reference value is calculated based on the average of the actual values ​​of the first natural frequencies of each blade of the several wind turbine generator sets, or based on the correction values ​​of the first natural frequencies of the blades of all the wind turbine generator sets that have been determined.

[0012] Optionally, in response to the fact that the number of wind turbine generator sets whose first-order natural frequencies of the blades have been determined is greater than or equal to half the number of the plurality of wind turbine generator sets, and the variance of the first-order natural frequencies of the blades of all the determined wind turbine generator sets is less than a preset correction threshold, the mean of the first-order natural frequencies of the blades of all the determined wind turbine generator sets is used as the reference value; otherwise, the reference value is calculated based on the average of the actual values ​​of the first-order natural frequencies of each blade in the plurality of wind turbine generator sets.

[0013] Optionally, the step of determining the correction value of the first-order natural frequency of each wind turbine blade based on multiple statistical values ​​of the actual value of the first-order natural frequency of each blade further includes: in response to the maximum value among the deviations of the average values ​​being less than a second predetermined threshold, determining the maximum value among the deviations between two preset quantiles of the actual value of the first-order natural frequency of each blade of the current wind turbine; in response to the maximum value among the deviations between the two preset quantiles of the actual value of the first-order natural frequency of each blade of the current wind turbine being greater than or equal to a third predetermined threshold, determining the theoretical value of the first-order natural frequency of the blade as the correction value of the first-order natural frequency of the current wind turbine blade; and in response to the maximum value among the deviations between the two preset quantiles of the actual value of the first-order natural frequency of each blade of the current wind turbine being less than a third predetermined threshold, determining the first-order natural frequency. The system determines whether the first difference between the average of the actual values ​​of the first-order natural frequencies of all blades of all wind turbine generator sets whose correction values ​​have not yet been determined falls within a preset range; in response to the first difference falling within the preset range, the reference value is determined as the correction value of the first-order natural frequency of the blades of the current wind turbine generator set; in response to the first difference not falling within the preset range, a second difference is determined between the average deviation of the average values ​​between any two blades of the current wind turbine generator set and the average of the actual values ​​of the first-order natural frequencies of all blades of all wind turbine generator sets whose correction values ​​have not yet been determined; in response to the second difference being less than a fourth predetermined threshold, the average of the actual values ​​of the first-order natural frequencies of all blades of all wind turbine generator sets whose correction values ​​have not yet been determined is determined as the correction value of the first-order natural frequency of the blades of the current wind turbine generator set.

[0014] Optionally, determining the theoretical value of the first natural frequency of the blade as the correction value of the first natural frequency of the blade of the current wind turbine generator set, or determining the correction value calculated based on the average of the actual values ​​of the first natural frequencies of each blade in the plurality of wind turbine generator sets as the correction value of the first natural frequency of the blade of the current wind turbine generator set, includes: in response to the maximum value of the deviation between two preset quantiles of the actual values ​​of the first natural frequencies of each blade of the current wind turbine generator set being less than or equal to a third predetermined threshold, and the maximum value of the standard deviation of the actual values ​​of the first natural frequencies of each blade of the current wind turbine generator set being less than or equal to a second predetermined threshold, or in response to the absolute value of the difference between the average of the minimum and second smallest values ​​of the average deviations between pairs of blades in the previous wind turbine generator set and the reference value being less than a fifth predetermined threshold, determining the reference value as the correction value of the first natural frequency of the blade of the current wind turbine generator set; otherwise, determining the theoretical value of the first natural frequency of the blade as the correction value of the first natural frequency of the blade of the current wind turbine generator set.

[0015] Optionally, the step of calculating the first-order natural frequency excess factor of each blade of each wind turbine generator set based on the corrected value and the actual value of the first-order natural frequency of each blade of each wind turbine generator set includes: for each blade of each wind turbine generator set in the plurality of wind turbine generator sets, calculating the deviation between the corrected value and the actual value of the first-order natural frequency, and calculating the ratio of the number of deviations greater than a sixth predetermined threshold to the total number of deviations within a preset time period; for each blade with the same number in all wind turbine generator sets in the plurality of wind turbine generator sets, calculating the average value and standard deviation of the ratio corresponding to the blade with the same number; and for each blade of each wind turbine generator set in the plurality of wind turbine generator sets, calculating the first-order natural frequency excess factor of the blade based on the ratio corresponding to the blade and the average value and / or standard deviation corresponding to the blade.

[0016] Optionally, the step of determining the distance factor of each wind turbine generator set based on the calculated local entropy of different frequency bands includes: for each blade, calculating the total local entropy of the blade based on the local entropy of multiple frequency bands of the blade; for each blade, calculating the average local entropy of the blade based on the total local entropy of the blade calculated multiple times within a preset time period; and for each wind turbine generator set, calculating the distance factor of the wind turbine generator set based on the average local entropy of each blade of the wind turbine generator set.

[0017] Optionally, the step of identifying faulty units among the multiple wind turbine generator sets based on the distance factor of each wind turbine generator set includes: for each wind turbine generator set, determining the maximum value among the mean local entropy values ​​of each blade of the wind turbine generator set, calculating the effective value of the total local entropy value of the blade corresponding to the maximum value, and calculating the ratio between the maximum value and the average of the mean local entropy values ​​of the remaining blades; for each wind turbine generator set, determining whether the wind turbine generator set belongs to a potentially faulty unit based on the ratio of the wind turbine generator set and the average of the ratios of the multiple wind turbine generator sets; for each potentially faulty unit among the determined potentially faulty units, determining whether the potentially faulty unit belongs to a suspected faulty unit based on the effective value of the potentially faulty unit, the average of the effective values ​​of the multiple wind turbine generator sets, and the variance; for each suspected faulty unit among the determined suspected faulty units, determining whether the suspected faulty unit belongs to a faulty unit based on the distance factor of the suspected faulty unit, the average of the distance factors of the multiple wind turbine generator sets, and the variance.

[0018] Optionally, the step of determining whether there are abnormal blades in the multiple wind turbine generator sets includes: for each wind turbine generator set, calculating a fusion feature based on the maximum value of the first-order natural frequency over-limit factor of each blade, the distance factor, and the blade length; in response to the fusion feature being less than a seventh predetermined threshold, determining that there are no abnormal blades; in response to the fusion feature being greater than or equal to the seventh predetermined threshold, for each faulty generator set, determining whether the faulty generator set has abnormal blades based on the first-order natural frequency over-limit factor of each blade of the faulty generator set, the local entropy of different frequency bands, and the blade length.

[0019] Optionally, the step of determining whether the faulty unit has abnormal blades includes: in response to a blade length less than a first blade length threshold, identifying blades in the faulty unit whose first-order natural frequency exceedance factor is greater than a blade abnormality threshold as abnormal blades; in response to a blade length greater than a second blade length threshold, identifying blades corresponding to the maximum value of the local entropy in different frequency bands of each branch of the faulty unit as abnormal blades; in response to a blade length greater than or equal to the first blade length threshold and less than or equal to the second blade length threshold, determining the number of blades in the faulty unit whose first-order natural frequency exceedance factor is greater than a blade abnormality threshold, and the number of blades corresponding to the maximum value of the local entropy in different frequency bands of each branch of the faulty unit, and identifying blades with the same number as abnormal blades.

[0020] In another general aspect, a blade anomaly identification device for wind turbine generators is provided. The blade anomaly identification device includes: a first-order natural frequency identification unit configured to: identify the actual value of the first-order natural frequency of each blade of each wind turbine generator in a wind farm based on low-frequency vibration data of each blade of each wind turbine generator, wherein the multiple wind turbine generators have blades of the same type; a first-order natural frequency correction unit configured to: determine a correction value of the first-order natural frequency of each blade of each wind turbine generator based on the actual first-order natural frequency of each blade of each wind turbine generator; and a candidate abnormal blade determination unit configured to: calculate based on the correction value and the actual value of the first-order natural frequency of each blade of each wind turbine generator. The system identifies the first-order natural frequency exceeding the limit factor of each blade of each wind turbine generator set, and identifies blades with a first-order natural frequency exceeding the limit factor greater than a preset threshold as candidate abnormal blades; the faulty unit identification unit is configured to: calculate the local entropy of different frequency bands of each blade of each wind turbine generator set based on the high-frequency vibration data of each blade of each wind turbine generator set, determine the distance factor of each wind turbine generator set based on the calculated different preferred local entropies, and identify the faulty unit among the multiple wind turbine generator sets based on the distance factor of each wind turbine generator set; the abnormal blade determination unit is configured to: determine whether there are abnormal blades among the multiple wind turbine generator sets based on the first-order natural frequency exceeding the limit factor, distance factor, blade length, and local entropy of different frequency bands of each blade of each wind turbine generator set.

[0021] In another general aspect, a computer-readable storage medium is provided that stores a computer program, which, when executed by a processor, implements the blade anomaly identification method as described above.

[0022] In another general aspect, a blade anomaly identification device for a wind turbine generator set is provided, the blade anomaly identification device comprising: a processor; and a memory storing a computer program, which, when executed by the processor, implements the blade anomaly identification method as described above.

[0023] The blade anomaly identification method, apparatus, and storage medium for wind turbine generators according to embodiments of this disclosure achieve accurate assessment of the blade's natural frequency through preprocessing of blade vibration data, correction of the blade's natural frequency, and statistical analysis of the blade's natural frequency deviation. Simultaneously, by constructing features from the high-frequency vibration data of the blades, the vibration energy deviation of the three blades is accurately identified. Finally, by using a blade length weighting factor, the calculation results of the two features are weighted and analyzed to complete the anomaly identification of blades of different lengths. The blade anomaly identification method, apparatus, and storage medium for wind turbine generators according to embodiments of this disclosure require low hardware costs, require a small number of fault samples, and have a simple data source. They are adaptable to application scenarios without SCADA data and have good applicability to both older wind farms and large wind turbine generators, exhibiting unique advantages in engineering applications.

[0024] On the other hand, the blade anomaly identification method, device, and storage medium of the wind turbine generator set according to the embodiments of this disclosure combine the blade damage mechanism and engineering data characteristics to construct blade anomaly identification feature factors, and use engineering experience to establish a damage assessment function based on blade length. This can adaptively identify common faults of blades of different sizes, effectively solve the problems of low accuracy and poor robustness of current blade anomaly identification, and improve the operating efficiency and power generation efficiency of the wind turbine generator set. Attached Figure Description

[0025] The above and other objects and features of the embodiments of this disclosure will become clearer from the following description taken in conjunction with the accompanying drawings illustrating the embodiments, wherein:

[0026] Figure 1 This is a flowchart illustrating a blade anomaly identification method according to an embodiment of the present disclosure;

[0027] Figure 2 This is a flowchart illustrating a method for determining a correction value for the first-order natural frequency of the blades of each wind turbine according to an embodiment of the present disclosure;

[0028] Figure 3 This is a flowchart illustrating a method for determining a correction value for the first-order natural frequency of each wind turbine blade based on various statistical values ​​of the actual value of the first-order natural frequency of each blade, according to an embodiment of the present disclosure.

[0029] Figure 4 This is a flowchart illustrating a method for calculating the first-order natural frequency overshoot factor of each blade of each wind turbine according to embodiments of the present disclosure;

[0030] Figure 5 This is a flowchart illustrating a method for determining the distance factor of each wind turbine according to embodiments of the present disclosure;

[0031] Figure 6 This is a flowchart illustrating a method for identifying faulty units among a plurality of wind turbine generators based on a distance factor for each wind turbine generator, according to an embodiment of the present disclosure.

[0032] Figure 7 This is a flowchart illustrating a method for determining whether there are abnormal blades in a plurality of wind turbine generators according to embodiments of the present disclosure;

[0033] Figure 8 This shows the first-order natural frequency over-limit factor R of each blade of each wind turbine generator in the wind farm. ij Example illustration;

[0034] Figure 9 This shows the first-order natural frequency over-limit factor R of each blade of each wind turbine generator in the wind farm. ij Another example illustration;

[0035] Figure 10 This is a block diagram of a blade anomaly identification device for a wind turbine generator set according to an embodiment of the present disclosure;

[0036] Figure 11 This is a block diagram of a blade anomaly identification device for a wind turbine generator set according to another embodiment of the present disclosure. Detailed Implementation

[0037] The following detailed embodiments are provided to aid the reader in gaining a comprehensive understanding of the methods, apparatus, and / or systems described herein. However, various changes, modifications, and equivalents of the methods, apparatus, and / or systems described herein will become apparent upon understanding this disclosure. For example, the order of operations described herein is merely illustrative and is not limited to those orders set forth herein, but may be changed as will become clear upon understanding this disclosure, except for operations that must occur in a specific order. Furthermore, for clarity and conciseness, descriptions of features known in the art may be omitted.

[0038] According to embodiments of this disclosure, a method for identifying blade anomalies in wind turbine generators is proposed, based on blade vibration data (low-frequency and high-frequency vibration data). Vibration acceleration sensors can be installed inside the wind turbine blades to detect blade vibration data. A data acquisition unit can periodically collect blade vibration data from the acceleration sensors and transmit the data from the hub to the nacelle switch via a wireless access point (AP). Finally, the nacelle switch can transmit the blade vibration data to an analysis platform via the wind turbine ring network for analysis to identify whether any blade anomalies exist. Therefore, the blade anomaly identification method according to embodiments of this disclosure can be executed in the control equipment of a wind farm or in higher-level control equipment.

[0039] Based on the above principles, the blade anomaly identification method according to embodiments of this disclosure collects blade vibration data covering different frequency bands and performs a series of analyses on the collected blade vibration data to achieve blade anomaly identification. The blade anomaly identification method according to embodiments of this disclosure combines the theoretical value of the blade's first-order natural frequency, analyzes the low-frequency components of the blade vibration data to establish a first-order natural frequency parameter library, records the actual first-order natural frequency of each wind turbine under initial operation, and uses a statistical model based on difference analysis to determine whether there are anomalies in the blades of a single unit. Furthermore, the blade anomaly identification method according to embodiments of this disclosure analyzes the high-frequency components of the blade vibration data, constructs features for the similarity of single blades and three blades, and uses the local entropy of a single blade to construct a distance factor to measure the similarity of the high-frequency vibration signals of three blades. Finally, the blade anomaly identification method according to embodiments of this disclosure establishes a damage assessment function based on blade length to identify abnormal features of blades of different sizes, thereby ultimately identifying abnormal states such as blade cracks, icing, whitening, and splitting.

[0040] The following is combined Figures 1 to 11 A detailed description is provided of a blade anomaly identification method, apparatus, and storage medium according to embodiments of the present disclosure.

[0041] Figure 1 This is a flowchart illustrating a blade anomaly identification method according to an embodiment of the present disclosure.

[0042] The theoretical value of the first natural frequency of a blade is determined during the blade design phase. However, during mass production and manufacturing, due to manufacturing deviations, the actual first natural frequency of the blades shipped from the factory deviates from the theoretical value. Generally, blade design specifications stipulate that the actual value of the first natural frequency can deviate from the theoretical value by ±5%. This means that when the first natural frequency of a blade fluctuates within ±5%, it is impossible to determine whether the blade has experienced an anomaly. Therefore, it is necessary to identify and correct the first natural frequency of each blade in a typhoon turbine generator set.

[0043] Reference Figure 1 In step S101, based on the low-frequency vibration data of each blade of each wind turbine in the wind farm, the actual value f of the first-order natural frequency of each blade of each wind turbine is identified. e The aforementioned multiple wind turbine generators have blades of the same type. In step S102, based on the actual first-order natural frequencies of each blade of each wind turbine generator, a correction value f for the first-order natural frequency of the blades of each wind turbine generator is determined. r '.

[0044] According to an embodiment of this disclosure, before step S101, the blade anomaly identification method may further include the following steps: acquiring vibration data of each blade of each wind turbine in multiple wind turbine generator sets; preprocessing the acquired vibration data to remove abnormal data and vibration data of the wind turbine generator set in the shutdown state; performing low-pass filtering and high-pass filtering on the preprocessed vibration data to obtain low-frequency vibration data and high-frequency vibration data of each blade of each wind turbine generator set.

[0045] By preprocessing the acquired vibration data, data with issues such as waveform drift, waveform distortion, and abnormal impacts can be removed from the blade vibration data. Specifically, waveform drift, distortion, and abnormal impacts can be eliminated based on the effective value, kurtosis value, and skewness value of the blade vibration data. For example, blade vibration data with excessively large or small effective values ​​or kurtosis values, or blade vibration data with skewness values ​​that deviate significantly from the normal range, can be removed from the acquired blade vibration data.

[0046] By preprocessing the acquired vibration data, data from the shutdown state can be eliminated. Specifically, during blade rotation, the greatest force is centrifugal force, which manifests as a signal dominated by the blade rotation period in the blade vibration data. Therefore, this signal can be used to filter shutdown data. Generally, the impeller speed is between 3 and 12 rpm, so blade vibration data obtained at speeds below 5 rpm can be eliminated. Here, 5 rpm is merely an example, and this disclosure is not limited to this.

[0047] Furthermore, since the first-order natural frequencies of blades are generally low, a low-pass filter is typically used to retain low-frequency vibration data in the 0–20 Hz range to filter out interference from invalid signals when identifying the actual first-order natural frequency of the blade. On the other hand, when calculating the high-frequency characteristics of blade vibration data, it is necessary to filter out low-frequency interference signals such as impeller rotation frequency; therefore, a high-pass filter can be used to filter out blade vibration data below 100 Hz. Here, the 0–20 Hz range and 100 Hz are merely examples, and this disclosure is not limited thereto.

[0048] According to embodiments of this disclosure, for any blade, the actual value of the first-order natural frequency of the blade can be determined within a preset range centered on the theoretical value of the blade's first-order natural frequency, based on the low-frequency vibration data of the blade and the harmonic data of a predetermined multiple of the rotor frequency data of the wind turbine generator set (i.e., the wind turbine generator set containing the blade) where the blade is located.

[0049] More specifically, taking the direction of blade flapping (or waving) as an example, the low-frequency vibration data of the blade is X={x1,x2,x3,…x n}, where n is the number of low-frequency vibration data points. A Fourier transform is performed on the low-frequency vibration data to obtain the spectrum, with the horizontal axis sequence being X. fft ={x fft1 ,x fft2 ,x fft3 ,…x fftk The ordinate sequence (also known as the spectrum sequence) is Y. fft ={y fft1 ,y fft2 ,y fft3 ,…y fftk},

[0050] Assume the theoretical first-order natural frequency (i.e., the theoretical value of the first-order natural frequency) in the direction of the blade flapping (or flaring) of the current model is f. e Considering that the accuracy error of the data acquisition equipment and the manufacturing deviation of the blade may cause the actual value of the first-order natural frequency of the blade to be greater than the design requirement, the blade anomaly identification method according to the embodiments of this disclosure is based on the following: e ±10%f e Within the range, calculate the impeller rotational frequency (f) to be analyzed. spd The harmonic numbers are n1 to n2.

[0051]

[0052] Next, iterate through each harmonic from n1 to n2. In the spectrum, using the frequency of each harmonic as the center, determine whether the amplitude of several points (e.g., but not limited to 5 points) to the left and right of the center is greater than the theoretical value of the first natural frequency. If so, set the amplitude of the several points to the left and right to 0; otherwise, do not process the several points to the left and right, thus obtaining the adjusted spectrum.

[0053] Finally, in the adjusted spectrum, the maximum amplitude value within the range of 0.9 times the theoretical first-order natural frequency to 1.1 times the theoretical first-order natural frequency is determined, and the frequency corresponding to the maximum amplitude value (i.e., the horizontal axis) is determined as the actual first-order natural frequency f. e Here, 0.9 and 1.1 are merely examples, and this disclosure is not limited thereto. Those skilled in the art can choose other appropriate values ​​according to actual needs.

[0054] According to embodiments of this disclosure, the actual value f of the first-order natural frequency of each blade of each wind turbine generator can be obtained. e Store in database A.

[0055] The following reference Figure 2 The method for determining the correction value of the first-order natural frequency of each wind turbine blade is described in detail.

[0056] Figure 2This is a flowchart illustrating a method for determining the correction value of the first-order natural frequency of the blades of each wind turbine according to an embodiment of the present disclosure.

[0057] According to embodiments of this disclosure, in order to determine the corrected value of the first-order natural frequency of the blade, it is necessary to first accumulate the actual values ​​of the first-order natural frequency of the blade over a period of time (e.g., but not limited to one month). On the other hand, the corrected value of the first-order natural frequency of the blade obtained through correction can be stored in a database B. For example, database B may have the structure shown in Table 1.

[0058] Table 1 Database B Structure

[0059]

[0060]

[0061] Reference Figure 2 In step S201, several wind turbine generator sets whose data volume meets preset requirements are determined from among the multiple wind turbine generator sets. These multiple wind turbine generator sets are the same as those mentioned in step S101. Meeting the preset data volume requirement means that the wind turbine generator set has at least two blades with more than 100 data entries (i.e., the actual values ​​of the first-order natural frequencies). Here, 100 entries is merely an example; this disclosure is not limited to this, and more or fewer entries can be selected.

[0062] Next, in step S202, multiple statistical values ​​of the actual values ​​of the first-order natural frequency of each blade in several wind turbine generator sets are determined. According to embodiments of this disclosure, these multiple statistical values ​​may include, for example, the mean (avg). ij variance std ij Lower quartile Q1 ij , median Q2 ij Upper quartile Q3 ij Where i represents the wind turbine generator number and j represents the blade number.

[0063] In step S203, the variance std in response to various statistical values ​​is... ij If the average value var is greater than or equal to a first predetermined threshold (e.g., but not limited to 0.05), the theoretical value f of the first-order natural frequency of the blade will be... theo The correction value f is determined as the first-order natural frequency of each wind turbine blade. r '(that is, the f mentioned above) e ).

[0064] On the other hand, in step S204, the variance std in response to various statistical values ​​is... ijIf the average value var is less than a first predetermined threshold, the correction value of the first natural frequency of each wind turbine blade is determined based on multiple statistical values ​​of the actual value of the first natural frequency of each blade.

[0065] Figure 3 This is a flowchart illustrating a method for determining a correction value for the first-order natural frequency of each wind turbine blade based on various statistical values ​​of the actual value of the first-order natural frequency of each blade, according to an embodiment of the present disclosure.

[0066] Reference Figure 3 In step S301, the average value of the actual first-order natural frequencies of all blades in several wind turbine generator sets whose correction values ​​for the first-order natural frequencies of the blades have not yet been determined is calculated.

[0067] In step S302, the maximum value avg_error and the mean value avg_mean of the average deviation between any two blades of the current wind turbine generator are determined.

[0068] In step S303, in response to avg_error being greater than or equal to a second predetermined threshold (e.g., but not limited to 0.02), the deviation between two predetermined quantiles (e.g., but not limited to q) based on the actual value of the first-order natural frequency of each blade of the current wind turbine is determined. ij =Q3 ij -Q1 ij The maximum value q in the above, and the maximum value std in the standard deviation of the actual values ​​of the first-order natural frequencies of each blade of the current wind turbine generator. max The average of the minimum and second minimum deviations between any two blades of the current wind turbine generator set. less The theoretical value of the first-order natural frequency f of the blade. theo The correction value f is determined as the first natural frequency of the current wind turbine blades. r ', or the reference value f c The correction value f is determined as the first natural frequency of the current wind turbine blades. r Here, the average value of the first-order natural frequency of each blade in several wind turbine generators can be used as the basis. Calculate the reference value f c Alternatively, it can be based on the correction value f of the first-order natural frequency of all known wind turbine blades. r 'Calculate the reference value f' c .

[0069] Specifically, if the correction value f of the first natural frequency of the blade has been determined... rThe number of wind turbine generators is greater than or equal to NA / 2, and the correction value f of the first-order natural frequency of the blades of all the wind turbine generators has been determined. r 'variance var f If the value is less than a preset correction threshold (e.g., but not limited to 0.06), then the correction value f of the first-order natural frequency of all determined wind turbine blades can be used. r The mean of ' is used as the reference value f c Here, NA represents the total number of wind turbine generators in the wind farm. Alternatively, it can be based on the average of the actual values ​​of the first-order natural frequencies of each blade in several wind turbine generators. Calculate the reference value f c In other words, if the correction value f for the first natural frequency of the blade has been determined... r The number of wind turbine generators is less than NA / 2, or the variance var f If the value is greater than or equal to a preset correction threshold, it can be based on the average value of the actual first-order natural frequency of each blade in several wind turbine generator sets. Calculate f c Alternatively, when the variance var f When the value is less than the preset correction threshold, the outlier in the correction values ​​of the first-order natural frequencies of all determined wind turbine blades can be removed first using the 3sigma principle. Then, the mean of the correction values ​​of the remaining determined first-order natural frequencies of wind turbine blades can be calculated as a reference value f. c .

[0070] Return to step S303, in response to the maximum value q being less than or equal to a third predetermined threshold (e.g., but not limited to 0.025), and the maximum value std max Less than or equal to a second predetermined threshold, or, in response to the average value avg less Compared with reference value f c If the absolute value of the difference between the two values ​​is less than a fifth predetermined threshold (e.g., but not limited to 0.02), the reference value f will be... c The correction value f is determined as the first natural frequency of the current wind turbine blades. r Otherwise, the theoretical value of the first-order natural frequency f of the blade will be used. theo The correction value f is determined as the first natural frequency of the current wind turbine blades. r '.

[0071] In other words, if q, std max avg less If either condition (1) or condition (2) is met, then the reference value f will be... c The correction value f is determined as the first natural frequency of the current wind turbine blades. r '.

[0072]

[0073] |avg less -f c |<0.02 (2)

[0074] Otherwise, f theo The correction value f is determined as the first natural frequency of the current wind turbine blades. r Here, the reference value f c It can be calculated based on the following equation:

[0075]

[0076] Among them, avg ij It can represent the average value of the actual value of the first natural frequency of the j-th blade of the i-th wind turbine generator.

[0077] According to embodiments of this disclosure, such as Figure 3 The method for determining the correction value of the first-order natural frequency of each wind turbine blade may further include the following steps.

[0078] In step S304, in response to avg_error being less than a second predetermined threshold, the deviation between two preset quantiles (e.g., but not limited to q) of the actual value of the first-order natural frequency of each blade of the current wind turbine generator is determined. ij =Q3 ij -Q1 ij The maximum value q in ).

[0079] In step S305, in response to q being greater than or equal to a third predetermined threshold (e.g., but not limited to 0.025), the theoretical value f of the first-order natural frequency of the blade is... theo The correction value f is determined as the first natural frequency of the current wind turbine blades. r '.

[0080] In step S306, in response to q being less than a third predetermined threshold, an average value is determined. Compared with reference value f c Whether the first difference between them falls within a preset range (e.g., but not limited to 0 to 0.02).

[0081] In step S307, in response to the first difference falling into a preset range, the reference value f is... c The correction value f is determined as the first natural frequency of the current wind turbine blades. r '.

[0082] On the other hand, in step S308, in response to the first difference not falling within a preset range, the mean value avg_mean and the average value of the average deviation between each pair of blades of the current wind turbine are determined. The second difference between them.

[0083] In step S309, in response to the second difference being less than a fourth predetermined threshold (e.g., but not limited to 0.03), the average value is... The correction value f is determined as the first natural frequency of the current wind turbine blades. r '.

[0084] Optionally, if the second difference is greater than or equal to the fourth predetermined threshold, the current wind turbine will be removed from the pool of wind turbines whose data volume meets the preset requirements. In other words, the first natural frequency of the blades of the current wind turbine will not be corrected; instead, the theoretical or actual value of the first natural frequency of the blades of the current wind turbine will be used directly.

[0085] According to embodiments of this disclosure, the corrected value of the first-order natural frequency of the blade is more accurate than the theoretical value, thus allowing for deviation analysis of the first-order natural frequency based on the corrected value. It is worth noting that due to the long-term rotation of the blade and the influence of harsh environments, the stability of blade vibration data is poor. In this case, using a single data alarm strategy or a single unit's data alarm strategy presents significant challenges in setting alarm thresholds, and the accuracy and robustness of anomaly identification are difficult to guarantee. Therefore, the blade anomaly identification method according to embodiments of this disclosure is based on a statistical model of difference analysis, using multiple data points to analyze the exceedance of the first-order natural frequency of each blade of each wind turbine unit.

[0086] Return to reference Figure 1 In step S103, the correction value f is based on the first-order natural frequency of each blade of each wind turbine generator set. r 'and the actual value f of the first natural frequency e 'Calculate the first-order natural frequency over-limit factor R of each blade of each wind turbine generator set.' ij And the first-order natural frequency overlimit factor R ij Leaves exceeding a preset threshold (e.g., but not limited to, 1) are identified as candidate abnormal leaves. See below for reference. Figure 4 The method for calculating the first-order natural frequency over-limit factor of each blade of each wind turbine generator set is described in detail.

[0087] Figure 4 This is a flowchart illustrating a method for calculating the first-order natural frequency over-limit factor of each blade of each wind turbine according to embodiments of the present disclosure.

[0088] Reference Figure 4 In step S401, for each blade of each wind turbine generator set in the multiple wind turbine generator sets, the correction value f of the first-order natural frequency is calculated. r 'and the actual value f of the first natural frequency rr The deviation e between them, and the proportion p of the number of deviations greater than a sixth predetermined threshold (e.g., but not limited to 1%) within a preset time period (e.g., but not limited to 5 days) to the total number of deviations. ij The aforementioned multiple wind turbine generator sets are the same as those mentioned in step S101. i represents the wind turbine generator set number, and j represents the blade number.

[0089] Here, the deviation e can be calculated using the following equation.

[0090]

[0091] In step S402, for each blade with the same number in all wind turbine generator sets, the average value of the proportion corresponding to the blade with the same number is calculated. j and standard deviation v j .

[0092] For example, for the j-th blade of all wind turbines in multiple wind turbine generator sets, the average value ave is calculated as follows. j and standard deviation v j .

[0093]

[0094] In step S403, for each blade of each wind turbine generator set in the multiple wind turbine generator sets, the first-order natural frequency exceedance factor R of the blade is calculated based on the proportion corresponding to the blade and the average value and / or standard deviation corresponding to the blade. ij .

[0095] Specifically, if the number of multiple wind turbine generators is less than or equal to a first quantity threshold (e.g., but not limited to 5), the first-order natural frequency excess factor R of the j-th blade of the i-th wind turbine generator can be calculated as follows. ij .

[0096]

[0097] On the other hand, if the number of multiple wind turbine generators exceeds the first quantity threshold, the first-order natural frequency over-limit factor R of the j-th blade of the i-th wind turbine generator can be calculated in the following way. ij .

[0098] First, if vj >0.15 and ave j If the value is greater than 0.5, then the first-order natural frequency excess factor R of the j-th blade of the i-th wind turbine generator can be calculated as follows: ij .

[0099]

[0100] Second, if v j >0.15 and ave j If the value is less than 0.5, then the first-order natural frequency excess factor R of the j-th blade of the i-th wind turbine generator can be calculated as follows: ij .

[0101]

[0102] Third, if v j <0.15 and 3×v j +ave j If the value is less than 1, the first-order natural frequency excess factor R of the j-th blade of the i-th wind turbine generator can be calculated as follows: ij .

[0103]

[0104] Fourth, if v j <0.15 and 3×v j +ave j If the value is greater than 1, then the first-order natural frequency excess factor R of the j-th blade of the i-th wind turbine can be calculated as follows: ij .

[0105]

[0106] By iterating through all the blades of multiple wind turbine generators, if the first-order natural frequency of the j-th blade of the i-th wind turbine generator exceeds the limit factor R... ij If the value exceeds a preset threshold, the leaf can be identified as a candidate abnormal leaf.

[0107] According to embodiments of this disclosure, normal blades, due to their intact overall structure, exhibit vibration data primarily based on impeller rotation frequency during rotation, with no significant anomalies in high-frequency vibration data. Referring to the vibration data of normal blades, when local cracks appear on a blade, its structural integrity and stiffness change. These changes may exacerbate nonlinear vibrations, meaning the vibration transmission to the crack becomes discontinuous, increasing transient components in the blade vibration data. These transient components can only be captured at higher sampling rates. Furthermore, friction at the crack, crack propagation, and the inability of damaged blades to effectively suppress high-frequency vibrations further complicate the frequency composition of the blade vibration data, resulting in enhanced broadband characteristics in the high-frequency components. Based on the above analysis, the blade anomaly identification method according to embodiments of this disclosure can further determine faulty units with abnormal blades based on the high-frequency vibration data of the blades.

[0108] Return to reference Figure 1 In step S104, based on the high-frequency vibration data of each blade of each wind turbine generator set, the local entropy of each blade of each wind turbine generator set in different frequency bands is calculated. i Local entropy based on different frequency bands i Determine the distance factor D for each wind turbine generator, and based on the distance factor D for each wind turbine generator, identify the faulty turbine generator among multiple wind turbine generators. See below for reference. Figure 5 and Figure 6 The specific description outlines the methods for determining the distance factor for each wind turbine and identifying faulty turbines.

[0109] Figure 5 This is a flowchart illustrating a method for determining the distance factor of each wind turbine according to embodiments of the present disclosure.

[0110] Reference Figure 5 In step S501, for each blade, the total local entropy value enr of the blade is calculated based on the local entropy of multiple frequency bands of the blade. p .

[0111] Specifically, the spectrum Y of the high-frequency data for any given blade fft Peak search can be performed at preset frequency intervals (e.g., but not limited to 10Hz) for values ​​with amplitudes greater than a preset amplitude threshold (e.g., but not limited to 0.005). Then, at the peak frequency f p Calculate f around the center. p The variance of the amplitude within the ±10Hz range. If the variance is greater than a preset variance threshold (e.g., but not limited to 0.002), then f will be... p The amplitude within the ±10Hz range is set to 0, and the processed spectrum Y is finally obtained. fftNext, the processed spectrum Y can be... fft "Divide into N equally spaced segments, denoted as Y1, Y2, ... Y..." N And calculate the local entropy eng for each frequency band (i.e., frequency zone) as follows. i , where i = 1, 2, ... N.

[0112] eng i =∑Y i 2

[0113] Thus, the local entropy E of each frequency band of each blade of the Xth wind turbine can be represented as follows: X :

[0114]

[0115] Among them, eng 1i Let i represent the local entropy of the i-th frequency band of the first blade.

[0116] Then, for any blade, the local entropy of a portion of its frequency bands can be summed to obtain the total local entropy value enr of the blade. p Where p represents the leaf number. For example, the total local entropy of a leaf, enr, can be calculated as follows: p :

[0117]

[0118] Where m,k∈[1,N],p∈[1,3].

[0119] In step S502, for each blade, the total local entropy value enr of the blade is calculated multiple times within a preset time period. p Calculate the mean local entropy of the blade. Here, the 3sigma principle can be used to remove individual outliers from the total local entropy values ​​of the leaf calculated multiple times within a preset time period. Then, the mean local entropy of the leaf can be calculated as follows.

[0120]

[0121] Where n represents the number of local entropy values ​​used to calculate the mean local entropy of the blade, and n is greater than 1.

[0122] In step S503, for each wind turbine generator set, based on the average local entropy of each blade of the wind turbine generator set... Calculate the distance factor D of the wind turbine generator set.

[0123] Specifically, for any wind turbine generator, the average local entropy of its three blades can be sorted to obtain the maximum value. Minimum value and median The distance factor D of the wind turbine generator set is calculated as follows.

[0124]

[0125] Figure 6 This is a flowchart illustrating a method for identifying faulty units among a plurality of wind turbine generators based on a distance factor for each wind turbine generator, according to an embodiment of the present disclosure.

[0126] Reference Figure 6 In step S601, for each wind turbine generator set, the mean local entropy of each blade of the wind turbine generator set is determined. The maximum value in Calculation and maximum value The effective value H of the total local entropy of the corresponding blade rms And calculate the maximum value. The ratio prop is the ratio between the mean of the local entropy of the remaining leaves and the mean of the local entropy of the other leaves.

[0127] Specifically, the maximum value can be calculated as follows: The effective value H of the total local entropy of the corresponding blade rms .

[0128]

[0129] In addition, the maximum value can be calculated in the following way. The ratio prop is the ratio between the mean of the local entropy of the remaining leaves and the mean of the local entropy of the other leaves.

[0130]

[0131] Next, in step S602, for each wind turbine generator set, based on the ratio prop and the average value prop of the ratios of multiple wind turbine generator sets... avg This determines whether the wind turbine generators are potentially faulty. The aforementioned multiple wind turbine generators are the same as those mentioned in step S101.

[0132] Specifically, for the i-th wind turbine generator, if its ratio prop i >1.1×prop avgIf the condition is met, the i-th wind turbine can be identified as a potentially faulty turbine; otherwise, it can be identified as a normal turbine. All potentially faulty turbines among the multiple wind turbines can constitute a sequence V of potentially faulty turbines. Here, the coefficient 1.1 is merely an example, and this disclosure is not limited thereto. Those skilled in the art can choose other coefficients greater than or less than 1.1 according to actual needs.

[0133] In step S603, for each of the identified potential faulty units, based on the H of the potential faulty unit... rms H of multiple wind turbine generator sets rms The average value H rms_avg and variance H rms_std To determine whether a potentially faulty unit belongs to the category of suspected faulty units.

[0134] Specifically, when H rms_std If the value is less than the first fault threshold (e.g., but not limited to 0.01), H rms_avg Greater than the second fault threshold (e.g., but not limited to 0.015), and H rms Greater than H rms_avg In this case, it can be determined that the potentially faulty unit belongs to the suspected faulty unit.

[0135] When H rms_std Less than the first fault threshold and H rms_avg When less than or equal to the second fault threshold, or when H rms_std If H is greater than the third fault threshold (e.g., but not limited to 0.1), rms >H rms_avg +H rms_std If so, it can be determined that the potentially faulty unit belongs to the suspected faulty unit.

[0136] When H rms_std If H is greater than or equal to the first fault threshold and less than or equal to the third fault threshold, then... rms >1.3×H rms_avg If so, it can be determined that the potentially faulty unit belongs to the suspected faulty unit category. Here, the coefficient 1.3 is only an example, and this disclosure is not limited to this. Those skilled in the art can choose other coefficients greater than or less than 1.3 according to actual needs.

[0137] In other cases, the potentially faulty unit can be determined to be a normal unit.

[0138] In step S604, for each of the identified suspected faulty wind turbine units, based on the distance factor D of the suspected faulty wind turbine unit and the average distance factor D of multiple wind turbine units, ... avg and variance D std To determine whether a suspected faulty unit is indeed a faulty unit.

[0139] Specifically, firstly, if the variance D std If the distance factor is greater than the first distance factor threshold (e.g., but not limited to 1) or less than the second distance factor threshold (e.g., but not limited to 0.4), then the distance factor alarm threshold D of the wind farm can be determined. ther =D avg +D std Otherwise, the distance factor alarm threshold D of the wind farm can be determined. ther =D avg Then, it can be determined whether the distance factor D of the suspected faulty unit is greater than the determined distance factor alarm threshold. If the distance factor D of the suspected faulty unit is greater than the determined distance factor alarm threshold, the suspected faulty unit can be determined to be a faulty unit; otherwise, the suspected faulty unit can be determined to be a normal unit.

[0140] Return to reference Figure 1 In step S105, based on the first-order natural frequency excess factor R of each blade of each wind turbine generator set... ij Distance factor D, blade length L, and local entropy eng in different frequency bands i To determine if there are any abnormal blades in multiple wind turbine generator sets. See below for reference. Figure 7 The specific description outlines a method for determining whether abnormal blades exist in multiple wind turbine generator sets.

[0141] Figure 7 This is a flowchart illustrating a method for determining whether there are abnormal blades in a plurality of wind turbine generators according to embodiments of the present disclosure.

[0142] Reference Figure 7 In step S701, for each wind turbine generator set, the first-order natural frequency over-limit factor R of each blade is used. ij The maximum value, distance factor D, and blade length L are used to calculate the fusion feature M. The fusion feature M can be calculated in the following way:

[0143]

[0144] Where R represents the first-order natural frequency overlimit factor R of each blade. ij The maximum value in the range, f(L), is expressed as follows:

[0145]

[0146] Wherein, k is a coefficient, which can take values ​​within a preset coefficient range (e.g., but not limited to 1 to 3). Optionally, k can be 1.2, but this disclosure is not limited thereto.

[0147] In step S702, in response to the fusion feature M being less than a seventh predetermined threshold (e.g., but not limited to 1), it is determined that there are no abnormal blades.

[0148] In step S703, in response to the fusion feature M being greater than or equal to a seventh predetermined threshold, for each faulty unit, the first-order natural frequency excess factor R of each blade of the faulty unit is calculated. ij and local entropy of different frequency bands i And the blade length L, to determine whether the faulty unit has abnormal blades.

[0149] Specifically, if the blade length L is less than the first blade length threshold (e.g., but not limited to 29), then the first-order natural frequency excess factor R in the faulty unit will be increased. ij Blades exceeding a blade anomaly threshold (e.g., but not limited to 1) are identified as anomalous blades. If the blade length L is greater than a second blade length threshold (e.g., but not limited to 70), the local entropy eng of different frequency bands of each blade of the faulty unit is then calculated. i The blade corresponding to the maximum value in the threshold is identified as the abnormal blade. If the blade length L is greater than or equal to the first blade length threshold and less than or equal to the second blade length threshold, then the first-order natural frequency over-limit factor R in the faulty unit is determined. ij The numbering of blades exceeding the blade anomaly threshold, and the local entropy (eng) of different frequency bands of each blade in the faulty unit. i The maximum value in the index corresponds to the blade number, and blades with the same number are identified as abnormal blades. However, if the two numbering systems are inconsistent, it can be determined that the faulty unit may have a blade abnormality, but the blade causing the abnormality cannot be located.

[0150] The applicability of the blade anomaly identification method according to embodiments of this disclosure is verified below through two specific examples.

[0151] In the first example, a wind farm consists of 16 wind turbine generators. Initial on-site inspections confirmed that the blades of all wind turbine generators were intact. Basic blade parameters are shown in Table 2.

[0152] Table 2 Basic Blade Parameters

[0153] Parameter name Reference value First-order oscillation natural frequency 0.46 First-order swing natural frequency 0.31 blade length 93m Impeller diameter 191m

[0154] In the initial stage of blade anomaly monitoring, the exceedance of the first-order natural frequency of the wind turbine generator was statistically analyzed. Taking the blade oscillation direction as an example, with a 1% deviation as the threshold, the exceedance of the first-order natural frequency of each blade is shown in Table 3.

[0155] Table 3. Exceeding Limits of the First-Order Natural Frequency Before Correction

[0156] Leaf number Over-limit percentage Leaf 1 100％ Leaf 2 99％ Leaf 3 100％

[0157] The blade anomaly identification method according to the embodiments of the present disclosure is used to correct the first-order natural frequency of the blade. The corrected first-order natural frequency exceeding the limit in the oscillation direction of each blade is shown in Table 4.

[0158] Table 4. Blade parameters exceeding limits after correction

[0159] Leaf number alarm percentage Leaf 1 53％ Leaf 2 51.5％ Leaf 3 54％

[0160] After first-order natural frequency correction, nearly half of the false alarms of first-order natural frequency exceeding the limit are eliminated. However, if alarms are generated using single data points, a large number of false alarms still exist. To further improve alarm accuracy, first-order natural frequency correction is performed sequentially on all wind turbine generators in the field. Then, the blade anomaly identification method according to the embodiments of this disclosure is used to obtain the first-order natural frequency exceeding factor R of each blade of each wind turbine generator through a horizontal comparison across the entire field. ij . Figure 8 This shows the first-order natural frequency over-limit factor R of each blade of each wind turbine generator in the wind farm. ij Example illustration. For example... Figure 8 As shown, the first-order natural frequency overlimit factor R of each blade is... ij All values ​​are less than the preset threshold. Therefore, the false alarm of the first-order natural frequency exceeding the limit has been eliminated.

[0161] In the second example, the No. 1 blade of a wind turbine generator set No. 10000007 in a certain wind farm reported an abnormal whitening in May. The specific parameters are shown in Table 5 below.

[0162] Table 5 Basic Parameters of the Blade

[0163] Parameter name Reference value First-order oscillation natural frequency 0.579 First-order swing natural frequency 0.358 blade length 80.8m Impeller diameter 165m

[0164] Analyzing the April data of this wind farm, the first-order natural frequency excess factor R of each blade of each wind turbine generator was calculated. ij The result is as follows Figure 9 As shown. Figure 9 This shows the first-order natural frequency over-limit factor R of each blade of each wind turbine generator in the wind farm. ij Another example is shown in the diagram. For example... Figure 9 As shown, wind turbine generators No. 10000003, 10000005, 10000006, 10000007, and 10000009 all exhibited an excess factor R of the first-order natural frequency. ij Exceeding the limit (i.e., R) ij The case where it is greater than 1).

[0165] The blade anomaly identification method according to the embodiments of this disclosure is used to further analyze the high-frequency vibration data of each blade, and calculate the effective value H of the total local entropy of each blade. rms The ratio prop and distance factor D are used to initially screen suspected faulty wind turbine units (i.e., wind turbine units No. 10000003, 10000005, 10000006, 10000007, and 10000009). The effective value H of the total local entropy of the blades of each wind turbine unit is also used. rms The ratio prop and distance factor D are shown in Table 6 below.

[0166] Table 6

[0167]

[0168]

[0169] Since the distance factor D of wind turbine generator No. 10000007 is 3.8753, which is greater than the distance factor alarm threshold, wind turbine generator No. 10000007 can be identified as a faulty unit. On the other hand, by calculating the fusion feature, the fusion feature M of wind turbine generator No. 10000007 is 3.559, which is greater than the seventh predetermined threshold. Furthermore, since the blade length exceeds 70 meters, the local entropy of different frequency bands of each blade of wind turbine generator No. 10000007 can be used to determine the fault. i The maximum value in the range is used to identify the abnormal blade. Further, based on the April data from this wind farm, the local entropy of different frequency bands of each blade of wind turbine generator set 10000007 can be determined to be {0.138, 0.025, 0.021}, thus identifying blade number 1 of wind turbine generator set 10000007 as an abnormal blade, consistent with the on-site conclusion (i.e., the information reported in May). In other words, the blade anomaly identification method according to the embodiments of this disclosure has high accuracy.

[0170] Figure 10 This is a block diagram of a blade anomaly identification device for a wind turbine generator set according to an embodiment of the present disclosure. Figure 10 The blade anomaly identification device can be implemented in the control equipment of a wind farm, or in a higher-level control equipment.

[0171] Reference Figure 10The blade anomaly identification device 1000 includes: a first-order natural frequency identification unit 1001, a first-order natural frequency correction unit 1002, a candidate abnormal blade determination unit 1003, a faulty unit identification unit 1004, and an abnormal blade determination unit 1005. The first-order natural frequency identification unit 1001 identifies the actual value f of the first-order natural frequency of each blade of each wind turbine in a wind farm based on the low-frequency vibration data of each blade of each wind turbine. e The aforementioned wind turbine generators all have blades of the same type. The first-order natural frequency correction unit 1002 determines the correction value f of the first-order natural frequency of each blade in each wind turbine generator based on the actual first-order natural frequency of each blade. r The candidate anomalous blade determination unit 1003 determines the blade based on the correction value f of the first-order natural frequency of each blade in each wind turbine generator set. r 'and the actual value f of the first natural frequency e 'Calculate the first-order natural frequency over-limit factor R of each blade of each wind turbine generator set.' ij And the first-order natural frequency overlimit factor R ij Blades exceeding a preset threshold are identified as candidate abnormal blades. The fault unit identification unit 1004 calculates the local entropy (eng) of different frequency bands of each blade in each wind turbine unit based on the high-frequency vibration data of each blade. i Local entropy based on different frequency bands i The distance factor D for each wind turbine is determined, and based on the distance factor D, faulty turbines among multiple wind turbines are identified. The abnormal blade determination unit 1005 determines the first-order natural frequency over-limit factor R of each blade in each wind turbine. ij Distance factor D, blade length L, and local entropy eng in different frequency bands i To determine whether there are abnormal blades in multiple wind turbine generator sets.

[0172] According to embodiments of this disclosure, the blade anomaly identification device 1000 may further include a data preprocessing unit. The data preprocessing unit acquires vibration data of each blade of each of the multiple wind turbine generator sets, preprocesses the acquired vibration data to remove abnormal data and vibration data from when the wind turbine generator set is shut down, and performs low-pass and high-pass filtering on the preprocessed vibration data to obtain low-frequency and high-frequency vibration data of each blade of each wind turbine generator set.

[0173] According to embodiments of this disclosure, for any blade, the first-order natural frequency identification unit 1001 can determine the actual value of the first-order natural frequency of the blade within a preset range centered on the theoretical value of the blade's first-order natural frequency, based on the low-frequency vibration data of the blade and the harmonic data of a predetermined multiple of the rotor frequency data of the wind turbine generator where the blade is located.

[0174] According to embodiments of this disclosure, the first-order natural frequency correction unit 1002 can determine a plurality of wind turbine generator sets whose data volume meets preset requirements from a plurality of wind turbine generator sets; determine multiple statistical values ​​of the actual value of the first-order natural frequency of each blade in the plurality of wind turbine generator sets; and respond to the variance std among the multiple statistical values. ij If the average value var is greater than or equal to a first predetermined threshold, the theoretical value f of the first-order natural frequency of the blade will be... theo The correction value f is determined as the first-order natural frequency of each wind turbine blade. r '; Response to the variance std in multiple statistics ij If the average value var is less than a first predetermined threshold, the correction value of the first natural frequency of each wind turbine blade is determined based on multiple statistical values ​​of the actual value of the first natural frequency of each blade.

[0175] Optionally, the first-order natural frequency correction unit 1002 can calculate the average of the actual values ​​of the first-order natural frequencies of all blades in several wind turbine generator sets for which the correction values ​​of the first-order natural frequencies of the blades have not yet been determined. Determine the maximum value of the average deviation between any two blades of the current wind turbine generator set, avg_error, and the mean value of the average deviation between any two blades, avg_mean; in response to avg_error being greater than or equal to a second predetermined threshold, determine the maximum value of the deviation between two predetermined quantiles of the actual value of the first-order natural frequency of each blade of the current wind turbine generator set, q, and the maximum value of the standard deviation of the actual value of the first-order natural frequency of each blade of the current wind turbine generator set, std. max The average of the minimum and second minimum deviations between any two blades of the current wind turbine generator set. less The theoretical value of the first-order natural frequency f of the blade. theo The correction value f is determined as the first natural frequency of the current wind turbine blades. r ', or the reference value f c The correction value f is determined as the first natural frequency of the current wind turbine blades. r ', where the average value of the first-order natural frequency of each blade in several wind turbine generator sets can be used as the basis for calculation. Calculate the reference value f cAlternatively, it can be based on the correction value f of the first-order natural frequency of all known wind turbine blades. r 'Calculate the reference value f' c .

[0176] Alternatively, if the correction value f for the first natural frequency of the blade has been determined... r The number of wind turbine generators is greater than or equal to NA / 2, and the correction value f of the first-order natural frequency of the blades of all the wind turbine generators has been determined. r 'variance var f If the value is less than a preset correction threshold (e.g., but not limited to 0.06), the first-order natural frequency correction unit 1002 can correct the first-order natural frequency f of all the wind turbine blades that have been determined. r The mean of ' is used as the reference value f c Here, NA represents the total number of wind turbine generators in the wind farm. Otherwise, the first-order natural frequency correction unit 1002 can be based on the average of the actual values ​​of the first-order natural frequencies of each blade in the wind turbine generators. Calculate the reference value f c .

[0177] Optionally, the first-order natural frequency correction unit 1002 may also, in response to avg_error being less than a second predetermined threshold, determine the maximum value q among the deviations between two preset quantiles of the actual value of the first-order natural frequency of each blade of the current wind turbine generator; and, in response to q being greater than or equal to a third predetermined threshold, adjust the theoretical value f of the first-order natural frequency of the blade. theo The correction value f is determined as the first natural frequency of the current wind turbine blades. r '; In response to q being less than a third predetermined threshold, determine the average value.' Compared with reference value f c Whether the first difference between them falls within a preset range; in response to the first difference falling within the preset range, the reference value f is adjusted. c The correction value f is determined as the first natural frequency of the current wind turbine blades. r '; In response to the first difference not falling within the preset range, determine the mean avg_mean and the average value of the average deviation between each pair of blades of the current wind turbine generator.' The second difference between them; in response to the second difference being less than a fourth predetermined threshold, the average value is... The correction value f is determined as the first natural frequency of the current wind turbine blades. r '.

[0178] Alternatively, the first-order natural frequency correction unit 1002 may also respond to a maximum value q being less than or equal to a third predetermined threshold, and the maximum value std maxLess than or equal to a second predetermined threshold, or, in response to the average value avg less Compared with reference value f c The absolute value of the difference between them is less than the fifth predetermined threshold, which will be the reference value f. c The correction value f is determined as the first natural frequency of the current wind turbine blades. r Otherwise, the theoretical value of the first-order natural frequency f of the blade is used. theo The correction value f is determined as the first natural frequency of the current wind turbine blades. r '.

[0179] According to embodiments of this disclosure, the candidate abnormal blade determination unit 1003 can calculate a correction value f of the first-order natural frequency for each blade of each wind turbine generator set in a plurality of wind turbine generator sets. r 'and the actual value f of the first natural frequency rr The deviation e between them, and the ratio p of the number of deviations greater than the sixth predetermined threshold within the preset time period to the total number of deviations. ij For each blade with the same number in all wind turbine generator sets, calculate the average proportion corresponding to the blades with the same number. j and standard deviation v j For each blade of each wind turbine in multiple wind turbine generator sets, calculate the first-order natural frequency exceedance factor R of the blade based on the proportion corresponding to the blade and the average value and / or standard deviation corresponding to the blade. ij .

[0180] According to embodiments of this disclosure, the faulty unit identification unit 1004 can calculate the total local entropy value enr of each blade based on the local entropy of multiple frequency bands of the blade. p For each blade, the total local entropy value enr is calculated multiple times within a preset time period. p Calculate the mean local entropy of the blade. For each wind turbine, based on the average local entropy of each blade of the wind turbine. Calculate the distance factor D of the wind turbine generator set.

[0181] Optionally, the fault identification unit 1004 can determine the average local entropy of each blade of each wind turbine for each wind turbine. The maximum value in Calculation and maximum value The effective value H of the total local entropy of the corresponding blade rms And calculate the maximum value. The ratio prop between the mean of the local entropy of the remaining blades and the mean of the local entropy of the other blades; for each wind turbine, the ratio prop is based on the average of the ratios of multiple wind turbines. avg Determine whether the wind turbine generator set belongs to the potentially faulty units; for each potentially faulty unit identified, based on the H of the potentially faulty unit... rms H of multiple wind turbine generator sets rms The average value H rms_avg and variance H rms_std Determine whether a potentially faulty turbine belongs to the category of suspected faulty turbines; for each suspected faulty turbine, based on the distance factor D of the suspected faulty turbine and the average distance factor D of multiple wind turbines... avg and variance D std To determine whether a suspected faulty unit is indeed a faulty unit.

[0182] According to embodiments of this disclosure, the abnormal blade determination unit 1005 can, for each wind turbine generator set, determine the abnormal blade based on the first-order natural frequency over-limit factor R of each blade. ij The maximum value, distance factor D, and blade length L are used to calculate the fusion feature M. If the fusion feature M is less than a seventh predetermined threshold, it is determined that no abnormal blades exist. If the fusion feature M is greater than or equal to the seventh predetermined threshold, for each faulty unit, the first-order natural frequency excess factor R of each blade in the faulty unit is calculated. ij and local entropy of different frequency bands i And the blade length L, to determine whether the faulty unit has abnormal blades.

[0183] Alternatively, the abnormal blade determination unit 1005 may also, in response to a blade length L being less than a first blade length threshold, determine the first-order natural frequency excess factor R in the faulty unit. ij Blades with a length L greater than the blade anomaly threshold are identified as abnormal blades; in response to a blade length L greater than the second blade length threshold, the local entropy eng of different frequency bands of each blade of the faulty unit is... i The blade corresponding to the maximum value in the value is identified as the abnormal blade; in response to a blade length L being greater than or equal to the first blade length threshold and less than or equal to the second blade length threshold, the first-order natural frequency over-limit factor R in the faulty unit is determined. ij The numbering of blades exceeding the blade anomaly threshold, and the local entropy (eng) of different frequency bands of each blade in the faulty unit. i The leaf number corresponding to the maximum value in the value is determined, and leaves with the same number are identified as abnormal leaves.

[0184] Figure 11 This is a block diagram of a blade anomaly identification device for a wind turbine generator set according to another embodiment of the present disclosure. Figure 11 The blade anomaly identification device can be implemented in the control equipment of a wind farm, or in a higher-level control equipment.

[0185] Reference Figure 11 The blade anomaly identification device 1100 includes a processor 1110 and a memory 1120. The processor 1110 may include (but is not limited to) a central processing unit (CPU), a digital signal processor (DSP), a microcomputer, a field-programmable gate array (FPGA), a system-on-a-chip (SoC), a microprocessor, an application-specific integrated circuit (ASIC), etc. The memory 1120 stores computer programs to be executed by the processor 1110. The memory 1120 includes high-speed random access memory and / or non-volatile computer-readable storage media. When the processor 1110 executes the computer program stored in the memory 1120, the blade anomaly identification method described above can be implemented.

[0186] The blade anomaly identification method for wind turbine generators according to embodiments of this disclosure can be programmed into a computer program and stored on a computer-readable storage medium. When the computer program is executed by a processor, the blade anomaly identification method described above can be implemented. Examples of computer-readable storage media include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disc storage, hard disk drive (HDD), solid-state drive (SSD), card storage (such as multimedia cards, secure digital (SD) cards, or ultra-fast digital (XD) cards), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state drive, and any other device configured to store computer programs and any associated data, data files, and data structures in a non-transitory manner and to provide the computer programs and any associated data, data files, and data structures to a processor or computer so that the processor or computer can execute the computer programs. In one example, the computer programs and any associated data, data files, and data structures are distributed across a networked computer system, such that the computer programs and any associated data, data files, and data structures are stored, accessed, and executed in a distributed manner through one or more processors or computers.

[0187] The blade anomaly identification method, apparatus, and storage medium for wind turbine generators according to embodiments of this disclosure achieve accurate assessment of the blade's natural frequency through preprocessing of blade vibration data, correction of the blade's natural frequency, and statistical analysis of the blade's natural frequency deviation. Simultaneously, by constructing features from the high-frequency vibration data of the blades, the vibration energy deviation of the three blades is accurately identified. Finally, by using a blade length weighting factor, the calculation results of the two features are weighted and analyzed to complete the anomaly identification of blades of different lengths. The blade anomaly identification method, apparatus, and storage medium for wind turbine generators according to embodiments of this disclosure require low hardware costs, require a small number of fault samples, and have a simple data source. They are adaptable to application scenarios without SCADA data and have good applicability to both older wind farms and large wind turbine generators, exhibiting unique advantages in engineering applications.

[0188] On the other hand, the blade anomaly identification method, device, and storage medium of the wind turbine generator set according to the embodiments of this disclosure combine the blade damage mechanism and engineering data characteristics to construct blade anomaly identification feature factors, and use engineering experience to establish a damage assessment function based on blade length. This can adaptively identify common faults of blades of different sizes, effectively solve the problems of low accuracy and poor robustness of current blade anomaly identification, and improve the operating efficiency and power generation efficiency of the wind turbine generator set.

[0189] While some embodiments of this disclosure have been shown and described, those skilled in the art will understand that modifications may be made to these embodiments without departing from the principles and spirit of this disclosure, which are defined by the claims and their equivalents.

Claims

1. A method for identifying blade anomalies in a wind turbine generator set, characterized in that, The blade anomaly identification method includes: Based on the low-frequency vibration data of each blade of each wind turbine in a wind farm, the actual value of the first-order natural frequency of each blade of each wind turbine is identified, wherein the multiple wind turbines have blades of the same type. Based on the actual first-order natural frequency of each blade of each wind turbine generator set, determine the correction value of the first-order natural frequency of each blade of each wind turbine generator set. Based on the correction value of the first natural frequency of each blade of each wind turbine generator set and the actual value of the first natural frequency, the first natural frequency excess factor of each blade of each wind turbine generator set is calculated, and blades with the first natural frequency excess factor greater than the preset threshold are identified as candidate abnormal blades. Based on the high-frequency vibration data of each blade of each wind turbine, the local entropy of each blade of each wind turbine is calculated in different frequency bands. Based on the calculated local entropy of different frequency bands, the distance factor of each wind turbine is determined. Based on the distance factor of each wind turbine, the faulty unit among the multiple wind turbines is identified. Based on the first-order natural frequency over-limit factor, distance factor, blade length, and local entropy of different frequency bands of each blade of each wind turbine generator set, it is determined whether there are abnormal blades in the multiple wind turbine generator sets.

2. The method for identifying blade anomalies in wind turbine generator sets as described in claim 1, characterized in that, The blade anomaly identification method further includes: Obtain vibration data of each blade of each of the multiple wind turbine generator sets; The acquired vibration data is preprocessed to remove abnormal data and vibration data of the wind turbine generator in the shutdown state. The preprocessed vibration data were subjected to low-pass and high-pass filtering to obtain the low-frequency and high-frequency vibration data of each blade of each wind turbine.

3. The method for identifying blade anomalies in wind turbine generators as described in claim 1, characterized in that, The steps for identifying the actual values ​​of the first-order natural frequencies of each blade in each wind turbine generator set include: For any blade, based on the low-frequency vibration data of the blade and the harmonic data of a predetermined multiple of the rotor frequency data of the wind turbine where the blade is located, the actual value of the first natural frequency of the blade is determined within a preset range centered on the theoretical value of the first natural frequency of the blade.

4. The method for identifying blade anomalies in wind turbine generators as described in claim 1, characterized in that, The steps for determining the correction value of the first-order natural frequency of each wind turbine blade include: From the plurality of wind turbine generator sets, determine a number of wind turbine generator sets whose data volume meets the preset requirements; Multiple statistical values ​​were used to determine the actual value of the first-order natural frequency of each blade in the aforementioned wind turbine generator sets. In response to the average variance of the various statistical values ​​being greater than or equal to a first predetermined threshold, the theoretical value of the first natural frequency of the blade is determined as the correction value of the first natural frequency of the blade of each wind turbine generator set. In response to the average variance being less than a first predetermined threshold, a correction value for the first natural frequency of each wind turbine blade is determined based on multiple statistical values ​​of the actual value of the first natural frequency of each blade.

5. The method for identifying blade anomalies in wind turbine generators as described in claim 4, characterized in that, The steps for determining the correction value of the first-order natural frequency of each wind turbine blade, based on multiple statistical values ​​of the actual first-order natural frequency of each blade, include: Calculate the average of the actual values ​​of the first-order natural frequencies of all blades in all wind turbine generator sets whose correction values ​​for the first-order natural frequencies of the blades have not yet been determined. Determine the maximum value and the mean value of the average deviation between any two blades of the current wind turbine generator set; In response to the maximum value among the average deviations being greater than or equal to a second predetermined threshold, the theoretical value of the first natural frequency of the blade is determined as a correction value for the first natural frequency of the blade of the current wind turbine generator set, based on the maximum value among the deviations between two preset quantiles of the actual value of the first natural frequency of the first natural frequency of the actual value of the first natural frequency of the blade of the current wind turbine generator set, the maximum value among the standard deviations of the actual value of the first natural frequency of the blade of the current wind turbine generator set, and the average of the minimum and second smallest values ​​among the average deviations between any two blades of the current wind turbine generator set. Alternatively, a reference value is determined as a correction value for the first natural frequency of the blade of the current wind turbine generator set. The reference value is calculated based on the average of the actual values ​​of the first natural frequency of the blade of each of the plurality of wind turbine generator sets, or based on the correction values ​​of the first natural frequency of the blades of all the wind turbine generator sets that have been determined.

6. The method for identifying blade anomalies in a wind turbine generator as described in claim 5, characterized in that, In response to the fact that the number of wind turbine generator sets whose first-order natural frequency correction values ​​of the blades have been determined is greater than or equal to half the number of the plurality of wind turbine generator sets, and the variance of the first-order natural frequency correction values ​​of the blades of all the determined wind turbine generator sets is less than a preset correction threshold, the mean of the first-order natural frequency correction values ​​of the blades of all the determined wind turbine generator sets is used as the reference value. Otherwise, the reference value is calculated based on the average of the actual values ​​of the first-order natural frequencies of each blade in the plurality of wind turbine generator sets.

7. The method for identifying blade anomalies in a wind turbine generator as described in claim 5, characterized in that, The steps for determining the correction value of the first-order natural frequency of each wind turbine blade, based on multiple statistical values ​​of the actual first-order natural frequency of each blade, also include: In response to the fact that the maximum value among the deviations of the average value is less than a second predetermined threshold, the maximum value among the deviations between two preset quantiles of the actual value of the first natural frequency of each blade of the current wind turbine is determined. In response to the fact that the maximum value of the deviation between two preset quantiles of the actual value of the first natural frequency of each blade of the current wind turbine generator is greater than or equal to a third predetermined threshold, the theoretical value of the first natural frequency of the blade is determined as the correction value of the first natural frequency of the blade of the current wind turbine generator. In response to the fact that the maximum value of the deviation between two preset quantiles of the actual value of the first natural frequency of each blade of the current wind turbine is less than a third predetermined threshold, it is determined whether the first difference between the average of the actual values ​​of the first natural frequencies of all blades of all wind turbines for which the correction value of the first natural frequency has not yet been determined and the reference value falls within a preset range. In response to the first difference falling into a preset range, the reference value is determined as the correction value of the first natural frequency of the blade of the current wind turbine generator set; In response to the first difference not falling within the preset range, a second difference is determined between the mean of the average deviation between the two blades of the current wind turbine generator set and the average of the actual values ​​of the first natural frequencies of all blades of all wind turbine generator sets whose correction values ​​for the first natural frequencies have not yet been determined. In response to the second difference being less than a fourth predetermined threshold, the average of the actual values ​​of the first natural frequencies of all blades of all wind turbine generator sets whose first natural frequency correction values ​​have not yet been determined is determined as the correction value of the first natural frequency of the blades of the current wind turbine generator set.

8. The method for identifying blade anomalies in a wind turbine generator as described in claim 5, characterized in that, The steps of determining the theoretical value of the first natural frequency of the blade as the correction value of the first natural frequency of the blade of the current wind turbine generator set, or determining the correction value calculated based on the average of the actual values ​​of the first natural frequencies of the blades of the several wind turbine generator sets as the correction value of the first natural frequency of the blades of the current wind turbine generator set, include: In response to the fact that the maximum value of the deviation between two preset quantiles of the actual value of the first natural frequency of each blade of the current wind turbine is less than or equal to a third predetermined threshold, and the maximum value of the standard deviation of the actual value of the first natural frequency of each blade of the current wind turbine is less than or equal to a second predetermined threshold, or in response to the fact that the absolute value of the difference between the average of the minimum and the second smallest values ​​of the average deviation between pairs of blades of the previous wind turbine and the reference value is less than a fifth predetermined threshold, the reference value is determined as the correction value of the first natural frequency of the blade of the current wind turbine. Otherwise, the theoretical value of the first natural frequency of the blade is determined as the correction value of the first natural frequency of the blade of the current wind turbine generator set.

9. The leaf anomaly identification method as described in claim 1, characterized in that, The steps for calculating the first-order natural frequency excess factor of each blade of each wind turbine generator set, based on the corrected value and the actual value of the first-order natural frequency of each blade, include: For each blade of each wind turbine generator set in the multiple wind turbine generator sets, calculate the deviation between the corrected value of the first natural frequency and the actual value of the first natural frequency, and count the ratio of the number of deviations greater than the sixth predetermined threshold to the total number of deviations within a preset time period. For each blade with the same number in all the wind turbine generator sets, calculate the average value and standard deviation of the proportion corresponding to the blade with the same number; For each blade of each wind turbine in the plurality of wind turbine generator sets, the first-order natural frequency over-limit factor of the blade is calculated based on the proportion corresponding to the blade and the average value and / or standard deviation corresponding to the blade.

10. The leaf anomaly identification method as described in claim 1, characterized in that, The steps for determining the distance factor for each wind turbine based on the calculated local entropy of different frequency bands include: For each blade, the total local entropy of the blade is calculated based on the local entropy of multiple frequency bands of the blade; For each blade, the mean local entropy of the blade is calculated based on the total local entropy of the blade calculated multiple times within a preset time period. For each wind turbine generator set, the distance factor of the wind turbine generator set is calculated based on the mean local entropy of each blade of the wind turbine generator set.

11. The leaf anomaly identification method as described in claim 10, characterized in that, The steps for identifying faulty units among the multiple wind turbine generator sets based on the distance factor of each wind turbine generator set include: For each wind turbine generator set, determine the maximum value among the mean local entropy values ​​of each blade of the wind turbine generator set, calculate the effective value of the total local entropy value of the blade corresponding to the maximum value, and calculate the ratio between the maximum value and the average value of the mean local entropy values ​​of the remaining blades. For each wind turbine generator set, based on the ratio of the wind turbine generator set and the average of the ratios of the multiple wind turbine generator sets, it is determined whether the wind turbine generator set belongs to the potentially faulty unit. For each of the identified potential faulty units, based on the effective value of the potential faulty unit and the average and variance of the effective values ​​of the multiple wind turbine units, it is determined whether the potential faulty unit belongs to the suspected faulty unit. For each suspected faulty unit among the identified suspected faulty units, based on the distance factor of the suspected faulty unit and the average and variance of the distance factors of the multiple wind turbine units, it is determined whether the suspected faulty unit belongs to the faulty unit.

12. The leaf anomaly identification method as described in claim 11, characterized in that, The steps for determining whether there are abnormal blades in the multiple wind turbine generator sets include: For each wind turbine generator, the fusion characteristics are calculated based on the maximum value of the first-order natural frequency over-limit factor of each blade, the distance factor, and the blade length. In response to the fusion feature being less than a seventh predetermined threshold, it is determined that there are no abnormal blades; In response to the fusion feature being greater than or equal to a seventh predetermined threshold, for each faulty unit, based on the first-order natural frequency over-limit factor of each blade of the faulty unit, the local entropy of different frequency bands, and the blade length, it is determined whether the faulty unit has abnormal blades.

13. The leaf anomaly identification method as described in claim 11, characterized in that, The steps for determining whether the faulty unit has abnormal blades include: In response to a blade length less than a first blade length threshold, blades in the faulty unit whose first-order natural frequency exceedance factor is greater than the blade abnormality threshold are identified as abnormal blades. In response to a blade length greater than a second blade length threshold, the blade corresponding to the maximum value of the local entropy in different frequency bands of each branch blade of the faulty unit is identified as an abnormal blade. In response to a blade length greater than or equal to a first blade length threshold and less than or equal to a second blade length threshold, the number of the blade in the faulty unit whose first-order natural frequency exceedance factor is greater than the blade abnormality threshold is determined, as well as the number of the blade corresponding to the maximum value of the local entropy of different frequency bands of each blade in the faulty unit, and the blades with the same number are identified as abnormal blades.

14. A blade anomaly identification device for a wind turbine generator set, characterized in that, The blade anomaly detection device includes: The first-order natural frequency identification unit is configured to: identify the actual value of the first-order natural frequency of each blade of each wind turbine in a wind farm based on the low-frequency vibration data of each blade of each wind turbine in a wind farm, wherein the multiple wind turbines have blades of the same type. The first-order natural frequency correction unit is configured to: determine the correction value of the first-order natural frequency of each blade of each wind turbine based on the actual first-order natural frequency of each blade of each wind turbine. The candidate abnormal blade determination unit is configured to: calculate the first-order natural frequency over-limit factor of each blade of each wind turbine based on the correction value of the first-order natural frequency of each blade of each wind turbine and the actual value of the first-order natural frequency, and identify blades whose first-order natural frequency over-limit factor is greater than a preset threshold as candidate abnormal blades. The faulty unit identification unit is configured to: calculate the local entropy of different frequency bands of each blade of each wind turbine based on the high-frequency vibration data of each blade of each wind turbine, determine the distance factor of each wind turbine based on the calculated different preferred local entropy, and identify the faulty unit among the multiple wind turbines based on the distance factor of each wind turbine. The abnormal blade determination unit is configured to determine whether there are abnormal blades in the multiple wind turbine generator sets based on the first-order natural frequency over-limit factor, distance factor, blade length, and local entropy of different frequency bands of each blade of each wind turbine generator set.

15. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the blade anomaly identification method as described in any one of claims 1 to 13.

16. A blade anomaly identification device for a wind turbine generator set, characterized in that, The blade anomaly detection device includes: processor; and The memory stores a computer program that, when executed by a processor, implements the blade anomaly identification method as described in any one of claims 1 to 13.

Images