Fault monitoring method and system for doubly-fed wind turbine based on frequency monitoring

Abstract

The application relates to the technical field of fault monitoring of wind turbines, in particular to a doubly-fed wind turbine fault monitoring method and system based on frequency monitoring, which comprises the following steps: acquiring operation parameter data of a doubly-fed wind turbine, generating a similar working condition time period of a gearbox in the doubly-fed wind turbine according to the operation parameter data based on a clustering algorithm; extracting target vibration data; generating modal components and central frequencies corresponding to each modal component according to the target vibration data, and generating modal component energy; generating a fusion modal energy value according to the modal component energy, and generating a unit fault early warning result. According to the application, only vibration data corresponding to the similar working condition time period is extracted, so that the interference of signal differences under different working conditions on vibration characteristics is excluded, a comprehensive criterion for unit faults based on the fusion modal energy value is used, the gearbox fault risk is early warned as accurately as possible, and false negatives are reduced.

Description

Technical Field

[0001] This application relates to the field of fault monitoring technology for wind turbine generators, and in particular to a fault monitoring method and system for doubly fed wind turbine generators based on frequency monitoring. Background Technology

[0002] Doubly fed wind turbines have become the mainstay of onshore and offshore wind power worldwide due to their variable speed and constant frequency operation capabilities, excellent power control performance, and high cost performance. They are one of the most widely used wind turbine types, especially in large wind farms of megawatt level and above. As the most core and vulnerable component of a wind turbine, the gearbox's operating status directly affects the safety and maintenance costs of the entire unit.

[0003] In existing technologies, gearbox fault monitoring generally adopts methods that directly extract features based on full-time vibration data, or relies on traditional decomposition methods such as EMD and wavelet transform, in order to extract fault-related characteristic frequency components. Vibration data is often mixed with different operating conditions, and direct full-time analysis or empirical segmentation is difficult to truly guarantee the consistency of the data under operating conditions. This can easily lead to the failure of the criteria for the same vibration feature under different operating conditions, resulting in false alarms. Traditional decomposition methods such as EMD and wavelet analysis of signals are prone to low fault monitoring accuracy due to mode aliasing.

[0004] Therefore, there is an urgent need to design a fault monitoring method and system for doubly fed wind turbines based on frequency monitoring. Summary of the Invention

[0005] Therefore, it is necessary to address the aforementioned technical problems by providing a method and system for monitoring doubly-fed wind turbine generator faults based on frequency monitoring. This method and system can solve the problems of false alarms caused by the failure of criteria for the same vibration feature under different operating conditions, as well as the problems of low fault monitoring accuracy caused by the susceptibility of traditional signal decomposition methods to mode mixing. It can extract vibration data only from the time period corresponding to similar operating conditions, eliminate the interference of signal differences under different operating conditions on vibration features, and achieve the most accurate early warning of gearbox fault risks and reduce missed alarms by using a comprehensive criterion based on fused modal energy values ​​for unit faults.

[0006] The technical solution of this invention is as follows:

[0007] A fault monitoring method for doubly-fed induction generator (DFIG) wind turbines based on frequency monitoring, the method comprising:

[0008] Obtain the operating parameter data of the doubly fed wind turbine, and generate similar operating condition time periods of the gearbox in the doubly fed wind turbine based on the operating parameter data using a clustering algorithm;

[0009] Target vibration data is extracted from the vibration monitoring data of the gearbox based on the similar operating condition time period;

[0010] Modal components and their corresponding center frequencies are generated based on the target vibration data, and modal component energy is generated based on the modal components and their corresponding center frequencies.

[0011] A fused mode energy value is generated based on the modal component energy, and a unit fault early warning result is generated based on the fused mode energy value.

[0012] Optionally, the operating parameter data of the doubly-fed induction generator (DFIG) wind turbine is obtained, and a similar operating condition time period of the gearbox in the DFIG wind turbine is generated based on the operating parameter data using a clustering algorithm, including:

[0013] Obtain the operating parameter data of the doubly fed wind turbine, and generate a multi-dimensional operating condition feature vector based on the operating parameter data;

[0014] The multidimensional working condition feature vector is normalized and weighted, and a weighted feature vector is generated.

[0015] The weighted feature vector is segmented according to a preset segmentation interval, and the target working condition segment is selected based on a clustering algorithm and preset screening conditions.

[0016] A similar operating condition time period is generated based on the target operating condition segment.

[0017] Optionally, the preset segmented intervals include wind speed intervals and pitch angle intervals;

[0018] The weighted feature vector is segmented according to a preset segmentation interval, and the target working condition segment is selected based on a clustering algorithm and preset screening conditions, including:

[0019] The weighted feature vector is segmented according to the wind speed range and the pitch angle range, and the data in each segment is clustered in a fine-grained manner using a clustering algorithm to obtain multiple operating condition clusters.

[0020] The target working condition segment is selected from each working condition cluster according to the preset filtering conditions;

[0021] Calculate the center vector of each target working condition segment, and select the target working condition segment from each target working condition segment based on the center vector.

[0022] Optionally, the vibration monitoring data includes first vibration data and second vibration data, wherein the first vibration data is acquired based on a first vibration sensor and the second vibration data is acquired based on a second vibration sensor; the first vibration sensor and the second vibration sensor are arranged opposite to each other on both sides of the input shaft bearing housing of the gearbox.

[0023] Target vibration data is extracted from the vibration monitoring data of the gearbox based on the similar operating condition time period, including:

[0024] Obtain the first vibration data and the second vibration data of the gearbox;

[0025] Differential vibration data is generated based on the first vibration data and the second vibration data;

[0026] Target vibration data is extracted from the differential vibration data based on the similar operating condition time period.

[0027] Optionally, generating modal components and their corresponding center frequencies based on the target vibration data, and generating modal component energy based on the modal components and their corresponding center frequencies, includes:

[0028] The target vibration data is decomposed based on the variational mode decomposition method, and multiple modal components and the center frequency corresponding to each modal component are generated.

[0029] The target modal components are selected based on each modal component, center frequency, and estimated main gear meshing frequency;

[0030] Modal component energy is generated based on the target modal component.

[0031] Optionally, a fused modal energy value is generated based on the modal component energy, and a unit fault early warning result is generated based on the fused modal energy value, including:

[0032] Generate a fused mode energy value based on the modal component energy;

[0033] Based on the fusion mode energy values ​​corresponding to each of the target vibration data, an energy growth rate and an energy fluctuation anomaly degree are generated; based on the energy growth rate and energy fluctuation anomaly degree, a unit fault early warning result is generated.

[0034] Optionally, the unit fault warning results include unit fault warnings and unit normal indications;

[0035] Based on the energy growth rate and energy fluctuation anomaly, a unit fault early warning result is generated, including:

[0036] Determine whether the energy growth rate is greater than the healthy growth rate and whether the energy fluctuation abnormality is greater than the healthy abnormality; if the determination is yes, generate a unit fault warning; if the determination is no, generate a unit normal indication.

[0037] Optionally, a frequency monitoring-based fault monitoring system for doubly-fed wind turbine generators is also provided, the system comprising:

[0038] The similar operating condition generation module is used to obtain the operating parameter data of the doubly fed wind turbine and generate similar operating condition time periods of the gearbox in the doubly fed wind turbine based on the operating parameter data using a clustering algorithm.

[0039] The vibration data filtering module is used to extract target vibration data from the vibration monitoring data of the gearbox based on the similar working condition time period.

[0040] The modal energy generation module is used to generate modal components and the center frequency corresponding to each modal component based on the target vibration data, and to generate modal component energy based on the modal components and the center frequency corresponding to each modal component.

[0041] The fault warning generation module is used to generate a fused mode energy value based on the modal component energy, and to generate a unit fault warning result based on the fused mode energy value.

[0042] Optionally, the similar operating condition generation module is further configured to: acquire operating parameter data of the doubly-fed wind turbine, generate a multi-dimensional operating condition feature vector based on the operating parameter data; normalize and assign weights to the multi-dimensional operating condition feature vector, and generate a weighted feature vector; segment the weighted feature vector according to a preset segmentation interval, and select a target operating condition segment based on a clustering algorithm and preset filtering conditions; and generate a similar operating condition time period based on the target operating condition segment.

[0043] Optionally, the similar operating condition generation module is further configured to: segment the weighted feature vector according to a preset segmentation interval, and select target operating condition segments based on a clustering algorithm and preset screening conditions, including: segmenting the weighted feature vector according to the wind speed interval and the pitch angle interval, and performing fine-grained clustering on the data in each segment using a clustering algorithm to obtain multiple operating condition clusters; selecting target operating condition segments from each of the operating condition clusters according to preset screening conditions; calculating the center vector of each of the target operating condition segments, and selecting target operating condition segments from each of the target operating condition segments based on the center vector.

[0044] Optionally, the vibration monitoring data includes first vibration data and second vibration data, wherein the first vibration data is acquired based on a first vibration sensor and the second vibration data is acquired based on a second vibration sensor; the first vibration sensor and the second vibration sensor are arranged opposite to each other on both sides of the input shaft bearing housing of the gearbox; the vibration data filtering module is further configured to: acquire the first vibration data and the second vibration data of the gearbox; generate differential vibration data based on the first vibration data and the second vibration data; and extract target vibration data from the differential vibration data according to the similar operating condition time period.

[0045] Optionally, the modal energy generation module is further configured to: decompose the target vibration data based on the variational mode decomposition method, and generate multiple modal components and the center frequency corresponding to each modal component; select target modal components based on each modal component, the center frequency and the estimated main gear meshing frequency; and generate modal component energy based on the target modal components.

[0046] Optionally, the fault warning generation module is further configured to: generate a fused modal energy value based on the modal component energy; generate an energy growth rate and an energy fluctuation anomaly degree based on the fused modal energy value corresponding to each of the target vibration data; and generate a unit fault warning result based on the energy growth rate and the energy fluctuation anomaly degree.

[0047] Optionally, the unit fault warning result includes a unit fault warning indication and a unit non-fault indication; the fault warning generation module is further used to: determine whether the energy growth rate is greater than the healthy growth rate and whether the energy fluctuation abnormality is greater than the healthy abnormality; if the determination is yes, then generate a unit fault warning; if the determination is no, then generate a unit normal indication.

[0048] Optionally, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps described in the above-described frequency monitoring-based doubly-fed wind turbine fault monitoring method.

[0049] Optionally, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the steps described in the above-described frequency monitoring-based double-fed wind turbine fault monitoring method.

[0050] The technical effects achieved by this invention are as follows:

[0051] The aforementioned frequency-monitoring-based fault monitoring method and system for doubly-fed induction generator (DFIG) wind turbines acquires operating parameter data of the DFIG wind turbines. Based on a clustering algorithm, it generates similar operating condition time periods for the gearbox within the DFIG wind turbine based on the operating parameter data. It then extracts target vibration data from the gearbox vibration monitoring data based on these similar operating condition time periods. Finally, it generates modal components and their corresponding center frequencies based on the target vibration data, and generates modal component energy based on the modal components and their corresponding center frequencies. The system then generates a fused modal energy value based on the modal component energy, and finally generates a fault warning result based on the fused modal energy value. This application aims to provide fundamental data for subsequent gearbox analysis, enabling frequency-based operational status analysis under the same operating conditions to improve the accuracy of subsequent fault monitoring. It fully considers the varying impacts of different operating conditions, such as wind speed, engine speed, and power, on gearbox vibration health. By acquiring operating parameter data of the doubly-fed induction generator (DFIG) wind turbine, and using a clustering algorithm, it generates similar operating condition time periods for the gearbox within the DFIG wind turbine based on this data. This allows for accurate identification of time periods where the gearbox operates under similar conditions, effectively avoiding interference from different operating conditions on vibration data characteristics and improving the reliability of subsequent feature extraction and anomaly detection. Specifically, target vibration data is extracted from the gearbox vibration monitoring data based on the similar operating condition time period. By extracting vibration data only from the time period corresponding to the similar operating conditions, interference from signal differences under different operating conditions on vibration characteristics is eliminated. Next, modal components and their corresponding center frequencies are generated based on the target vibration data. Modal component energy is then generated based on the modal components and their corresponding center frequencies, thus separating each physical characteristic modal component with a clearly defined center frequency. The energy of each mode is calculated, accurately extracting the gear meshing frequency and its harmonics, effectively avoiding frequency aliasing or information loss. Finally, a fused modal energy value is generated based on the modal component energy, and a unit fault early warning result is generated based on the fused modal energy value. By using a comprehensive criterion for unit faults based on the fused modal energy value, the gearbox fault risk can be predicted as accurately as possible, reducing missed reports and improving the timeliness of unit fault monitoring and maintenance response. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating a frequency monitoring-based fault monitoring method for doubly-fed wind turbines in one embodiment.

[0053] Figure 2 This is a flowchart illustrating the process of generating similar working condition time periods in one embodiment;

[0054] Figure 3 This is a flowchart illustrating the process of selecting the target operating condition segment in one embodiment;

[0055] Figure 4This is a schematic diagram of the process for extracting target vibration data in one embodiment;

[0056] Figure 5 This is a schematic diagram of the process for generating modal component energy in one embodiment;

[0057] Figure 6 This is a schematic diagram of the process for generating unit fault early warning results in one embodiment;

[0058] Figure 7 This is a schematic diagram of the process for generating a normal unit indication in one embodiment;

[0059] Figure 8 This is a structural block diagram of a frequency monitoring-based fault monitoring system for a doubly fed wind turbine in one embodiment. Detailed Implementation

[0060] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0061] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0062] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0063] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0064] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0065] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0066] In one embodiment, a terminal is provided, the terminal being configured to: acquire operating parameter data of a doubly-fed induction generator (DFIG) wind turbine; generate a similar operating condition time period for the gearbox in the DFIG wind turbine based on the operating parameter data using a clustering algorithm; extract target vibration data from vibration monitoring data of the gearbox based on the similar operating condition time period; generate modal components and center frequencies corresponding to each modal component based on the target vibration data, and generate modal component energy based on the modal components and center frequencies corresponding to each modal component; generate a fused modal energy value based on the modal component energy, and generate a unit fault early warning result based on the fused modal energy value.

[0067] The terminal may be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices.

[0068] In one embodiment, such as Figure 1 As shown, a fault monitoring method for doubly-fed induction generator (DFIG) wind turbines based on frequency monitoring is provided. The method includes:

[0069] Step S100: Obtain the operating parameter data of the doubly fed wind turbine, and generate similar operating condition time periods of the gearbox in the doubly fed wind turbine based on the operating parameter data using a clustering algorithm;

[0070] Step S200: Extract target vibration data from the vibration monitoring data of the gearbox according to the similar working condition time period;

[0071] Step S300: Generate modal components and the center frequency corresponding to each modal component based on the target vibration data, and generate modal component energy based on the modal components and the center frequency corresponding to each modal component;

[0072] Step S400: Generate a fused mode energy value based on the modal component energy, and generate a unit fault early warning result based on the fused mode energy value.

[0073] In this embodiment, to provide basic data for subsequent gearbox analysis and achieve frequency-based operational status analysis under the same operating conditions, thereby improving the accuracy of subsequent fault monitoring, the different impacts of various operating conditions such as wind speed, engine speed, and power on gearbox vibration health are fully considered. This allows for the acquisition of operating parameter data for the doubly-fed induction generator (DFIG) wind turbine. Based on this operating parameter data, a clustering algorithm is used to generate similar operating condition time periods for the gearbox within the DFIG wind turbine. This enables accurate identification of time periods where the gearbox is under similar operating conditions, effectively avoiding interference from different operating conditions on vibration data characteristics and improving the reliability of subsequent feature extraction and anomaly detection. The method is targeted and specific. Then, based on the similar operating condition time period, target vibration data is extracted from the gearbox vibration monitoring data. By extracting vibration data only for the time period corresponding to similar operating conditions, interference from signal differences under different operating conditions on vibration characteristics is eliminated. Next, modal components and their corresponding center frequencies are generated based on the target vibration data. Modal component energy is then generated based on the modal components and their corresponding center frequencies, achieving the separation of each physical characteristic modal component with a clear center frequency. The energy of each mode is calculated, accurately extracting the gear meshing frequency and its harmonics, effectively avoiding frequency aliasing or information loss. Finally, a fused modal energy value is generated based on the modal component energy, and a unit fault early warning result is generated based on the fused modal energy value. By using the fused modal energy value as a comprehensive criterion for unit faults, the method achieves the most accurate early warning of gearbox fault risks, reduces missed reports, and improves the timeliness of unit fault monitoring and maintenance response.

[0074] In one embodiment, such as Figure 2 As shown, step S100: Obtain the operating parameter data of the doubly-fed induction generator (DFIG) wind turbine, and generate similar operating condition time periods for the gearbox in the DFIG wind turbine based on the operating parameter data using a clustering algorithm, including:

[0075] Step S110: Obtain the operating parameter data of the doubly fed wind turbine, and generate a multi-dimensional operating condition feature vector based on the operating parameter data;

[0076] Step S120: Normalize and assign weights to the multidimensional working condition feature vector, and generate a weighted feature vector;

[0077] Step S130: Divide the weighted feature vector into segments according to the preset segmentation interval, and select the target working condition segment based on the clustering algorithm and preset screening conditions;

[0078] Step S140: Generate a similar working condition time period based on the target working condition segment.

[0079] In this embodiment, in step S110, the operating parameter data of the doubly-fed wind turbine is acquired. This operating parameter data includes raw operating parameter data such as wind speed, main shaft speed, generator speed, active power, and pitch angle. Based on these raw operating parameters, the transmission ratio and main shaft acceleration are calculated to form a multi-dimensional operating condition feature vector. The transmission ratio reflects the actual gearbox speed change characteristics and transmission state, and is obtained by dividing the generator speed by the main shaft speed. The main shaft acceleration is obtained by calculating the derivative of the main shaft speed and is used to capture anomalies such as sudden changes in operating conditions and dynamic responses.

[0080] In step S120, the operating condition feature vector is normalized to have a mean of 0 and a variance of 1. The normalization method is z-score standardization or min-max normalization. Then, principal component analysis or correlation analysis is used, combined with historical vibration response data, to assign weights to each feature, resulting in a weighted feature vector.

[0081] Then, the weighted feature vector is segmented according to the preset segmentation interval, and the target working condition segment is selected based on the clustering algorithm and preset filtering conditions. The time period corresponding to the final working condition segment is extracted and a similar working condition time period is generated.

[0082] For example, the multidimensional working condition feature vector is represented as follows: ,in, These represent wind speed, main shaft speed, generator speed, active power, pitch angle, transmission ratio, and main shaft acceleration, respectively.

[0083] The weighted feature vector is represented as: In vectors to These represent the weights of wind speed, main shaft speed, generator speed, active power, pitch angle, transmission ratio, and main shaft acceleration, respectively. to The values ​​are 0.2, 0.2, 0.25, 0.15, 0.05, 0.1, and 0.05, respectively.

[0084] In this embodiment, not only are wind speed, spindle speed, generator speed, active power, and pitch angle collected, but the transmission ratio and spindle acceleration, which can capture more detailed differences in operating conditions, are also calculated. Through normalization and weight configuration, the clustering results are made to better fit the actual gearbox. The weights can also be adjusted to improve flexibility and adaptability to different gearboxes.

[0085] It should be noted that the examples of the multidimensional operating condition feature vector and weighted feature vector mentioned above are only for illustrating the technical solution of this application and are not intended to limit it. In practice, different numbers of parameters can be monitored based on the resources that the managers of the doubly fed wind turbine can allocate. This application does not limit this.

[0086] In another embodiment, step S120 exemplarily illustrates a method for assigning weights to each feature using correlation analysis combined with historical vibration response data:

[0087] First, collect historical operating condition characteristic data and historical vibration response data within a preset historical period. The historical period can be set to one month, six months, or longer; the larger the data volume, the better. Historical operating condition characteristic data includes wind speed, main shaft speed, generator speed, active power, and pitch angle. Historical vibration response data includes vibration energy, vibration RMS, and main meshing frequency component energy. When collecting data, it is required that the operating parameters and vibration response characteristics at the same moment correspond one-to-one to facilitate subsequent sequence calculations.

[0088] Then, invalid or outlier points in the operating parameters or vibration characteristics that are outside the reasonable physical range are removed to clean up anomalies. Interpolation or forward / backward filling can be used, or samples containing missing values ​​can be directly discarded to handle missing values. Ensure that the timestamps of all parameters are strictly aligned to guarantee the integrity of each data point. Then, normalize the processed historical operating characteristic data and historical vibration response data.

[0089] Next, for each operating condition parameter, a correlation analysis method is used to calculate its correlation coefficient with the target vibration response characteristics. The correlation coefficient is calculated using the Pearson correlation coefficient to obtain the correlation coefficient between each operating condition parameter and the target vibration response characteristics. Then, the absolute value of the correlation coefficient is taken and normalized to obtain the weight of each operating condition parameter. The formula for calculating the Pearson correlation coefficient is prior art and should be known to those skilled in the art; therefore, it will not be elaborated upon in this application. The formula for normalizing the absolute value of the correlation coefficient is as follows:

[0090] ,in, The weights of each operating condition parameter, Let be the correlation coefficient between the i-th operating condition parameter to be calculated and the vibration response characteristics. M is the total number of operating condition parameters. This represents the sum of the absolute values ​​of the correlation coefficients between all operating parameters and vibration response characteristics, where x is the index number of the operating parameter.

[0091] After obtaining the weights of each working condition parameter, the weights can be assigned to the normalized working condition feature vectors that are subsequently detected, so as to obtain a weighted feature vector.

[0092] Therefore, by setting weights for operating parameters, the different impacts of various operating characteristics such as wind speed, rotational speed, and power on gearbox vibration health are fully considered. This allows features with different influencing effects to play a greater role in subsequent operating condition similarity calculations and cluster analysis, making the clustering results more physically realistic and discriminative. It avoids irrelevant or weakly correlated features from affecting the truly important features, improving the accuracy of operating condition segmentation. Furthermore, through weight allocation, subsequent cluster comparison analysis will focus more on those features most sensitive to vibration, enabling earlier and more accurate detection of early gearbox anomalies and trend changes, increasing attention to key features. Moreover, by combining weight allocation with normalization, the influence of dimensions is eliminated, and the influence of large numerical features on distance or similarity is further eliminated through weight allocation.

[0093] In one embodiment, such as Figure 3 As shown, the preset segmented intervals include wind speed intervals and pitch angle intervals;

[0094] Step S130: Divide the weighted feature vector into segments according to preset segmentation intervals, and select target working condition segments based on clustering algorithms and preset screening conditions, including:

[0095] Step S131: Divide the weighted feature vector into segments according to the wind speed range and the pitch angle range, and perform fine-grained clustering on the data in each segment using a clustering algorithm to obtain multiple operating condition clusters;

[0096] Step S132: Select the target working condition segment from each of the working condition clusters according to the preset filtering conditions;

[0097] Step S133: Calculate the center vector of each target working condition segment, and select the target working condition segment from each target working condition segment based on the center vector.

[0098] In this embodiment, in step S131, both the wind speed range and the pitch angle range are preset. For example, all data is divided into multiple segments based on the wind speed and pitch angle ranges. The wind speed segments include, but are not limited to: 0-5 m / s, 5-10 m / s, 10-20 m / s; the pitch angle ranges include, but are not limited to: 0-10°, 10-20°, 20-90°. The data within each segment is considered a subset. Next, a clustering algorithm such as K-means, Gaussian Mixture Model, or density clustering (DBSCAN) is used. A weighted feature vector is input, the Euclidean distance between data points is calculated, and the optimal number of clusters is automatically selected to divide the data into several operating condition clusters.

[0099] In step S132, when screening target working condition segments, working condition segments that are continuous in time and whose duration meets a set threshold are selected from the samples in the working condition cluster. Non-continuous and isolated sample points are removed. That is, the preset screening condition is that the working condition segments are continuous in time and whose duration meets a set threshold. The set threshold is preset, such as 5 minutes.

[0100] In step S133, the Euclidean distance between the feature vector of the target working condition segment and the center vector of its cluster is calculated, and several consecutive sampling times with the smallest distance are selected as the representative working condition segments of the working condition cluster, i.e., the final working condition segments.

[0101] Finally, the final operating condition segment is set as the target operating condition segment. This is achieved by first setting preset segmentation intervals for wind speed range and pitch angle range, and then clustering the features to improve clustering accuracy.

[0102] Furthermore, setting wind speed ranges is primarily based on the fact that wind speed is a key indicator of wind energy input intensity, directly affecting the load and the dynamic response of the turbine, which in turn directly impacts the gearbox's condition. Setting pitch angle ranges, on the other hand, considers that pitch angle is a primary adjustment parameter in the wind turbine's control strategy, directly influencing blade stress and power generation.

[0103] When setting wind speed ranges, a fixed range method can be used, specifically dividing the data into zones based on the aforementioned 0-5m / s, 5-10m / s, and 10-20m / s. Alternatively, an adaptive quantile method can be employed, segmenting the wind speed data according to the 25%, 50%, and 75% quantile points of the data distribution. This method is suitable for wind farms with uneven wind speed distribution. For pitch angle ranges, segmentation can also be based on actual adjustment points, such as discrete angles in blade adjustment control, or by subdividing the ranges with frequent pitch operations. Of course, the segmentation thresholds for wind speed and pitch angle ranges can be dynamically adjusted based on the actual wind turbine operating distribution, or set by maintenance experience, to adapt to different wind farms and turbine models.

[0104] Based on the settings of wind speed range and pitch angle range, each data point is classified according to the range to which both wind speed and pitch angle belong, forming a two-dimensional segmented grid of wind speed-pitch. For example, wind speed 0-5m / s and pitch 0-10°, wind speed 5-10m / s and pitch 10-20°, etc., each combination is a segmented unit.

[0105] For example, if the wind speed range is 0-14 m / s and the pitch angle range is 0-16°, then the wind speed range can be set to four segments: 0-4, 4-8, 8-12, and 12-16 m / s, and the pitch angle range can be set to three segments: 0-6, 6-12, and 12-18°. This results in a total of 4 x 3, or 12 segment units. If the wind speed is 5.6 m / s and the pitch angle is 8.4°, then it should be classified into the segment unit [4-8, 6-12].

[0106] Furthermore, the steps of fine-grained clustering using a clustering algorithm are illustrated using a segment with wind speeds of 4-8 m / s and pitch angles of 6-12°. Within this segment, a subset of 2500 data points was extracted. K-means clustering was then performed on the weighted feature vectors of these 2500 data points, trying different numbers of clusters P. k The value is calculated for each cluster's silhouette coefficient, and the P-value corresponding to the highest silhouette coefficient is selected. k Value. For example, try P respectively. k When the clustering coefficients are 2, 3, 4, and 5, calculate the silhouette coefficient for each clustering, assuming P... k The profile coefficient is highest when P = 3, therefore P is selected. k =3, divided into 3 clusters. Then, the cluster label for each sample is obtained. For example, 1500 items belong to cluster 1, 600 items belong to cluster 2, and 400 items belong to cluster 3. It should be noted that in the K-means clustering method, the number of clusters is generally represented by K. To distinguish it from the modal component K in the variational mode decomposition method below, this embodiment uses the number of clusters P. k express.

[0107] Next, the target operating condition segment is selected from each operating condition cluster according to the preset screening criteria. Taking cluster 1 with 1500 records as an example, checking the timestamps reveals a continuous sample period from 08:30 to 09:10, with samples taken every 5 seconds, totaling 480 records. The remaining data is scattered, such as a few minutes of data, which does not meet the minimum continuous duration requirement of 10 minutes. Therefore, only the 40-minute period from 08:30 to 09:10 is retained as the valid operating condition segment, and other isolated points are removed, thus selecting the target operating condition segment.

[0108] Then, within the continuous segment from 08:30 to 09:10, the Euclidean distance between each data point and the central feature vector of cluster 1 is calculated. The data points are sorted from smallest to largest distance, and the 100 consecutive data points with the smallest distance are selected as the representative working condition segment of cluster 1. At the same time, representative working condition segments from other clusters are also selected. In other words, the selected target working condition segments are the representative working condition segments corresponding to each cluster. The representative working condition segments corresponding to a cluster are similar working states.

[0109] Furthermore, in step S140, the generated similar working condition time period is also a time period corresponding to each representative working condition segment of a cluster, so as to ensure that the working state corresponding to the similar working condition time period is the same.

[0110] Therefore, by acquiring the operating parameter data of the doubly-fed induction generator (DFIG) wind turbine, and then performing normalization, weight allocation, segmentation, clustering, and filtering, the time periods corresponding to similar operating states of the gearbox under specific conditions can be obtained. This provides basic data for subsequent gearbox analysis, enabling frequency-based operating state analysis under the same operating conditions, thereby improving the accuracy of subsequent fault monitoring. The specific conditions include those where wind speed, engine speed, load, and pitch angle are approximately constant.

[0111] In one embodiment, the vibration monitoring data includes first vibration data and second vibration data. The first vibration data is acquired based on a first vibration sensor, and the second vibration data is acquired based on a second vibration sensor. The first and second vibration sensors are disposed opposite to each other on both sides of the input shaft bearing housing of the gearbox. Exemplarily, the first and second vibration sensors are symmetrically mounted along the input shaft axis. The first and second vibration sensors are both 100mm from the centerline of the bearing housing. The mounting surfaces are selected at the outer circumference of the bearing housing to ensure close contact with the surface of the input shaft bearing housing, thereby improving the vibration transmission quality.

[0112] Furthermore, both the first and second vibration sensors employ triaxial accelerometers, using the gearbox input shaft axis as a reference. The X-axis of both sensors is parallel to this axis, ensuring efficient acquisition of axial vibrations. The Y and Z axes of both sensors are explicitly orthogonally mounted in a plane perpendicular to the axis, enabling precise capture of radial and tangential vibration characteristics generated at the gear meshing point. During installation, mounting fixtures are used to assist in installation, and the angles are calibrated using a level and laser rangefinder, with the angle deviation controlled within ±2° to ensure accuracy. Dedicated magnetic bases or bolts are used to secure the sensors during installation to prevent long-term aging and loosening of adhesives from affecting signal quality.

[0113] like Figure 4 As shown, step S200: Extracting target vibration data from the vibration monitoring data of the gearbox based on the similar operating condition time period, including:

[0114] Step S210: Obtain the first vibration data and the second vibration data of the gearbox;

[0115] Step S220: Generate differential vibration data based on the first vibration data and the second vibration data;

[0116] Step S230: Extract target vibration data from the differential vibration data according to the similar working condition time period.

[0117] In this embodiment, in the gearbox, the early gear wear signal is weak and the vibration energy is low, often masked by environmental vibration or noise. By placing the first vibration sensor and the second vibration sensor on both sides of the input shaft bearing seat of the gearbox and generating differential vibration data based on the first vibration data and the second vibration data, common-mode vibration noise interference from wind turbine tower or environmental vibration is effectively suppressed, thereby highlighting the weak vibration signal at the local gear meshing point, improving the sensor's sensitivity to early gear wear faults, effectively extracting the vibration characteristics of near-field vibration generated in the gear wear area, weakening far-field common-mode interference, and significantly improving the signal-to-noise ratio. This solves the problem in the prior art where single-sensor monitoring is prone to mixing in a large amount of background interference signals, resulting in spectrum chaos and increased difficulty in characteristic frequency identification.

[0118] In addition, the input shaft of the gearbox usually bears a large load, and the wear of gear meshing is concentrated in this area. The bearing housing directly supports the bearing, and the vibration signal transmission path is the shortest and the loss is the smallest. Therefore, the first vibration sensor and the second vibration sensor are set opposite to each other on both sides of the bearing housing of the input shaft of the gearbox, so that the measuring point position is close to the position where the fault is likely to occur under actual operating conditions. Thus, reliable data support is provided by setting the sensor installation position.

[0119] Differential vibration data is generated based on the first vibration data and the second vibration data using the following formula: ,in, For differential vibration data, This is the first vibration data. This is the second vibration data. Through differential processing, the far-field interference signals captured by sensors that are relatively close to each other are roughly the same. Subtracting the two signals can effectively cancel each other out, thus significantly highlighting the local vibration characteristics.

[0120] After acquiring the differential vibration data, target vibration data is extracted from the differential vibration data according to the similar working condition time period, and then subsequent fault monitoring and analysis are performed based on the target vibration data.

[0121] In one embodiment, after a fixed maintenance period, the common-mode suppression effect of the first and second vibration sensors needs to be tested, and the position adjustment is determined based on the test results. The fixed maintenance period is a pre-set timeframe, such as one month. By setting this fixed maintenance period, continuous testing is eliminated, saving computational resources.

[0122] First, regarding the first vibration data Second vibration data After removing the DC component and trend term, and eliminating acquisition noise, a Fast Fourier Transform is performed to obtain the complex amplitude. and . and They represent the frequencies respectively. The vibration intensity is then calculated. Next, the differential signal amplitude is calculated. ,pass Let's calculate it. Then, we'll calculate the common-mode rejection ratio:

[0123]

[0124] In the formula, Common-mode rejection ratio (CMRR) is used to represent the frequency response ratio (FRR). Below, the energy ratio of the differential signal to the common-mode signal, The higher the value, the more effectively common-mode interference can be suppressed.

[0125] By comparing the result with a preset common-mode suppression threshold, it is determined whether the result is less than the common-mode suppression threshold. If it is less than the threshold, it indicates that the common-mode suppression effect has decreased, and a self-test / maintenance reminder is triggered. The common-mode suppression threshold is preset by those skilled in the art, and no specific limitations or examples are provided.

[0126] Therefore, this embodiment ensures that fault warnings are not affected by environmental factors, structural transmission, installation changes, or other issues not related to the equipment itself by setting up evaluation and self-calibration steps, thus greatly reducing false alarms.

[0127] In one embodiment, such as Figure 5 As shown, step S300: generating modal components and the center frequency corresponding to each modal component based on the target vibration data, and generating modal component energy based on the modal components and the center frequency corresponding to each modal component, including:

[0128] Step S310: Decompose the target vibration data based on the variational mode decomposition method, and generate multiple modal components and the center frequency corresponding to each modal component;

[0129] Step S320: Select the target modal components based on each modal component, center frequency, and estimated main gear meshing frequency;

[0130] Step S330: Generate modal component energy based on the target modal component.

[0131] In this embodiment, in step S310, the target vibration data is... Variational mode decomposition (VMD) is performed, with the number of decomposition modes set to K. The target vibration data is decomposed into K modal components, and the center frequency f of each modal component is obtained. Variational mode decomposition is an adaptive signal decomposition technique that can adaptively decompose non-stationary complex signals into a set of modal components with different center frequencies. It is a mature technique in the prior art, therefore the formula will not be elaborated in this application.

[0132] In step S320, the meshing frequency of the main gear is estimated based on the gearbox structural parameters and operating conditions. and its several harmonics And from the modal components obtained by decomposition, the center frequency and and its several harmonics The closest modal component is designated as the target modal component. This is achieved by first monitoring the gear's rotational speed, and then estimating the main gear meshing frequency based on the product of the rotational speed and the number of teeth. The frequencies of multiple harmonics are generally... ,in, It is the harmonic order ( =1, 2, 3...).

[0133] Therefore, VMD can accurately extract the characteristic components of gear meshing frequency and its harmonic frequencies, effectively avoiding mode aliasing and solving the problems of low stability and poor accuracy caused by traditional EMD or wavelet analysis in existing technologies.

[0134] Next, in step S330, each selected target modal component is processed based on the following formula. Calculate modal component energy :

[0135] Where [t0,t1] is the analysis time window, and k represents the index of the target modal component. This represents the modal component energy of the k-th target modal component.

[0136] Furthermore, each of the similar operating condition time periods corresponds to one target vibration data point, and each target vibration data point corresponds to multiple modal component energies. .

[0137] In one embodiment, such as Figure 6 As shown, step S400: generating a fused mode energy value based on the modal component energy, and generating a unit fault early warning result based on the fused mode energy value, specifically includes:

[0138] Step S410: Based on the following formula, calculate the energy of multiple modal components corresponding to the target vibration data. Generate fusion modal energy values :

[0139] ,in, To fuse modal energy values, The number of target modal components, where one target vibration data point corresponds to one fused modal energy value. .

[0140] Step S420: Generate the energy growth rate and energy fluctuation anomaly degree based on the fused modal energy values ​​corresponding to each of the target vibration data;

[0141] Step S430: Generate unit fault early warning results based on the energy growth rate and energy fluctuation anomaly.

[0142] In this embodiment, in step S420, the energy growth rate represents the growth trend of the fusion value and is generated based on the following formula:

[0143] ;

[0144] In the formula, For energy growth rate, For the index of fused modal energy values, The total number of fused modal energy values, The mean of the index of the fused modal energy values. For the first One fusion mode energy value, This represents the mean of the fused modal energy values. The calculation formula is .calculate Its function is to decentralize and ensure the numerical stability of trend fitting.

[0145] The energy growth rate is a statistical characteristic of the overall operating status of the gearbox. It is not sensitive to individual anomalies or occasional signal fluctuations, and can accurately capture gradually abnormal risk signals from a large amount of data, reducing the false alarm rate. Many mechanical failures show a steady increase in key characteristics before they occur, but they do not necessarily jump to the alarm threshold immediately. Based on the energy growth rate, these small but continuous changes can be captured in advance, enabling early warning and gaining a maintenance window. It is especially suitable for gradual hidden dangers such as bearing or gear wear, increased clearance, and deteriorated lubrication.

[0146] The energy fluctuation anomaly degree is generated based on the following formula:

[0147] In the formula, The energy fluctuation anomaly degree corresponding to the energy value of the g-th fusion mode. for The standard deviation of the fused modal energy values. The energy fluctuation anomaly is used to measure the degree of outlier relative to the global mean and fluctuation range. It can quickly detect acute faults such as sudden impact, instability, and component breakage. When the trend is stable and increasing, a sudden and drastic deviation of individual points from the mean can also serve as supplementary evidence for secondary judgment, achieving highly sensitive detection of sudden events in local time periods.

[0148] Therefore, using the slope of the growth trend for anomaly screening can effectively monitor the continuous changes in equipment status, with the advantages of robustness and early warning. On the other hand, using the degree of energy fluctuation anomaly screening for local anomalies can keenly capture sudden and extreme state changes, and achieve precise location of single-point risks. By combining the two, the advantages of both trend and local health monitoring can be combined, providing comprehensive and multi-level anomaly diagnosis capabilities for gearboxes in doubly fed wind turbine units.

[0149] In one embodiment, such as Figure 7 As shown, step S430: Generate unit fault early warning results based on the energy growth rate and energy fluctuation anomaly degree, including:

[0150] Step S431: Determine whether the energy growth rate is greater than the health growth rate, or whether the energy fluctuation abnormality is greater than the health abnormality.

[0151] Step S432: If the determination is yes, then generate a unit fault warning;

[0152] Step S433: If the determination is negative, generate a normal unit indication.

[0153] In this embodiment, the unit fault early warning includes three scenarios: a first regular early warning, a second regular early warning, and a highest confidence early warning. In step S432, if the determination is yes, several scenarios are included: First, the energy growth rate is greater than the healthy growth rate, but the energy fluctuation anomaly is less than or equal to the healthy anomaly; in this case, a first regular early warning is generated. Second, the energy growth rate is less than or equal to the healthy growth rate, and the energy fluctuation anomaly is greater than the healthy anomaly; in this case, a second regular early warning is generated. Third, the energy growth rate is greater than the healthy growth rate, and the energy fluctuation anomaly is greater than the healthy anomaly; in this case, a highest confidence early warning is generated.

[0154] The first routine warning indicates that the health of the gearbox has been deteriorating over multiple consecutive periods, usually indicating progressive failures such as wear and failure. At this time, check the energy change curves of each major modal component, such as the main frequency and harmonics, to eliminate the overall trend deviation caused by the abnormality of individual components. Also check whether there are systematic changes in operating conditions such as load, temperature, and wind speed during this period to eliminate false trends caused by changes in operating conditions.

[0155] The second routine warning indicates that the gearbox's condition has suddenly or extremely abnormaled within a specific time period, which could be due to a sudden malfunction, impact, misoperation, or extreme operating conditions. At this time, check the original differential vibration waveform and time / frequency domain data for strong impacts, spikes, or abnormal noise. Additionally, detailed spectrum analysis can be performed to observe for new frequency components, broadband noise, or enhanced characteristic frequencies, determining whether it is a genuine equipment fault or occasional interference.

[0156] At the highest confidence warning level, if both trend anomaly and single-point fluctuation anomaly are met, a comprehensive check should be performed.

[0157] In step S433, if the determination is negative, it means that the energy growth rate is less than or equal to the healthy growth rate and the energy fluctuation abnormality is less than or equal to the healthy abnormality. At this time, a normal unit indication is generated.

[0158] In this embodiment, the health growth rate and the health anomaly degree are pre-set based on gearbox data corresponding to historical health periods. The historical health period refers to a historical time interval during which the equipment has been confirmed to be without anomalies and in good operating condition. Data during this period can be considered a health baseline for subsequent statistical comparison of trends and fluctuations. The data obtained from the historical health period... When performing a sequence, calculate as described above. The method calculates the energy growth rate corresponding to the historical health period, and calculates the mean based on each energy growth rate. and standard deviation Ultimately passed The health growth rate is calculated. The health abnormality degree is generally set to 2.

[0159] In one embodiment, such as Figure 8 As shown, a fault monitoring system for doubly-fed induction generator (DFIG) wind turbines based on frequency monitoring is also provided. The system includes:

[0160] The similar operating condition generation module is used to obtain the operating parameter data of the doubly fed wind turbine and generate similar operating condition time periods of the gearbox in the doubly fed wind turbine based on the operating parameter data using a clustering algorithm.

[0161] The vibration data filtering module is used to extract target vibration data from the vibration monitoring data of the gearbox based on the similar working condition time period.

[0162] The modal energy generation module is used to generate modal components and the center frequency corresponding to each modal component based on the target vibration data, and to generate modal component energy based on the modal components and the center frequency corresponding to each modal component.

[0163] The fault warning generation module is used to generate a fused mode energy value based on the modal component energy, and to generate a unit fault warning result based on the fused mode energy value.

[0164] In one embodiment, the similar operating condition generation module is further configured to: acquire operating parameter data of the doubly-fed wind turbine, generate a multi-dimensional operating condition feature vector based on the operating parameter data; normalize and assign weights to the multi-dimensional operating condition feature vector, and generate a weighted feature vector; segment the weighted feature vector according to a preset segmentation interval, and select a target operating condition segment based on a clustering algorithm and preset screening conditions; and generate a similar operating condition time period based on the target operating condition segment.

[0165] In one embodiment, the similar operating condition generation module is further configured to: segment the weighted feature vector according to a preset segmentation interval, and select target operating condition segments based on a clustering algorithm and preset screening conditions, including: segmenting the weighted feature vector according to the wind speed interval and the pitch angle interval, and performing fine-grained clustering on the data in each segment using a clustering algorithm to obtain multiple operating condition clusters; selecting target operating condition segments from each of the operating condition clusters according to preset screening conditions; calculating the center vector of each of the target operating condition segments, and selecting target operating condition segments from each of the target operating condition segments based on the center vector.

[0166] In one embodiment, the vibration monitoring data includes first vibration data and second vibration data, wherein the first vibration data is acquired based on a first vibration sensor and the second vibration data is acquired based on a second vibration sensor; the first vibration sensor and the second vibration sensor are disposed opposite to each other on both sides of the input shaft bearing housing of the gearbox; the vibration data filtering module is further configured to: acquire the first vibration data and the second vibration data of the gearbox; generate differential vibration data based on the first vibration data and the second vibration data; and extract target vibration data from the differential vibration data based on the similar operating condition time period.

[0167] In one embodiment, the modal energy generation module is further configured to: decompose the target vibration data based on the variational modal decomposition method and generate multiple modal components and the center frequency corresponding to each modal component; select target modal components based on each modal component, the center frequency and the estimated main gear meshing frequency; and generate modal component energy based on the target modal components.

[0168] In one embodiment, the fault warning generation module is further configured to: generate a fused modal energy value based on the modal component energy; generate an energy growth rate and an energy fluctuation anomaly degree based on the fused modal energy value corresponding to each of the target vibration data; and generate a unit fault warning result based on the energy growth rate and the energy fluctuation anomaly degree.

[0169] In one embodiment, the unit fault warning result includes a unit fault warning indication and a unit non-fault indication; the fault warning generation module is further configured to: determine whether the energy growth rate is greater than the healthy growth rate and whether the energy fluctuation abnormality is greater than the healthy abnormality; if the determination is yes, then generate a unit fault warning; if the determination is no, then generate a unit normal indication.

[0170] It should be noted that the information interaction and execution process between the above modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0171] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0172] It should be noted that the information interaction and execution process between the above modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0173] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0174] This application also provides a network device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the steps in any of the above method embodiments.

[0175] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0176] This application provides a computer program product that, when run on a mobile terminal, enables the mobile terminal to implement the steps described in the above-described method embodiments.

[0177] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-described embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographic device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0178] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0179] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0180] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0181] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0182] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

[0183] One embodiment of this application also provides a computer device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the steps in any of the above-described methods.

[0184] The computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above description is an example of a computer device and does not constitute a limitation on the computer device. It may include more or fewer components than described above, or a combination of certain components, or different components, such as input / output devices, network access devices, etc.

[0185] The processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0186] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard drive or RAM. In other embodiments, the memory may be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory may include both internal and external storage units of the computer device. The memory is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory can also be used to temporarily store data that has been output or will be output.

[0187] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0188] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A fault monitoring method for doubly-fed induction generator (DFIG) wind turbines based on frequency monitoring, characterized in that, The method includes: Obtain the operating parameter data of the doubly fed wind turbine, and generate similar operating condition time periods of the gearbox in the doubly fed wind turbine based on the operating parameter data using a clustering algorithm; Target vibration data is extracted from the vibration monitoring data of the gearbox based on the similar operating condition time period; Modal components and their corresponding center frequencies are generated based on the target vibration data, and modal component energy is generated based on the modal components and their corresponding center frequencies. A fused mode energy value is generated based on the modal component energy, and a unit fault early warning result is generated based on the fused mode energy value.

2. The fault monitoring method for doubly-fed wind turbines based on frequency monitoring according to claim 1, characterized in that, Obtain the operating parameter data of the doubly-fed induction generator (DFIG) wind turbine, and generate similar operating condition time periods for the gearbox in the DFIG wind turbine based on the operating parameter data using a clustering algorithm, including: Obtain the operating parameter data of the doubly fed wind turbine, and generate a multi-dimensional operating condition feature vector based on the operating parameter data; The multidimensional working condition feature vector is normalized and weighted, and a weighted feature vector is generated. The weighted feature vector is segmented according to a preset segmentation interval, and the target working condition segment is selected based on a clustering algorithm and preset screening conditions. A similar operating condition time period is generated based on the target operating condition segment.

3. The fault monitoring method for doubly-fed wind turbines based on frequency monitoring according to claim 2, characterized in that, The preset segmented intervals include wind speed intervals and pitch angle intervals; The weighted feature vector is segmented according to a preset segmentation interval, and the target working condition segment is selected based on a clustering algorithm and preset screening conditions, including: The weighted feature vector is segmented according to the wind speed range and the pitch angle range, and the data in each segment is clustered in a fine-grained manner using a clustering algorithm to obtain multiple operating condition clusters. The target working condition segment is selected from each working condition cluster according to the preset filtering conditions; Calculate the center vector of each target working condition segment, and select the target working condition segment from each target working condition segment based on the center vector.

4. The fault monitoring method for doubly-fed wind turbines based on frequency monitoring according to claim 1, characterized in that, The vibration monitoring data includes first vibration data and second vibration data. The first vibration data is acquired based on a first vibration sensor, and the second vibration data is acquired based on a second vibration sensor. The first vibration sensor and the second vibration sensor are arranged opposite each other on both sides of the input shaft bearing housing of the gearbox. Target vibration data is extracted from the vibration monitoring data of the gearbox based on the similar operating condition time period, including: Obtain the first vibration data and the second vibration data of the gearbox; Differential vibration data is generated based on the first vibration data and the second vibration data; Target vibration data is extracted from the differential vibration data based on the similar operating condition time period.

5. The fault monitoring method for doubly-fed wind turbines based on frequency monitoring according to claim 1, characterized in that, Based on the target vibration data, modal components and their corresponding center frequencies are generated, and modal component energy is generated based on the modal components and their corresponding center frequencies, including: The target vibration data is decomposed based on the variational mode decomposition method, and multiple modal components and the center frequency corresponding to each modal component are generated. The target modal components are selected based on each modal component, center frequency, and estimated main gear meshing frequency; Modal component energy is generated based on the target modal component.

6. The fault monitoring method for doubly-fed wind turbines based on frequency monitoring according to claim 1, characterized in that, A fused mode energy value is generated based on the modal component energy, and a unit fault early warning result is generated based on the fused mode energy value, including: Generate a fused mode energy value based on the modal component energy; The energy growth rate and energy fluctuation anomaly are generated based on the fused modal energy values ​​corresponding to the vibration data of each target. The unit fault early warning result is generated based on the energy growth rate and energy fluctuation anomaly.

7. The fault monitoring method for doubly-fed wind turbines based on frequency monitoring according to claim 6, characterized in that, The unit fault warning results include unit fault warnings and unit normal indications; Based on the energy growth rate and energy fluctuation anomaly, a unit fault early warning result is generated, including: Determine whether the energy growth rate is greater than the health growth rate and whether the energy fluctuation abnormality is greater than the health abnormality. If the determination is yes, then a unit fault warning is generated; If the determination is negative, a normal unit indication is generated.

8. A fault monitoring system for doubly-fed induction generator (DFIG) wind turbines based on frequency monitoring, characterized in that, The system includes: The similar operating condition generation module is used to obtain the operating parameter data of the doubly fed wind turbine and generate similar operating condition time periods of the gearbox in the doubly fed wind turbine based on the operating parameter data using a clustering algorithm. The vibration data filtering module is used to extract target vibration data from the vibration monitoring data of the gearbox based on the similar working condition time period. The modal energy generation module is used to generate modal components and the center frequency corresponding to each modal component based on the target vibration data, and to generate modal component energy based on the modal components and the center frequency corresponding to each modal component. The fault warning generation module is used to generate a fused mode energy value based on the modal component energy, and to generate a unit fault warning result based on the fused mode energy value.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images