Wind turbine generator abnormal operation identification method and device, medium and equipment

Abstract

The invention relates to the technical field of electric power, in particular to a wind turbine generator abnormal operation identification method and device, a medium and equipment. According to the method, wind turbine generators are grouped according to generator set models to construct identification groups, time domain and frequency domain features of working condition parameters in the identification groups are independently extracted for different abnormal identification targets, and a comprehensive feature matrix is constructed. And after principal component analysis dimensionality reduction, dividing a target operation condition cluster by using a clustering algorithm, and finally detecting an abnormal unit in the same target operation condition cluster by identifying deviation of parameters and group statistical distribution. According to the method, the dynamic working condition reference is established through unsupervised clustering, the problem of poor adaptability of a fixed threshold value is overcome, and the sensitivity to early compound faults is remarkably improved by utilizing a multi-parameter feature fusion and group comparison mechanism, so that accurate anomaly recognition can be realized without relying on annotated data, the false alarm rate and the operation and maintenance cost are effectively reduced, and the accuracy of anomaly recognition is improved. And reliable technical support is provided for predictive maintenance of the wind turbine generator.

Description

Technical Field

[0001] This application relates to the field of power technology, and in particular to a method, device, medium and equipment for identifying abnormal operation of wind turbine units. Background Technology

[0002] In related technologies, wind turbine anomaly monitoring mainly relies on fixed threshold alarms from SCADA (Supervisory Control and Data Acquisition) systems and traditional statistical analysis. The former is prone to false alarms and missed alarms due to its inability to adapt to dynamic operating conditions, while the latter lacks sufficient analytical capabilities for multi-parameter coupling and nonlinear relationships. Although some studies have employed supervised machine learning methods, these rely on large amounts of labeled data, while the scarcity of actual fault samples limits the model's generalization ability. These technologies generally suffer from poor adaptability, weak multi-parameter analysis capabilities, and strong dependence on labeled data, making it difficult to achieve accurate early anomaly identification. Summary of the Invention

[0003] In view of this, this application provides a method, device, medium and equipment for identifying abnormal operation of wind turbine units, which realizes accurate identification of early abnormalities of wind turbine units, effectively overcomes the problems of high false alarm rate of fixed threshold method and dependence on labeled data of supervised learning, significantly improves fault detection sensitivity and reduces operation and maintenance costs.

[0004] According to one aspect of this application, a method for identifying abnormal operation of a wind turbine is provided, comprising: Acquire the runtime sequence data of multiple wind turbine units within the monitoring period, wherein the runtime sequence data of the wind turbine units includes the time sequence data of the wind turbine units under different operating parameters; The wind turbines are grouped according to their model number to obtain at least one identification group; Based on the preset anomaly identification target corresponding to the identification group, the identification parameters and operating condition parameters of the identification group are determined from the operating parameters; Feature extraction is performed on the time-series data of the target wind turbines in the identification group under the operating parameters to construct a comprehensive feature matrix of the identification group. Clustering is performed on the comprehensive feature matrix to obtain the target operating condition cluster; Within the target operating condition cluster, wind turbine units with abnormal operation are identified based on the identification parameters.

[0005] According to another aspect of this application, a wind turbine abnormal operation identification device is provided, comprising: The acquisition module is used to acquire the runtime sequence data of multiple wind turbines within the monitoring period. The runtime sequence data of the wind turbines includes the time sequence data of the wind turbines under different operating parameters. The processing module is configured to: group the wind turbines according to their turbine models to obtain at least one identification group; determine the identification parameters and operating condition parameters of the identification group from the operating parameters based on the preset anomaly identification targets corresponding to the identification group; extract features from the time-series data of the target wind turbines within the identification group under the operating condition parameters to construct a comprehensive feature matrix of the identification group; and perform clustering processing on the comprehensive feature matrix to obtain a target operating condition cluster. The identification module is used to identify wind turbine units with abnormal operation within the target operating condition cluster based on the identification parameters.

[0006] According to another aspect of this application, a readable storage medium is provided, on which a program or instructions are stored, which, when executed by a processor, implement the steps of the above-described wind turbine abnormal operation identification method.

[0007] According to another aspect of this application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the steps of the above-described wind turbine abnormal operation identification method.

[0008] By employing the above technical solutions, this application provides a method, apparatus, medium, and equipment for identifying abnormal operation of wind turbine units. It constructs identification groups by grouping wind turbine units according to their model, and independently extracts the time-domain and frequency-domain features of operating condition parameters within each identification group for different anomaly identification targets, constructing a comprehensive feature matrix. After dimensionality reduction using principal component analysis, a clustering algorithm is used to divide the target operating condition clusters. Finally, within the same target operating condition cluster, abnormal units are detected by the deviation of the identification parameters from the group's statistical distribution. This application establishes a dynamic operating condition benchmark through unsupervised clustering, overcoming the poor adaptability of fixed thresholds. Furthermore, by utilizing multi-parameter feature fusion and group comparison mechanisms, it significantly improves the sensitivity to early complex faults, thus achieving accurate anomaly identification without relying on labeled data. This effectively reduces false alarm rates and maintenance costs, providing reliable technical support for predictive maintenance of wind turbine units.

[0009] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0010] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating the wind turbine abnormal operation identification method provided in an embodiment of this application is shown. Figure 2 A flowchart illustrating a wind turbine abnormal operation identification method according to another embodiment of this application is shown; Figure 3 A structural block diagram of the wind turbine abnormal operation identification device provided in an embodiment of this application is shown. Detailed Implementation

[0011] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.

[0012] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0013] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “attached” to another element, it can be directly connected or attached to the other element, or there may be intermediate elements. Furthermore, “connected” or “attached” as used herein can include wireless connections or wireless interconnections. The term “and / or” as used herein includes all or any unit and all combinations of one or more associated listed items.

[0014] Exemplary embodiments according to this application will now be described in more detail with reference to the accompanying drawings. However, these exemplary embodiments may be implemented in many different forms and should not be construed as being limited to the embodiments set forth herein. It should be understood that these embodiments are provided so that the disclosure of this application is thorough and complete, and that the concept of these exemplary embodiments is fully conveyed to those skilled in the art.

[0015] This application provides a method for identifying abnormal operation of wind turbine units, such as... Figure 1 As shown, the method includes: Step 101: Obtain the operating sequence data of multiple wind turbine units within the monitoring period.

[0016] The wind turbine's operating sequence data includes time-series data of the wind turbine under different operating parameters.

[0017] Step 102: Group the wind turbines according to their model number to obtain at least one identification group.

[0018] Step 103: Based on the preset anomaly identification target corresponding to the identification group, determine the identification parameters and operating condition parameters of the identification group from the operating parameters.

[0019] Step 104: Extract features from the time-series data of the target wind turbines within the identification group under operating parameters, and construct a comprehensive feature matrix for the identification group.

[0020] Step 105: Perform clustering on the comprehensive feature matrix to obtain the target operating condition cluster.

[0021] Step 106: Within the target operating condition cluster, identify wind turbine units with abnormal operation based on the identification parameters.

[0022] The wind turbine abnormal operation identification method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the wind turbine abnormal operation identification method, but is not limited to the above forms.

[0023] This embodiment constructs identification groups by grouping wind turbine units according to their model. For different anomaly identification targets, time-domain and frequency-domain features of operating condition parameters within each identification group are independently extracted to construct a comprehensive feature matrix. After dimensionality reduction using principal component analysis, clustering algorithms are used to divide the target operating condition clusters. Finally, within the same target operating condition cluster, anomaly units are detected by the deviation between the identification parameters and the group's statistical distribution. This embodiment establishes a dynamic operating condition benchmark through unsupervised clustering, overcoming the poor adaptability of fixed thresholds. Furthermore, by utilizing multi-parameter feature fusion and group comparison mechanisms, it significantly improves the sensitivity to early-stage complex faults, thus achieving accurate anomaly identification without relying on labeled data. This effectively reduces false alarm rates and maintenance costs, providing reliable technical support for predictive maintenance of wind turbine units.

[0024] Another embodiment of this application provides a method for identifying abnormal operation of wind turbine units, such as... Figure 2 As shown, the method includes: Step 201: Obtain the operating sequence data of multiple wind turbine units within the monitoring period.

[0025] The wind turbine's operating sequence data includes time-series data of the wind turbine under different operating parameters.

[0026] In this step, the operating sequence data of multiple wind turbine units within the monitoring period are obtained from the SCADA system.

[0027] Specifically, a wind farm or a regional SCADA system typically monitors multiple, or even different, wind turbine models. SCADA systems are designed to collect operational data from different wind turbines and store it in a central database. Therefore, this step retrieves operational data from different wind turbine models from the SCADA system. Furthermore, this step acquires data for a predetermined monitoring period, such as 24 hours or a week, obtaining continuous or equally spaced sampling of operational data from all wind turbines within that monitoring period, thus obtaining the operational sequence data for each wind turbine. The operational sequence data for each wind turbine includes time-series data for different operating parameters within that monitoring period. Operating parameters include, but are not limited to: ambient temperature, generator speed, turbine output power, wind speed, blade angle to the wind, gearbox temperature, bearing temperature, gearbox bearing vibration value, gearbox oil temperature, etc. Gearbox bearing vibration value refers to the effective value of the frequency band amplitude collected by the SCADA system, such as the effective value of velocity.

[0028] Furthermore, after obtaining the wind turbine's operating sequence data, the operating sequence data is preprocessed by denoising, missing value imputation, and standardization in order to update the wind turbine's operating sequence data.

[0029] Step 202: Group the wind turbines according to their model number to obtain at least one identification group.

[0030] In this step, all wind turbines are divided into different identification groups according to their turbine model. Each identification group consists of target wind turbines of the same model. It is worth noting that all subsequent steps are performed within these identification groups to ensure comparability.

[0031] Step 203: Based on the preset anomaly identification target corresponding to the identification group, determine the identification parameters and operating condition parameters of the identification group from the operating parameters.

[0032] In this step, after grouping by unit model, a specific anomaly identification target is pre-set for each identification group based on the anomaly identification requirements in the actual application scenario. Therefore, among all operating parameters, the identification parameters for the identification group are determined according to the anomaly identification target, and the operating parameters associated with the anomaly identification target are selected as the operating condition parameters for the identification group.

[0033] For specific examples, for an anomaly identification group targeting abnormal bearing temperature, the identification parameter is the bearing temperature, and the operating parameters are those strongly correlated with the bearing temperature, such as ambient temperature, generator speed, and fan output power. For anomaly identification group targeting abnormal gearbox bearing vibration, the identification parameter is the gearbox bearing vibration value, and the operating parameters are those related to the gearbox bearing vibration, such as wind speed, generator speed, and fan output power.

[0034] Step 204: The time-series data of the target wind turbine under operating parameters is segmented using a sliding time window to obtain the first time-series data segment of the target wind turbine's operating parameters within the time window; features of the first time-series data segment of different operating parameters of the target wind turbine within the time window are extracted and spliced ​​to obtain the feature vector of the target wind turbine within the time window; based on the feature vectors of the target wind turbine in different time windows, the turbine feature matrix of the target wind turbine is constructed; based on the turbine feature matrices of different target wind turbines within the identification group, the comprehensive feature matrix of the identification group is constructed.

[0035] In this step, features are extracted from the time-series data of the target wind turbines within the identification group under operating parameters to construct a comprehensive feature matrix for the identification group. It is worth noting that the feature extraction focuses solely on operating parameters to enable accurate clustering of operating conditions in subsequent steps. Identification parameters, however, are not included in clustering but are only used for comparisons within the target operating condition cluster.

[0036] Specifically, for any operating parameter of any target wind turbine within the identification group, a sliding time window is used to segment the time series data of the target wind turbine under that operating parameter, obtaining the first time series data segment of the target wind turbine's operating parameter within each time window. For example, each time window is 4 hours, with a step size of 1 hour.

[0037] Then, based on the physical characteristics of the operating condition parameters and the mechanism by which they affect the identification parameters, the type of features to be extracted for the operating condition parameters is determined. These feature types include time-domain features and frequency-domain features.

[0038] Specifically, if the operating parameters are steady-state parameters (i.e., the first preset type), and the operating parameters themselves do not have significant dynamic frequency components, then only the time-domain features of the operating parameters are extracted. For example, steady-state parameters include ambient temperature, generator speed, wind turbine output power, blade angle, etc. The time-domain features include, but are not limited to, the mean, variance, and extreme values ​​of the target wind turbine's operating parameters within the first time-series data segment of the time window. Here, the mean is used to represent the average operating level within the time window, the standard deviation is used to represent the operating stability within the time window, and the maximum or minimum value is used to represent extreme operating conditions within the time window.

[0039] If the operating condition parameter is a dynamic parameter (second preset type), and its different frequency components directly lead to different response patterns in the identified parameter, then in addition to extracting the time-domain features of the operating condition parameter, its frequency-domain features are also extracted. For example, a dynamic parameter might be wind speed. Specifically, for operating condition parameters that are dynamic, the first time-series data segment of the target wind turbine under that operating condition parameter within the time window is processed using FFT (Fast Fourier Transform) to extract the frequency-domain features of the target wind turbine under that operating condition parameter within the time window from the FFT processing result. The FFT processing result is a spectrum, from which frequency-domain feature values ​​are extracted. Frequency-domain features include, but are not limited to, centroid frequency, mean square frequency, frequency standard deviation, spectral peak value, band energy, and spectral entropy. Here, the centroid frequency is the centroid location of the spectrum, used to represent the dominant frequency region where energy is concentrated; the mean square frequency is the second moment of the frequency distribution in the spectrum; the frequency standard deviation is used to represent the dispersion of the frequency distribution; the spectral peak is the N largest amplitude values ​​in the spectrum and their corresponding frequencies; the band energy is obtained by dividing the spectrum into several key bands, such as the low-frequency band, the meshing frequency band, and the high-frequency band, and taking the sum of the squares of the amplitudes in each band as the signal energy in the band, which is used to detect specific faults; the spectral entropy is used to represent the complexity of the spectrum. A high entropy value indicates complex frequency components, which may indicate a deterioration in the state.

[0040] Next, the feature values ​​of all operating parameters of the target wind turbine within the same time window are concatenated into a long vector in a fixed order to obtain the feature vector of the target wind turbine within that time window. For example, the feature vector of the target wind turbine A in the [T1, T2] time window is [time domain feature 1, time domain feature 2, frequency domain feature 3, ...].

[0041] Furthermore, the feature vectors of the same target wind turbine in different time windows are stacked vertically row by row in chronological order to obtain the turbine's feature matrix. It's worth noting that in the turbine feature matrix, each row represents the operating state of the target wind turbine in different time windows, and each column represents different features of extracted operating parameters. Each row in the turbine feature matrix is ​​used as an instance of the operating state of the target wind turbine within its corresponding time window. This transforms the operating state of the target wind turbine in each time window into a structured feature matrix, providing a data foundation for subsequent steps.

[0042] For example, the characteristic matrix of target wind turbine A is represented as follows: .

[0043] Therefore, the feature matrices of all target wind turbines within the identification group are stacked vertically row by row to form the comprehensive feature matrix of the identification group. It can be understood that the comprehensive feature matrix of the identification group can represent the operating status of different target wind turbines within the identification group in different time windows, and is composed of instances of the operating status of different target wind turbines in different time windows.

[0044] Step 205: Dimensionality reduction is performed on the comprehensive feature matrix to obtain a low-dimensional feature matrix; the principal component vectors of different rows in the low-dimensional feature matrix are used as the operating state instances of different target wind turbines in different time windows; the operating state instances are clustered to obtain the target operating condition cluster.

[0045] In this step, principal component analysis (PCA) is used to compress the high-dimensional comprehensive feature matrix into a low-dimensional space while preserving as much of the key information of the original data as possible.

[0046] Specifically, after dimensionality reduction of the comprehensive feature matrix using principal component analysis (PCA), a low-dimensional feature matrix is ​​obtained. This low-dimensional feature matrix has the same number of rows as the comprehensive feature matrix, and the rows of the low-dimensional feature matrix correspond to those of the comprehensive feature matrix, representing the operating states of different target wind turbines within different time windows. However, the number of columns in the low-dimensional feature matrix is ​​the same as the number of principal components set during PCA. The columns of the low-dimensional feature matrix are no longer the original features with explicit physical meaning, but rather different values ​​of the same type of principal components obtained through PCA. Each principal component value in the low-dimensional feature matrix is ​​obtained by PCA from the eigenvector corresponding to its row in the comprehensive feature matrix. It should be noted that the principal component vector composed of the principal component values ​​of each row in the low-dimensional feature matrix is ​​used as an instance of the operating state of the target wind turbine within the time window to which that row belongs, so as to update the operating state instance using principal components.

[0047] For example, to facilitate clustering in subsequent steps, the number of principal components can be set to 2. That is, the low-dimensional feature matrix includes two columns: the first column represents the value of principal component 1 for different target wind turbines in different time windows, and the second column represents the value of principal component 2 for different target wind turbines in different time windows.

[0048] For example, a low-dimensional feature matrix can be represented as: , in, and These represent different values ​​of principal component 1. , These represent different values ​​for principal component 2.

[0049] In this embodiment, the reduced-dimensional feature matrix eliminates redundancy and reduces noise, enabling subsequent clustering algorithms to group running state instances more efficiently and accurately.

[0050] Furthermore, the K-means clustering algorithm is used to group all running state instances in the low-dimensional feature matrix, so as to classify running state instances with similar running conditions in the low-dimensional feature matrix into the same target running condition cluster.

[0051] Specifically, the K-means algorithm randomly initializes a preset number of centroids, and then iterates to assign each running state instance to the cluster containing the nearest centroid, and recalculates the centroid position until it stabilizes, thus obtaining a preset number of running state clusters (i.e., clustering results).

[0052] It is worth mentioning that this step first uses the elbow rule to calculate the sum of squared clustering errors (i.e., clustering index) of the operating condition clusters when the number of center points is different preset numbers, and then plots the curves. The preset number value corresponding to the inflection point of the curve is selected as the optimal number of operating condition clusters, and the optimal number of operating condition clusters is taken as the target number.

[0053] Then, using the target number, the K-means algorithm is employed to cluster all operating state instances in the low-dimensional feature matrix to obtain target operating condition clusters. At this point, each operating state instance in the low-dimensional feature matrix is ​​assigned a cluster label to its respective target operating condition cluster. For example, cluster labels include operating conditions such as "high-power stable operation" and "low-wind-speed turbulent operation".

[0054] It is worth mentioning that the target operating condition cluster includes all operating state instances from different target wind turbine units and different time windows, but under extremely similar operating conditions, ensuring the reliability of subsequent anomaly identification.

[0055] Step 206: Based on the second time-series data segment of the identification parameters corresponding to the operating state instances within the target operating condition cluster, determine the distribution parameters of the identification parameters within the target operating condition cluster; based on the distribution parameters of the identification parameters within the target operating condition cluster, determine the abnormal threshold of the identification parameters within the target operating condition cluster; compare the second time-series data segment of the identification parameters corresponding to the operating state instances within the target operating condition cluster with the distribution parameters of the identification parameters within the target operating condition cluster to determine the abnormal operating state instances within the target operating condition cluster; based on the abnormal operating state instances, determine the wind turbine units with abnormal operation.

[0056] Among them, the second time-series data segment of the identification parameters corresponding to the operating status instance is the time-series data of the identification parameters of the target wind turbine to which the operating status instance belongs within the time window to which the operating status instance belongs.

[0057] In this step, for any operating state instance within any target operating condition cluster, the second time-series data segment of the identification parameters corresponding to that operating state instance is obtained. The second time-series data segment of the identification parameters corresponding to the operating state instance is the time-series data of the identification parameters of the target wind turbine to which that operating state instance belongs within the time window of that operating state instance. Then, the statistical distribution of the second time-series data segments of the identification parameters corresponding to all operating state instances within the target operating condition cluster is calculated to obtain the distribution parameters of the identification parameters within the target operating condition cluster, such as the mean and standard deviation. Based on these distribution parameters, an anomaly threshold is pre-set, for example, the anomaly threshold is the mean plus twice the standard deviation.

[0058] Then, check each operating state instance within the target operating condition cluster: if there is data exceeding the abnormal threshold in the second time-series data segment of the identification parameters corresponding to a certain operating state instance, it indicates that the identification parameter behavior of the target wind turbine to which the operating state instance belongs has deviated under this specific operating state that is highly similar to the operating conditions of other operating state instances, and then mark the operating state instance as an abnormal operating state instance.

[0059] To eliminate transient interference, this step checks whether there are multiple consecutive operating state instances of the same target wind turbine that are all marked as abnormal operating state instances in chronological order, and whether these abnormal operating state instances belong to the same target operating condition cluster. If such a situation exists, the target wind turbine is determined to be an abnormally operating wind turbine.

[0060] This embodiment compares the same parameters under the same unit model and operating conditions, minimizing the impact of individual differences and fluctuations in operating conditions of wind turbine units. This enables it to keenly detect subtle but persistent deviations in group behavior caused by potential faults, achieving true early warning.

[0061] Step 207: Based on the wind turbines operating abnormally under different anomaly identification targets, determine the faulty wind turbine and the cause of the faulty wind turbine, and visualize and alarm the anomaly and fault identification results.

[0062] In this step, after anomaly identification is completed, the final anomaly and fault identification results are generated based on the wind turbines determined to be operating abnormally under different anomaly identification targets. These results are then visualized and trigger alarms to assist in fault handling.

[0063] Specifically, a fault diagnosis rule base is pre-configured, defining the correlation between typical equipment fault modes and one or more anomaly identification targets. For example, the correlation could be: if the same wind turbine is simultaneously identified as abnormal by both the bearing temperature anomaly identification target and the gearbox bearing vibration anomaly identification target, then the wind turbine is diagnosed as having an early bearing wear fault, with a high confidence level. If the wind turbine is only identified as abnormal by the bearing temperature anomaly identification target, while the gearbox bearing vibration anomaly identification target shows normal operation, then the temperature sensor collecting the wind turbine's temperature is diagnosed as having measurement inaccuracies or poor heat dissipation, with a medium confidence level, and so on.

[0064] Furthermore, iterate through all the wind turbines that were determined to be operating abnormally in step 206, check their abnormal states under different abnormality identification targets, and automatically trigger the fault diagnosis rule base to perform coupling analysis, find matching correlations, and generate a comprehensive fault diagnosis conclusion.

[0065] Therefore, based on the wind turbines identified as operating abnormally under different anomaly identification targets and the fault diagnosis conclusions, a structured anomaly and fault identification result report is generated. This report clearly lists the turbine ID of the operating abnormal wind turbine, the corresponding anomaly identification target, the turbine ID of the faulty wind turbine, possible fault cause inferences, and the fault confidence level. Visualization presents the anomaly and fault identification results intuitively through a human-machine interface. For example, in the global wind farm map, normal turbines are marked in green, while operating abnormal wind turbines and faulty wind turbines are highlighted in yellow or red according to their severity. The alarm mechanism is an active intervention triggered by visualization, automatically sending alarm information to the operation and maintenance team through SCADA system pop-ups, emails, and SMS messages. This transforms early fault symptoms hidden in the data into intuitive visual signals and timely alarm notifications, greatly shortening fault detection and response time, effectively avoiding unplanned downtime, reducing operation and maintenance costs, and achieving closed-loop management from condition monitoring to predictive maintenance.

[0066] It is worth mentioning that a single abnormal identification parameter can be caused by a variety of reasons. However, in this embodiment, when multiple mutually verifying identification parameters simultaneously show abnormalities, the probability of confirming a real equipment fault is greatly increased. Furthermore, this embodiment can effectively distinguish between sensor faults and mechanical body faults. For example, an abnormal temperature alone may point to a sensor, while simultaneous abnormalities in temperature and vibration strongly suggest bearing wear, providing maintenance personnel with a clear direction for troubleshooting.

[0067] It should be noted that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0068] Furthermore, such as Figure 3 As shown, as a specific implementation of the above-mentioned wind turbine abnormal operation identification method, this application embodiment provides a wind turbine abnormal operation identification device 300, which includes: an acquisition module 301, a processing module 302 and an identification module 303.

[0069] Among them, the acquisition module 301 is used to acquire the running sequence data of multiple wind turbines within the monitoring period. The running sequence data of the wind turbines includes the time sequence data of the wind turbines under different operating parameters. The processing module 302 is used to group wind turbines according to their turbine model to obtain at least one identification group; and to determine the identification parameters and operating condition parameters of the identification group from the operating parameters based on the preset anomaly identification targets corresponding to the identification group; and to extract features from the time series data of the target wind turbines in the identification group under the operating condition parameters to construct a comprehensive feature matrix of the identification group; and to perform clustering processing on the comprehensive feature matrix to obtain the target operating condition cluster. The identification module 303 is used to identify wind turbine units with abnormal operation based on identification parameters within the target operating condition cluster.

[0070] Optionally, the processing module 302 is specifically used to segment the time-series data of the target wind turbine under operating parameters using a sliding time window to obtain the first time-series data segment of the target wind turbine's operating parameters within the time window; extract and splice the features of the first time-series data segment of different operating parameters of the target wind turbine within the time window to obtain the feature vector of the target wind turbine within the time window; construct the unit feature matrix of the target wind turbine based on the feature vector of the target wind turbine in different time windows; and construct the comprehensive feature matrix of the identification group based on the unit feature matrices of different target wind turbines within the identification group.

[0071] Optionally, the processing module 302 is specifically used to extract the time domain features of the first time-series data segment of the operating parameters of the target wind turbine within the time window if the operating parameters are of the first preset type; and to extract the time domain features and frequency domain features of the first time-series data segment of the operating parameters of the target wind turbine within the time window if the operating parameters are of the second preset type.

[0072] Optionally, the processing module 302 is specifically used to perform dimensionality reduction processing on the comprehensive feature matrix to obtain a low-dimensional feature matrix; to use the principal component vectors of different rows in the low-dimensional feature matrix as the operating state instances of different target wind turbine units in different time windows; and to perform clustering processing on the operating state instances to obtain the target operating condition cluster.

[0073] Optionally, the processing module 302 is specifically used to cluster the running state instances according to different preset quantities to obtain the clustering results corresponding to the preset quantities; calculate the clustering index of the clustering results; determine the target quantity from the preset quantities based on the clustering index; and cluster the running state instances according to the target quantity to obtain the target running condition cluster.

[0074] Optionally, the identification module 303 is specifically used to determine the distribution parameters of the identification parameters within the target operating condition cluster based on the second time-series data segment of the identification parameters corresponding to the operating state instance within the target operating condition cluster. The second time-series data segment of the identification parameters corresponding to the operating state instance is the time-series data of the identification parameters of the target wind turbine to which the operating state instance belongs within the time window of the operating state instance. Based on the distribution parameters of the identification parameters within the target operating condition cluster, the module determines the abnormal threshold of the identification parameters within the target operating condition cluster. By comparing the second time-series data segment of the identification parameters corresponding to the operating state instance within the target operating condition cluster with the distribution parameters of the identification parameters within the target operating condition cluster, the module determines the abnormal operating state instance within the target operating condition cluster. Based on the abnormal operating state instance, the module determines the wind turbine with abnormal operation.

[0075] Optionally, the identification module 303 is specifically used to determine the target wind turbine to which the abnormal operation instance belongs; if the target wind turbine has multiple consecutive abnormal operation state instances in time sequence, and all abnormal operation state instances of the target wind turbine belong to the same target operating condition cluster, the target wind turbine is regarded as an abnormal wind turbine.

[0076] Specific limitations regarding the wind turbine abnormal operation identification device can be found in the limitations of the wind turbine abnormal operation identification method described above, and will not be repeated here. Each module in the aforementioned wind turbine abnormal operation identification device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0077] Based on the above, Figures 1 to 2 Accordingly, embodiments of this application also provide a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. Figures 1 to 2 The method for identifying abnormal operation of wind turbine units is shown.

[0078] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, or portable hard drive) and includes several instructions to cause a computer device (such as a personal computer, server, or network device) to execute the methods described in the various implementation scenarios of this application.

[0079] Based on the above, Figures 1 to 2 The method shown, and Figure 3To achieve the above objectives, the present application also provides a computer device, specifically a personal computer, server, network device, etc., as shown in the virtual device embodiment. This computer device includes a storage medium and a processor; the storage medium stores a computer program; the processor executes the computer program to achieve the above-described objectives. Figures 1 to 2 The method for identifying abnormal operation of wind turbine units is shown.

[0080] Optionally, the computer device may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB ports, card reader ports, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Bluetooth interfaces, Wi-Fi interfaces), etc.

[0081] Those skilled in the art will understand that the computer device structure provided in this embodiment does not constitute a limitation on the computer device, and may include more or fewer components, or combine certain components, or have different component arrangements.

[0082] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages and stores the hardware and software resources of a computer device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software within the physical device.

[0083] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platform, or the embodiments of this application can be implemented by hardware.

[0084] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application. Those skilled in the art will understand that the modules in the apparatus of the embodiment can be distributed within the apparatus of the embodiment as described, or can be modified to be located in one or more apparatuses different from this embodiment. The modules of the above-described embodiment can be combined into one module, or further divided into multiple sub-modules.

[0085] The serial numbers in this application are for descriptive purposes only and do not represent the superiority or inferiority of any particular implementation scenario. The above disclosures are merely a few specific implementation scenarios of this application; however, this application is not limited thereto, and any variations conceived by those skilled in the art should fall within the protection scope of this application.

Claims

1. A method for identifying abnormal operation of a wind turbine generator set, characterized in that, The method includes: Acquire the runtime sequence data of multiple wind turbine units within the monitoring period, wherein the runtime sequence data of the wind turbine units includes the time sequence data of the wind turbine units under different operating parameters; The wind turbines are grouped according to their model number to obtain at least one identification group; Based on the preset anomaly identification target corresponding to the identification group, the identification parameters and operating condition parameters of the identification group are determined from the operating parameters; Feature extraction is performed on the time-series data of the target wind turbines in the identification group under the operating parameters to construct a comprehensive feature matrix of the identification group. Clustering is performed on the comprehensive feature matrix to obtain the target operating condition cluster; Within the target operating condition cluster, wind turbine units with abnormal operation are identified based on the identification parameters.

2. The method for identifying abnormal operation of wind turbine units according to claim 1, characterized in that, The step of extracting features from the time-series data of the target wind turbines within the identification group under the specified operating parameters, and constructing a comprehensive feature matrix for the identification group, specifically includes: A sliding time window is used to segment the time series data of the target wind turbine under the operating parameters to obtain the first time series data segment of the operating parameters of the target wind turbine within the time window. Extract and stitch together the features of the first time-series data segment of the different operating parameters of the target wind turbine within the time window to obtain the feature vector of the target wind turbine within the time window; Based on the feature vectors of the target wind turbine in different time windows, construct the turbine feature matrix of the target wind turbine; Based on the unit feature matrices of different target wind turbines within the identification group, a comprehensive feature matrix of the identification group is constructed.

3. The method for identifying abnormal operation of wind turbine units according to claim 2, characterized in that, The extraction of features from the first time-series data segment of the target wind turbine's operating parameters within the time window specifically includes: If the operating condition parameter is a first preset type, extract the time domain features of the operating condition parameter of the target wind turbine in the first time series data segment within the time window; If the operating condition parameter is a second preset type, extract the time domain features and frequency domain features of the operating condition parameter of the target wind turbine in the first time series data segment within the time window.

4. The method for identifying abnormal operation of wind turbine units according to claim 2, characterized in that, The clustering process of the comprehensive feature matrix to obtain the target operating condition cluster specifically includes: The comprehensive feature matrix is ​​subjected to dimensionality reduction processing to obtain a low-dimensional feature matrix; The principal component vectors of different rows in the low-dimensional feature matrix are used as instances of the operating status of different target wind turbines in different time windows. The running state instances are clustered to obtain the target running condition cluster.

5. The method for identifying abnormal operation of wind turbine units according to claim 4, characterized in that, The clustering process of the running state instances to obtain the target running condition cluster specifically includes: The running state instances are clustered according to different preset numbers to obtain the clustering results corresponding to the preset number; Calculate the clustering index of the clustering results; Based on the clustering index, a target quantity is determined from the preset quantity; Based on the target number, the running state instances are clustered to obtain the target running condition cluster.

6. The method for identifying abnormal operation of wind turbine units according to claim 5, characterized in that, The step of identifying wind turbine units with abnormal operation based on the identification parameters within the target operating condition cluster specifically includes: Based on the second time-series data segment of the identification parameter corresponding to the operating state instance within the target operating condition cluster, the distribution parameter of the identification parameter within the target operating condition cluster is determined. The second time-series data segment of the identification parameter corresponding to the operating state instance is the time-series data of the identification parameter of the target wind turbine to which the operating state instance belongs within the time window to which the operating state instance belongs. Based on the distribution parameters of the identification parameters within the target operating condition cluster, an abnormal threshold for the identification parameters within the target operating condition cluster is determined; By comparing the second time-series data segment of the identification parameter corresponding to the running state instance within the target running condition cluster with the distribution parameter of the identification parameter within the target running condition cluster, the abnormal running state instance within the target running condition cluster is determined. Based on the example of the abnormal operating state, the wind turbine unit with the abnormal operation is identified.

7. The method for identifying abnormal operation of wind turbine units according to claim 6, characterized in that, The step of determining the wind turbine unit with abnormal operation based on the abnormal operation state instance specifically includes: Determine the target wind turbine to which the abnormal operation instance belongs; If the target wind turbine has multiple consecutive instances of abnormal operating state in chronological order, and all instances of abnormal operating state of the target wind turbine belong to the same target operating condition cluster, the target wind turbine is considered as the wind turbine with abnormal operation.

8. A wind turbine abnormal operation identification device, characterized in that, The device includes: The acquisition module is used to acquire the runtime sequence data of multiple wind turbines within the monitoring period. The runtime sequence data of the wind turbines includes the time sequence data of the wind turbines under different operating parameters. The processing module is configured to: group the wind turbines according to their turbine models to obtain at least one identification group; determine the identification parameters and operating condition parameters of the identification group from the operating parameters based on the preset anomaly identification targets corresponding to the identification group; extract features from the time-series data of the target wind turbines within the identification group under the operating condition parameters to construct a comprehensive feature matrix of the identification group; and perform clustering processing on the comprehensive feature matrix to obtain a target operating condition cluster. The identification module is used to identify wind turbine units with abnormal operation within the target operating condition cluster based on the identification parameters.

9. A readable storage medium having a program or instructions stored thereon, characterized in that, When the program or instructions are executed by the processor, they implement the steps of the wind turbine abnormal operation identification method as described in any one of claims 1 to 7.

10. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the program, it implements the wind turbine abnormal operation identification method as described in any one of claims 1 to 7.

Images