Fault mode identification method and system for wind turbine generator gearbox

Abstract

The invention discloses a fault mode identification method and system for a wind turbine generator gearbox. The fault mode recognition method comprises the steps that SCADA data associated with a wind turbine generator are obtained, the SCADA data comprise associated data used for conducting fault mode recognition on a gearbox of the wind turbine generator, and the associated data comprise temperature related data associated with the gearbox and oil pressure related data associated with gearbox oil; and inputting the SCADA data into a predetermined wind turbine generator gearbox fault detection model to obtain a fault mode recognition result of the wind turbine generator gearbox, the fault mode recognition result comprising a fault mode recognition result of at least one wind turbine generator component related to the wind turbine generator gearbox. Through the method, the multi-component failure mode of the gearbox can be accurately identified, and the interpretability of the correlation between the failure related characteristics and the failure mode is improved.

Description

Technical Field

[0001] This disclosure generally relates to the fields of wind power and data processing, and more specifically, to a method and system for fault mode identification of wind turbine gearboxes. Background Technology

[0002] In related fields, as the core transmission component of wind turbine generators, the wind turbine gearbox (or simply gearbox) functions by converting the low-speed mechanical energy of the wind turbine into the high-speed electrical energy input required by the generator through precise mechanical design. This achieves torque gain and rational distribution, ensuring efficient and stable energy transmission. The gearbox mainly consists of the gearbox body (gear system, bearing system, and housing structural components) and the lubrication system. Unlike generators and main bearings, which use grease lubrication, gearboxes, due to their high energy density transmission and multi-interface lubrication requirements, must rely on an oil lubrication system to ensure reliable gear meshing and bearing operation. Therefore, the gearbox lubrication system consists of key components such as lubricating oil, oil pump, filter, cooler, and oil distributor. It not only needs to provide a continuous oil supply but also needs to perform cooling and pressure regulation functions. Any temporary lack of lubrication can lead to irreversible damage to gears or bearings.

[0003] Due to the aforementioned characteristics of gearboxes, gearbox fault diagnosis is far more complex than that of main bearings, generators, and other equipment. Existing gearbox fault diagnosis solutions are primarily data-driven, especially deep learning-based solutions. Deep learning-based data analysis requires extensive labeled data for training; black-box models cannot correlate fault features with physical damage mechanisms, and their cross-scenario generalization ability is insufficient to adapt to the diverse failure scenarios of gearboxes. Therefore, to address the problems in existing solutions, an improved method is needed for gearbox fault diagnosis. Summary of the Invention

[0004] The embodiments of this disclosure provide a fault mode identification method and system for wind turbine gearboxes, aiming to solve the fault diagnosis problem of wind turbine gearboxes. By analyzing only SCADA data to obtain the mapping relationship between fault modes and associated fault features, and constructing a model for predicting faults in wind turbine gearboxes, fault mode identification and fault diagnosis are achieved. This improves the accuracy, efficiency, objectivity, and comprehensiveness of fault identification in wind turbine gearboxes, and enhances the interpretability of the relationship between fault modes and their associated features.

[0005] In one general aspect, a fault mode identification method for wind turbine gearboxes is provided. The method includes: acquiring SCADA data associated with the wind turbine, the SCADA data including associated data for fault mode identification of the wind turbine gearbox, the associated data including temperature-related data associated with the gearbox and oil pressure-related data associated with the gearbox fluid; inputting the SCADA data into a predetermined wind turbine gearbox fault detection model to obtain a fault mode identification result for the wind turbine gearbox, wherein the predetermined wind turbine gearbox fault detection model is constructed by: acquiring historical fault data associated with the wind turbine gearbox, the historical fault data including historical temperature data, historical oil pressure data, and historical fault indication data associated with the gearbox; determining fault modes associated with the gearbox and corresponding associated fault features based on the historical fault data; and constructing the predetermined wind turbine gearbox fault detection model based on the mapping relationship between the fault modes and corresponding associated fault features.

[0006] Optionally, the step of constructing the predetermined wind turbine gearbox fault detection model based on the fault mode and the corresponding associated fault features through the mapping relationship between the two may include: establishing a mapping relationship between the fault mode and the associated fault features based on the fault mode and the corresponding associated fault features; constructing the predetermined wind turbine gearbox fault detection model based on the mapping relationship between the two, wherein the fault mode identification result includes the fault mode identification result of at least one wind turbine component related to the wind turbine gearbox.

[0007] Optionally, the failure mode may include gearbox body component failure and gearbox external associated component failure, and the gearbox body component failure may include at least one of broken teeth failure, tooth surface wear failure, gearbox bearing failure, and gearbox internal copper ring wear failure, and the associated component failure may include at least one of temperature control valve failure, motor pump failure, poor lubrication failure, oil circuit blockage failure, cooling fan failure, and three-way valve failure.

[0008] Optionally, the wear failure of the copper ring inside the gearbox, the poor lubrication failure, and the oil circuit blockage failure are all related to the pressure characteristics of the gearbox oil pump outlet and the pressure characteristics of the gearbox oil inlet. Among them, the tooth breakage failure, tooth surface wear failure, gearbox bearing failure, and three-way valve failure are all related to the temperature characteristics of the gearbox oil sump, the temperature characteristics of the bearings or teeth of each stage of the multi-stage transmission of the gearbox, and the temperature characteristics of the gearbox oil inlet.

[0009] Optionally, for each fault mode, the fault risk of each fault mode can be determined by dividing the multiple associated fault features corresponding to each fault mode into multiple associated fault feature groups and determining the boundary conditions of each associated feature group in the multiple associated feature groups. The associated fault features include at least one of the following: gearbox oil pump outlet pressure characteristics, gearbox oil inlet pressure characteristics, gearbox oil sump temperature characteristics, temperature characteristics of each bearing or tooth of the gearbox multi-stage transmission, gearbox oil inlet temperature characteristics, wind turbine nacelle acceleration x-axis characteristics, wind turbine nacelle acceleration y-axis characteristics, and gearbox fault indication characteristics.

[0010] Optionally, the fault mode identification method may further include: for faults in the gearbox body component, dividing the gearbox oil pump outlet pressure characteristics, gearbox inlet pressure characteristics, gearbox oil sump temperature characteristics, gearbox low-speed shaft temperature characteristics, first gearbox high-speed shaft temperature characteristics, and second gearbox high-speed shaft temperature characteristics into multiple associated fault feature groups, and determining the fault risk of the gearbox body component fault based on the satisfaction of the boundary conditions of each associated feature group in the multiple associated feature groups, wherein the multiple associated fault feature groups include the first associated fault feature group to the fifth associated fault feature group, and the first associated fault feature group includes multiple gearbox low-speed shaft temperature characteristics, the second associated fault feature group includes multiple first gearbox high-speed shaft temperature characteristics, the third associated fault feature group includes multiple second gearbox high-speed shaft temperature characteristics, the fourth associated fault feature group includes gearbox oil sump temperature characteristics, multiple gearbox low-speed shaft temperature characteristics, multiple first gearbox high-speed shaft temperature characteristics, and multiple second gearbox high-speed shaft temperature characteristics, and the fifth associated fault feature group includes gearbox oil pump outlet pressure characteristics and gearbox inlet pressure characteristics.

[0011] Optionally, the fault mode identification method may further include: the predetermined wind turbine gearbox fault detection model uses a preset classifier and / or a preset neural network model to obtain the fault mode type of the wind turbine gearbox.

[0012] In another general aspect, a system for fault mode identification of wind turbine gearboxes is provided, the system comprising: a main controller for the wind turbine, configured to be connected to the wind turbine; a switch located on the tower of the wind turbine and configured to be connected to the main controller; an edge computing device on the tower of the wind turbine, wherein a predetermined wind turbine gearbox fault detection model is arranged in the edge computing device, and the edge computing device is configured to: connect to the main controller via the switch and receive SCADA data from the main controller, wherein the SCADA data is identified using the predetermined wind turbine gearbox fault detection model to determine fault modes of the wind turbine gearbox, wherein the SCADA data includes correlation data for fault mode identification of the wind turbine gearbox. The associated data includes temperature-related data associated with the gearbox and oil pressure-related data associated with the gearbox fluid. The predetermined wind turbine gearbox fault detection model is constructed as follows: Historical fault data associated with the wind turbine gearbox is acquired, including historical temperature data, historical oil pressure data, and historical fault indication data associated with the gearbox; the historical fault data is analyzed to determine fault modes associated with the gearbox and corresponding associated fault characteristics; a mapping relationship is established between the fault modes and associated fault characteristics based on the fault modes and associated fault characteristics; and the predetermined wind turbine gearbox fault detection model is constructed based on the mapping relationship.

[0013] Optionally, the operation of the edge computing device receiving SCADA data from the main controller may include: receiving temperature-related data associated with the gearbox from a temperature sensor associated with the gearbox via the switch, and receiving oil pressure-related data associated with the gearbox oil from a pressure sensor associated with the gearbox oil, wherein the temperature sensor associated with the gearbox includes at least one of a gearbox oil sump temperature sensor, a gearbox oil inlet temperature sensor, and temperature sensors of each stage bearing or tooth of a multi-stage gearbox transmission, and the pressure sensor associated with the gearbox oil includes a gearbox oil pump outlet pressure sensor and / or a gearbox oil inlet pressure sensor.

[0014] Optionally, the gearbox oil sump temperature sensor is arranged adjacent to the gearbox oil sump, the gearbox oil inlet temperature sensor is arranged adjacent to the gearbox oil inlet, the temperature sensors of each bearing or tooth of the multi-stage transmission of multiple gearboxes are arranged adjacent to their corresponding bearings or teeth, the gearbox oil pump outlet pressure sensor is arranged adjacent to the gearbox oil pump outlet, and the gearbox oil inlet pressure sensor is arranged adjacent to the gearbox oil inlet.

[0015] In another general aspect, a computer-readable storage medium is provided that, when instructions in the computer-readable storage medium are executed by at least one processor, causes the at least one processor to perform the fault diagnosis and / or fault mode identification method for a wind turbine gearbox as described above.

[0016] In another general aspect, a computer program product is provided, comprising a computer program / instructions that, when executed by a processor, implement the fault diagnosis and / or fault mode identification method for wind turbine gearboxes as described above.

[0017] In another general aspect, a computer device is provided, the computer device comprising: at least one processor; at least one memory storing computer-executable instructions, wherein, when executed by the at least one processor, the computer-executable instructions cause the at least one processor to perform the fault diagnosis and / or fault mode identification method for a wind turbine gearbox as described above.

[0018] In another general aspect, a wind turbine is provided that includes the computer equipment described above.

[0019] The fault mode identification method and system for wind turbine gearboxes according to embodiments of this disclosure, compared with existing diagnostic models, achieves superior results with clear feature mapping relationships, strong model interpretability, accurate diagnostic results, and improved efficiency by constructing a mapping relationship between fault features and fault modes, and by using a model trained solely on SCADA data to predict faults in wind turbine gearboxes. Furthermore, the fault mode identification / fault diagnosis method for wind turbine gearboxes according to embodiments of this disclosure can simultaneously diagnose failures of multiple components related to the gearbox, significantly improving the overall system's operation and maintenance efficiency. Moreover, by deploying this lightweight fault diagnosis model at edge devices such as the wind turbine main control unit, the efficiency of fault diagnosis can be greatly improved, while reducing system operation and maintenance costs. This achieves high-precision, low-false-alarm, early-warning, and low-cost gearbox health management, significantly improving the reliability and economic efficiency of wind turbine operation and maintenance. Attached Figure Description

[0020] The above and other objects and features of the embodiments of this disclosure will become clearer from the following description taken in conjunction with the accompanying drawings illustrating the embodiments, wherein: Figure 1 This is a flowchart illustrating a fault mode identification / fault diagnosis method for a wind turbine gearbox according to an embodiment of the present disclosure; Figure 2This is a schematic diagram illustrating the system architecture and model deployment according to this disclosure; Figure 3 This is a flowchart illustrating an example of a fault mode identification / fault diagnosis method for a wind turbine gearbox according to an embodiment of the present disclosure; Figure 4 This is a block diagram illustrating a fault diagnosis device for a wind turbine gearbox according to an embodiment of the present disclosure; Figure 5 This is a block diagram illustrating a computer device according to an embodiment of the present disclosure. Detailed Implementation

[0021] The following detailed embodiments are provided to aid the reader in gaining a comprehensive understanding of the methods, apparatus, and / or systems described herein. However, various changes, modifications, and equivalents of the methods, apparatus, and / or systems described herein will become apparent upon understanding this disclosure. For example, the order of operations described herein is merely illustrative and is not limited to those orders set forth herein, but may be changed as will become clear upon understanding this disclosure, except for operations that must occur in a specific order. Furthermore, for clarity and conciseness, descriptions of features known in the art may be omitted.

[0022] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings, examples of which are illustrated in the drawings, wherein the same reference numerals always refer to the same parts. The embodiments will now be described with reference to the accompanying drawings in order to explain this disclosure.

[0023] In related technologies, the gearbox lubrication system must not only ensure the precise delivery of lubricating oil to the gear meshing area and critical lubrication points such as bearings, but also suppress heat accumulation in the gearbox through cooling to prevent failure due to overheating. Therefore, it is necessary to monitor the relevant operating status in real time using temperature, pressure, and differential pressure sensors. Gearbox fault monitoring may require simultaneous monitoring of multiple subsystems, including the gear system, bearing system, and lubrication system, as well as multiple components within each subsystem. Failure in any link (including but not limited to components) can lead to malfunctions. For example, long-term poor lubrication of the gearbox (e.g., oil contamination, oil pump failure, filter blockage) will accelerate gear and bearing wear.

[0024] Existing gearbox fault diagnosis solutions include: general mechanical fault diagnosis methods, vibration signal-based diagnostic solutions, and deep learning-based data analysis solutions. For general mechanical fault diagnosis methods, due to differences in lubrication mechanisms (e.g., differences between grease lubrication and oil lubrication), it is difficult to identify specific faults in highly complex equipment like gearboxes. For vibration signal-based diagnostic solutions, high-precision vibration sensors need to be installed separately, which not only increases hardware costs (not standard equipment) but also makes vibration signals susceptible to environmental noise interference, requiring complex noise reduction preprocessing. For deep learning-based data analysis solutions, this method heavily relies on engineers' experience for manual feature selection, and the diagnostic model requires a large amount of labeled data for training, making it difficult to adapt to the diverse failure scenarios of gearboxes.

[0025] To address the aforementioned and other technical problems existing in the prior art, this disclosure proposes a fault mode identification method and system for wind turbine gearboxes, which can at least achieve the following: In terms of mechanism adaptability, the solution of this disclosure fully considers the dynamic characteristics of the gearbox hydraulic system, thereby accurately identifying the unique oil pressure and temperature coupled failure modes of the gearbox; in terms of engineering implementation, the solution of this disclosure does not require the additional deployment of sensors and signal processing equipment required by vibration monitoring solutions, thereby significantly reducing system complexity and maintenance costs; in terms of algorithm design, it solves the problems of traditional SCADA analysis methods lacking a systematic feature construction framework, over-reliance on human experience, and requiring a large amount of labeled data.

[0026] The following reference Figures 1 to 5 A detailed description is provided of a fault mode identification method and system for wind turbine gearboxes according to embodiments of the present disclosure.

[0027] First, refer to Figures 1 to 3 A detailed description of a fault mode identification / fault diagnosis method for a wind turbine gearbox according to embodiments of the present disclosure.

[0028] Figure 1 This is a flowchart illustrating a fault mode identification / fault diagnosis method for a wind turbine gearbox according to an embodiment of the present disclosure. Figure 2 This is a schematic diagram illustrating the system architecture and model deployment according to this disclosure. Figure 3 This is a flowchart illustrating an example of a fault mode identification / fault diagnosis method for a wind turbine gearbox according to an embodiment of the present disclosure.

[0029] Reference Figure 1 According to an embodiment of this disclosure, in step S101, SCADA data associated with the wind turbine is acquired.

[0030] Here, SCADA data may include associated data used for fault (hereinafter also referred to as failure) diagnosis of wind turbine gearboxes. Furthermore, associated data may include temperature-related data associated with the gearbox and oil pressure-related data associated with the gearbox fluid. For example, temperature data may include inlet temperature, sump temperature, low-speed shaft temperature, and high-speed shaft temperature (e.g., first high-speed shaft temperature, second high-speed shaft temperature) solely from the corresponding SCADA system.

[0031] As an example, step S101 may further include: receiving temperature-related data associated with the gearbox from a temperature sensor associated with the gearbox via a switch located on the tower of the wind turbine, and receiving oil pressure-related data associated with the gearbox oil from a pressure sensor associated with the gearbox oil.

[0032] For example, temperature sensors associated with the gearbox may include at least one of a gearbox oil sump temperature sensor, a gearbox oil inlet temperature sensor, and temperature sensors for each stage of bearings or teeth in a multi-stage gearbox drive. Pressure sensors associated with the gearbox fluid may include a gearbox oil pump outlet pressure sensor and / or a gearbox oil inlet pressure sensor.

[0033] Gearbox fault diagnosis can be achieved by using only various temperature and pressure-related data associated with the gearbox from the SCADA system (e.g., only gearbox hydraulic system data (oil pressure, oil temperature, bearing temperature, etc.), without using vibration data from vibration sensors, and excluding irrelevant variables). This makes it easier and cheaper to deploy models built based on such data in edge devices of wind turbines / wind turbines (e.g., wind turbine main control PLCs), without requiring vibration data from additional separately deployed vibration sensors, thereby reducing diagnostic costs by eliminating the need for additional equipment such as vibration sensors.

[0034] Furthermore, the gearbox oil sump temperature sensor can be arranged adjacent to the gearbox oil sump, the gearbox inlet temperature sensor can be arranged adjacent to the gearbox inlet, the temperature sensors for each stage of bearings or teeth in multiple gearbox multi-stage transmissions can be arranged adjacent to their corresponding stage bearings or teeth, the gearbox oil pump outlet pressure sensor can be arranged adjacent to the gearbox oil pump outlet, and the gearbox inlet pressure sensor can be arranged adjacent to the gearbox inlet. Regarding the adjacently arranged sensors, taking the temperature sensor as an example, it can be arranged by inserting a conductive material, such as a silver sheet, into a non-functional support near the corresponding component, for example, connected to an external device via a wired connection. The external device determines the temperature of the corresponding sensing part by measuring the resistance of the conductive material.

[0035] By placing various sensors near the corresponding components, the real-time performance and accuracy of the sensor data collected by these sensors can be maximized without affecting the normal operation of the components.

[0036] According to an embodiment of this disclosure, in step S102, SCADA data is input into a predetermined wind turbine gearbox fault detection model to obtain the fault diagnosis result of the wind turbine gearbox.

[0037] Here, the fault diagnosis results include the fault diagnosis results of at least one wind turbine component related to the wind turbine gearbox.

[0038] As an example, a predetermined wind turbine gearbox fault detection model can be arranged in an edge computing device located within the wind turbine tower (e.g., inside and / or outside the tower). For instance, a predetermined wind turbine gearbox fault detection model can be arranged in an edge computing device located at the base of the wind turbine tower.

[0039] As an example, refer to Figure 3 This illustrates an exemplary system architecture and model deployment according to this disclosure. Figure 3 In the middle, the unit end includes wind turbines WTG1, WTG2, WTG3 to WTGN. Figure 3 The device topologies within wind turbines WTG1 and WTG2 are also shown, which are deployed in the corresponding edge computing devices according to the above model of this disclosure.

[0040] For example, an exemplary system hardware architecture consists of edge computing devices deployed on the wind turbine tower (e.g., at the base of the tower), switches (e.g., tower base switches), a wind turbine PLC controller (wind turbine main controller), and a site-side component (e.g., such as...). Figure 3 The equipment shown in the figure consists of edge computing monitoring equipment, edge management equipment, central switch, packet switch, and wind farm local server / transfer server.

[0041] Specifically, each edge computing device communicates in real time with the corresponding wind turbine master controller via a tower-based switch to obtain gearbox operating data. The edge computing device has a pre-trained wind turbine gearbox fault detection (e.g., failure identification) model built-in, enabling data analysis and pattern recognition to be performed on the edge computing device itself. Each tower-based switch aggregates the identification results to the site-end equipment via the industrial network for centralized monitoring by maintenance personnel.

[0042] As an example, a system for fault mode identification of a wind turbine gearbox according to embodiments of the present disclosure may include: a main controller of the wind turbine, a switch, and an edge computing device for the wind turbine tower.

[0043] Here, the main controller of the wind turbine can be connected to the wind turbine. For example, a switch can be located on the tower of the wind turbine and connected to the main controller. In addition, for the edge computing device on the tower of the wind turbine, a predetermined wind turbine gearbox fault detection model can be arranged in the edge computing device, and the edge computing device can be connected to the main controller via the switch and receive SCADA data from the main controller.

[0044] For example, SCADA data includes associated data used for fault mode identification of wind turbine gearboxes, including temperature-related data associated with the gearbox and oil pressure-related data associated with the gearbox fluid.

[0045] According to this disclosure, the predetermined wind turbine gearbox fault detection model can be used to identify SCADA data in order to determine the fault mode of the wind turbine gearbox.

[0046] As an example, the predetermined wind turbine gearbox fault detection model can be constructed as follows: Historical fault data associated with the wind turbine gearbox is acquired, including historical temperature data, historical oil pressure data, and historical fault indication data associated with the gearbox; the historical fault data is analyzed to determine the fault modes associated with the gearbox and their corresponding associated fault characteristics; a mapping relationship is established between the fault modes and associated fault characteristics based on these relationships; and the predetermined wind turbine gearbox fault detection model is constructed based on this mapping relationship.

[0047] Additionally, the operation of the edge computing device receiving SCADA data from the main controller may include: receiving temperature-related data associated with the gearbox from a temperature sensor associated with the gearbox via a switch, and receiving oil pressure-related data associated with the gearbox oil from a pressure sensor associated with the gearbox oil. Here, the temperature sensor associated with the gearbox may include at least one of a gearbox oil sump temperature sensor, a gearbox inlet temperature sensor, and temperature sensors for each stage bearing or tooth of a multi-stage gearbox transmission, and the pressure sensor associated with the gearbox oil may include a gearbox oil pump outlet pressure sensor and / or a gearbox inlet pressure sensor.

[0048] For example, a gearbox oil sump temperature sensor can be arranged adjacent to the gearbox oil sump. For example, a gearbox inlet temperature sensor can be arranged adjacent to the gearbox inlet. For example, temperature sensors for each stage of bearings or teeth in a multi-stage gearbox transmission can be arranged adjacent to their corresponding bearings or teeth. For example, a gearbox oil pump outlet pressure sensor can be arranged adjacent to the gearbox oil pump outlet, and a gearbox inlet pressure sensor can be arranged adjacent to the gearbox inlet.

[0049] It should be noted that the operations performed by the various components in the above system can be compared with those described in the reference. Figure 1 The related content is similar, so I will not repeat it here.

[0050] According to the system architecture described above in this disclosure, compared with the traditional centralized architecture of cloud services (hereinafter referred to as the cloud solution), the technical advantages of this system architecture include: Improved distributed real-time performance. By processing data locally and promptly through edge computing devices, network transmission latency is eliminated, significantly improving timeliness compared to traditional cloud solutions.

[0051] Enhanced autonomy and reliability. The system's edge nodes possess offline diagnostic capabilities, enabling them to maintain local operation even during network outages. This significantly enhances the system's robustness compared to traditional cloud-based solutions.

[0052] Enhanced scalability. Modular edge nodes support a flexible "one machine, one model" deployment mode, effectively avoiding the computing power bottleneck of traditional cloud solutions when expanding the unit. The system architecture disclosed in this paper forms a decentralized intelligent diagnostic closed loop.

[0053] By deploying the gearbox fault detection model of wind turbines in an edge computing device, the response efficiency of gearbox fault detection can be greatly improved and the response time can be shortened. For example, the real-time response can be shortened to less than one second, thereby realizing gearbox fault diagnosis with a response time of seconds or even milliseconds, which greatly improves the efficiency of fault identification.

[0054] For example, an edge computing device can be connected to the main controller of a wind turbine via a switch located on the tower of the wind turbine, and the edge computing device can receive SCADA data from the main controller.

[0055] By using SCADA data from the main controller to train the wind turbine gearbox fault detection model, it is possible to obtain gearbox-related operating data in real time with the corresponding wind turbine main controller. This allows the trained model to adapt to the most recent gearbox operation, further improving the accuracy, real-time performance, and efficiency of the model's predictions.

[0056] According to embodiments of this disclosure, referring to Figure 2A predetermined wind turbine gearbox fault detection model can be constructed through the following steps S201 to S204: In step S201, historical fault data associated with the wind turbine gearbox is obtained. The historical fault data includes historical temperature data associated with the gearbox, historical oil pressure data associated with the gearbox oil, and historical fault indication data associated with the gearbox.

[0057] For example, historical fault data may include gearbox failure datasets for a specific wind turbine model over a specific time period, such as gearbox failure datasets for mainstream medium-speed wind turbine models over a predetermined time period (e.g., several years). Here, even in application scenarios where the historical fault data is a small sample of historical gearbox failure datasets, an effective failure identification model can still be trained.

[0058] As an example, the acquired unit data can be preprocessed, such as performing data cleanup based on task logic consistency checks / boundary value checks to remove physically infeasible data caused by sensor failures or abnormal operating conditions (e.g., temperature data showing 850℃, exceeding 100℃, or remaining consistently unchanged or intermittently unchanging; invalid data such as wind speed, temperature, pressure, and power values ​​exceeding boundary ranges). Furthermore, for handling missing values, it is necessary to ensure the integrity and time alignment of time-series data and remove unusable (NA) and missing (NULL) data. Additionally, the acquired data represents data from a stable system state; unstable data from the initial grid connection phase is removed, and stable operating data based on the real-time unit status of the SCADA system, where the unit is grid-connected and its power exceeds 10%, is used as modeling sample data (i.e., historical fault data).

[0059] In step S202, based on historical fault data, the fault modes (or failure modes) associated with the gearbox and the corresponding associated fault characteristics are determined.

[0060] As an example, failure modes may include gearbox body component failures and failures of associated components outside the gearbox. For example, gearbox body component failures may include at least one of the following: broken teeth failure, tooth surface wear failure, gearbox bearing failure, and wear failure of the gearbox internal copper rings (e.g., the copper rings of each stage of the planetary gear pin shafts in the gearbox body, such as the first-stage planetary gear pin shafts, second-stage planetary gear pin shafts, third-stage planetary gear pin shafts, etc.). For example, associated component failures may include at least one of the following: temperature control valve failure, motor pump failure, poor lubrication failure, oil passage blockage failure, cooling fan failure, and three-way valve failure.

[0061] For example, when the failure mode is a gearbox body component failure, the associated failure characteristics include gearbox oil pump outlet pressure characteristics, gearbox oil inlet pressure characteristics, gearbox oil sump temperature characteristics, gearbox low-speed shaft temperature characteristics, first gearbox high-speed shaft temperature characteristics, and second gearbox high-speed shaft temperature characteristics.

[0062] By identifying the aforementioned failure modes and associated failure characteristics, failure diagnosis of multiple components related to the gearbox can be covered, improving system operation and maintenance efficiency. For example, the method disclosed herein can achieve failure diagnosis of both the gearbox body (e.g., broken teeth, tooth surface wear, bearing failure, etc.) and the lubrication system (e.g., temperature control valve, motor pump, oil circuit blockage, etc.), covering more than 90% of common failures.

[0063] For example, wear failure of the copper ring inside the gearbox, poor lubrication failure, and oil circuit blockage failure can all be associated with the outlet pressure characteristics of the gearbox oil pump and the inlet pressure characteristics of the gearbox.

[0064] For example, broken tooth failure, tooth surface wear failure, gearbox bearing failure, and three-way valve failure can all be related to the temperature characteristics of the gearbox oil sump, the temperature characteristics of the bearings or teeth at each stage of the multi-stage transmission of the gearbox (e.g., the temperature of each high-speed shaft or tooth of the gearbox, the temperature of each low-speed shaft or tooth of the gearbox), and the temperature characteristics of the gearbox oil inlet.

[0065] By analyzing various temperature and pressure characteristics, the failure of multiple components can be determined simultaneously using several features, thereby significantly improving the efficiency of fault diagnosis. Furthermore, by determining the correlation between failure modes and various features, the diagnostic accuracy of the corresponding failure modes is further improved. For example, by constructing high-precision failure features, the diagnostic accuracy rate can be increased to over 90%.

[0066] In step S203, the mapping relationship between the fault mode and the associated fault characteristics is determined.

[0067] By constructing a mapping relationship between at least some of the aforementioned associated fault features and specific fault modes, the interpretability of the model output can be improved. This greatly enhances the interpretability of the output results of the lightweight fault detection model for gearboxes, improves the reliability of diagnostic results, and further shortens training time and improves data training efficiency, which is conducive to the deployment of this fault detection model in wind turbine / wind turbine-related edge equipment.

[0068] As an example, step S203 may further include: inputting the fault mode and associated fault features into a preset classifier and / or a preset neural network (e.g., a convolutional neural network (CNN) model) to obtain a mapping relationship between the fault mode and the associated fault features.

[0069] By determining the mapping relationship between failure modes and associated failure features, the diagnosis results of failure modes are more accurate compared with traditional deep learning model solutions.

[0070] Furthermore, for example, for each failure mode, the failure risk (e.g., different levels of failure risk) can be determined by dividing the multiple associated failure features corresponding to each failure mode into multiple associated failure feature groups and determining the failure risk of each failure mode based on the boundary conditions of each associated feature group in the multiple associated feature groups.

[0071] For example, associated fault characteristics may include at least one of the following: gearbox oil pump outlet pressure characteristics, gearbox oil inlet pressure characteristics, gearbox oil sump temperature characteristics, temperature characteristics of each bearing or tooth of the gearbox multi-stage transmission (e.g., gearbox high-speed shaft 1 temperature, gearbox high-speed shaft 2 temperature, gearbox low-speed shaft temperature), gearbox oil inlet temperature characteristics, wind turbine nacelle acceleration x-axis characteristics, wind turbine nacelle acceleration y-axis characteristics, and gearbox fault indication characteristics.

[0072] As examples, specific examples of the above-mentioned features are as follows: Features for the gearbox oil pump outlet pressure may include, for example: pressure threshold at low oil pump speed (specific range of ambient temperature), abnormal standard deviation of the average value after daily grouping and clustering of pressure, overall upward or downward trend of pressure, sudden increase in oil pump outlet pressure in oil sump temperature grouping, oil sump temperature > 45 degrees, high-speed shaft temperature < 65 degrees, sudden drop in oil pump outlet pressure exceeding 2 bar; Features for the temperature of gearbox high-speed shaft 2 may include, for example: temperature change rate of high-speed shaft 2, abnormal fluctuation, excessively rapid rise, temperature rise rate, temperature rise threshold, duration of subsequence, absolute increment, increment per unit time, load-temperature correlation gradient index (corresponding to specific power generation and / or specific operating conditions).

[0073] For example, the various forms of the parameters (temperature and pressure) and their corresponding anomaly judgment criteria according to this disclosure are exemplified below: For fixed temperature thresholds: if the temperature exceeds the normal operating range, the abnormal judgment and threshold setting include, for example: if the temperature at any point on the bearing or tooth surface is greater than 75 degrees, the body will be damaged.

[0074] Regarding the temperature change rate: the temperature change rate is obtained by dividing the temperature rise or fall of two adjacent data points by the time. The abnormal judgment and threshold setting include, for example: if the temperature change rate is >0.06℃ / s, the main body is damaged, the oil pump fails, or the heat dissipation is abnormal.

[0075] Regarding the intensity of temperature fluctuations: Calculate the rolling data (e.g., standard deviation) of each temperature feature (including low-speed bearing (referred to as low-speed shaft) 1, high-speed bearing (referred to as high-speed shaft) 1, high-speed bearing 2, and oil temperature, etc.) to determine their fluctuations. The anomaly judgment and threshold setting include, for example, if the fluctuation intensity exceeds 0.025, then there is a lubrication abnormality or mechanical loosening.

[0076] For cases of rapid temperature rise: a time window is used to determine the temperature change per minute. Anomaly detection and threshold settings include, for example: when the load is >80%, the temperature change is >3.5℃, indicating overload or oil film failure.

[0077] For temperature rise exceeding the threshold: the difference between the bearing temperature and the oil bath temperature exceeds the range under normal operating conditions; in each power range, if the oil bath temperature is 45-50℃ and the average temperature rise exceeds 23℃, the bearing will be damaged; if the oil bath temperature is 40-45℃ and the average temperature rise exceeds 25℃, the bearing will be damaged.

[0078] Regarding the temperature rise rate: the difference between two adjacent temperature rise data points is divided by the time to obtain the temperature rise rate; its anomaly judgment and threshold setting include, for example: when the load is >80%, the temperature change is >3.5℃, the load is overloaded or the oil film fails.

[0079] For the duration of a monotonically increasing temperature subsequence, the abnormal judgment and threshold setting are as follows: when it exceeds 1 hour, an early warning is triggered, indicating that the cooling system is malfunctioning, continuously overloaded, or that internal local wear generates heat.

[0080] For the absolute increment of the monotonically increasing temperature subsequence, the anomaly detection and threshold setting include, for example, triggering a bearing wear warning if the cumulative temperature is >15℃.

[0081] For the unit time increment of the monotonically increasing temperature subsequence, the anomaly judgment and threshold setting are as follows: when the increment is >0.03℃ / s and ≤0.05℃ / s, it is determined as an early fault; when the increment is >0.05℃ / s, it is determined as a late fault.

[0082] For the load-temperature related gradient index: heating rate / active power change rate, its anomaly judgment and threshold setting include, for example: a heating rate exceeding 0.1℃ / s indicates abnormal friction (such as planetary gear off-center loading).

[0083] Regarding the comparison of the temperature relationship between the oil sump and the oil inlet, the abnormal judgment and threshold settings are as follows: if the percentage of inlet temperatures lower than the oil sump temperature by 6 degrees Celsius is less than 1%, the temperature control valve is stuck. If the percentage of inlet temperatures lower than the oil sump temperature by 8 or 12 degrees Celsius but higher than 6 degrees Celsius is less than 1%, the cooling fan speed is malfunctioning.

[0084] For pressure exceeding threshold detection (too high: oil circuit blockage; too low: pump failure), pressure cycles are grouped and clustered. Continuously rising pressure indicates a risk of lubrication system blockage, while continuously falling pressure indicates decreased pump efficiency or seal failure.

[0085] If the oil pump outlet pressure does not change, it means the oil pump is not operating at high speed. Additionally, the fault indication is also considered a feature.

[0086] As another example, the characteristics used to determine the failure of a thermostatic valve may include, for example, a failure flag of 1 if the oil sump temperature, the inlet temperature, the low-speed shaft 1 temperature, the high-speed shaft 1 temperature, and the high-speed shaft 2 temperature all exceed a threshold.

[0087] In addition, although the three parameters of wind turbine nacelle acceleration x-axis characteristics, wind turbine nacelle acceleration y-axis characteristics, and gearbox fault indication characteristics come from the SCADA system rather than from additionally deployed vibration sensors, these data can still reflect vibration-related characteristics and can be applied to fault diagnosis of gearbox body components.

[0088] As an example, the process of determining the failure risk level of a specific component can be illustrated as follows: For gearbox tooth surface wear, the features associated with this failure model include the following feature groups 1 to 5: (1) Feature group 1: The temperature of low speed axis 1 exceeds the threshold, the temperature of low speed axis 1 is in the form of a monotonic subsequence (e.g., monotonically increasing), the temperature rise of low speed axis 1 exceeds the threshold, and the temperature rise history of low speed axis 1 is abnormal. (2) Feature group 2: The temperature of high speed axis 1 exceeds the threshold, the temperature of high speed axis 1 is in the form of a monotonic subsequence, the temperature rise of high speed axis 1 exceeds the threshold, and the temperature rise history of high speed axis 1 is abnormal. (3) Feature group 3: The temperature of high-speed axis 2 exceeds the threshold, the temperature of high-speed axis 2 is in the form of a monotonic subsequence, the temperature rise of high-speed axis 2 exceeds the threshold, and the temperature rise history of high-speed axis 2 is abnormal. (4) Feature group 4: The temperature of the gearbox oil sump exceeds the threshold, the temperature of the gearbox low-speed shaft 1 exceeds the threshold, the temperature of the gearbox high-speed shaft 1 exceeds the threshold, the temperature of the gearbox high-speed shaft 2 exceeds the threshold, the temperature rise of the low-speed shaft 1 exceeds the threshold, the temperature rise of the high-speed shaft 1 exceeds the threshold, the temperature rise of the high-speed shaft 2 exceeds the threshold, the temperature rise history of the low-speed shaft 1 is abnormal, the temperature rise history of the high-speed shaft 1 is abnormal, the temperature rise history of the high-speed shaft 2 is abnormal. (5) Feature group 5: The pressure at the gearbox inlet is lower than the threshold, and the pressure at the gearbox oil pump outlet exceeds the threshold.

[0089] The fault risk level determined based on the fulfillment of conditions for the above five characteristic groups includes: (1) For feature group 1, feature group 2, and feature group 3, if there are two or more conditions that are 1 (here, a condition is marked as 1 if it is met, and marked as 0 if it is not met), then return a high risk level (which can be represented by "high"); if there is one condition that is 1, then return a low risk level (which can be represented by "low"); if there is no condition that is 1, then return 0. (2) For feature group 4, if there are 4 or more conditions that are 1, then return high; if there are 3 conditions that are 1, then return medium risk level (which can be represented by "medium"); if there are 1 or 2 conditions that are 1, then return low risk level; if there are no conditions that are 1, then return 0. (3) If there are two or more conditions of 1 for feature group 5, then return a low risk level. Select the highest level from all the returned high risk or low risk level information as the result of determining the risk level of this type of failure.

[0090] By judging the fault risk level of each fault mode, the corresponding risk level can be output more accurately while accurately outputting the fault mode, thus improving the early warning effect of fault diagnosis. For example, if the output result is a low-risk diagnosis result for a specific fault mode, it means that an early warning can be achieved for that specific fault mode, which is more conducive to avoiding the more serious consequences that such faults may cause through early warning.

[0091] In step S204, a predetermined wind turbine gearbox fault detection model is constructed based on historical fault data and mapping relationships.

[0092] As an example, regarding copper ring wear, under the characteristic group of decreased inlet pressure, initial rise followed by fall in oil pump outlet pressure, and a rise in the maximum temperature of the second high-speed bearing (referred to as high-speed shaft 2), the corresponding fault mode is: decreased inlet pressure and clogged gearbox oil filter. As another example, regarding the temperature control valve, under the characteristic group of simultaneous rapid increases in inlet temperature, oil sump temperature, low-speed shaft temperature, first high-speed bearing temperature (referred to as high-speed shaft 1), and high-speed shaft 2 temperature, exceeding the threshold under the same operating conditions, while the inlet temperature is not significantly lower than the oil sump temperature, the corresponding fault mode is: temperature control valve failure. As yet another example, regarding gearbox motor pump failure, after grouping oil sump temperatures, if the oil pump outlet pressure jumps by more than 2 bar between 30 and 35 degrees Celsius, the motor pump is normal. If this characteristic group is not met, the corresponding fault mode is: gearbox motor pump failure.

[0093] By using the above method to construct a predetermined wind turbine gearbox fault detection model, it is possible to establish a mapping relationship between features related to gearbox fault detection and fault modes. This not only improves the accuracy of fault / failure mode prediction but also enhances the interpretability of the model output and improves prediction efficiency.

[0094] According to embodiments of this disclosure, the fault mode identification / fault diagnosis method may further include the following processing: for faults in the gearbox body component, the fault risk of the gearbox body component is determined by dividing the gearbox oil pump outlet pressure characteristics, gearbox oil inlet pressure characteristics, gearbox oil sump temperature characteristics, gearbox low-speed shaft temperature characteristics, first gearbox high-speed shaft temperature characteristics, and second gearbox high-speed shaft temperature characteristics into multiple associated fault feature groups, and by satisfying the boundary conditions of each associated feature group in the multiple associated feature groups.

[0095] Here, multiple associated fault feature groups may include the first associated fault feature group to the fifth associated fault feature group, and the first associated fault feature group may include multiple gearbox low-speed shaft temperature features, the second associated fault feature group may include multiple first gearbox high-speed shaft temperature features, the third associated fault feature group may include multiple second gearbox high-speed shaft temperature features, the fourth associated fault feature group may include gearbox oil sump temperature features, multiple gearbox low-speed shaft temperature features, multiple first gearbox high-speed shaft temperature features, and multiple second gearbox high-speed shaft temperature features, and the fifth associated fault feature group may include gearbox oil pump outlet pressure features and gearbox oil inlet pressure features.

[0096] By using the first to fifth associated fault feature groups, faults in gearbox components and their corresponding risks can be accurately diagnosed, improving the operational efficiency of the gearbox. This achieves high-precision, low-false-alarm, early-warning, and low-cost gearbox (e.g., [missing information]) health management, thereby significantly improving the reliability and operational economy of wind turbine units.

[0097] According to this disclosure, the mapping relationship between the key features of the failure mechanism (such as temperature (oil inlet, oil sump, shaft temperature, etc.), oil pressure (oil inlet, oil pump outlet, etc.) and key parameters such as fault indication) and the failure modes is obtained, which enables the model trained based on the input data to accurately predict gearbox-related faults and improve the accuracy of fault diagnosis.

[0098] According to this disclosure, in the process of fault diagnosis, there is no need to install additional vibration sensors or other related monitoring equipment. Instead, the analysis is performed by directly utilizing the existing SCADA data (temperature, pressure, etc.) of the wind turbine. This reduces data dependence, is well adapted to the existing wind turbine system, and thus reduces hardware modification costs.

[0099] According to this disclosure, the intelligent diagnostic scheme based on mechanism and SCADA data, which is driven by the above-described method of this disclosure, solves the industry problems such as the insensitivity of traditional vibration monitoring to lubrication system failure and the lag in early warning of progressive faults. It achieves high-precision, low-false-alarm, early warning and low-cost gearbox health management, thereby significantly improving the reliability and operation and maintenance economy of wind turbine units.

[0100] For example, in diagnosing progressive fault types, oil analysis and pressure-temperature correlation determination can be achieved by combining related factors such as oil pressure and temperature, thereby enabling early warning. Furthermore, by combining failure identification and prediction based on oil pressure and temperature, early warning of lubrication system failures (e.g., long-term poor lubrication reducing lifespan) can be achieved.

[0101] Furthermore, unlike diagnostic methods that output failure probabilities, the solution disclosed herein can display specific failure modes and their corresponding risk levels. For example, an exemplary output result could be: "High-speed shaft bearing wear with a high risk level." Therefore, the solution disclosed herein can achieve precise location of the faulty component, which is beneficial for related maintenance decisions. In addition, the fault diagnosis solution disclosed herein is highly targeted, focusing only on gearbox-related faults (including main body faults and related auxiliary component faults), such as specific failure modes including lubrication systems and gear / bearing wear.

[0102] Unlike existing solutions that involve extracting the mean and variance of the oil pump outlet pressure from SCADA, inputting the data into a predetermined model (e.g., a random forest model), and outputting an "80% probability of anomaly," requiring maintenance personnel to combine other alarm information to guess the cause of the fault (e.g., motor damage or oil circuit blockage), this disclosure determines the matching mapping "pressure did not jump by 2 bar when oil temperature was 30-35℃ → motor pump failure" through a fault detection model, and then directly outputs "gearbox motor pump failure, high risk," and associates it with historical data of similar failures, allowing maintenance personnel to immediately check the motor pump circuit or replace components.

[0103] Next, refer to Figure 4 This invention describes a fault diagnosis apparatus for a wind turbine gearbox according to embodiments of the present disclosure.

[0104] Figure 4 This is a block diagram illustrating a fault diagnosis device 400 for a wind turbine gearbox according to an embodiment of the present disclosure.

[0105] Reference Figure 4 The fault diagnosis device 400 for wind turbine gearbox according to embodiments of the present disclosure may include a data acquisition module 410 and a fault detection module 420.

[0106] According to embodiments of this disclosure, the data acquisition module 410 can perform the following: acquire SCADA data associated with the wind turbine. For example, the SCADA data includes associated data for fault diagnosis of the wind turbine gearbox, and the associated data includes temperature-related data associated with the gearbox and oil pressure-related data associated with the gearbox oil.

[0107] According to embodiments of this disclosure, the fault detection module 420 can perform the following: input SCADA data into a predetermined wind turbine gearbox fault detection model to obtain fault diagnosis results for the wind turbine gearbox. Here, the fault diagnosis results include fault diagnosis results for at least one wind turbine component related to the wind turbine gearbox.

[0108] It should be noted that the operations performed on the above structural frames can be compared with those in the reference section. Figure 1 The related content is similar, so I will not repeat it here.

[0109] Figure 5 This is a block diagram illustrating a computer device 500 according to an embodiment of the present disclosure.

[0110] Reference Figure 5 The computer device 500 according to embodiments of the present disclosure may include a processor 510 and a memory 520. The processor 510 may include (but is not limited to) a central processing unit (CPU), a digital signal processor (DSP), a microcomputer, a field-programmable gate array (FPGA), a system-on-a-chip (SoC), a microprocessor, an application-specific integrated circuit (ASIC), etc. The memory 520 may store computer-executable instructions to be executed by the processor 510. The memory 520 includes high-speed random access memory and / or a non-volatile computer-readable storage medium. When the processor 510 executes the computer-executable instructions stored in the memory 520, the fault mode identification / fault diagnosis method for wind turbine gearboxes described above can be implemented.

[0111] According to embodiments of this disclosure, a wind turbine generator is provided, which includes the computer device 500 as described above.

[0112] The fault mode identification / fault diagnosis method for wind turbine gearboxes according to embodiments of this disclosure can be written as a computer program / instructions to form a computer program product and stored on a computer-readable storage medium. When the computer program / instructions are executed by a processor, the fault mode identification / fault diagnosis method for wind turbine gearboxes as described above can be implemented. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device / server, the electronic device / server is enabled to perform the fault mode identification / fault diagnosis method for wind turbine gearboxes as described above. Examples of computer-readable storage media include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disc storage, hard disk drive (HDD), solid-state drive (SSD), card storage (such as multimedia cards, secure digital (SD) cards, or ultra-fast digital (XD) cards), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state drive, and any other device configured to store computer programs and any associated data, data files, and data structures in a non-transitory manner and to provide the computer programs and any associated data, data files, and data structures to a processor or computer so that the processor or computer can execute the computer programs. In one example, the computer programs and any associated data, data files, and data structures are distributed across a networked computer system, such that the computer programs and any associated data, data files, and data structures are stored, accessed, and executed in a distributed manner through one or more processors or computers.

[0113] The fault mode identification method and system for wind turbine gearboxes according to embodiments of the present disclosure achieve fault diagnosis by using a model trained solely on SCADA data to predict faults in wind turbine gearboxes. This improves the accuracy, efficiency, objectivity, and comprehensiveness of fault identification for wind turbine gearboxes, and enhances the interpretability of the relationship between fault modes and their associated features.

[0114] On the other hand, the fault mode identification / fault diagnosis method for wind turbine gearboxes according to the embodiments of this disclosure can realize the simultaneous diagnosis of failures of multiple components related to the gearbox, which greatly improves the overall system operation and maintenance efficiency.

[0115] On the other hand, by deploying this lightweight fault diagnosis model on edge devices such as wind turbine main control, localized real-time diagnosis can be achieved, reducing data transmission latency and improving response speed. This can greatly improve the efficiency of fault diagnosis and reduce system operation and maintenance costs. In this way, high-precision, low-false-alarm, early warning, and low-cost gearbox health management can be achieved, which can significantly improve the reliability and operation and maintenance economy of wind turbine units.

[0116] While some embodiments of this disclosure have been disclosed and described, those skilled in the art should understand that modifications and variations may be made to these embodiments without departing from the schemes and spirit of this disclosure, which are defined by the claims and their equivalents.

Claims

1. A fault mode identification method for wind turbine gearboxes, characterized in that, The fault mode identification method includes: Acquire SCADA data associated with the wind turbine, the SCADA data including associated data for fault mode identification of the wind turbine gearbox, the associated data including temperature-related data associated with the gearbox and oil pressure-related data associated with the gearbox oil; The SCADA data is input into a predetermined wind turbine gearbox fault detection model to obtain the fault mode identification results of the wind turbine gearbox. The predetermined wind turbine gearbox fault detection model is constructed using the following method: Acquire historical fault data associated with the wind turbine gearbox, including historical temperature data, historical oil pressure data, and historical fault indication data associated with the gearbox; Based on the historical faults, the fault modes associated with the gearbox and the corresponding associated fault characteristics are determined; Based on the aforementioned fault modes and their corresponding associated fault characteristics, the predetermined wind turbine gearbox fault detection model is constructed through the mapping relationship between the two.

2. The fault mode identification method as described in claim 1, characterized in that, The step of constructing the predetermined wind turbine gearbox fault detection model based on the fault mode and the corresponding associated fault features through the mapping relationship between the two includes: Based on the fault modes and the corresponding associated fault features, a mapping relationship between the fault modes and the associated fault features is established. Based on the aforementioned mapping relationship, the predetermined wind turbine gearbox fault detection model is constructed using this mapping relationship. The fault mode identification results include the fault mode identification results of at least one wind turbine component related to the wind turbine gearbox.

3. The fault mode identification method as described in claim 1, characterized in that, The failure modes include gearbox body component failure and gearbox external related component failure. The gearbox body component failure includes at least one of broken teeth failure, tooth surface wear failure, gearbox bearing failure, and gearbox internal copper ring wear failure. The related component failure includes at least one of temperature control valve failure, motor pump failure, poor lubrication failure, oil circuit blockage failure, cooling fan failure, and three-way valve failure.

4. The fault mode identification method as described in claim 3, characterized in that, Gearbox internal copper ring wear, poor lubrication, and oil circuit blockage are all related to the gearbox oil pump outlet pressure characteristics and gearbox oil inlet pressure characteristics. Among them, broken tooth failure, tooth surface wear failure, gearbox bearing failure, and three-way valve failure are all related to the temperature characteristics of the gearbox oil sump, the temperature characteristics of the bearings or teeth at each stage of the multi-stage transmission of the gearbox, and the temperature characteristics of the gearbox oil inlet.

5. The fault mode identification method as described in claim 1, characterized in that, For each failure mode, the failure risk of each failure mode is determined by dividing the multiple associated failure features corresponding to each failure mode into multiple associated failure feature groups and by considering whether the boundary conditions of each associated feature group are satisfied. Among them, the associated fault characteristics include at least one of the following: gearbox oil pump outlet pressure characteristics, gearbox oil inlet pressure characteristics, gearbox oil sump temperature characteristics, temperature characteristics of each bearing or tooth of the gearbox multi-stage transmission, gearbox oil inlet temperature characteristics, wind turbine nacelle acceleration x-axis characteristics, wind turbine nacelle acceleration y-axis characteristics, and gearbox fault indication characteristics.

6. The fault mode identification method as described in claim 1, characterized in that, The fault mode identification method further includes: for gearbox body component faults, dividing the gearbox oil pump outlet pressure characteristics, gearbox oil inlet pressure characteristics, gearbox oil sump temperature characteristics, gearbox low-speed shaft temperature characteristics, first gearbox high-speed shaft temperature characteristics, and second gearbox high-speed shaft temperature characteristics into multiple associated fault feature groups, and determining the fault risk of the gearbox body component fault based on the satisfaction of the boundary conditions of each associated feature group. The plurality of associated fault feature groups include a first associated fault feature group to a fifth associated fault feature group. The first associated fault feature group includes a plurality of gearbox low-speed shaft temperature features, the second associated fault feature group includes a plurality of first gearbox high-speed shaft temperature features, the third associated fault feature group includes a plurality of second gearbox high-speed shaft temperature features, the fourth associated fault feature group includes gearbox oil sump temperature features, a plurality of gearbox low-speed shaft temperature features, a plurality of first gearbox high-speed shaft temperature features, and a plurality of second gearbox high-speed shaft temperature features, and the fifth associated fault feature group includes gearbox oil pump outlet pressure features and gearbox oil inlet pressure features.

7. The fault mode identification method as described in claim 1, characterized in that, The fault mode identification method further includes: The predetermined wind turbine gearbox fault detection model uses a preset classifier and / or a preset neural network model to obtain the fault mode type of the wind turbine gearbox.

8. A system for fault mode identification in wind turbine gearboxes, characterized in that, The system includes: The main controller of the wind turbine is configured to connect to the wind turbine. A switch, located in the tower of the wind turbine, is configured to connect to the main controller; An edge computing device for the wind turbine tower, wherein a predetermined wind turbine gearbox fault detection model is arranged in the edge computing device, and the edge computing device is configured to connect to the main controller via the switch and receive SCADA data from the main controller. Specifically, the predetermined wind turbine gearbox fault detection model is used to identify the SCADA data in order to determine the fault mode of the wind turbine gearbox. The SCADA data includes correlation data used for fault mode identification of wind turbine gearboxes. This correlation data includes temperature-related data associated with the gearbox and oil pressure-related data associated with the gearbox fluid. The predetermined wind turbine gearbox fault detection model is constructed as follows: Historical fault data associated with the wind turbine gearbox is acquired, including historical temperature data, historical oil pressure data, and historical fault indication data associated with the gearbox; the historical fault data is analyzed to determine the fault modes associated with the gearbox and their corresponding associated fault characteristics; a mapping relationship is established between the fault modes and the associated fault characteristics based on the fault modes and their corresponding associated fault characteristics; and the predetermined wind turbine gearbox fault detection model is constructed based on the mapping relationship.

9. The system as described in claim 8, characterized in that, The operation of the edge computing device receiving SCADA data from the main controller includes: The switch receives temperature-related data associated with the gearbox from a temperature sensor associated with the gearbox, and oil pressure-related data associated with the gearbox oil from a pressure sensor associated with the gearbox oil. The temperature sensors associated with the gearbox include at least one of a gearbox oil sump temperature sensor, a gearbox oil inlet temperature sensor, and temperature sensors for each stage of bearings or teeth in a multi-stage gearbox transmission. The pressure sensors associated with the gearbox oil include a gearbox oil pump outlet pressure sensor and / or a gearbox oil inlet pressure sensor.

10. The system as described in claim 9, characterized in that, The gearbox oil sump temperature sensor is arranged adjacent to the gearbox oil sump. The gearbox oil inlet temperature sensor is arranged adjacent to the gearbox oil inlet. Temperature sensors for each stage bearing or tooth in a multi-stage transmission of multiple gearboxes are arranged adjacent to their corresponding bearings or teeth. The gearbox oil pump outlet pressure sensor is arranged adjacent to the gearbox oil pump outlet, and the gearbox inlet pressure sensor is arranged adjacent to the gearbox inlet.

11. A computer device, characterized in that, include: At least one processor; At least one memory that stores computer-executable instructions. The computer-executable instructions, when executed by the at least one processor, cause the at least one processor to perform the fault mode identification method for wind turbine gearboxes as described in any one of claims 1 to 7.

Images