Wind turbine generator fault positioning method, electronic device and storage medium

Abstract

The present application relates to the technical field of wind turbine fault positioning, and specifically provides a wind turbine generator fault positioning method, an electronic device and a storage medium, aiming to solve the technical problem that in the prior art, the fault is mostly positioned by staff, thereby resulting in low accuracy of the fault positioning method. To this end, the wind turbine generator fault positioning method of the present application comprises: obtaining input variables related to the wind turbine generator; inputting the measured values of the input variables into a trained generator state monitoring model to obtain reconstructed values of the input variables; determining a reconstruction error based on the reconstructed values and the measured values of the input variables; and positioning the fault of the wind turbine generator based on the reconstruction error. In this way, the accuracy of the wind turbine generator fault positioning is improved, and the stability of the wind turbine generator is ensured.

Description

Technical Field

[0001] This invention relates to the field of wind turbine generator fault location technology, specifically providing a wind turbine generator fault location method, electronic equipment, and storage medium. Background Technology

[0002] Wind turbines are mostly installed in remote areas such as high mountains, deserts, and offshore, operating in harsh environments with complex and variable forces. Long-term operation under variable speed and load conditions easily leads to failures in critical components such as generators, causing downtime. As an indispensable key component of the wind turbine's drivetrain system, the generator's performance directly affects the performance and reliability of the drivetrain and even the entire turbine. Therefore, timely and effective fault location of the generator to ensure the safe, reliable, and efficient operation of the unit is an effective way to reduce downtime, lower maintenance costs, and extend the unit's service life. However, current technologies mostly rely on manual fault location, resulting in low accuracy and failing to meet practical needs.

[0003] Accordingly, there is a need in this field for a new fault location scheme for wind turbine generators to solve the above problems. Summary of the Invention

[0004] To overcome the aforementioned deficiencies, this invention is proposed to provide solutions or at least partially solve the above-mentioned technical problems. This invention provides a method for locating faults in a wind turbine generator, electronic equipment, and a storage medium.

[0005] In a first aspect, the present invention provides a method for fault location of a wind turbine generator, the method comprising: acquiring input variables related to the wind turbine generator; inputting the measured values ​​of the input variables into a trained generator state monitoring model to obtain reconstructed values ​​of the input variables; determining a reconstruction error based on the reconstructed values ​​and measured values ​​of the input variables; and locating the fault of the wind turbine generator based on the reconstruction error.

[0006] In one implementation, the input variables related to the wind turbine generator include wind speed, generator active power, generator reactive power, generator speed, actual torque, generator drive end bearing temperature, generator non-drive end bearing temperature, generator stator U-phase coil temperature, generator stator V-phase coil temperature, generator stator W-phase coil temperature, grid-side three-phase voltage, and grid-side three-phase current.

[0007] In one embodiment, the reconstruction error includes the reconstruction error of the input variable and the overall reconstruction error of the generator condition monitoring model; determining the reconstruction error based on the reconstructed value and the measured value of the input variable includes: determining the reconstruction error of the input variable based on the absolute value of the residual between the reconstructed value and the measured value; and determining the overall reconstruction error of the generator condition monitoring model based on the reconstruction error of the input variable.

[0008] In one embodiment, fault location of the wind turbine generator based on the reconstruction error includes: calculating the correlation coefficient between the overall reconstruction error of the generator condition monitoring model and the reconstruction error of the input variable; determining abnormal state parameters of the wind turbine generator based on the correlation coefficient; and locating the fault location of the generator based on the abnormal state parameters of the wind turbine generator.

[0009] In one embodiment, determining the abnormal state parameters of the wind turbine generator based on the correlation coefficient includes: using the input variable corresponding to the maximum absolute value of the correlation coefficient as the abnormal state parameter of the wind turbine generator.

[0010] In one embodiment, the method further includes: monitoring the generator faults of the wind turbine based on the reconstruction error and alarm threshold, and extracting fault samples; and diagnosing early faults of the wind turbine generator based on the fault samples.

[0011] In one embodiment, extracting the fault sample includes: when the reconstruction error exceeds the alarm threshold, acquiring the status data after the alarm and using the status data as the fault sample.

[0012] In one implementation, diagnosing early faults of a wind turbine generator based on the fault samples includes: inputting the fault samples into a trained generator fault diagnosis model to obtain the final fault diagnosis result.

[0013] In a second aspect, an electronic device is provided, comprising a processor and a storage device adapted to store a plurality of program codes adapted to be loaded and executed by the processor to perform the wind turbine generator fault location method described in any of the preceding claims.

[0014] In a third aspect, a computer-readable storage medium is provided, wherein a plurality of program codes are stored therein, the program codes being adapted to be loaded and run by a processor to perform the wind turbine generator fault location method described in any of the preceding claims.

[0015] The above-described technical solutions of the present invention have at least one or more of the following beneficial effects:

[0016] The wind turbine generator fault location method of this invention first acquires input variables related to the wind turbine generator. Then, the measured values ​​of these input variables are input into a trained generator condition monitoring model to obtain reconstructed values. Next, the reconstruction error is determined based on the reconstructed and measured values. Finally, the fault location of the wind turbine generator is performed based on the reconstruction error. In this way, the reconstruction error can be determined using the reconstructed values ​​output by the generator condition monitoring model, thereby locating the early fault location of the wind turbine generator. Using a network model as the medium for determining the fault location further improves the accuracy of fault location. Attached Figure Description

[0017] The disclosure of this invention will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Furthermore, similar numbers in the drawings are used to denote similar components, wherein:

[0018] Figure 1 This is a schematic flowchart of the main steps of a wind turbine generator fault location method according to an embodiment of the present invention;

[0019] Figure 2 This is a flowchart illustrating a method for fault monitoring and fault diagnosis of a wind turbine generator in one embodiment;

[0020] Figure 3 This is a scatter plot of wind speed and power under normal operating conditions of a wind turbine in one embodiment.

[0021] Figure 4 This is a histogram of reconstruction error and reconstruction error frequency for the test set of the generator condition monitoring model (LSTM-DAE) in one embodiment;

[0022] Figure 5 This is a schematic diagram of an early fault detection method for a generator in one embodiment;

[0023] Figure 6 This is a schematic diagram of the confusion matrix of early fault diagnosis results of a generator based on a generator fault diagnosis model (XGBoost model) in one embodiment;

[0024] Figure 7 This is a schematic diagram of the structure of an electronic device in one embodiment. Detailed Implementation

[0025] Some embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0026] In the description of this invention, "module" and "processor" can include hardware, software, or a combination of both. A module can include hardware circuitry, various suitable sensors, communication ports, memory, and may also include software components, such as program code, or a combination of software and hardware. A processor can be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and / or signal processing capabilities. The processor can be implemented in software, in hardware, or a combination of both. Non-transitory computer-readable storage media includes any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" means all possible combinations of A and B, such as only A, only B, or A and B. The terms "at least one A or B" or "at least one of A and B" have a similar meaning to "A and / or B" and can include only A, only B, or A and B. The singular terms "a" or "this" can also include plural forms.

[0027] Currently, most existing technologies rely on staff to locate faults in wind turbine generators, resulting in low accuracy of fault location methods that are difficult to meet practical needs.

[0028] To address this, this application proposes a method, electronic device, and storage medium for fault location of wind turbine generators. First, input variables related to the wind turbine generator are acquired. Then, the measured values ​​of these input variables are input into a trained generator condition monitoring model to obtain reconstructed values ​​of the input variables. Next, the reconstruction error is determined based on the reconstructed and measured values ​​of the input variables. Finally, the fault location of the wind turbine generator is performed based on the reconstruction error. In this way, the reconstruction error can be determined using the reconstructed values ​​output by the generator condition monitoring model, thereby locating the early fault location of the wind turbine generator. Using a network model as the medium for determining the fault location further improves the accuracy of the identified fault location.

[0029] See appendix Figure 1 , Figure 1 This is a schematic flowchart of the main steps of a wind turbine generator fault location method according to an embodiment of the present invention.

[0030] like Figure 1 As shown, the wind turbine generator fault location method in this embodiment of the invention mainly includes the following steps S101-S103.

[0031] Step S101: Obtain the input variables related to the wind turbine generator.

[0032] In one specific implementation, the input variables related to the wind turbine generator include wind speed, generator active power, generator reactive power, generator speed, actual torque, generator drive end bearing temperature, generator non-drive end bearing temperature, generator stator U-phase coil temperature, generator stator V-phase coil temperature, generator stator W-phase coil temperature, grid-side three-phase voltage, and grid-side three-phase current.

[0033] Step S102: Input the measured values ​​of the input variables into the trained generator state monitoring model to obtain the reconstructed values ​​of the input variables.

[0034] In one embodiment, an LSTM Seq2Seq network is introduced on the basis of the noise reduction autoencoder to construct a generator state monitoring model based on a long short-term memory noise reduction autoencoder (LSTM-DAE), but it is not limited to this.

[0035] After constructing the generator condition monitoring model, the model is further trained based on the training dataset and the test dataset to obtain the trained generator condition monitoring model.

[0036] By inputting the measured values ​​of the input variables into the trained generator condition monitoring model, the reconstructed values ​​of the input variables can be obtained.

[0037] Step S103: Determine the reconstruction error based on the reconstructed value and the measured value of the input variable.

[0038] In one specific implementation, the reconstruction error includes the reconstruction error of the input variable and the overall reconstruction error of the generator condition monitoring model; determining the reconstruction error based on the reconstructed value and the measured value of the input variable includes: determining the reconstruction error of the input variable based on the absolute value of the residual between the reconstructed value and the measured value; and determining the overall reconstruction error of the generator condition monitoring model based on the reconstruction error of the input variable.

[0039] Specifically, after inputting the measured values ​​of the input variables into the generator condition monitoring model, reconstructed values ​​can be obtained. The absolute value of the residual between the reconstructed values ​​and the measured values ​​is then determined and used as the reconstruction error of the input variables. Furthermore, the square root of the square of the reconstruction errors of each input variable is taken as the overall reconstruction error of the generator condition monitoring model.

[0040] Step S104: Based on the reconstruction error, locate the fault in the wind turbine generator.

[0041] In one specific implementation, fault location of the wind turbine generator based on the reconstruction error includes: calculating the correlation coefficient between the overall reconstruction error of the generator condition monitoring model and the reconstruction error of the input variable; determining the abnormal state parameters of the wind turbine generator based on the correlation coefficient; and locating the fault location of the generator based on the abnormal state parameters of the wind turbine generator.

[0042] Specifically, the Spearman correlation coefficient can be used to calculate the correlation coefficient between the overall reconstruction error of the generator condition monitoring model and the reconstruction error of the input variables.

[0043] For example, when the input variables are 16 variables, including wind speed, generator active power, generator reactive power, generator speed, actual torque, generator drive-end bearing temperature, generator non-drive-end bearing temperature, generator stator U-phase coil temperature, generator stator V-phase coil temperature, generator stator W-phase coil temperature, grid-side three-phase voltage, and grid-side three-phase current, the Spearman correlation coefficient is used to calculate the overall reconstruction error σ of the generator condition monitoring model and the reconstruction errors σ1, σ2, ..., σ of the 16 input variables. 16 The correlation coefficient between them, where the Spearman correlation coefficient is calculated using the following formula:

[0044]

[0045] In the formula, n is the sequence length, and d i Grade difference. Generally, it is considered that if 1 ≥ |r s |≥0.8, strong correlation; 0.8>|r s |≥0.5, moderate correlation; 0.5>|r s |≥0.3, low correlation;|r s |<0.3, weak correlation.

[0046] In one specific implementation, determining the abnormal state parameters of the wind turbine generator based on the correlation coefficient includes: using the input variable corresponding to the maximum absolute value of the correlation coefficient as the abnormal state parameter of the wind turbine generator.

[0047] Specifically, the input variable corresponding to the maximum absolute value of the correlation coefficient is the abnormal state parameter of the wind turbine generator.

[0048] Further, the fault location of the generator can be determined based on the abnormal state parameters of the wind turbine generator. In one embodiment, the corresponding part of the generator corresponding to the abnormal state parameters can be identified as the fault location, but this is not limited to this. In some complex scenarios, users or staff can further locate or troubleshoot the generator fault location based on the abnormal state parameters of the wind turbine generator. This can further improve the accuracy of fault location and thus enhance the safety and reliability of the wind turbine generator.

[0049] Based on steps S101-S104 above, firstly, input variables related to the wind turbine generator are acquired. Then, the measured values ​​of these input variables are input into the trained generator condition monitoring model to obtain reconstructed values. Next, the reconstruction error is determined based on the reconstructed and measured values. Finally, the wind turbine generator is fault located based on the reconstruction error. In this way, the reconstruction error can be determined using the reconstructed values ​​output by the generator condition monitoring model, thereby locating early fault locations in the wind turbine generator. Using a network model as the medium for determining the fault location further improves the accuracy of the identified fault location.

[0050] In addition, existing research on condition monitoring and fault diagnosis of key components of wind turbines mainly includes the following two methods: (1) physical model-based methods; (2) data-driven methods (vibration, oil, sound signals and SCADA operation data). However, as wind turbines become larger and more complex, it will be difficult to establish physical models of key components of wind turbines. And methods based on vibration signals, oil and sound signals require additional hardware equipment for acquisition, storage and transmission, which is difficult to promote on a large scale. Based on SCADA operation data, the method of using machine learning and other methods to build a normal behavior model of wind turbine generators and combining it with predictive residual analysis to realize fault detection and fault diagnosis of wind turbine generators is currently a research hotspot. Domestic and foreign scholars have carried out a lot of research on the fault warning and diagnosis of key equipment such as generators and gearboxes from the aspects of condition monitoring and fault diagnosis, but the following problems still exist:

[0051] (1) Existing status monitoring methods either consider multiple status parameters of the equipment but ignore the timing of SCADA data, or consider the timing but a single status parameter cannot fully and accurately reflect the operating status of the equipment, and fail to fully integrate the two features.

[0052] (2) Fault diagnosis methods based on a single model cannot fully extract fault information, resulting in low fault diagnosis accuracy, weak generalization ability and poor stability when performing model transfer.

[0053] Thus, this application can also perform fault detection and diagnosis on generators through the following embodiments, thereby further improving the safety and reliability of wind turbine generators.

[0054] In one specific embodiment, the method further includes: monitoring the generator faults of the wind turbine based on the reconstruction error and alarm threshold, and extracting fault samples; and diagnosing early faults of the wind turbine generator based on the fault samples.

[0055] In one embodiment, the overall reconstruction error of the generator condition monitoring model can be used as an example of the reconstruction error shown, but is not limited thereto.

[0056] Specifically, fault detection and diagnosis of wind turbine generators are performed using reconstruction errors and alarm thresholds. Specifically, faults in wind turbine generators are monitored based on reconstruction errors and alarm thresholds, fault samples are extracted, and further, early-stage faults in wind turbine generators are diagnosed based on these fault samples.

[0057] In one specific implementation, extracting the fault sample includes: when the reconstruction error exceeds the alarm threshold, acquiring the status data after the alarm and using the status data as the fault sample.

[0058] Specifically, when the reconstruction error is detected to exceed the alarm threshold, the status data after the alarm is obtained and used as a fault sample.

[0059] In one embodiment, the probability density function of the reconfiguration error can be calculated using a kernel density estimation method. Based on the probability density function and a set confidence level, the alarm threshold for generator faults can be obtained. Since the method of further determining the alarm threshold using kernel density estimation is a conventional technique, it will not be elaborated upon here.

[0060] In one specific implementation, diagnosing early faults of wind turbine generators based on the fault samples includes: inputting the fault samples into a trained generator fault diagnosis model to obtain the final fault diagnosis result.

[0061] In one embodiment, an XGBoost ensemble learning algorithm can be used to construct an XGBoost multi-class generator fault diagnosis model, which is then trained to obtain a well-trained generator fault diagnosis model. Fault samples are then input into the trained generator fault diagnosis model to obtain the final fault diagnosis result.

[0062] The above methods can significantly improve the performance, fault detection accuracy and reliability of the wind turbine generator condition monitoring model. They can effectively solve the problems of low early warning accuracy, insufficient early warning time and difficulty in obtaining fault samples in existing studies. They have high reconstruction accuracy, superior fault early warning capability and better diagnostic accuracy.

[0063] In one specific embodiment, Figure 2 The document demonstrates the process of fault detection, diagnosis, and location methods for wind turbine generators, which will be explained in detail through the following steps S201-S209.

[0064] S201: Based on the distribution characteristics of abnormal data in the wind speed-power scatter plot of wind turbines, the quartile method is used to remove abnormal data caused by fluctuations in operating conditions, curtailment, and hardware failures in data acquisition and transmission. First, data with obvious abnormal states are directly removed: 1. Normal shutdown, idling, or startup data of wind turbines with wind speeds less than 3 m / s or greater than 25 m / s; 2. Abnormal shutdown data of wind turbines with wind speeds greater than 3 m / s and power less than or equal to 0. Next, the quartile method is used to remove abnormal data caused by curtailment, fluctuations in operating conditions, and hardware failures in data acquisition and transmission. Finally, through data cleaning, the operating data of wind turbines under normal operating conditions (healthy state) is obtained, as follows: Figure 3 As shown.

[0065] S202: Based on the obtained operating data of the wind turbine under normal operating conditions (healthy state), and according to the operating principle of the wind turbine, 16 monitoring variables that can reflect the operating status of the generator are selected as input variables for the model, including wind speed, generator active power, generator reactive power, generator speed, actual torque, generator drive end bearing temperature, generator non-drive end bearing temperature, generator stator U-phase coil temperature, generator stator V-phase coil temperature, generator stator W-phase coil temperature, grid-side three-phase voltage, and grid-side three-phase current.

[0066] S203: After data cleaning and feature selection, the training and test sets for the generator condition monitoring model are obtained. The training and test set data are then normalized to eliminate the dimensional influence between heterogeneous data from multiple sources in the wind turbine SCADA system, thereby reducing the difficulty of model training. The normalization formula is shown below:

[0067]

[0068] S204: Based on the noise reduction autoencoder, an LSTM Seq2Seq network is introduced to construct a generator state monitoring model based on a long short-term memory noise reduction autoencoder (LSTM-DAE). Figure 2The training process of the proposed generator condition monitoring model is shown in the figure. The proposed generator condition monitoring model is trained and tested based on the training set and test set obtained by S203.

[0069] S205: Based on the generator condition monitoring model and test set trained in step S204, obtain the distribution law of reconstruction error under normal generator operation (health state). Calculate the probability density function of the test set reconstruction error using the kernel density estimation method. Based on the probability density function and the set confidence level, obtain the generator fault alarm threshold. Figure 4 The reconstruction error of the generator condition monitoring model (LSTM-DAE) test set and the frequency histogram of reconstruction error are shown.

[0070] S206: Based on the generator condition monitoring model trained in step S204 and the fault alarm threshold calculated in step S205, online condition monitoring of the wind turbine generator is performed to detect early generator faults. Based on the detected generator faults, a fault sample dataset is constructed for generator fault diagnosis in the next stage. Figure 2 The process of online condition monitoring of the proposed generator condition monitoring model is demonstrated in the paper. Figure 5 The paper presents a real-world case study based on wind turbine generator failures and demonstrates the validation results of the proposed generator condition monitoring model. The results show that the proposed generator condition monitoring model has the capability to provide early warning of generator failures and can detect failures in advance and issue alarms.

[0071] S207: Based on the extracted fault sample dataset, an XGBoost multi-class generator fault diagnosis model is constructed using the XGBoost ensemble learning algorithm. Figure 2 The training process of the proposed generator fault diagnosis model is shown in the figure.

[0072] S208: Based on the generator fault diagnosis model trained in step S207, perform online fault diagnosis on the wind turbine generator to achieve early fault diagnosis of the generator. Figure 6 The paper presents a real-world case study based on wind turbine generator failures and demonstrates the validation results of the proposed generator fault diagnosis model. The results show that the proposed generator fault diagnosis model has superior fault diagnosis capabilities and high accuracy and reliability in fault diagnosis.

[0073] S209: Based on the generator condition monitoring model (LSTM-DAE) trained in S204 using the Long Short-Term Memory Denoising Autoencoder (LSTM-DAE), obtain the model output (reconstructed values ​​of the 16 input variables such as wind speed) from the model inputs (16 input variables). Then, the overall reconstruction error σ of the monitoring model (LSTM-DAE) and the reconstruction errors σ1, σ2, ..., σ of each input variable can be calculated. 16Based on the Spearman correlation coefficient, the overall reconstruction error σ and the errors of each reconstruction component σ1, σ2, ..., σ are calculated. 16 The correlation coefficient between the variables is used to determine the input variable corresponding to the maximum absolute value of the correlation coefficient. This value corresponds to the abnormal state parameter of the wind turbine generator, which can then be used to further locate the fault location of the generator. The formula for calculating the Spearman correlation coefficient is as follows:

[0074]

[0075] In the formula, n is the sequence length, and d i Grade difference. Generally, it is considered that if 1 ≥ |r s |≥0.8, strong correlation; 0.8>|r s |≥0.5, moderate correlation; 0.5>|r s |≥0.3, low correlation;|r s |<0.3, weak correlation.

[0076] The quartile method removes outlier noise points and accumulation points from wind turbine operation data. Based on the wind turbine operating principle, monitoring variables reflecting generator operating status are selected as input variables for the generator condition monitoring model. A normal generator behavior model is constructed using a noise-reducing autoencoder and a long short-term memory network to reconstruct normal generator operation data and obtain the reconstruction error under healthy generator conditions. Statistical methods are used to explore the distribution law of reconstruction error under healthy generator conditions. The probability density function of reconstruction error is calculated using kernel density estimation, and a fault alarm threshold is set to achieve online detection of early generator faults. Based on the detected generator faults, a fault sample dataset is constructed for the next stage of generator fault diagnosis. Based on the extracted fault sample dataset, an XGBoost multi-class generator fault diagnosis model is constructed using the XGBoost ensemble learning algorithm. Based on the Spearman correlation coefficient, abnormal generator state parameters can be accurately determined, further locating the fault location. This significantly improves the performance, accuracy, and reliability of the generator condition monitoring model, effectively solving the problems of low early warning accuracy, insufficient early warning time, and difficulty in obtaining fault samples in existing research. It has high reconstruction accuracy, superior fault early warning capability, and better diagnostic accuracy.

[0077] It should be noted that although the steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effects of the present invention, different steps do not necessarily have to be executed in such an order. They can be executed simultaneously (in parallel) or in other orders, and these variations are all within the scope of protection of the present invention.

[0078] Those skilled in the art will understand that all or part of the processes in the method of the above embodiment of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium can include any entity or device capable of carrying the computer program code, a medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content included in the computer-readable storage medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0079] Furthermore, the present invention also provides an electronic device. In one embodiment of the electronic device according to the present invention, such as Figure 7 As shown, the electronic device includes a processor 71 and a storage device 72. The storage device can be configured to store a program for executing the wind turbine generator fault location method of the above-described method embodiments. The processor can be configured to execute the program in the storage device, which includes, but is not limited to, a program for executing the wind turbine generator fault location method of the above-described method embodiments. For ease of explanation, only the parts related to the embodiments of the present invention are shown. For specific technical details not disclosed, please refer to the method section of the embodiments of the present invention.

[0080] Furthermore, the present invention also provides a computer-readable storage medium. In one embodiment of the computer-readable storage medium according to the present invention, the computer-readable storage medium can be configured to store a program for executing the wind turbine generator fault location method of the above-described method embodiments. This program can be loaded and run by a processor to implement the above-described wind turbine generator fault location method. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The computer-readable storage medium can be a storage device device comprising various electronic devices. Optionally, in the embodiments of the present invention, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0081] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for locating faults in a wind turbine generator, characterized in that, The method includes: Obtain the input variables related to the wind turbine generator; The measured values ​​of the input variables are input into the trained generator condition monitoring model to obtain the reconstructed values ​​of the input variables; The reconstruction error is determined based on the reconstructed and measured values ​​of the input variables. The reconstruction error includes the reconstruction error of the input variables and the overall reconstruction error of the generator condition monitoring model. Determining the reconstruction error based on the reconstructed and measured values ​​of the input variables includes: determining the reconstruction error of the input variables based on the absolute value of the residual between the reconstructed and measured values; and determining the overall reconstruction error of the generator condition monitoring model based on the reconstruction error of the input variables. Based on the reconstruction error, fault location is performed on the wind turbine generator, including: Calculate the correlation coefficient between the overall reconstruction error of the generator condition monitoring model and the reconstruction error of the input variables; Based on the correlation coefficient, the abnormal state parameters of the wind turbine generator are determined; The fault location of the generator is located based on the abnormal state parameters of the wind turbine generator.

2. The method for locating faults in a wind turbine generator according to claim 1, characterized in that, The input variables related to the wind turbine generator include wind speed, generator active power, generator reactive power, generator speed, actual torque, generator drive end bearing temperature, generator non-drive end bearing temperature, generator stator U-phase coil temperature, generator stator V-phase coil temperature, generator stator W-phase coil temperature, grid-side three-phase voltage, and grid-side three-phase current.

3. The method for locating faults in a wind turbine generator according to claim 1, characterized in that, Determining abnormal state parameters of wind turbine generators based on the correlation coefficient includes: using the input variable corresponding to the maximum absolute value of the correlation coefficient as the abnormal state parameter of the wind turbine generator.

4. The method for locating faults in a wind turbine generator according to claim 1, characterized in that, The method further includes: Based on the reconstruction error and alarm threshold, the generator faults of the wind turbine are monitored, and fault samples are extracted. Early faults in wind turbine generators are diagnosed based on the aforementioned fault samples.

5. The method for locating faults in a wind turbine generator according to claim 4, characterized in that, The step of extracting the fault sample includes: when the reconstruction error exceeds the alarm threshold, acquiring the status data after the alarm and using the status data as the fault sample.

6. The wind turbine generator fault location method according to claim 4, characterized in that, Diagnosing early faults of wind turbine generators based on the fault samples includes: inputting the fault samples into a trained generator fault diagnosis model to obtain the final fault diagnosis result.

7. An electronic device comprising a processor and a storage device, said storage device being adapted to store a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by the processor to perform the wind turbine generator fault location method according to any one of claims 1 to 6.

8. A computer-readable storage medium storing a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by a processor to perform the wind turbine generator fault location method according to any one of claims 1 to 6.

Images