High tower wind turbine generator fault early warning and diagnosis method and system

Abstract

The invention discloses a high-tower wind turbine generator fault early warning and diagnosis method and system. The method comprises the following steps: preprocessing SCADA data of the high-tower wind turbine generator; decomposing the preprocessed SCADA data through an improved EEMD algorithm, extracting feature information of at least two time scales, and splicing I MF components of at least two time points into a feature matrix; inputting the feature matrix into an enhanced vision Transformer (Vi T) model to obtain a prediction result of the real-time state of the wind turbine generator; and calculating a residual matrix based on the prediction result and the SCADA data, and when a root mean square error of the residual matrix exceeds a preset early warning threshold, triggering fault early warning. According to the technical scheme, by improving the EEMD algorithm and enhancing the visual Transform model, the accuracy and real-time performance of fault early warning and diagnosis of the high-tower wind turbine generator system are improved, the intelligent level of the system is enhanced, and the operation and maintenance cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a method and system for early warning and diagnosing faults of a high-tower wind turbine generator set. Background Art

[0002] As an important form of wind power generation, high-tower wind turbines usually have towers over 100 meters high, aiming to utilize the stable wind speed at higher altitudes to improve power generation efficiency. However, as the tower height increases, wind turbines face more complex failure modes and higher maintenance costs during operation. Traditional fault detection methods often rely on manual inspections and regular maintenance, which have problems such as long response time, limited detection range, and high labor costs. Therefore, there is an urgent need for an efficient and intelligent fault warning and diagnosis method to ensure the safe and stable operation of high-tower wind turbines.

[0003] Existing methods often face challenges such as high data noise, difficulty in feature extraction, and high model complexity when processing high-dimensional, multi-scale SCADA data. Therefore, how to effectively extract useful features from SCADA data and use advanced machine learning models for fault diagnosis has become an urgent issue to be solved in the industry. Summary of the invention

[0004] The present invention provides a high-tower wind turbine fault warning and diagnosis method and system, which are used to improve the accuracy and real-time performance of high-tower wind turbine fault warning and diagnosis by improving the EEMD algorithm and strengthening the visual Transformer model, thereby enhancing the intelligence level of the system and reducing the operation and maintenance costs.

[0005] According to a first aspect of the present invention, a high tower wind turbine fault warning and diagnosis method is provided, the high tower wind turbine fault warning and diagnosis method comprising:

[0006] Preprocessing of SCADA data for high tower wind turbines;

[0007] The preprocessed SCADA data is decomposed by the improved EEMD algorithm to extract the characteristic information of at least two time scales, and the IMF components of at least two time points are spliced ​​into a characteristic matrix.

[0008] Inputting the feature matrix into an enhanced visual Transformer (Vi T) model to obtain a prediction result of the real-time state of the wind turbine;

[0009] A residual matrix is ​​calculated based on the prediction result and the SCADA data, and a fault warning is triggered when a root mean square error of the residual matrix exceeds a preset warning threshold;

[0010] By analyzing the abnormal variables in the residual matrix, the fault type of the high-tower wind turbine is identified and the fault diagnosis is performed.

[0011] In one embodiment, the decomposing the pre-processed SCADA data by using the improved EEMD algorithm includes:

[0012] Set a sliding window and perform extreme value extension based on the extreme value points and historical data in the window;

[0013] Dynamically adjust the termination conditions of decomposition and eliminate information leakage through historical data backtracking.

[0014] In one embodiment, it further includes:

[0015] A multi-layer asymmetric convolution module is used to extract local features from the feature matrix, and a deformable attention mechanism is combined to improve the computational efficiency of the model;

[0016] The parameters are optimized through multiple rounds of training using the Adam optimizer and the mean square error (MSE) loss function.

[0017] In one embodiment, the fault warning triggering mechanism includes:

[0018] By analyzing the deviation between the predicted value and the actual value in the residual matrix, the tolerance of the prediction error is dynamically adjusted based on historical data and normal operation status;

[0019] Based on the trend and magnitude of the residual matrix changes, the possibility of computational failure is determined.

[0020] In one embodiment, the identifying the fault type of the high tower wind turbine and performing fault diagnosis includes:

[0021] By analyzing the abnormal patterns in the residual matrix, the faulty variables are determined, and then the fault type of the wind turbine is inferred;

[0022] Based on the fault type, a detailed diagnosis report of the high tower wind turbine fault is output.

[0023] In one embodiment, it further includes:

[0024] The root mean square error (RMSE) calculation formula is as follows:

[0025]

[0026] Among them, y i is the observed value at the i-th moment in the actual SCADA data, is the predicted value, N is the total number of data points, and α is the adjustment coefficient.

[0027] According to a second aspect of the present invention, there is provided a high tower wind turbine fault warning and diagnosis system, comprising:

[0028] A preprocessing module is used to preprocess the SCADA data of high-tower wind turbines;

[0029] An extraction module is used to decompose the pre-processed SCADA data by using an improved EEMD algorithm, extract feature information of at least two time scales, and splice the IMF components of at least two time points into a feature matrix;

[0030] A prediction module, used for inputting the feature matrix into an enhanced visual Transformer (Vi T) model to obtain a prediction result of the real-time state of the wind turbine;

[0031] An early warning module is used to calculate a residual matrix based on the prediction result and the SCADA data, and trigger a fault early warning when a root mean square error of the residual matrix exceeds a preset early warning threshold;

[0032] The diagnosis module is used to identify the fault type of the high-tower wind turbine and perform fault diagnosis by analyzing the abnormal variables in the residual matrix.

[0033] According to a third aspect of the present invention, there is provided an electronic device, the electronic device comprising: a communication interface, a processor, and a memory;

[0034] Wherein, the memory is used to store program instructions, and when the program instructions are executed by the processor that is communicatively connected to the memory through the communication interface, any of the above-mentioned high-tower wind turbine fault warning and diagnosis methods is implemented.

[0035] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a computer (e.g., a processor in a computer), implement any of the above-mentioned high-tower wind turbine fault warning and diagnosis methods.

[0036] In summary, the present invention provides a method and system for early warning and diagnosis of faults of high-tower wind turbines, the method comprising: preprocessing SCADA data of high-tower wind turbines; decomposing the preprocessed SCADA data by an improved EEMD algorithm, extracting feature information of at least two time scales, and splicing the IMF components of at least two time points into a feature matrix; inputting the feature matrix into an enhanced visual Transformer (ViT) model to obtain a prediction result of the real-time state of the wind turbine; calculating a residual matrix based on the prediction result and the SCADA data, and triggering a fault warning when the root mean square error of the residual matrix exceeds a preset warning threshold; identifying the fault type of the high-tower wind turbine and performing fault diagnosis by analyzing abnormal variables in the residual matrix. The technical solution of the present application uses an improved EEMD algorithm to decompose SCADA data and extract multi-scale features, which helps to capture subtle changes in the operating state of the wind turbine and improve the sensitivity of fault detection. Combined with the enhanced visual Transformer model, multi-layer asymmetric convolution modules and deformable attention mechanisms are used to enhance the model's learning ability for complex data patterns and improve the accuracy of fault diagnosis. By analyzing the residual matrix, dynamically adjusting the prediction error tolerance, and combining the change trend analysis, the fault warning strategy is optimized to reduce false alarms and missed alarms and improve the reliability of the warning system. The system automatically identifies the fault type and generates a diagnostic report to assist operation and maintenance personnel to quickly locate the problem, formulate targeted maintenance measures, reduce operation and maintenance costs, and extend equipment life.

[0037] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0038] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0040] Figure 1 A flow chart of a high tower wind turbine fault warning and diagnosis method provided by an embodiment of the present invention;

[0041] Figure 2A flow chart of another high tower wind turbine fault warning and diagnosis method provided by an embodiment of the present invention;

[0042] Figure 3 A flow chart of another high tower wind turbine fault warning and diagnosis method provided by an embodiment of the present invention;

[0043] Figure 4 A flow chart of another high tower wind turbine fault warning and diagnosis method provided by an embodiment of the present invention;

[0044] Figure 5 A flow chart of another high tower wind turbine fault warning and diagnosis method provided by an embodiment of the present invention;

[0045] Figure 6 A structural diagram of a high tower wind turbine fault warning and diagnosis system provided by an embodiment of the present invention;

[0046] Figure 7 A structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.

[0048] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0049] like Figure 1As shown, the present invention provides a high tower wind turbine fault warning and diagnosis method, the high tower wind turbine fault warning and diagnosis method comprising:

[0050] In step S11, the SCADA data of the high tower wind turbine is preprocessed;

[0051] In step S12, the preprocessed SCADA data is decomposed by an improved EEMD algorithm to extract feature information of at least two time scales, and the IMF components of at least two time points are spliced ​​into a feature matrix;

[0052] In step S13, the feature matrix is ​​input into an enhanced visual Transformer (Vi T) model to obtain a prediction result of the real-time state of the wind turbine;

[0053] In step S14, a residual matrix is ​​calculated based on the prediction result and the SCADA data, and when the root mean square error of the residual matrix exceeds a preset warning threshold, a fault warning is triggered;

[0054] In step S15, the fault type of the high-tower wind turbine is identified and fault diagnosis is performed by analyzing the abnormal variables in the residual matrix.

[0055] In one embodiment, with the development of data acquisition technology and artificial intelligence algorithms, fault diagnosis methods based on SCADA (Supervisory Control and Data Acquisition) system data have gradually attracted attention. The SCADA system can monitor the operating status of wind turbines in real time and collect a large amount of operating data. By conducting in-depth analysis of these data, early warning and accurate diagnosis of unit faults can be achieved.

[0056] The SCADA system is a key system for monitoring and data collection during the operation of wind turbines. It can record various operating parameters of wind turbines in real time, such as speed, power, temperature, vibration, etc. However, the original SCADA data often has problems such as noise interference, missing values, and outliers. If the above data quality problems are not properly handled, they will seriously affect the accuracy and reliability of subsequent data analysis and model training. For example, noise data may mask subtle changes in the operating status of wind turbines, making it difficult to effectively extract fault features; missing values ​​and outliers may lead to incompleteness and inconsistency of data sequences, affecting the overall quality and availability of data. Therefore, preprocessing of SCADA data is the basis and prerequisite of the entire fault diagnosis process. Its purpose is to clean and organize the original data into high-quality data sets suitable for subsequent analysis.

[0057] By setting reasonable thresholds and rules, obviously wrong or unreasonable data points can be identified and removed. For example, for wind speed data, if negative values ​​or maximum values ​​that exceed the physical range appear, they can be judged as wrong data and removed. At the same time, some persistent abnormal data segments that are obviously caused by sensor failures should also be marked and cleaned. Use appropriate methods to fill missing data. Common methods include mean filling, median filling, interpolation, etc. For example, for power data that is missing in a short period of time, linear interpolation can be used to fill it based on the data of the adjacent moments before and after to ensure the continuity of the data; for data that is missing for a long time, it may be necessary to combine the correlation of other related parameters for comprehensive estimation and filling. Use statistical analysis methods or machine learning algorithms to identify outliers in the data. For example, based on the Box-Plot method, data points that exceed 1.5 times the interquartile range are regarded as outliers; or machine learning algorithms such as Isolation Forest are used to detect abnormal samples in high-dimensional data. For detected outliers, you can choose to delete, modify or conduct further analysis to determine the cause of their occurrence according to the specific situation.

[0058] EEMD (Ensemble Empirical Mode Decomposition) is an adaptive signal decomposition method that can decompose complex nonlinear and non-stationary signals into multiple intrinsic mode functions (IMFs). Each IMF component represents the characteristics of the signal at different time scales, arranged from high frequency to low frequency, so as to effectively capture the multi-scale change information of the signal. However, the traditional EEMD algorithm has some shortcomings, such as endpoint effect and modal aliasing, which may affect the accuracy and stability of the decomposition results, thereby reducing the effectiveness of fault feature extraction. Therefore, improving the EEMD algorithm is a key link in improving fault diagnosis performance.

[0059] The endpoints of the signal are extended by using methods such as mirror extension and polynomial fitting extension to reduce the decomposition error at the endpoints. For example, for a one-dimensional power signal, mirror symmetric extension is performed at its starting and ending ends respectively, so that the decomposition process can also obtain relatively smooth and accurate IMF components near the endpoints, avoiding the distortion of the decomposition results caused by the endpoint effect. The modal aliasing phenomenon between different IMF components is reduced by optimizing the white noise addition strategy and decomposition parameters. For example, in the ensemble averaging process of EEMD, the amplitude and number of additions of white noise are reasonably adjusted so that each IMF component can more clearly separate the characteristic information of different frequency bands, thereby improving the physical meaning and interpretability of the decomposition results.

[0060] After the improved EEMD decomposition is completed, feature information of at least two time scales is extracted from the multiple IMF components obtained by decomposition. For example, the first few IMF components in the high-frequency band (such as IMF1 and IMF2) are selected to reflect the rapid change characteristics of the wind turbine operating status, such as instantaneous fault impact, etc.; at the same time, the IMF components in the low-frequency band (such as IMF3 and IMF4) are selected to characterize the long-term trend change characteristics of the wind turbine, such as the slow wear and fatigue of the unit. By splicing the IMF components of at least two time points into a feature matrix, the feature information of different time points can be fully utilized to construct a multi-dimensional feature space that comprehensively reflects the operating status of the wind turbine. For example, for the vibration signal of a certain wind turbine, assuming that the improved EEMD decomposition is performed at two time points t1 and t2 respectively, and the corresponding IMF components are obtained, the IMF1, IMF2 at time t1 and the IMF1, IMF2 at time t2 are arranged and combined in a certain order into a feature matrix as the feature vector of the subsequent model input, so that the model can capture the vibration characteristic change pattern of the wind turbine at different times, thereby more accurately predicting its status and identifying potential faults.

[0061] The core advantage of the Transformer model is that it can capture the global dependencies between elements at different positions in the sequence data through the self-attention mechanism, unlike the traditional convolutional neural network (CNN) which is limited by the local receptive field. When processing the multi-dimensional feature sequence data of wind turbines, the global modeling capability is of great significance for accurately grasping the complex relationships between the various components of the wind turbines and the dynamic changes in the operating status. For example, during the operation of a wind turbine, the temperature change of the gearbox may have complex nonlinear correlations with multiple factors such as the load and speed of the generator. The self-attention mechanism of the Transformer model can automatically learn and weight the mutual influence between different factors, so as to more comprehensively understand the operating status characteristics of the wind turbine.

[0062] For the ViT model in the visual field, we have enhanced and improved it to better suit the wind turbine fault diagnosis task. On the one hand, we optimized the input encoding method of the model, effectively reorganized and encoded the multi-dimensional feature matrix of the wind turbine in space, so that it is more in line with the Transformer model's processing requirements for visual features; on the other hand, we adjusted the model's hyperparameters such as depth and width to balance the model's expressiveness and computational complexity, ensuring that the model can run efficiently while ensuring prediction accuracy. For example, in terms of input encoding, the feature matrix originally arranged in time series is processed in blocks, and position embedding vectors are added, so that the model can simultaneously perceive the time sequence and spatial structure information of the features; in terms of model structure adjustment, the performance of the model on the wind turbine fault diagnosis dataset under different depth and width settings is compared through experiments, and finally a model architecture that can fully learn feature representation without overfitting is determined.

[0063] The constructed feature matrix is ​​input into the enhanced visual Transformer (ViT) model. The model can automatically extract the key patterns and rules in the feature matrix through learning and training of a large amount of historical normal operation data and fault data, and then predict the real-time status of the wind turbine. The prediction results are usually output in the form of probability distribution, indicating the possibility of the wind turbine being in different operating states (such as normal state, different types of fault states). For example, assuming that the model outputs that the probability of the wind turbine being in a normal state is 0.8, the probability of being in a gearbox fault state is 0.1, and the probability of being in a generator fault state is 0.1, then according to the principle of maximum probability, it can be preliminarily judged that the wind turbine is currently in a normal operating state, but it also prompts the need to further monitor and analyze the relevant parameters of the gearbox and generator to confirm whether there are potential fault hazards.

[0064] Based on the prediction results of the enhanced visual Transformer (Vi T) model and the original SCADA data, the residual matrix is ​​calculated. Each element in the residual matrix represents the difference between the predicted value and the actual measured value at the corresponding position. The specific calculation method can be a simple difference calculation, or it can be a normalization or other mathematical transformation of the difference to highlight the degree of deviation between the prediction and the actual. For example, for the speed data of the wind turbine, at a certain time t, the speed predicted by the model is ω_pre(t), and the speed actually collected by the SCADA system is ω_real(t), then the corresponding residual value can be expressed as e(t) = ω_real(t)-ω_pre(t), and the residual values ​​of all relevant parameters at different times are arranged in a matrix form to form a residual matrix. The residual matrix can intuitively reflect the inconsistency between the model prediction and the actual operating state, which is often a potential signal of a wind turbine failure.

[0065] When the root mean square error (RMSE) of the residual matrix exceeds the preset warning threshold, a fault warning is triggered. The root mean square error (RMSE) calculation formula is as follows:

[0066]

[0067] Among them, y i is the observed value at the i-th moment in the actual SCADA data, is the predicted value, N is the total number of data points, and α is the adjustment coefficient.

[0068] The setting of the warning threshold is the key to the fault warning link, and it is necessary to comprehensively consider factors such as the fluctuation range of the wind turbine during normal operation, the statistical characteristics of the model prediction error, and the empirical data of actual fault cases. For example, by analyzing a large amount of historical normal operation data, the RMSE distribution of the residual matrix is ​​calculated, its mean and standard deviation are determined, and then the warning threshold is set to the mean plus several times the standard deviation (such as the mean plus 3 times the standard deviation). During the operation of the wind turbine, once the RMSE of the residual matrix exceeds the threshold, it indicates that the operating status of the wind turbine may have changed abnormally, and there may be a risk of failure, which requires further inspection and diagnosis in a timely manner.

[0069] After the fault warning is triggered, the abnormal variables in the residual matrix are analyzed in depth. Abnormal variables usually refer to variables corresponding to parameters whose residual values ​​significantly deviate from the normal range, and are often closely related to the specific fault types of wind turbines. By analyzing the distribution characteristics, change trends and mutual correlations of these abnormal variables, it is possible to preliminarily determine the components or systems where the fault may occur. For example, if the residual values ​​of the gearbox oil temperature, oil pressure and gearbox vibration-related parameters in the residual matrix show large deviations at the same time and show a certain correlation change pattern, such as the oil temperature residual continues to be positive, the oil pressure residual continues to be negative and the vibration residual amplitude increases, it can be inferred that the gearbox may have hidden faults, such as poor lubrication, gear wear and other problems.

[0070] Combined with the mechanism model of wind turbines, historical fault data and expert experience knowledge, the mapping relationship between fault types and abnormal features of residual matrices is established. Machine learning algorithms (such as decision trees, support vector machines, deep neural networks, etc.) or data mining techniques are used to classify and identify abnormal patterns in the residual matrix, so as to accurately diagnose specific fault types. For example, a fault sample database is constructed based on historical fault data, which contains residual matrix feature samples corresponding to different types of faults. These samples are then trained using deep neural networks. When a new fault warning occurs, the current residual matrix is ​​input into the trained neural network model. The model can automatically output the most likely fault type, such as gearbox fault, generator stator winding short circuit fault, etc. At the same time, the severity and development trend of the fault can be further evaluated and analyzed in combination with the mechanism model of wind turbines, providing maintenance personnel with detailed fault diagnosis information and guiding them to perform targeted maintenance operations.

[0071] The fault diagnosis method for high-tower wind turbines builds a systematic, efficient and accurate wind turbine fault diagnosis framework by comprehensively using a series of advanced technical means such as SCADA data preprocessing, improved EEMD algorithm decomposition and feature extraction, enhanced visual Transformer (ViT) model state prediction, residual matrix calculation and fault warning, and fault type identification and diagnosis. It can make full use of the massive SCADA data in the operation process of wind turbines, deeply mine the multi-scale feature information contained in the data, realize the accurate prediction of the real-time state of wind turbines, and the early warning and accurate diagnosis of faults, effectively improve the operating reliability and economy of high-tower wind turbines, and has important practical application value and broad market prospects. In practical applications, this method can be further optimized and customized according to the models, operating environments and specific needs of different wind turbines to better meet the fault diagnosis needs of the wind power industry and promote the healthy and sustainable development of the wind power industry.

[0072] The technical solution in this embodiment uses an improved EEMD algorithm to decompose SCADA data and extract multi-scale features, which helps to capture subtle changes in the operating status of wind turbines and improve the sensitivity of fault detection. Combined with the enhanced visual Transformer model, multi-layer asymmetric convolution modules and deformable attention mechanisms are used to enhance the model's learning ability for complex data patterns and improve the accuracy of fault diagnosis. By analyzing the residual matrix, dynamically adjusting the prediction error tolerance, combining with the change trend analysis, optimizing the fault warning strategy, reducing false alarms and missed alarms, and improving the reliability of the early warning system. The system automatically identifies the fault type and generates a diagnostic report to assist operation and maintenance personnel in quickly locating problems, formulating targeted maintenance measures, reducing operation and maintenance costs, and extending equipment life.

[0073] In one embodiment, Figure 2 As shown, step S12 includes the following steps S21-S22:

[0074] In step S21, a sliding window is set, and extreme value extension is performed based on extreme value points and historical data in the window;

[0075] In step S22, the termination condition of the decomposition is dynamically adjusted to eliminate information leakage by backtracking historical data.

[0076] In one embodiment, in the fault warning and diagnosis process of high-tower wind turbines, when processing SCADA data, an improved empirical mode decomposition (EEMD) algorithm is used to decompose the preprocessed data. The process includes setting a sliding window for extreme value extension and dynamically adjusting the termination condition of the decomposition to eliminate information leakage. When processing time series data, the sliding window technology helps to capture the local characteristics of the data by applying a fixed-size window to the data, moving it step by step and analyzing it. In EEMD decomposition, the window size should be set according to the characteristics of the data and the analysis objectives. A window that is too small may not capture enough contextual information, and a window that is too large may increase the amount of calculation. The step size determines the speed at which the window moves. A smaller step size can obtain more detailed local features, but the amount of calculation is larger; a larger step size has higher calculation efficiency, but may ignore some detailed information.

[0077] In each sliding window, local extreme points are identified, and these extreme points are used for extreme extension to construct upper and lower envelopes. Local maximum and minimum values ​​are identified by comparing the size relationship between adjacent data points. All local maxima (constructing the upper envelope) and all local minima (constructing the lower envelope) are connected using methods such as cubic spline interpolation. The mean of the upper and lower envelopes is calculated as the local trend, and the local oscillation component is obtained by subtracting the mean. By performing the above processing in each window, the local characteristics of the data can be effectively captured, and the accuracy and reliability of EEMD decomposition can be improved.

[0078] In the EEMD decomposition process, it is necessary to set the termination condition to determine when to stop decomposing the signal. Common termination conditions include standard deviation (SD), S-Number criterion and threshold method. Standard deviation means that when the standard deviation of two consecutive decomposition results is less than the preset threshold, the decomposition is stopped. S-Number criterion means that when the number of zero-crossing points and extreme points of consecutive decomposition results is equal or the difference does not exceed 1, the decomposition is stopped. The threshold method refers to setting the change threshold of the mean of the upper and lower envelopes. When the change amplitude is less than the preset threshold, the decomposition is stopped.

[0079] In practical applications, using historical data backtracking to set the decomposition termination condition may lead to information leakage, that is, the model obtains future information during the training process, affecting the prediction performance. To eliminate information leakage, the entire data set is decomposed by EEMD, ensuring that all data use the same number of I MFs, and then divided into training sets, validation sets, and test sets. This can avoid inconsistent numbers of I MFs decomposed on different data sets, resulting in a decrease in model performance. Ensure that only current and historical data are used during model training and validation to prevent future information from affecting the generalization ability of the model.

[0080] Assuming that the SCADA data of a high-tower wind turbine has obvious seasonal fluctuations, when using the sliding window technology for EEMD decomposition, selecting a window size that includes complete seasonal changes can effectively capture seasonal characteristics. By performing extreme value extension in each window, constructing upper and lower envelopes, reducing endpoint effects, and improving decomposition accuracy. During the decomposition process, the termination conditions are dynamically adjusted, such as according to the standard deviation changes of historical data, to ensure the adaptability and accuracy of the decomposition process. Through the above method, information leakage can be effectively eliminated and the reliability of the fault warning model can be improved.

[0081] In the fault warning and diagnosis of high-tower wind turbines, the improved EEMD algorithm is used to decompose the SCADA data. Combining the sliding window and dynamic termination condition adjustment technology, it can effectively extract the multi-scale characteristics of the data and improve the prediction performance of the model. By reasonably setting the sliding window, extreme value extension method and decomposition termination condition, combined with historical data analysis, information leakage can be effectively eliminated to ensure the generalization ability and reliability of the model.

[0082] In one embodiment, Figure 3 As shown, the following steps S31-S32 are also included:

[0083] In step S31, a multi-layer asymmetric convolution module is used to extract local features from the feature matrix, and a deformable attention mechanism is combined to improve the computational efficiency of the model;

[0084] In step S32, the Adam optimizer and the mean square error (MSE) loss function are used to perform parameter optimization through multiple rounds of training.

[0085] In one embodiment, in a high-tower wind turbine fault diagnosis method, a multi-layer asymmetric convolution module is used to extract local features from a feature matrix. An asymmetric convolution module generally refers to a convolution kernel having different sizes in different directions. For example, a combination of 1×3 and 3×1 convolution kernels can expand the receptive field of the convolution operation without increasing too much computational effort, and capture richer local feature information. In the wind turbine fault diagnosis scenario, the feature matrix contains multi-dimensional operating parameter data from different sensors, such as temperature, pressure, vibration, etc. Local feature patterns may exist at different time scales and spatial dimensions, and these local features can be extracted layer by layer through multi-layer asymmetric convolution modules. For example, for the vibration signal feature matrix of a wind turbine gearbox, the asymmetric convolution module can capture the tiny impact features during the gear meshing process and the periodic fluctuation features of the shaft rotation. These local features are often sensitive indicators of early faults.

[0086] The traditional attention mechanism needs to perform comprehensive two-to-two interactive calculations on the entire feature sequence during the calculation process. Although it can capture global dependencies, it has a huge amount of calculation and low efficiency when processing large-scale feature data. The deformable attention mechanism adjusts the sampling position by introducing a learnable offset, so that the attention calculation can be more focused on the key feature area, reducing unnecessary computational overhead. In the application of wind turbine fault diagnosis, feature information with high fault correlation can be screened out more quickly from a large amount of SCADA data feature matrix, thereby improving the diagnosis speed. For example, when processing the feature matrix composed of multi-point vibration monitoring data of wind turbine blades, the deformable attention mechanism can quickly locate the specific blade position and its corresponding vibration frequency component that may have problems, without having to perform the same degree of complex calculations on all data points, thereby achieving a significant improvement in computational efficiency.

[0087] Using the Adam (Adaptive Moment Estimation) optimizer to optimize model parameters is a key step in this fault diagnosis method. The Adam optimizer is an optimization algorithm based on adaptive learning rate. It combines the advantages of traditional stochastic gradient descent (SGD) and momentum acceleration characteristics, and can automatically adjust the learning rate according to different parameters. In the training process of the wind turbine fault diagnosis model, facing complex nonlinear feature mapping relationships and a large amount of training data, the Adam optimizer can quickly converge to a better parameter solution. For example, in the early stage of model training, the Adam optimizer can quickly adjust parameters with a large learning rate to rapidly reduce the model loss function; as the training progresses, it will automatically reduce the learning rate and make fine parameter adjustments to avoid oscillations caused by excessive learning rates near the minimum value of the loss function, thereby ensuring the stability and efficiency of model training.

[0088] The mean square error (MSE) loss function is used as the objective function of model training to measure the difference between the model prediction output and the actual wind turbine operating status label. In the fault diagnosis scenario, the MSE loss function can intuitively reflect the model's prediction accuracy for the normal operating status and various fault states of the wind turbine. Through multiple rounds of training, the model parameters are continuously optimized to minimize the MSE loss function value, so that the model can gradually learn the precise mapping relationship between the feature matrix and the wind turbine operating status. For example, during the training process, if the error between the model's predicted value and the actual value for a certain type of fault state is large, the MSE loss function will give a larger loss value, prompting the Adam optimizer to adjust the model parameters so that the subsequent prediction results are constantly close to the actual situation, ultimately improving the accuracy and reliability of the entire fault diagnosis model.

[0089] In the high-tower wind turbine fault diagnosis method of this embodiment, by adopting a multi-layer asymmetric convolution module combined with a deformable attention mechanism, efficient local feature extraction and calculation acceleration of the wind turbine operating state feature matrix are achieved; at the same time, the multi-round training parameter optimization strategy based on the Adam optimizer and the MSE loss function ensures that the model can accurately learn complex fault modes, thereby improving the overall performance of wind turbine fault diagnosis. This innovative technical solution provides strong support for the intelligent maintenance and reliable operation of wind turbines, and can effectively reduce fault downtime and improve power generation efficiency and economic benefits in practical applications.

[0090] In one embodiment, Figure 4 As shown, step S14 includes the following steps S41-S42:

[0091] In step S41, by analyzing the deviation between the predicted value and the actual value in the residual matrix, the tolerance of the prediction error is dynamically adjusted based on historical data and normal operation status;

[0092] In step S42, the possibility of calculation failure is implemented according to the variation trend and amplitude of the residual matrix.

[0093] In one embodiment, in the high tower wind turbine fault warning trigger mechanism, it is necessary to deeply analyze historical data, which contains rich information about the operation of wind turbines under different working conditions, including the normal operating parameter range under various environmental conditions (such as different wind speeds, wind directions, temperatures, etc.) and abnormal data characteristics corresponding to previous fault cases. Through the mining and statistical analysis of a large amount of historical data, a relatively reasonable prediction error benchmark range can be preliminarily determined.

[0094] The operating status of wind turbines is not static. Even under similar external environments, the parameters of wind turbines during normal operation will show certain dynamic changes due to factors such as wear and aging of the turbines themselves and short-term load fluctuations. Therefore, it is necessary to continuously collect and evaluate data under normal operating conditions in order to capture these subtle changes in a timely manner and provide a real-time basis for dynamically adjusting the prediction error tolerance.

[0095] For example, when analyzing the power output data of a wind turbine, it is assumed that historical data shows that within a certain wind speed range, the normal fluctuation range of power output is ±5%. However, in actual operation, with seasonal changes or slight wear of internal components of the unit, the power fluctuation during normal operation may gradually expand to ±7%. At this time, if the fault is still judged based on the fixed tolerance of ±5% of historical data, frequent false alarms may occur. Therefore, by dynamically adjusting the tolerance in combination with real-time normal operation data and appropriately relaxing it to ±7%, the actual operating characteristics of the current wind turbine can be more accurately reflected, avoiding unnecessary misjudgments.

[0096] Wind turbines operate in a complex and changeable natural environment, and are affected by factors such as wind and sand erosion, rain erosion, and temperature changes for a long time. Their performance and status will gradually change. In addition, the operating load of wind turbines is not constant and will be adjusted in real time with factors such as wind speed and grid demand. The mechanism of dynamically adjusting the tolerance of prediction errors can respond to these changes in a timely manner. When the wind turbine is in a stable operation period, the fluctuations of various parameters are small and stable within a certain range, the tolerance can be appropriately tightened to improve the sensitivity of fault warning so as to detect small abnormal changes in time. On the contrary, when the wind turbine has just completed maintenance or the operating environment has changed significantly (such as entering a season with frequent wind speed fluctuations), the parameters may fluctuate significantly in the short term. At this time, the tolerance should be appropriately relaxed to avoid misjudging the normal adjustment process as a fault. For example, for the gearbox oil temperature parameters of a wind turbine, in the high temperature environment of summer, the normal range of the oil temperature may increase, and the tolerance for prediction errors should also be increased accordingly; and in the period after maintenance, due to the different degrees of running-in of various components, the oil temperature may fluctuate in the short term. At this time, the dynamic adjustment of the tolerance can effectively avoid false alarms while not missing any real signs of failure.

[0097] The residual matrix contains the deviation information between the predicted value and the actual value. Through the comprehensive analysis of its change trend and amplitude, the possibility of fault occurrence can be accurately calculated. The change trend reflects the direction and rate of evolution of the deviation over time. For example, if the residual value continues to increase, it indicates that the operating state of the wind turbine may be gradually deviating from the normal track and there is a potential risk of failure; the amplitude indicates the absolute size of the deviation. A larger residual amplitude means that the difference between the current state and the normal mode is more significant, and the possibility of failure is higher. Specifically, when the residual value of a key parameter of the wind turbine (such as the generator speed) is monitored to show an upward trend at multiple consecutive time points, and the increase exceeds a certain threshold, it can be considered that the system corresponding to the parameter may be abnormal. For example, the residual value gradually rises from a level close to zero to 1.5 times the upper limit of the tolerance, and the upward trend is still continuing, which strongly suggests that there may be problems with the control mechanism of the generator speed, such as control algorithm parameter drift, actuator failure, etc.

[0098] When calculating the actual fault probability, it is necessary to comprehensively consider various factors such as the residual characteristics of different parameters in the residual matrix, the correlation between the parameters, and the operating conditions of the wind turbine. Different parameters have different sensitivities to the operating status of the wind turbine. The abnormality of some parameters may directly indicate a serious fault, while the slight fluctuation of some parameters may only be a random disturbance of normal operation. For example, for the gearbox vibration parameters and oil temperature parameters, the residual change of the vibration parameters may more directly reflect the mechanical failure of the gearbox, so it has a higher weight in the corresponding fault probability calculation; while the residual change of the oil temperature parameter may be affected by many factors (such as ambient temperature, load, etc.), and the weight is relatively low. At the same time, the correlation between the gearbox and other components (such as generators, blades, etc.) must also be considered. When the gearbox is abnormal, it may have a chain reaction on the operation of the generator. Therefore, these correlation effects need to be comprehensively evaluated when calculating the fault probability.

[0099] In addition, the operating conditions of wind turbines will also affect the judgment of fault probability. Under high-load operation, the residual tolerance of certain parameters can be appropriately relaxed, and the calculation threshold of fault probability can also be adjusted accordingly to adapt to complex operating environments.

[0100] The high-tower wind turbine fault warning trigger mechanism in the technical solution of this embodiment can effectively adapt to the dynamic changes in the wind turbine operating state and improve the accuracy and reliability of fault warning by dynamically adjusting the prediction error tolerance and calculating the fault probability based on the residual matrix change trend and amplitude. It not only fully taps the value of historical data and normal operation data, but also comprehensively considers a variety of actual operation factors, providing strong support for early warning and preventive maintenance of wind turbine faults, helping to reduce fault downtime, ensuring the safe and stable operation of wind turbines, and improving the overall economic benefits and operation management level of wind farms.

[0101] In one embodiment, Figure 5 As shown, step S15 includes the following steps S51-S52:

[0102] In step S51, by analyzing the abnormal pattern in the residual matrix, the variable with the fault is determined, and then the fault type of the wind turbine is inferred;

[0103] In step S52, based on the fault type, a detailed diagnosis report of the high tower wind turbine fault is output.

[0104] In one embodiment, in the fault diagnosis process of high-tower wind turbines, the residual matrix is ​​the core data basis. Through the systematic analysis of the residual matrix, the variable with the fault can be accurately located, and the specific fault type can be further inferred. Abnormal patterns in the residual matrix are usually manifested in the significant deviation of the residual value from the normal range, specific change trends, and abnormal correlations with other variable residuals. For example, when the gearbox of a wind turbine fails, the residual values ​​of the gearbox oil temperature, oil pressure, and vibration-related parameters may increase significantly at the same time, and the change trends of these residual values ​​may show a certain correlation, such as the oil temperature residual continues to rise, the oil pressure residual fluctuates and decreases, and the vibration residual amplitude increases and the frequency component changes. These abnormal patterns are in sharp contrast to the residual characteristics of the normal operation of the gearbox, so they can be effectively identified.

[0105] If an abnormal pattern is detected in the residual matrix, the specific variable that has failed can be determined. This is achieved by setting reasonable thresholds and judgment rules. For example, for the residual sequence of each monitored variable, its statistical characteristics such as mean and standard deviation are calculated. When the residual value exceeds the mean plus several times the standard deviation, it can be judged that the variable has failed. After determining the fault variable, the possible fault type is inferred based on the mechanism model of the wind turbine, the fault mode database, and expert experience knowledge. Different fault types often correspond to specific variable combinations and abnormal patterns. For example, if it is monitored that the stator temperature residual of the generator is continuously high, the stator current residual fluctuates periodically, and the rotor vibration residual increases, it can be inferred that the generator may have a stator winding short circuit fault.

[0106] Based on the determined fault type, a high tower wind turbine fault diagnosis report is generated. The report usually includes fault overview, fault detailed analysis, fault impact assessment, and maintenance recommendations.

[0107] The fault overview briefly describes the time when the fault occurred, the wind turbine unit number, the fault type, and the preliminary judgment of the fault severity, so that maintenance personnel can quickly understand the overall situation of the fault.

[0108] The detailed fault analysis is to elaborate on the abnormal performance of each relevant variable when the fault occurs, including the changing trend and amplitude of the residual value and the correlation analysis with other variables; combined with the operating principle and structure of the wind turbine, analyze the specific causes that may lead to the fault, such as component wear, poor lubrication, influence of external environmental factors, etc.; provide a description of the process of fault development to help maintenance personnel fully understand the evolution mechanism of the fault.

[0109] Fault impact assessment is to evaluate the potential impact of faults on wind turbine power generation efficiency, operational stability, other components, and possible chain reactions to the operation of the entire wind farm. For example, a gearbox failure may cause a drop in power generation, vibration transmission may cause fatigue damage to adjacent components, and may even force the wind turbine to shut down for maintenance, affecting the overall power generation and economic benefits of the wind farm.

[0110] The maintenance recommendations provide specific maintenance measures based on the type and severity of the fault, including a list of parts that need to be replaced, maintenance operation steps, tools and equipment required for maintenance, and safety precautions. At the same time, maintenance time estimates and maintenance cost estimates are provided to enable wind farm operators to make reasonable maintenance plan arrangements and resource allocation.

[0111] Fault diagnosis reports should be presented in a combination of intuitive and easy-to-understand language and charts. For example, use a line chart to show the changing trend of residual values, use a table to list fault variables and their abnormal characteristics, and use a flow chart to represent fault inference logic and maintenance steps, so that maintenance personnel can quickly and accurately obtain key information and put it into practice. In addition, the report should also have a certain degree of traceability and verifiability. Record the data source, analysis method and judgment basis based on the fault diagnosis process so that they can be verified and validated in the subsequent maintenance process; at the same time, provide data support for subsequent fault case analysis and model optimization, and continuously improve the performance and accuracy of the fault diagnosis system.

[0112] The high-tower wind turbine fault type identification and diagnosis report generation mechanism in the technical solution of this embodiment can accurately locate the fault variables and infer the fault type by deeply analyzing the abnormal patterns in the residual matrix, and then generate a fault diagnosis report with detailed content and clear structure. It can not only provide clear fault information and maintenance guidance for wind turbine maintenance personnel, which helps to repair faults in time, reduce downtime and reduce maintenance costs, but also improve the intelligence level and reliability of the wind turbine fault diagnosis system by continuously accumulating fault diagnosis cases and optimizing diagnosis models, providing solid guarantee for the safe and stable operation of high-tower wind turbines and the sustainable development of the wind power industry.

[0113] In one embodiment, Figure 6 FIG. 1 is a block diagram of a high tower wind turbine fault warning and diagnosis system according to an exemplary embodiment. Figure 6 As shown, the high tower wind turbine fault warning and diagnosis system includes a preprocessing module 61, an extraction module 62, a prediction module 63, a warning module 64 and a diagnosis module 65.

[0114] The preprocessing module 61 is used to preprocess the SCADA data of the high tower wind turbine;

[0115] The extraction module 62 is used to decompose the pre-processed SCADA data by using the improved EEMD algorithm, extract feature information of at least two time scales, and splice the IMF components of at least two time points into a feature matrix;

[0116] The prediction module 63 is used to input the feature matrix into the enhanced visual Transformer (Vi T) model to obtain the prediction result of the real-time state of the wind turbine;

[0117] The early warning module 64 is used to calculate a residual matrix based on the prediction result and the SCADA data, and trigger a fault early warning when the root mean square error of the residual matrix exceeds a preset early warning threshold;

[0118] The diagnosis module 65 is used to identify the fault type of the high-tower wind turbine and perform fault diagnosis by analyzing the abnormal variables in the residual matrix.

[0119] The preprocessing module 61, the extraction module 62, the prediction module 63, the warning module 64 and the diagnosis module 65 included in the high tower wind turbine fault warning and diagnosis system block diagram are controlled to execute the high tower wind turbine fault warning and diagnosis method described in any of the above embodiments.

[0120] like Figure 7 As shown, the present invention provides an electronic device 700, which includes: a communication interface, a processor 701, and a memory 702;

[0121] The memory 702 is used to store program instructions. When the program instructions are executed by the processor 701 that is communicatively connected to the memory 702 through the communication interface, the SCADA data of the high-tower wind turbine is preprocessed; the preprocessed SCADA data is decomposed by an improved EEMD algorithm, feature information of at least two time scales is extracted, and IMF components of at least two time points are spliced ​​into a feature matrix; the feature matrix is ​​input into an enhanced visual Transformer (Vi T) model to obtain a prediction result of the real-time state of the wind turbine; a residual matrix is ​​calculated based on the prediction result and the SCADA data, and a fault warning is triggered when the root mean square error of the residual matrix exceeds a preset warning threshold; and the fault type of the high-tower wind turbine is identified and fault diagnosis is performed by analyzing abnormal variables in the residual matrix.

[0122] The present invention provides a computer-readable storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, SCADA data of a high-tower wind turbine is preprocessed; the preprocessed SCADA data is decomposed by an improved EEMD algorithm, feature information of at least two time scales is extracted, and IMF components of at least two time points are spliced ​​into a feature matrix; the feature matrix is ​​input into an enhanced visual Transformer (Vi T) model to obtain a prediction result of a real-time state of the wind turbine; a residual matrix is ​​calculated based on the prediction result and the SCADA data, and a fault warning is triggered when a root mean square error of the residual matrix exceeds a preset warning threshold; and the fault type of the high-tower wind turbine is identified and the fault diagnosis is performed by analyzing abnormal variables in the residual matrix.

[0123] It should be understood that the specific features, operations and details described hereinabove about the method of the present invention may also be similarly applied to the device and system of the present invention, or, vice versa. In addition, each step of the method of the present invention described above may be performed by the corresponding parts or units of the device or system of the present invention.

[0124] It should be understood that each module / unit of the system of the present invention can be implemented in whole or in part by software, hardware, firmware or a combination thereof. Each module / unit can be embedded in a processor of a computer device or independent of a processor in the form of hardware or firmware, or can be stored in a memory of a computer device in the form of software for the processor to call to perform the operation of each module / unit. Each module / unit can be implemented as an independent component or module, or two or more modules / units can be implemented as a single component or module.

[0125] In one embodiment, a computer device is provided, which includes a memory and a processor, and the memory stores computer instructions executable by the processor, and the computer instructions instruct the processor to execute each step of the method of the embodiment of the present invention when executed by the processor. The computer device can be a server, a terminal, or any other electronic device with necessary computing and / or processing capabilities in a broad sense. In one embodiment, the computer device may include a processor, a memory, a network interface, a communication interface, etc. connected through a system bus. The processor of the computer device can be used to provide necessary computing, processing and / or control capabilities. The memory of the computer device may include a non-volatile storage medium and an internal memory. An operating system, a computer program, etc. may be stored in or on the non-volatile storage medium. The internal memory can provide an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface and the communication interface of the computer device can be used to connect and communicate with external devices through a network. The steps of the method of the present invention are executed by the processor.

[0126] The present invention may be implemented as a computer-readable storage medium having a computer program stored thereon, which causes the steps of the method of an embodiment of the present invention to be executed when executed by a processor. In one embodiment, the computer program is distributed on a plurality of computer devices or processors coupled to a network so that the computer program is stored, accessed, and executed in a distributed manner by one or more computer devices or processors. A single method step / operation, or two or more method steps / operations, may be performed by a single computer device or processor or by two or more computer devices or processors. One or more method steps / operations may be performed by one or more computer devices or processors, and one or more other method steps / operations may be performed by one or more other computer devices or processors. One or more computer devices or processors may perform a single method step / operation, or perform two or more method steps / operations.

[0127] Those skilled in the art will appreciate that the method steps of the present invention can be completed by instructing related hardware such as a computer device or a processor through a computer program, and the computer program can be stored in a non-transitory computer-readable storage medium, and when the computer program is executed, the steps of the present invention are executed. Depending on the circumstances, any reference to memory, storage, database or other medium herein may include non-volatile and / or volatile memory. Examples of non-volatile memory include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EPROM), flash memory, magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state disk, etc. The SCADA data is decomposed using the improved EEMD algorithm to extract multi-scale features, which helps to capture subtle changes in the operating status of the wind turbine and improve the sensitivity of fault detection. Combined with the enhanced visual TRANSFORMER model, multi-layer asymmetric convolution modules and deformable attention mechanisms are used to enhance the model's learning ability for complex data patterns and improve the accuracy of fault diagnosis. By analyzing the residual matrix, dynamically adjusting the prediction error tolerance, combined with the change trend analysis, the fault warning strategy is optimized, false alarms and missed alarms are reduced, and the reliability of the warning system is improved. The system automatically identifies the fault type and generates a diagnostic report to assist operation and maintenance personnel to quickly locate the problem, formulate targeted maintenance measures, reduce operation and maintenance costs, and extend the life of the equipment. Examples of volatile memory include random access memory (RAM), external cache memory, etc.

[0128] The various technical features described above can be combined arbitrarily. Although all possible combinations of these technical features are not described, any combination of these technical features should be considered to be covered by this specification as long as there is no contradiction in such combination.

[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A high tower wind turbine fault warning and diagnosis method, characterized in that: include: Preprocessing of SCADA data for high tower wind turbines; The preprocessed SCADA data is decomposed by the improved EEMD algorithm to extract the characteristic information of at least two time scales, and the IMF components of at least two time points are spliced ​​into a characteristic matrix; Inputting the feature matrix into an enhanced visual Transformer (ViT) model to obtain a prediction result of the real-time state of the wind turbine; A residual matrix is ​​calculated based on the prediction result and the SCADA data, and a fault warning is triggered when a root mean square error of the residual matrix exceeds a preset warning threshold; By analyzing the abnormal variables in the residual matrix, the fault type of the high-tower wind turbine is identified and the fault diagnosis is performed.

2. The high tower wind turbine fault warning and diagnosis method according to claim 1, characterized in that: The decomposition of the pre-processed SCADA data by the improved EEMD algorithm includes: Set a sliding window and perform extreme value extension based on the extreme value points and historical data in the window; Dynamically adjust the termination conditions of decomposition and eliminate information leakage through historical data backtracking.

3. The high tower wind turbine fault warning and diagnosis method according to claim 1, characterized in that: Also includes: A multi-layer asymmetric convolution module is used to extract local features from the feature matrix, and a deformable attention mechanism is combined to improve the computational efficiency of the model; The parameters are optimized through multiple rounds of training using the Adam optimizer and the mean square error (MSE) loss function.

4. The high tower wind turbine fault warning and diagnosis method according to claim 1, characterized in that: The fault warning triggering mechanism includes: By analyzing the deviation between the predicted value and the actual value in the residual matrix, the tolerance of the prediction error is dynamically adjusted based on historical data and normal operation status; Based on the trend and magnitude of the residual matrix changes, the possibility of computational failure is determined.

5. The high tower wind turbine fault warning and diagnosis method according to claim 1, characterized in that: The method of identifying the fault type of a high-tower wind turbine and performing fault diagnosis comprises: By analyzing the abnormal patterns in the residual matrix, the faulty variables are determined, and then the fault type of the wind turbine is inferred; Based on the fault type, a detailed diagnosis report of the high tower wind turbine fault is output.

6. The high tower wind turbine fault warning and diagnosis method according to claim 1, characterized in that: Also includes: The root mean square error (RMSE) calculation formula is as follows: Among them, y i is the observed value at the i-th moment in the actual SCADA data, is the predicted value, N is the total number of data points, and α is the adjustment coefficient.

7. A high tower wind turbine fault warning and diagnosis system, characterized in that: include: A preprocessing module is used to preprocess the SCADA data of high-tower wind turbines; An extraction module is used to decompose the preprocessed SCADA data by using an improved EEMD algorithm, extract characteristic information of at least two time scales, and splice the IMF components of at least two time points into a characteristic matrix; A prediction module, used for inputting the feature matrix into an enhanced visual Transformer (ViT) model to obtain a prediction result of the real-time state of the wind turbine; An early warning module is used to calculate a residual matrix based on the prediction result and the SCADA data, and trigger a fault early warning when a root mean square error of the residual matrix exceeds a preset early warning threshold; The diagnosis module is used to identify the fault type of the high-tower wind turbine and perform fault diagnosis by analyzing the abnormal variables in the residual matrix.

8. The high tower wind turbine fault warning and diagnosis system according to claim 7, characterized in that: The preprocessing module, the extraction module, the prediction module, the warning module and the diagnosis module are controlled to execute the high-tower wind turbine fault warning and diagnosis method according to any one of claims 1-6.

9. An electronic device, characterized in that: include: Communication interface, processor, memory; Wherein, the memory is used to store program instructions, and when the program instructions are executed by the processor that is communicatively connected to the memory through the communication interface, the electronic device implements the high tower wind turbine fault warning and diagnosis method described in any one of claims 1 to 6.

10. A computer-readable storage medium having program instructions stored thereon, characterized in that: When the program instructions are executed by a computer, the computer implements the high-tower wind turbine fault warning and diagnosis method according to any one of claims 1 to 6.

Images