Method for evaluating health state of large components of cabin of wind turbine generator in blade icing state

Abstract

The invention provides a method for evaluating the health state of a large part of a cabin of a wind turbine generator in a blade icing state. Relates to the technical field of wind turbine generator health state evaluation, in particular to a health state evaluation technology for wind turbine generator cabin large components in a blade icing state. The method comprises the following steps: constructing a multi-dimensional comprehensive feature matrix through feature extraction and time alignment processing of SCADA (Supervisory Control And Data Acquisition) data, voiceprint and vibration signals of a large cabin component of a wind turbine generator under a cold weather condition; the meteorological parameters and the blade icing dynamic characteristics are combined, a Transform network is adopted to achieve blade icing state identification, and a comprehensive characteristic matrix under the blade icing state is output; a Transform-KAN model is innovatively constructed, a Transform network is utilized to extract health features of large components of a cabin in a blade icing state, and health state classification is completed through the KAN network. According to the method, multi-source data are fused, deep learning and feature extraction technologies are applied, the influence of blade icing on part health is accurately evaluated, and a basis is provided for operation and maintenance decision making.

Description

Technical Field

[0001] The present invention relates to the field of wind power generation technology, and in particular to a method for evaluating the health status of large components in a wind turbine nacelle when blades are frozen. Specifically, the present invention collects data from different sensors and combines data processing technology to comprehensively evaluate the health status of large components in the nacelle of a wind turbine, and can timely predict and diagnose failures or performance degradation of wind turbines when blades are frozen, providing effective maintenance decision support. Background Art

[0002] With the widespread use of wind turbines in cold regions, the problem of blade icing under low temperature weather conditions has become increasingly prominent. Blade icing not only significantly reduces the operating efficiency of wind turbines, but may also cause unit failure. Therefore, accurately assessing the health of wind turbines when blades are frozen and taking corresponding measures in a timely manner to prevent major losses have become key technical problems that the wind power industry needs to overcome.

[0003] At present, most of the methods for evaluating the health status of key components of wind turbines rely on a single data source, or only integrate acoustic and vibration signals, and fail to effectively integrate with other data in the SCADA system. For example, vibration sensors can detect vibration changes caused by blade icing, but their sensitivity and accuracy are easily affected by factors such as wind speed and rotation speed in complex meteorological conditions, resulting in deviations in the diagnosis of icing conditions. Although data such as wind speed and unit power in the SCADA system can reflect the overall operation of the unit, they cannot directly reflect the icing condition of the blade surface, and the accuracy of their data may also be affected under extreme climatic conditions. Therefore, existing methods fail to effectively combine multiple data sources and cannot comprehensively and accurately evaluate the impact of blade icing on the health status of key components of wind turbines.

[0004] Most of the current health status assessment models are based on data analysis under conventional meteorological conditions, and lack in-depth research on the health status of large nacelle components under cold meteorological conditions, especially under blade icing conditions. The impact of blade icing on wind turbine operation is not limited to the blades themselves, but may also have a negative impact on other important components in the nacelle, such as gearboxes and generators. However, most existing studies only focus on the blade icing phenomenon itself, and rarely involve comprehensive health assessments of large nacelle components under icing conditions. Under blade icing conditions, traditional health assessment models have limitations, mainly reflected in the failure to fully integrate multi-feature data and the failure to effectively integrate the complex relationship between multi-source information such as soundprints and vibrations and large nacelle components, resulting in limited accuracy and credibility of the evaluation results. Summary of the invention

[0005] The purpose of the present invention is to provide a method for evaluating the health status of large components of a wind turbine nacelle when blades are in an icy state, so as to solve the problems raised in the above-mentioned background technology.

[0006] In view of the above technical problems, the present invention proposes the following technical solutions: a method for evaluating the health status of large components in the nacelle of a wind turbine when the blades are in an icy state. The health status evaluation method of the present invention mainly includes the following contents: data collection and fusion of large components in the nacelle under cold weather conditions, automatic identification of the blade icing state, and establishment of a health status evaluation model for the blades in an icy state. The method includes the following steps:

[0007] S1 First, obtain the SCADA (Supervisory Control and Data Acquisition) data, voiceprints, and vibration signals of the large cabin components (main shaft, gearbox, generator, blades, etc.) of the wind turbine under cold weather conditions. Secondly, the SCADA data is cleaned and standardized, including denoising, standardization, missing data filling, and outlier detection to ensure data quality and provide a reliable basis for subsequent analysis. Then, the voiceprint and vibration signals are denoised, standardized, and feature extracted, mainly extracting frequency domain features, time domain features, and statistical features to reflect the operating characteristics of the wind turbine under different working conditions, and extracting key information from the data. Finally, the processed SCADA data, voiceprints, and vibration signals are feature fused, mainly using fusion methods such as weighted averaging and dimensionality reduction mechanisms, and fully combining information from different data sources to generate a comprehensive feature matrix of the same time scale, thereby improving the accuracy and robustness of subsequent model predictions.

[0008] S2 inputs the comprehensive feature matrix obtained in S1, meteorological data (such as temperature, humidity, wind speed, etc.) and blade surface icing characteristics into a pre-trained blade icing identification model to obtain a blade icing identification result. The blade icing identification model is based on a Transformer network, and utilizes the comprehensive feature matrix obtained in S1, combined with meteorological conditions and blade surface icing characteristics, to effectively distinguish between the blade icing state and the normal working state. Specifically, the model learns the feature distribution under each state based on the comprehensive feature matrix obtained in S1, so as to determine whether there is icing on the blade. The output of the model is the determination result of the blade icing state and the comprehensive feature matrix under the blade icing state, and provides a basis for subsequent health status assessment.

[0009] S3 inputs the comprehensive feature matrix under the blade icing state obtained in S2 into the Transformer-KAN model for health status assessment. The Transformer-KAN model classifies the health status of large components in the nacelle of the wind turbine, wherein the health status characteristics of large components in the nacelle are extracted based on the Transformer network, and the extracted health status characteristics are specifically classified based on the KAN (Kolmogorov-Arnold Networks) model. And evaluate whether there is a risk of failure of large components in the nacelle due to blade icing. Subsequently, the model assigns a health status score to each large component in the nacelle. The scoring result reflects the health status of large components in the nacelle under the blade icing state. The scoring levels can be divided into health, minor failure, moderate failure and severe failure. The evaluation results provide data support for subsequent operation and maintenance decisions.

[0010] Furthermore, the SCADA data in S1 mainly include gearbox oil temperature, gearbox bearing temperature, average wind speed, generator bearing temperature, generator power, speed, etc.; the soundprint and vibration signals include main shaft vibration, gearbox vibration, generator vibration, cabin soundprint, blade soundprint, etc.

[0011] Furthermore, frequency domain features, time domain features and statistical features are extracted from the voiceprint and vibration signal in S1. Among them, the time domain features include mean, root mean square, kurtosis, etc., which are used to reflect the volatility, amplitude and change of the signal; the frequency domain features include spectrum, main frequency, frequency bandwidth, etc., which can reveal the vibration mode, frequency component and potential fault frequency of the equipment; the statistical features include peak factor, frequency domain entropy, signal energy, etc., which can reveal the stability, complexity and potential fault mode of the equipment.

[0012] Furthermore, the features from different signal sources (SCADA data, voiceprints, and vibration signals) in S1 are fused to improve the accuracy and robustness of model prediction. Common fusion methods include weighted averaging and KPCA (Kernel Principal Component Analysis) dimensionality reduction. Weighted averaging generates a comprehensive feature vector by assigning weights to different signal sources and combining their respective importance. It is suitable for situations where signal sources contribute unevenly; while KPCA dimensionality reduction maps data from high-dimensional space to low-dimensional space through nonlinear mapping, extracts the most representative features, reduces redundant information, and improves computational efficiency. These fusion methods can effectively combine the advantages of multi-source data, provide comprehensive and accurate feature representation, and enhance the performance of subsequent analysis and prediction.

[0013] Furthermore, the SCADA data after feature fusion in S1, as well as the voiceprint and vibration signals, are time-aligned to ensure the consistency of these data in the time dimension. This process is achieved by synchronizing the timestamps of different data sources, thereby generating time-aligned multi-source fusion data, providing a unified time reference for subsequent deep learning models.

[0014] Furthermore, the ice formation characteristics of the blade surface in S2 are obtained by the following technical means: the ice layer type is divided into frost ice, transparent ice or mixed ice by using infrared thermal imager and visible light camera images through convolutional neural network (CNN); the ice layer thickness is directly measured by the ultrasonic sensor array on the leading edge of the blade, or dynamically estimated by combining environmental parameters and Macklin ice growth model; the infrared reflectivity is extracted by near-infrared spectrometer to obtain the mean and standard deviation of the reflectivity in the 800-1200nm band; the acoustic characteristics are analyzed from the soundprint signal to obtain the energy attenuation rate and spectral entropy in the 2-5kHz frequency band; the surface texture is quantified by the root mean square value of the high-frequency component (5-10kHz) of the vibration signal and the wavelet packet energy entropy.

[0015] Furthermore, when constructing the blade icing identification model described in S2, the core of the Transformer architecture is the multi-head self-attention mechanism, which can process different parts of the input data in parallel to capture long-distance dependencies and local features in the data. By decomposing the input data into multiple "attention heads", each head focusing on different aspects of the input sequence, the model can more fully understand the complex relationship between the comprehensive feature data. In this way, the model can efficiently process the comprehensive feature matrix and extract feature information related to the blade icing state.

[0016] In order to retain the position information in the input sequence, the model introduces positional encoding. Positional encoding adds a fixed vector to each position, allowing the model to perceive the position relationship in the sequence. The introduction of positional encoding ensures that the model can retain time order information when processing sequence data, which is particularly important for dynamic monitoring of blade icing status.

[0017] In each layer of the encoder, in addition to the multi-head self-attention mechanism, there is also a feed-forward neural network (FFN). FFN performs the same fully connected operation on the feature vector at each position to further extract high-level feature representations. Through FFN, the model can perform nonlinear transformations on the output of the multi-head self-attention mechanism, thereby enhancing the model's expressiveness.

[0018] During the model training process, the cross entropy loss function is used to measure the difference between the predicted results and the true labels. The cross entropy loss function can effectively evaluate the classification performance of the model and optimize the model parameters by minimizing the difference between the predicted probability distribution and the true label. In order to efficiently update the model parameters, the Adam (Adaptive MomentEstimation) optimizer is used. The Adam optimizer combines the advantages of momentum and adaptive learning rate, and can converge quickly and stabilize the training process. By adjusting the learning rate and hyperparameters, the model can achieve a high accuracy within a limited training time.

[0019] Through the above design, the Transformer network can effectively process the comprehensive feature matrix and effectively identify the blade icing state and normal working state. The model learns the relationship between the comprehensive feature data through the self-attention mechanism, retains the sequence information in combination with the position encoding, and extracts high-level features through the feedforward neural network. This architecture not only improves the expressiveness of the model, but also enhances its adaptability to complex data.

[0020] Furthermore, in S3, the health status feature extraction is performed on the comprehensive feature matrix through the Transformer network. The Transformer encoder captures the global dependencies in the data through the multi-head self-attention mechanism and FFN to generate context-related feature representations. The multi-head self-attention mechanism divides the input data into multiple "heads" for parallel calculation, and the feature representations learned by each head are then spliced ​​and linearly transformed, so that the features of different subspaces in the multi-source fusion data can be captured. FFN further processes these feature representations and increases the nonlinear ability of the model through two linear transformations and a nonlinear activation function. In addition, internal covariate shift is reduced by layer normalization, model convergence is accelerated, and overfitting is prevented by the Dropout layer. In order to extract global features and reduce feature dimensions, the output feature map of the Transformer is pooled into a fixed-size feature vector through average pooling. The pooled feature map is flattened into a one-dimensional vector for input into the subsequent fully connected layer.

[0021] Furthermore, based on the features extracted by the Transformer network in S3, the KAN model is further introduced to classify the health status. The KAN model performs nonlinear mapping through basic linear transformation and piecewise polynomial basis function (B-spline) to improve the expressiveness and interpretability of the model. The KAN model classifies the extracted health status features into healthy, mild fault, moderate fault and severe fault. In the KAN model, the input features are first mapped to the output space through linear transformation to increase the linear expression ability of the model.

[0022] The input features are mapped to multiple piecewise polynomial spaces through the B-spline basis function. Each piecewise polynomial can capture the nonlinear relationship of the input data in different intervals.

[0023] Through the combination of basic activation functions and piecewise polynomial basis functions, the input features are nonlinearly mapped to generate richer feature representations. Finally, the KAN model outputs a vector, each element of which represents the predicted probability that the sample belongs to the corresponding health state.

[0024] Thus, the output vector of the Transformer-KAN model is obtained. Each element represents the predicted probability that the sample belongs to the corresponding health state.

[0025] Furthermore, in order to comprehensively evaluate the performance of the Transformer-KAN model in S3, indicators such as accuracy, precision, recall, F1 score and confusion matrix are used. These indicators can measure the performance of the model in blade icing recognition and health status classification tasks from multiple perspectives. Through learning rate scheduling and regularization technology, combined with the Simulated Firefly Optimization Algorithm (SFOA) for optimization, the model training process can be further optimized to prevent overfitting and accelerate convergence.

[0026] Furthermore, the health status of each sample is determined based on the predicted probability output by the Transformer-KAN model described in S3. Samples are classified into specific health status categories, including health, minor fault, moderate fault, and major fault, by setting thresholds or using classification strategies (such as maximum probability selection). This classification result provides a scientific basis for operation and maintenance decisions and ensures the efficient operation and maintenance of wind turbines.

[0027] The beneficial effects of adopting this patent are: by integrating SCADA data with voiceprints and vibration signals, the impact of blade icing on large components of the wind turbine nacelle is fully considered, and the accuracy and reliability of health status assessment are improved. The introduction of Transformer network and KAN model can effectively process complex multi-source data and improve the expressiveness and interpretability of the model. Time alignment and feature extraction technology are used to ensure data consistency and model robustness. By optimizing algorithms and performance evaluation indicators, the training efficiency and classification performance of the model are further improved. It provides a scientific basis for the operation and maintenance decisions of wind turbines under cold weather conditions, reduces the risk of failures caused by blade icing, and ensures the safety and efficient operation of the unit.

[0028] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically listed below. It should be understood that the above general description and the detailed description below are only exemplary and cannot limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions implemented in the present invention, the following briefly introduces the drawings required for the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0030] Figure 1 It is a business execution flow chart of the method for evaluating the health status of large components in the nacelle of a wind turbine generator set when the blades are in an icing state according to the present invention;

[0031] Figure 2 This is a flowchart of the business execution of data preprocessing and feature fusion described in the present invention;

[0032] Figure 3 It is a business execution flow chart of the health status evaluation of large components of a wind turbine based on the Transformer-KAN model according to the present invention; DETAILED DESCRIPTION

[0033] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following embodiments are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0034] The health status evaluation method of large components in the nacelle of a wind turbine under blade icing conditions is as follows: Figure 1 As shown in the figure, the health status evaluation method of large components of the wind turbine nacelle under blade icing conditions accurately identifies the blade icing condition and evaluates the health status of large components of the wind turbine nacelle under blade icing conditions through data collection, preprocessing, fusion and analysis, combined with a deep learning model, to ensure the safe and efficient operation of the unit under cold weather conditions. The specific implementation plan is as follows:

[0035] S1 Figure 2As shown, first, the SCADA data, voiceprints, and vibration signals of the large components of the wind turbine cabin (main shaft, gearbox, generator, blades, etc.) under cold weather conditions are obtained. Secondly, the SCADA data is cleaned and standardized, including denoising, standardization, missing data filling, and outlier detection to ensure data quality and provide a reliable basis for subsequent analysis. Then, the voiceprint and vibration signals are denoised, standardized, and feature extracted, mainly extracting frequency domain features, time domain features, and statistical features to reflect the operating characteristics of the wind turbine under different working conditions, and extracting key information from the data. Finally, the processed SCADA data, voiceprints, and vibration signals are feature fused, mainly using fusion methods such as weighted average and dimensionality reduction mechanisms, and fully combining information from different data sources to generate a comprehensive feature matrix of the same time scale, thereby improving the accuracy and robustness of subsequent model predictions.

[0036] S2 inputs the comprehensive feature matrix obtained in S1, meteorological data (such as temperature, humidity, wind speed, etc.) and blade surface icing characteristics into a pre-trained blade icing identification model to obtain a blade icing identification result. The blade icing identification model is based on a Transformer network, and utilizes the comprehensive feature matrix obtained in S1, combined with meteorological conditions and blade surface icing characteristics, to effectively distinguish between the blade icing state and the normal working state. Specifically, the model learns the feature distribution under each state based on the comprehensive feature matrix obtained in S1, so as to determine whether there is icing on the blade. The output of the model is the determination result of the blade icing state and the comprehensive feature matrix under the blade icing state, and provides a basis for subsequent health status assessment.

[0037] S3 Figure 3 As shown, the comprehensive feature matrix under the blade icing state obtained in S2 is input into the Transformer-KAN model for health status assessment. The Transformer-KAN model classifies the health status of large components in the nacelle of the wind turbine, wherein the health status characteristics of the large components in the nacelle are extracted based on the Transformer network, and the extracted health status characteristics are specifically classified based on the KAN model. And evaluate whether there is a risk of failure of large components in the nacelle due to blade icing. Subsequently, the model assigns a health status score to each large component in the nacelle. The scoring result reflects the health status of the large components in the nacelle under the blade icing state. The scoring levels can be divided into health, minor fault, moderate fault and severe fault, etc. The evaluation results provide data support for subsequent operation and maintenance decisions.

[0038] Furthermore, the sampling interval of the SCADA data in the S1 is 1 min, which mainly includes gearbox oil temperature, gearbox drive end bearing temperature, gearbox non-drive end bearing temperature, average wind speed, generator front bearing temperature, generator rear bearing temperature, generator stator winding temperature, average generator power, etc.; the sampling frequency of sound, vibration and speed signals is 12.8 kHz, which mainly includes: horizontal vibration of the main shaft front bearing, horizontal vibration of the main shaft rear bearing, axial vibration of the gearbox low-speed shaft, axial vibration of the gearbox high-speed shaft, vertical vibration of the generator drive end, vertical vibration of the generator non-drive end, cabin sound print, blade sound print, etc.

[0039] Furthermore, frequency domain features, time domain features and statistical features are extracted from the voiceprint and vibration signal in S1. Among them, the time domain features include mean, root mean square, kurtosis, etc., which are used to reflect the volatility, amplitude and change of the signal; the frequency domain features include spectrum, main frequency, frequency bandwidth, etc., which can reveal the vibration mode, frequency component and potential fault frequency of the equipment; the statistical features include peak factor, frequency domain entropy, signal energy, etc., which can reveal the stability, complexity and potential fault mode of the equipment.

[0040] Furthermore, the features from different signal sources (SCADA data, voiceprints, and vibration signals) in S1 are fused to improve the accuracy and robustness of model prediction. Common fusion methods include weighted averaging and KPCA dimensionality reduction. Weighted averaging generates a comprehensive feature vector by assigning weights to different signal sources and combining their respective importance, which is suitable for situations where signal sources contribute unevenly; while KPCA dimensionality reduction maps data from high-dimensional space to low-dimensional space through nonlinear mapping, extracts the most representative features, reduces redundant information, and improves computational efficiency. These fusion methods can effectively combine the advantages of multi-source data, provide comprehensive and accurate feature representation, and enhance the performance of subsequent analysis and prediction.

[0041] Furthermore, the SCADA data after feature fusion in S1, as well as the voiceprint and vibration signals, are time-aligned to ensure the consistency of these data in the time dimension. This process is achieved by synchronizing the timestamps of different data sources, thereby generating time-aligned multi-source fusion data, providing a unified time reference for subsequent deep learning models.

[0042] Furthermore, the characteristics of ice formation on the blade surface in S2 are mainly manifested as: the ice layer is unevenly distributed on the leading edge and upper surface of the blade, and is morphologically classified as frost ice (loose and porous), transparent ice (dense and smooth) or mixed ice (multi-phase composite) by CNN, and its formation is synergistically affected by temperature (-15℃ to 0℃), humidity (>85%RH) and wind speed (>4m / s). The thickness of the ice layer is 0.5-10mm measured by ultrasonic or 0.1-2mm / h estimated by the model, resulting in a 5%-30% increase in the local mass of the blade. The dynamic growth of ice layer leads to abnormal fluctuation of infrared reflectivity (decreased by 20%-40%), the mean value drops to 40%-60% (normal 70%-85%), and the standard deviation increases to 8%-12%; accompanied by the shift of acoustic signal spectrum characteristics (high frequency energy attenuation>15dB), the energy attenuation rate of 2-5kHz frequency band is greater than 15%, and the spectrum entropy is 1.2-1.8 (normal 0.6-1.0); the surface roughness increases significantly (Ra>50μm), the vibration root mean square value of 5-10kHz frequency band is 0.8-1.5g (normal 0.3-0.6g), and the wavelet packet energy entropy is 4.5-6.0 (normal 2.5-3.5). The above features are processed by time alignment and input into the Transformer network to support the construction of blade icing identification model.

[0043] Furthermore, when constructing the blade icing recognition model described in S2, the core of the Transformer architecture is the multi-head self-attention mechanism, which can process different parts of the input data in parallel to capture long-distance dependencies and local features in the data. By decomposing the input data into multiple "attention heads", each head focusing on different aspects of the input sequence, the model can more comprehensively understand the complex relationship between multi-source data. Specifically, the calculation formula of the multi-head self-attention mechanism is: MultiHead(Q,K,V)=Concat(head1,head2,...,head h )W O in, Here, Q, K, and V represent query, key, and value matrices, respectively. k is the dimension of the key, h is the number of attention heads, and W O is the output projection matrix. In this way, the model can efficiently process the comprehensive feature data and extract the feature information related to the blade icing state.

[0044] In order to retain the position information in the input sequence, the model introduces position encoding. Position encoding adds a fixed vector to each position, allowing the model to perceive the position relationship in the sequence. The calculation formula for position encoding is: Among them, pos represents the position, i represents the dimension, and d model is the dimension of the model. The introduction of position encoding ensures that the model can retain the temporal order information when processing sequence data, which is particularly important for the dynamic monitoring of blade icing status.

[0045] In each layer of the encoder, in addition to the multi-head self-attention mechanism, a feed-forward neural network is also included. FFN performs the same fully connected operation on the feature vector at each position to further extract high-level feature representations. Its calculation formula is: FFN(x)=max(0,xW1+b1)W2+b2 Among them, W1, W2, b1 and b2 are the parameters of the network. Through FFN, the model can perform nonlinear transformation on the output of the multi-head self-attention mechanism, thereby enhancing the expressiveness of the model.

[0046] During the model training process, the cross entropy loss function is used to measure the difference between the predicted results and the true labels. The cross entropy loss function can effectively evaluate the classification performance of the model and optimize the model parameters by minimizing the difference between the predicted probability distribution and the true label. The specific formula is: Among them, y is the true label, is the predicted output of the model. In order to efficiently update the model parameters, the Adam optimizer is used. The Adam optimizer combines the advantages of momentum and adaptive learning rate, which can converge quickly and stabilize the training process. By adjusting the learning rate and hyperparameters, the model can achieve a higher accuracy within a limited training time.

[0047] Through the above design, the Transformer network can effectively process the comprehensive feature matrix and effectively identify the blade icing state and normal working state. The model learns the relationship between the comprehensive feature data through the self-attention mechanism, retains the sequence information in combination with the position encoding, and extracts high-level features through the feedforward neural network. This architecture not only improves the expressiveness of the model, but also enhances its adaptability to complex data.

[0048] Furthermore, the multi-source fusion data after time alignment in the S3 is subjected to feature extraction through the Transformer network. The Transformer encoder captures the global dependencies in the data through the multi-head self-attention mechanism and FFN to generate context-related feature representations. The multi-head self-attention mechanism divides the input data into multiple "heads" for parallel calculation, and the feature representations learned by each head are then spliced ​​and linearly transformed, so that the features of different subspaces in the multi-source fusion data can be captured. FFN further processes these feature representations and increases the nonlinear ability of the model through two linear transformations and a nonlinear activation function. In addition, layer normalization is used to reduce internal covariate shift, accelerate model convergence, and prevent overfitting through the Dropout layer. In order to extract global features and reduce feature dimensions, the output feature map of the Transformer is pooled into a fixed-size feature vector through average pooling. The pooled feature map is flattened into a one-dimensional vector for input into the subsequent fully connected layer.

[0049] Furthermore, based on the features extracted by the Transformer network in S3, the KAN model is further introduced to classify the health status. The KAN model performs nonlinear mapping through basic linear transformation and B-spline basis function to improve the expression ability and interpretability of the model. The KAN model classifies the extracted health status features into healthy, mild fault, moderate fault and severe fault. In the KAN model, the input features are first mapped to the output space through linear transformation to increase the linear expression ability of the model. The calculation formula of linear transformation is: BaseLinear=Wx+b Among them, x is the input feature, W is the basic weight matrix, and b is the bias vector.

[0050] The input features are mapped to multiple piecewise polynomial spaces through the B-spline basis function. Each piecewise polynomial can capture the nonlinear relationship of the input data in different intervals. The calculation formula of the B-spline basis function is: Among them, x is the input feature, c i are the coefficients of the piecewise polynomial, B i (x) is the i-th B-spline basis function.

[0051] Through the combination of basic activation functions and piecewise polynomial basis functions, the input features are nonlinearly mapped to generate richer feature representations. Finally, the KAN model outputs a vector, each element of which represents the predicted probability that the sample belongs to the corresponding health state. The basic activation function can use nonlinear functions such as SiLU, and the calculation formula is: Nonlinear(x)=BaseLinear(Activation(x))+B-spline(x) Among them, Activation(x) is a basic activation function, such as SiLU function.

[0052] The output vector of the Transformer-KAN model is obtained. Each element represents the predicted probability that the sample belongs to the corresponding health state.

[0053] Furthermore, in order to comprehensively evaluate the performance of the Transformer-KAN model in S3, indicators such as accuracy, precision, recall, F1 score and confusion matrix are used. These indicators can measure the performance of the model in blade icing recognition and health status classification tasks from multiple perspectives. Through learning rate scheduling and regularization technology, combined with the firefly optimization algorithm for optimization, the model training process can be further optimized to prevent overfitting and accelerate convergence.

[0054] Furthermore, the health status of each sample is determined based on the predicted probability output by the Transformer-KAN model described in S3. Samples are classified into specific health status categories, including health, minor fault, moderate fault, and major fault, by setting thresholds or using classification strategies (such as maximum probability selection). This classification result provides a scientific basis for operation and maintenance decisions and ensures the efficient operation and maintenance of wind turbines.

[0055] 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 make equivalent replacements for some or all of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

1. A method for evaluating the health status of large components in the nacelle of a wind turbine when blades are frozen. Its features are as follows: including: S1: Acquire SCADA data, soundprints, and vibration signals of large components of the wind turbine nacelle under cold weather conditions; perform data cleaning and standardization on the SCADA data, and perform denoising, standardization, and feature extraction on the soundprints and vibration signals; perform feature fusion on the processed SCADA data, soundprints, and vibration signals and generate a comprehensive feature matrix of the same time scale; S2: Input the comprehensive feature matrix obtained in S1, the pre-acquired meteorological data, and the blade surface icing features into a pre-trained blade icing identification model to obtain a comprehensive feature matrix under the blade icing state; the blade icing identification model is constructed based on a Transformer network; S3: The comprehensive feature matrix of the blade icing state obtained in S2 is input into the Transformer-KAN model for health status assessment; the Transformer-KAN model classifies the health status of large components of the wind turbine nacelle, wherein the health status characteristics of the large components of the nacelle are extracted based on the Transformer network, and the extracted health status characteristics are specifically classified based on the KAN model, and it is evaluated whether there is a risk of failure of large components of the nacelle due to blade icing.

2. According to the method for evaluating the health status of large components of a wind turbine nacelle under blade icing conditions described in claim 1, it is characterized by the following: the SCADA data in step S1 includes gearbox oil temperature, gearbox bearing temperature, average wind speed, generator bearing temperature, generator power, and rotation speed; the soundprint and vibration signal include main shaft vibration, gearbox vibration, generator vibration, nacelle soundprint, and blade soundprint.

3. According to the method for evaluating the health status of large components of a wind turbine cabin under blade icing conditions described in claim 1, it is characterized by the following: in step S1, the SCADA data is cleaned and standardized including denoising, standardization, missing data filling and outlier detection; the features extracted from the soundprint and vibration signal include time domain features, frequency domain features and statistical features.

4. The method for evaluating the health status of large components of a wind turbine nacelle when blades are frozen according to claim 1, Its characteristics are as follows: In S1, the feature fusion method includes weighted average and KPCA (Kernel Principal Component Analysis) dimensionality reduction; The SCADA data after feature fusion, as well as the voiceprint and vibration signals, are time-aligned to provide a unified time reference.

5. The method for evaluating the health status of large components of a wind turbine nacelle when blades are frozen according to claim 1, characterized in that: The meteorological data in step S2 include temperature, humidity, wind speed, air pressure and dew point temperature, which are derived from a SCADA system or an external meteorological station; the icing characteristics of the blade surface include ice layer type, thickness, infrared reflectivity, acoustic properties and surface texture, which are obtained through a multimodal sensor or a physical model.

6. According to the method for evaluating the health status of large components of a wind turbine nacelle under blade icing conditions according to claim 1, it is characterized in that: in S2, during the training process of the blade icing identification model based on the Transformer network, a cross entropy loss function is used to measure the difference between the predicted result and the true label, and the model parameters are updated through the Adam (Adaptive Moment Estimation) optimizer to improve the classification performance of the model.

7. The method for evaluating the health status of large components of a wind turbine nacelle when blades are frozen according to claim 1, characterized in that: In S3, a Transformer network is introduced to extract health status features of large cabin components; The Transformer encoder captures global dependencies through multi-head self-attention mechanism and FFN, and optimizes feature representation through layer normalization, Dropout layer and average pooling.

8. The method for evaluating the health status of large components of a wind turbine nacelle when blades are frozen according to claim 1, characterized in that: In S3, the KAN model is introduced to classify the health status, and the samples are classified into healthy, mild fault, moderate fault and severe fault categories through linear transformation and piecewise polynomial basis function (B-spline) for nonlinear mapping.

9. The method for evaluating the health status of large components of a wind turbine nacelle based on multi-source data under blade icing conditions according to claim 1, characterized in that: In S3, accuracy, precision, recall, F1 score and confusion matrix are used to evaluate the model performance, and the simulated firefly optimization algorithm (SFOA) is used for optimization.

Images