Wind driven generator fault monitoring method and device based on operation data

Through the combination of magnetostrictive sensors and Bayesian neural network model, the strain and deformation characteristics of the rotating components of wind turbines are extracted, and the problem of insufficient early fault prediction capabilities in the existing technology is solved, and efficient fault detection and early warning is achieved. Improves the operating reliability and maintenance efficiency of wind turbines.

CN118934499AActive Publication Date: 2024-11-12YE COUNTY GUOBO DASHIYA WIND POWER CO LTD

Patent Information

Application Number
CN202411216554.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-11-12
Estimated Expiration
2044-09-02

AI Technical Summary

Technical Problem

Existing wind turbine fault monitoring technologies are difficult to achieve early failure prediction, and due to changes in environmental noise and operating conditions, false alarms or missed reports are prone to occur, and the model lacks prediction capabilities in high-dimensional data and nonlinear failure modes.

Method used

Magnetostrictive sensors are used to detect the tiny strain and deformation data of rotating components, and extract the time domain, frequency domain and time frequency domain characteristics through signal noise reduction, outlier value removal and adaptive filtering optimization, combining Bayesian neural network model and history The data is pre-detected, and the fault warning signal is generated in real time.

Benefits of technology

It improves the accuracy and robustness of wind turbine fault detection, can effectively identify and prevent early failures, reduce downtime risks and maintenance costs, and improves operating efficiency and safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118934499A_ABST
    Figure CN118934499A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of wind driven generators, in particular to a wind driven generator fault monitoring method and device based on operation data, and the method comprises the following steps: S1, arranging and installing a sensor: arranging a magnetostrictive sensor on a rotating part; s2, data acquisition: acquiring a strain signal and deformation data in real time; s3, data preprocessing: preprocessing the strain signal and the deformation data; s4, fault feature extraction: extracting fault feature parameters; s5, model construction and training: constructing a wind driven generator fault pre-detection model to identify early fault signs and predict the probability of fault occurrence; and S6, real-time monitoring and fault early warning: monitoring the strain signal and the deformation data of the rotating part in real time, and generating a fault early warning signal. According to the invention, mastering of real-time health states of key components of the wind driven generator is improved, and occurrence of potential faults is effectively prevented.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbine generators, and in particular to a method and device for monitoring wind turbine generator faults based on operating data. Background Art

[0002] With the rapid development of wind power technology, wind turbines have become an important part of the global energy supply. However, key components of wind turbines, such as main shafts, gearboxes and bearings, are extremely susceptible to fatigue and wear during long-term high-load and high-speed operation. Failure of these components will not only lead to a decrease in the operating efficiency of wind turbines, but may also cause serious equipment damage, resulting in high maintenance costs and unforeseen downtime. Therefore, real-time monitoring and fault warning of key components of wind turbines have become an important means to ensure the safe and stable operation of wind power equipment.

[0003] Existing wind turbine fault monitoring technologies mainly rely on monitoring data from vibration sensors and temperature sensors. These methods can usually only detect problems when a fault has occurred or is about to occur, and it is difficult to accurately predict early faults. In addition, existing technologies mostly rely on fixed threshold fault detection mechanisms, lack dynamic adjustment capabilities, and are easily affected by environmental noise and changes in operating conditions, resulting in false alarms or missed alarms. At the same time, most existing fault detection models use simple machine learning algorithms, which cannot fully utilize the historical data and complex operating characteristics of wind turbines, resulting in insufficient prediction capabilities of the model under high-dimensional data and nonlinear fault modes, making it difficult to meet the needs of practical applications.

[0004] The present invention aims to overcome the deficiencies of the prior art and proposes a wind turbine fault monitoring method based on operating data, thereby improving the operating reliability and maintenance efficiency of the wind turbine and reducing the risk of faults. Summary of the invention

[0005] The invention provides a wind turbine fault monitoring method and equipment based on operation data.

[0006] The wind turbine fault monitoring method based on operation data comprises the following steps:

[0007] S1, sensor arrangement and installation: magnetostrictive sensors are arranged on the rotating parts of the wind turbine to detect the tiny strain and deformation of the rotating parts during operation;

[0008] S2, data acquisition: real-time acquisition of strain signals and deformation data of rotating parts of wind turbines through magnetostrictive sensors;

[0009] S3, data preprocessing: preprocessing the collected strain signals and deformation data, including signal noise reduction, outlier removal and data smoothing, and optimizing the original strain signals and deformation data using adaptive filtering algorithms;

[0010] S4, fault feature extraction: based on the preprocessed strain signal and deformation data, extract fault feature parameters, including frequency domain features, time domain features and time-frequency domain features, and generate feature vectors for model input;

[0011] S5, model construction and training: Based on the constructed feature vector, a wind turbine fault pre-detection model is constructed, and the model is trained in combination with the wind turbine historical strain signal and deformation data and the fault case library to identify early fault signs and predict the probability of fault occurrence, including:

[0012] S51, feature vector input: using the generated feature vector as input data of the model, and performing standardization and normalization on the input data;

[0013] S52, model construction: constructing a wind turbine fault pre-detection model based on the processed feature vector;

[0014] S53, model training and validation: dividing the historical strain signals and deformation data of the wind turbine into a training set and a validation set, using the data of the training set to train the model through cross-validation and hyperparameter optimization, and using the validation set to validate the model;

[0015] S6, real-time monitoring and fault warning: During the operation of the wind turbine, the strain signals and deformation data of the rotating parts are monitored in real time, and predicted through the trained wind turbine fault pre-detection model. When signs of potential faults are detected, a fault warning signal is generated, and the fault type, location and development trend are output.

[0016] Optionally, the arrangement and installation of sensors in S1 includes:

[0017] S11, sensor placement location selection: arranging magnetostrictive sensors on the surface or predetermined positions inside the rotating parts (main shaft, gear box and bearings) of the wind turbine;

[0018] S12, sensor fixing method: The magnetostrictive sensor is fixed to the rotating part by non-contact or contact mounting. The non-contact mounting adopts magnetic bracket or adhesion technology, and the contact mounting adopts mechanical clamp or fastener fixing;

[0019] S13, sensor calibration and adjustment: After the magnetostrictive sensor is installed, preliminary calibration and zero point adjustment are performed to compensate for displacement or error during the installation process.

[0020] Optionally, the data collection in S2 includes:

[0021] S21, strain signal acquisition: the strain signals of the rotating parts of the wind turbine are collected in real time through magnetostrictive sensors, including axial strain, radial strain and torsional strain;

[0022] S22, deformation data collection: collecting deformation data of the rotating component through a magnetostrictive sensor, including bending deformation, stretching deformation and shear deformation of the component;

[0023] S23, data acquisition synchronization and storage: the collected strain signals and deformation data are time-stamped and stored in a local or cloud data storage unit in real time.

[0024] Optionally, the data preprocessing in S3 includes:

[0025] S31, signal denoising: using wavelet transform to perform signal denoising on the collected strain signal and deformation data;

[0026] S32, outlier elimination: identifying and eliminating outliers in strain signals and deformation data by statistical methods;

[0027] S33, data smoothing: using sliding average algorithm to smooth the signal;

[0028] S34, adaptive filtering algorithm optimization: the pre-processed strain signal and deformation data are optimized using the LMS adaptive filtering algorithm.

[0029] Optionally, the fault feature extraction in S4 includes:

[0030] S41, time domain feature extraction: extracting time domain feature parameters based on the preprocessed strain signal and deformation data, the time domain features include mean, root mean square (RMS), skewness and kurtosis;

[0031] S42, frequency domain feature extraction: performing Fourier transform on the preprocessed strain signal and deformation data to extract frequency domain feature parameters, including main frequency, frequency center, bandwidth and spectrum energy;

[0032] S43, time-frequency domain feature extraction: short-time Fourier transform is used to extract time-frequency domain features;

[0033] S44, feature vector generation: feature fusion is performed on the extracted time domain, frequency domain and time-frequency domain feature parameters to generate a feature vector for input into a fault pre-detection model.

[0034] Optionally, the feature vector input in S51 includes:

[0035] S511, standardization of feature vectors: standardizing the generated feature vectors to eliminate the dimensional differences between the features;

[0036] S512, normalization processing of feature vectors: normalization processing is performed on the standardized feature vectors to scale the feature values ​​to a uniform range;

[0037] S513, construction of model input data: the standardized and normalized feature vector V i ″ is input into the wind turbine fault prediction model.

[0038] Optionally, the wind turbine fault pre-detection model in S52 adopts a Bayesian neural network (BNN) model, and the Bayesian neural network (BNN) model includes:

[0039] S521, Bayesian reasoning process: In the fault detection of wind turbines, a weighted prior distribution is introduced, which is expressed as:

[0040]

[0041] Among them, p(w|D) is the posterior distribution of weight w after given data D, p(D|w) is the likelihood function, p(w) is the prior distribution of weight, Z is the normalization constant, α, γ k is the adjustment coefficient, w k is the model weight corresponding to the historical fault data, K is the number of historical fault data or reference fault modes;

[0042] S522, likelihood estimation based on adaptive kernel function: an adaptive kernel function is introduced to adaptively adjust the variance according to the position of the data point, which is expressed as:

[0043]

[0044] Among them, p(D|w) is the likelihood function, is a normal distribution, is the variance, is the base variance, μ is the mean vector of the data, l is the kernel width parameter, N is the number of data samples, and y i is the observed value of the i-th sample;

[0045] S523, prediction distribution calculation: The prediction distribution is approximated by Monte Carlo sampling, processes complex data structures, and provides predictions of fault occurrence, expressed as:

[0046]

[0047] Among them, p(y * |x *,D) is the predicted distribution, which means that given the input x * and data D, the output is y * The probability of M is the number of samplings, w m is the weight sampled from the posterior distribution p(w|D);

[0048] S524, Fault Detection and Uncertainty Quantification: Provides a quantitative assessment of uncertainty by calculating the variance of the predicted values, expressed as:

[0049]

[0050] Among them, Uncertainty(x * ) is the uncertainty measure, which means that when input x * The variance of the prediction is (x * ) is the mean of the predicted values, f(x * ,w m ) is the mth sampling weight w m The predicted value under ;

[0051] S525, model update and adaptive adjustment: The model is updated in real time through incremental learning and adaptive adjustment of learning rate, which is expressed as:

[0052]

[0053] Among them, w t+1 is the updated weight, w t is the current weight, η is the learning rate, D t is the data of the current time step t, η0 is the initial learning rate, and λ is the learning rate attenuation factor.

[0054] Optionally, the model training and verification in S53 includes:

[0055] S531, data set division: dividing the historical strain signals and deformation data of the wind turbine into a training set and a validation set according to a predetermined ratio;

[0056] S532, cross validation: During the model training process, use k-fold cross validation to divide the training set into k subsets, use k-1 subsets for training, and use the remaining subset for validation. Repeat k times, each time selecting a different subset as the validation set, and finally take the average value as the evaluation standard of model performance;

[0057] S533, Hyperparameter Optimization: Grid search is used for hyperparameter optimization, expressed as:

[0058]

[0059] Among them, θ * is the best hyperparameter combination, arg min is to find the parameter value that minimizes the subsequent expression, θ is the current hyperparameter combination, Θ grid is a predefined hyperparameter grid, k is the number of cross-validation folds, is the loss function, is the output value predicted on the j-fold validation set using the hyperparameter combination θ, y j is the true label on the j-th fold validation set.

[0060] Optionally, the real-time monitoring and fault warning in S6 includes:

[0061] S61, real-time signal monitoring: During the operation of the wind turbine, the strain signals and deformation data of the rotating parts (such as the main shaft, gearbox, bearings, etc.) are monitored in real time;

[0062] S62, comparative analysis of data and model prediction results: input the real-time collected strain signals and deformation data into the trained wind turbine fault pre-detection model, and output the current fault state or health risk probability;

[0063] S63, fault warning generation and information output: When the real-time prediction results show the existence of a potential fault (the predicted fault probability exceeds the set detection threshold), a fault warning signal is immediately generated, including the type of fault (such as bearing fault, gearbox fault, etc.), the location of the fault, and the development trend of the fault (such as the speed at which the fault deteriorates).

[0064] The wind turbine fault monitoring device based on operation data is used to implement the above-mentioned wind turbine fault monitoring method based on operation data, and includes the following units:

[0065] Sensor arrangement and installation unit: magnetostrictive sensors are arranged on the rotating parts of the wind turbine to detect the tiny strains and deformations of the rotating parts during operation;

[0066] Data acquisition unit: collects strain signals and deformation data of rotating parts of wind turbines in real time through magnetostrictive sensors;

[0067] Data preprocessing unit: preprocess the collected strain signals and deformation data, including signal noise reduction, outlier removal and data smoothing, and optimize the original strain signals and deformation data using adaptive filtering algorithms;

[0068] Fault feature extraction unit: extracts fault feature parameters based on the preprocessed strain signal and deformation data, including frequency domain features, time domain features and time-frequency domain features, and generates feature vectors for model input;

[0069] Model building and training unit: Build a wind turbine fault pre-detection model based on the constructed feature vector, and train the model in combination with the wind turbine historical strain signal and deformation data and the fault case library to identify early fault signs and predict the probability of fault occurrence;

[0070] Real-time monitoring and fault warning unit: During the operation of the wind turbine, the strain signals and deformation data of the rotating parts are monitored in real time, and predictions are made through the trained wind turbine fault pre-detection model. When signs of potential faults are detected, a fault warning signal is generated, and the fault type, location and development trend are output.

[0071] Beneficial effects of the present invention:

[0072] The present invention can effectively detect the tiny strain and deformation generated by the components during high-load operation by accurately arranging magnetostrictive sensors on the rotating parts of the wind turbine. The stability and measurement accuracy of the sensors in complex operating environments are ensured by optimizing the installation position and fixing method of the sensors. The integrity and timeliness of the data are guaranteed by real-time acquisition of multi-dimensional strain signals and deformation data, combined with a high-precision clock for synchronous marking and storage, and reliable data support is provided for fault monitoring. This precise monitoring capability greatly improves the grasp of the real-time health status of key components of the wind turbine and effectively prevents the occurrence of potential faults.

[0073] The present invention greatly improves the signal-to-noise ratio and the accuracy of feature extraction by performing signal denoising, outlier elimination, data smoothing and adaptive filtering optimization on the collected strain signals and deformation data. It combines time domain, frequency domain and time-frequency domain feature extraction methods to comprehensively capture key fault features in the signal, ensures that the generated feature vector can reflect potential problems in the operation of the wind turbine, provides high-quality input data for subsequent fault pre-detection models, significantly improves the detection accuracy and robustness of the model, and can more effectively identify and predict early faults of wind turbines.

[0074] The present invention adopts a Bayesian neural network model, combines historical strain signals and deformation data for model training, and introduces an adaptive kernel function and a dynamic weight update mechanism, thereby enhancing the adaptability and prediction ability of the model when processing nonlinear and high-dimensional data. Hyperparameter optimization and cross-validation are performed through grid search to ensure the generalization ability and stability of the model on different data sets. In practical applications, the model performs instant fault prediction and early warning by real-time monitoring of the strain signals and deformation data of rotating parts, effectively reducing the risk of downtime and maintenance costs. The detection threshold set based on historical data ensures the reliability of fault early warning, and comprehensively improves the operating efficiency and safety of wind turbines. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0076] Figure 1 A schematic diagram of a monitoring method flow chart of an embodiment of the present invention;

[0077] Figure 2 Schematic diagram of functional units of a device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0078] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0079] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0080] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0081] like Figure 1 As shown, the wind turbine fault monitoring method based on operation data includes the following steps:

[0082] S1, sensor arrangement and installation: magnetostrictive sensors are arranged on the rotating parts of the wind turbine to detect the tiny strain and deformation of the rotating parts during operation;

[0083] S2, data acquisition: real-time acquisition of strain signals and deformation data of rotating parts of wind turbines through magnetostrictive sensors;

[0084] S3, data preprocessing: preprocess the collected strain signals and deformation data, including signal noise reduction, outlier removal and data smoothing, and use adaptive filtering algorithm to optimize the original strain signals and deformation data to eliminate environmental interference and measurement noise, ensuring the accuracy of subsequent fault analysis;

[0085] S4, fault feature extraction: based on the preprocessed strain signal and deformation data, extract fault feature parameters, including frequency domain features, time domain features and time-frequency domain features, and generate feature vectors for model input;

[0086] S5, model construction and training: Based on the constructed feature vector, a wind turbine fault pre-detection model is constructed, and the model is trained in combination with the wind turbine historical strain signal and deformation data and the fault case library to identify early fault signs and predict the probability of fault occurrence, including:

[0087] S51, feature vector input: the generated feature vector is used as the input data of the model, and the input data is standardized and normalized to ensure that the feature vector has a consistent scale and range before being input into the model, so as to improve the efficiency and accuracy of model training;

[0088] S52, model construction: constructing a wind turbine fault pre-detection model based on the processed feature vector;

[0089] S53, model training and validation: divide the historical strain signals and deformation data of wind turbines into training sets and validation sets, use the data of the training set to train the model through cross-validation and hyperparameter optimization to improve the generalization ability of the model and ensure that the model performs well on new data. Use the validation set to validate the model and evaluate its accuracy and stability in fault detection and prediction;

[0090] S6, real-time monitoring and fault warning: During the operation of the wind turbine, the strain signals and deformation data of the rotating parts are monitored in real time, and predicted through the trained wind turbine fault pre-detection model. When signs of potential faults are detected, a fault warning signal is generated, and the fault type, location and development trend are output to provide timely maintenance suggestions for operation and maintenance personnel;

[0091] Through the above steps, early detection and early warning of wind turbine faults are achieved, and the real-time operation data of key components can be accurately acquired and processed. Through an efficient fault pre-detection model, timely fault warnings can be provided in actual operation, thereby improving the operating reliability and maintenance efficiency of wind turbines and reducing the risk of failures.

[0092] Sensor arrangement and installation in S1 include:

[0093] S11, sensor placement location selection: magnetostrictive sensors are placed on the surface or predetermined positions inside the rotating parts (main shaft, gearbox and bearings) of the wind turbine. The predetermined positions are selected based on the stress concentration areas or fatigue-prone areas of the rotating parts to ensure that the sensors can accurately detect the tiny strains and deformations generated by the rotating parts during operation;

[0094] S12, sensor fixing method: The magnetostrictive sensor is fixed on the rotating part by non-contact or contact mounting. The non-contact mounting adopts magnetic bracket or adhesion technology, and the contact mounting adopts mechanical clamp or fastener to ensure the stability and measurement accuracy of the sensor in high-speed operation environment;

[0095] S13, sensor calibration and adjustment: After the magnetostrictive sensor is installed, preliminary calibration and zero point adjustment are performed to compensate for displacement or error during installation and ensure the accuracy and consistency of the measurement data;

[0096] Through the above content, it is ensured that the magnetostrictive sensor can stably and accurately detect tiny strains and deformations on the rotating parts of the wind turbine, and can maintain the measurement accuracy of the sensor in high-speed and complex operating environments, thereby providing reliable data support for fault monitoring and improving the accuracy of fault detection and the effectiveness of early warning.

[0097] Data collection in S2 includes:

[0098] S21, strain signal acquisition: The strain signals of the rotating parts of the wind turbine are collected in real time through magnetostrictive sensors, including axial strain, radial strain and torsional strain. By sensing the strain changes of the rotating parts under different working conditions (such as starting, acceleration, deceleration and constant speed operation), the stress state generated by the parts during operation is captured;

[0099] S22, deformation data collection: magnetostrictive sensors are used to collect deformation data of rotating parts, including bending deformation, telescopic deformation and shear deformation of parts, to monitor the deformation of rotating parts under high load and high speed in real time, ensuring that subtle structural changes caused by fatigue, overload or abnormal operation are captured;

[0100] S23, data acquisition synchronization and storage: To ensure the synchronization of strain signals and deformation data, a high-precision clock is used to time-stamp the collected strain signals and deformation data, and the collected strain signals and deformation data are synchronously stored in a local or cloud data storage unit in real time;

[0101] Through the above contents, accurate monitoring of the rotating components of wind turbines under various operating conditions is ensured. By capturing axial strain, radial strain, torsional strain signals as well as bending, expansion and shear deformation, comprehensive structural health information is provided. High-precision time synchronization and data storage ensure the integrity and timeliness of the data, providing a reliable basis for fault detection and analysis, and improving the accuracy of fault monitoring and early warning capabilities.

[0102] Data preprocessing in S3 includes:

[0103] S31, signal denoising: wavelet transform is used to perform signal denoising on the collected strain signal and deformation data, which is expressed as:

[0104]

[0105] Among them, d j,k is the detail coefficient, a J is the approximation coefficient, ψ j,k (t) and φ J (t) are wavelet basis function and scale function, J is the maximum scale of wavelet decomposition, j is the scale parameter, and k is the translation parameter;

[0106] S32, outlier elimination: Identify and eliminate outliers in strain signals and deformation data by statistical methods, using the 3σ principle, that is, for a set of data X i , calculate its mean μ and standard deviation σ, any value X that deviates from the mean by more than 3 times the standard deviation i are considered as outliers and removed, expressed as:

[0107] |X i -μ|>3σ then X i is an outlier;

[0108] S33, data smoothing: The signal is smoothed using a sliding average algorithm to eliminate short-term fluctuations in the data and make the signal smoother and more continuous, expressed as:

[0109]

[0110] Among them, Y(t) is the smoothed signal, X(t) is the original signal, and N is the size of the sliding window;

[0111] S34, adaptive filtering algorithm optimization: The pre-processed strain signal and deformation data are optimized using the LMS adaptive filtering algorithm, and the filter coefficients are adjusted according to the real-time collected signals to minimize the influence of noise, which is expressed as:

[0112] w n+1 =w n +2μe(n)X(n);

[0113] Among them, w n is the filter coefficient, μ is the step size factor, e(n) is the error signal, and X(n) is the input signal;

[0114] Through the above content, the signal-to-noise ratio and the accuracy of feature extraction are improved, providing a more reliable and accurate data basis for subsequent fault analysis, and ensuring the accuracy and real-time performance of wind turbine fault monitoring.

[0115] Fault feature extraction in S4 includes:

[0116] S41, time domain feature extraction: based on the preprocessed strain signal and deformation data, extract the time domain feature parameters. The time domain features include mean, root mean square (RMS), skewness and kurtosis, which are expressed as:

[0117]

[0118] Among them, Mean is the mean, X i represents the i-th signal sample, N is the number of samples;

[0119]

[0120] Among them, RMS is the root mean square value;

[0121]

[0122] Among them, μ is the mean, σ is the standard deviation, and Skewness is the skewness;

[0123]

[0124] Among them, Kurtosis is the kurtosis;

[0125] S42, frequency domain feature extraction: Perform Fourier transform on the preprocessed strain signal and deformation data to extract frequency domain feature parameters, including main frequency, frequency center, bandwidth and spectrum energy, expressed as:

[0126]

[0127] Among them, X(f) is the frequency domain signal, x(t) is the time domain signal, f is the frequency, and j is the imaginary unit;

[0128]

[0129] Among them, f c is the frequency center, |X(f)| is the spectrum amplitude;

[0130] S43, time-frequency domain feature extraction: Short-time Fourier transform is used to extract time-frequency domain features, expressed as:

[0131]

[0132] Where X(t,f) is the time-frequency domain signal, w(t-τ) is the window function, and τ is the time delay;

[0133] S44, feature vector generation: feature fusion is performed on the extracted time domain, frequency domain and time-frequency domain feature parameters to generate a feature vector for input of the fault pre-detection model, which is expressed as:

[0134] V=[T1,T2,...,T m ,F1,F2,...,F n ,SF1,SF2,...,SF p ];

[0135] Among them, T1, T2, …, T m represents the extracted time domain feature parameters, F1, F2, …, F n Represents the extracted frequency domain feature parameters, SF1, SF2,…, SF p Represents the extracted time-frequency domain feature parameters;

[0136] Through the above content, the full range of signal information can be captured to ensure that key fault features are not missed, thereby improving the accuracy and robustness of the fault detection model. Combined with a variety of feature extraction methods, it can more effectively identify and predict early faults of wind turbines and improve the reliability and safety of the system.

[0137] The feature vector input in S51 includes:

[0138] S511, standardization of feature vectors: Standardize the generated feature vectors to ensure that different features have the same mean and standard deviation to eliminate the dimensional differences between the features, expressed as:

[0139]

[0140] Among them, V i is the i-th feature in the feature vector, μ i For feature V i The mean value, σ i For feature V i The standard deviation of V i ′ is the standardized eigenvalue;

[0141] S512, normalization of feature vectors: normalize the standardized feature vectors and scale the feature values ​​to a uniform range to enhance the training effect of the model, which is expressed as:

[0142]

[0143] Among them, V i ″ is the normalized eigenvalue, min(V i ′) and max(V i ′) are the characteristics V i The minimum and maximum values ​​of ′;

[0144] S513, construction of model input data: the standardized and normalized feature vector V i “Input into the wind turbine fault prediction model to ensure that the model can be trained and predicted based on the cleaned and optimized data;

[0145] Through the above content, the consistency and comparability of the model input data are ensured, the deviation caused by the dimensional differences between different features is effectively avoided, the training efficiency and stability of the model are enhanced, and it is ensured that the model can show good learning ability and fault detection accuracy on diversified input data, thereby improving the overall reliability and accuracy of wind turbine fault monitoring.

[0146] The wind turbine fault prediction model in S52 adopts the Bayesian neural network (BNN) model, which includes:

[0147] S521, Bayesian reasoning process: In the fault detection of wind turbines, in order to better capture the historical fault modes, the Bayesian reasoning process is improved to introduce weighted prior distribution, so that the model is more biased towards known fault modes and improves the prediction ability in complex environments, which is expressed as:

[0148]

[0149] Among them, p(w|D) is the posterior distribution of weight w after given data D, p(D|w) is the likelihood function, which represents the probability of data D given weight w, p(w) is the prior distribution of weight, Z is the normalization constant, which ensures that the prior distribution is a probability distribution, α,γ k is the adjustment coefficient, which is used to adjust the relationship between the weight and the historical data weight, w k is the model weight corresponding to the historical fault data, K is the number of historical fault data or reference fault modes;

[0150] S522, likelihood estimation based on adaptive kernel function: Adaptive kernel function is introduced to adaptively adjust the variance according to the location of the data point, so that the model can process complex fault data more accurately, which is expressed as:

[0151]

[0152] Among them, p(D|w) is the likelihood function, which represents the probability of data D under weight w. is a normal distribution, indicating that the observed value y i In the predicted value f(x i ,w) nearby probability, is the variance, is the base variance, μ is the mean vector of the data, l is the kernel width parameter, N is the number of data samples, and y i is the observed value of the i-th sample;

[0153] S523, prediction distribution calculation: The prediction distribution is approximated by Monte Carlo sampling, processes complex data structures, and provides predictions of fault occurrence, expressed as:

[0154]

[0155] Among them, p(y * |x * ,D) is the predicted distribution, which means that given the input x * and data D, the output is y * The probability of M is the number of samplings, w m is the weight sampled from the posterior distribution p(w|D);

[0156] S524, Fault Detection and Uncertainty Quantification: Provides a quantitative assessment of uncertainty by calculating the variance of the predicted values, expressed as:

[0157]

[0158]

[0159] Among them, Uncertainty(x * ) is the uncertainty measure, which means that when input x * The variance of the prediction is (x * ) is the mean of the predicted values, f(x * ,w m ) is the mth sampling weight w m The predicted value under ;

[0160] S525, model update and adaptive adjustment: through incremental learning and adaptive adjustment of learning rate, the model is updated in real time to maintain the high efficiency of the model, which is expressed as:

[0161]

[0162] Among them, w t+1 is the updated weight, wt is the current weight, η is the learning rate, which is adaptively adjusted with time step t, D t is the data of the current time step t, η0 is the initial learning rate, and λ is the learning rate attenuation factor;

[0163] Through the above content, the adaptability and accuracy of the model in processing nonlinear and high-dimensional data are enhanced, while the uncertainty quantification of the prediction results is provided, which helps to make more reliable decisions when detecting faults at an early stage. It can dynamically adjust itself to cope with changing fault modes, improve the robustness of the system and the accuracy of fault warning, and thus provide strong support for the efficient operation and maintenance of wind turbines.

[0164] Model training and validation in S53 include:

[0165] S531, data set division: dividing the historical strain signals and deformation data of the wind turbine into a training set and a validation set according to a predetermined ratio;

[0166] S532, cross validation: During the model training process, k-fold cross validation is used to divide the training set into k subsets, and k-1 subsets are used for training. The remaining subset is used for validation. This is repeated k times, and a different subset is selected as the validation set each time. The average value is finally taken as the evaluation standard of the model performance, which is expressed as:

[0167]

[0168] Among them, k is the number of cross-validation folds, the value is 5 or 10, Accuracy j is the model accuracy on the j-fold validation set;

[0169] S533, Hyperparameter Optimization: Grid search is used for hyperparameter optimization, expressed as:

[0170]

[0171] Among them, θ * is the best hyperparameter combination, arg min is to find the parameter value that minimizes the subsequent expression, θ is the current hyperparameter combination, Θ grid is a predefined hyperparameter grid containing all hyperparameter combinations to be evaluated, k is the number of cross-validation folds, is the loss function used to measure the model prediction value and the true value y j The difference between (mean square error), is the output value predicted on the j-fold validation set using the hyperparameter combination θ, y j is the true label on the j-th fold validation set;

[0172] Through the above content, all possible hyperparameter combinations can be comprehensively evaluated to avoid missing the best configuration. At the same time, cross-validation effectively prevents overfitting problems, thereby ensuring the stable performance of the model on different data sets, improving the reliability and prediction accuracy of the model in practical applications, and providing strong technical support for early fault detection and prevention of wind turbines.

[0173] Real-time monitoring and fault warning in S6 include:

[0174] S61, real-time signal monitoring: During the operation of the wind turbine, the strain signals and deformation data of the rotating parts (such as the main shaft, gearbox, bearings, etc.) are monitored in real time, and these signals are continuously collected through sensors to ensure timely acquisition of dynamic stress and deformation information of the rotating parts under different operating conditions;

[0175] S62, comparative analysis of data and model prediction results: input the real-time collected strain signals and deformation data into the trained wind turbine fault pre-detection model, and output the current fault state or health risk probability;

[0176] S63, fault warning generation and information output: When the real-time prediction results show the existence of a potential fault (the predicted fault probability exceeds the set detection threshold), a fault warning signal is immediately generated, including the type of fault (such as bearing fault, gearbox fault, etc.), the location of the fault, and the development trend of the fault (such as the speed of fault deterioration);

[0177] Through the above content, downtime and maintenance costs are reduced, the service life of the equipment is extended, and the real-time prediction model ensures the reliability and accuracy of the early warning through comparative analysis with the preset thresholds, making the operation of wind turbines more efficient, intelligent and safe.

[0178] The detection threshold is set based on statistical analysis of historical data, including:

[0179] Calculate the mean and standard deviation of features: extract strain signals and deformation data from historical data, calculate the mean and standard deviation, and use them to describe the feature distribution under normal operating conditions;

[0180] Set probability threshold: According to the mean and standard deviation, determine the probability distribution of the feature value in the normal state, and identify anomalies by setting the probability threshold, which is expressed as:

[0181] T=μ+z α / 2 σ;

[0182] Where α is the confidence level, which is 0.05 (corresponding to a 95% confidence interval) or 0.01 (corresponding to a 99% confidence interval), and z α / 2is the critical value of the standard normal distribution, T is the detection threshold, and feature values ​​exceeding this threshold will be considered abnormal.

[0183] like Figure 2 As shown, the wind turbine fault monitoring device based on operation data is used to implement the above-mentioned wind turbine fault monitoring method based on operation data, and includes the following units:

[0184] Sensor arrangement and installation unit: magnetostrictive sensors are arranged on the rotating parts of the wind turbine to detect the tiny strains and deformations of the rotating parts during operation;

[0185] Data acquisition unit: collects strain signals and deformation data of rotating parts of wind turbines in real time through magnetostrictive sensors;

[0186] Data preprocessing unit: preprocess the collected strain signals and deformation data, including signal noise reduction, outlier removal and data smoothing, and optimize the original strain signals and deformation data using adaptive filtering algorithms;

[0187] Fault feature extraction unit: extracts fault feature parameters based on the preprocessed strain signal and deformation data, including frequency domain features, time domain features and time-frequency domain features, and generates feature vectors for model input;

[0188] Model building and training unit: Build a wind turbine fault pre-detection model based on the constructed feature vector, and train the model in combination with the wind turbine historical strain signal and deformation data and the fault case library to identify early fault signs and predict the probability of fault occurrence;

[0189] Real-time monitoring and fault warning unit: During the operation of the wind turbine, the strain signals and deformation data of the rotating parts are monitored in real time, and predictions are made through the trained wind turbine fault pre-detection model. When signs of potential faults are detected, a fault warning signal is generated, and the fault type, location and development trend are output.

[0190] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0191] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A wind turbine fault monitoring method based on operation data, characterized in that: The following steps are involved: S1, sensor arrangement and installation: magnetostrictive sensors are arranged on the rotating parts of the wind turbine to detect the tiny strain and deformation of the rotating parts during operation; S2, data acquisition: real-time acquisition of strain signals and deformation data of rotating parts of wind turbines through magnetostrictive sensors; S3, data preprocessing: preprocessing the collected strain signals and deformation data, including signal noise reduction, outlier removal and data smoothing, and optimizing the original strain signals and deformation data using adaptive filtering algorithms; S4, fault feature extraction: based on the preprocessed strain signal and deformation data, extract fault feature parameters, including frequency domain features, time domain features and time-frequency domain features, and generate feature vectors for model input; S5, model construction and training: Based on the constructed feature vector, a wind turbine fault pre-detection model is constructed, and the model is trained in combination with the wind turbine historical strain signal and deformation data and the fault case library to identify early fault signs and predict the probability of fault occurrence, including: S51, feature vector input: using the generated feature vector as input data of the model, and performing standardization and normalization on the input data; S52, model construction: constructing a wind turbine fault pre-detection model based on the processed feature vector; S53, model training and validation: dividing the historical strain signals and deformation data of the wind turbine into a training set and a validation set, using the data of the training set to train the model through cross-validation and hyperparameter optimization, and using the validation set to validate the model; S6, real-time monitoring and fault warning: During the operation of the wind turbine, the strain signals and deformation data of the rotating parts are monitored in real time, and predicted through the trained wind turbine fault pre-detection model. When signs of potential faults are detected, a fault warning signal is generated, and the fault type, location and development trend are output.

2. The wind turbine fault monitoring method based on operation data according to claim 1 is characterized in that: The arrangement and installation of sensors in S1 include: S11, sensor placement location selection: placing a magnetostrictive sensor on a surface or a predetermined position inside a rotating component of a wind turbine; S12, sensor fixing method: The magnetostrictive sensor is fixed to the rotating part by non-contact or contact mounting. The non-contact mounting adopts magnetic bracket or adhesion technology, and the contact mounting adopts mechanical clamp or fastener fixing; S13, sensor calibration and adjustment: After the magnetostrictive sensor is installed, preliminary calibration and zero point adjustment are performed to compensate for displacement or error during the installation process.

3. The wind turbine fault monitoring method based on operation data according to claim 1 is characterized in that: The data collection in S2 includes: S21, strain signal acquisition: the strain signals of the rotating parts of the wind turbine are collected in real time through magnetostrictive sensors, including axial strain, radial strain and torsional strain; S22, deformation data collection: collecting deformation data of the rotating component through a magnetostrictive sensor, including bending deformation, stretching deformation and shear deformation of the component; S23, data acquisition synchronization and storage: the collected strain signals and deformation data are time-stamped and stored in a local or cloud data storage unit in real time.

4. The wind turbine fault monitoring method based on operation data according to claim 1, characterized in that: The data preprocessing in S3 includes: S31, signal denoising: using wavelet transform to perform signal denoising on the collected strain signal and deformation data; S32, outlier elimination: identifying and eliminating outliers in strain signals and deformation data by statistical methods; S33, data smoothing: using sliding average algorithm to smooth the signal; S34, adaptive filtering algorithm optimization: the pre-processed strain signal and deformation data are optimized using the LMS adaptive filtering algorithm.

5. The wind turbine fault monitoring method based on operation data according to claim 1, characterized in that: The fault feature extraction in S4 includes: S41, time domain feature extraction: extracting time domain feature parameters based on the preprocessed strain signal and deformation data, the time domain features include mean, root mean square value, skewness and kurtosis; S42, frequency domain feature extraction: performing Fourier transform on the preprocessed strain signal and deformation data to extract frequency domain feature parameters, including main frequency, frequency center, bandwidth and spectrum energy; S43, time-frequency domain feature extraction: short-time Fourier transform is used to extract time-frequency domain features; S44, feature vector generation: feature fusion is performed on the extracted time domain, frequency domain and time-frequency domain feature parameters to generate a feature vector for input into a fault pre-detection model.

6. The wind turbine fault monitoring method based on operation data according to claim 5, characterized in that: The feature vector input in S51 includes: S511, standardization of feature vectors: standardizing the generated feature vectors to eliminate the dimensional differences between the features; S512, normalization processing of feature vectors: normalization processing is performed on the standardized feature vectors to scale the feature values ​​to a uniform range; S513, construction of model input data: the standardized and normalized feature vector V i ″ is input into the wind turbine fault prediction model.

7. The wind turbine fault monitoring method based on operation data according to claim 6 is characterized in that: The wind turbine fault pre-detection model in S52 adopts a Bayesian neural network model, and the Bayesian neural network model includes: S521, Bayesian reasoning process: In the fault detection of wind turbines, a weighted prior distribution is introduced, which is expressed as: Among them, p(w|D) is the posterior distribution of weight w after given data D, p(D|w) is the likelihood function, p(w) is the prior distribution of weight, Z is the normalization constant, α, γ k is the adjustment coefficient, w k is the model weight corresponding to the historical fault data, K is the number of historical fault data or reference fault modes; S522, likelihood estimation based on adaptive kernel function: an adaptive kernel function is introduced to adaptively adjust the variance according to the position of the data point, which is expressed as: Among them, p(D|w) is the likelihood function, is a normal distribution, is the variance, is the base variance, μ is the mean vector of the data, l is the kernel width parameter, N is the number of data samples, and y i is the observed value of the i-th sample; S523, prediction distribution calculation: The prediction distribution is approximated by Monte Carlo sampling, processes complex data structures, and provides predictions of fault occurrence, expressed as: Among them, p(y * |x * ,D) is the predicted distribution, which means that given the input x * and data D, the output is y * The probability of M is the number of samplings, w m is the weight sampled from the posterior distribution p(w|D); S524, Fault Detection and Uncertainty Quantification: Provides a quantitative assessment of uncertainty by calculating the variance of the predicted values, expressed as: Among them, Uncertainty(x * ) is the uncertainty measure, which means that when input x * The variance of the prediction is is the mean of the predicted values, f(x * ,w m ) is the mth sampling weight w m The predicted value under ; S525, model update and adaptive adjustment: The model is updated in real time through incremental learning and adaptive adjustment of learning rate, which is expressed as: Among them, w t+1 is the updated weight, w t is the current weight, η is the learning rate, D t is the data of the current time step t, η0 is the initial learning rate, and λ is the learning rate attenuation factor.

8. The wind turbine fault monitoring method based on operation data according to claim 7 is characterized in that: The model training and verification in S53 includes: S531, data set division: dividing the historical strain signals and deformation data of the wind turbine into a training set and a validation set according to a predetermined ratio; S532, cross validation: During the model training process, use k-fold cross validation to divide the training set into k subsets, use k-1 subsets for training, and use the remaining subset for validation. Repeat k times, each time selecting a different subset as the validation set, and finally take the average value as the evaluation standard of model performance; S533, Hyperparameter Optimization: Grid search is used for hyperparameter optimization, expressed as: Among them, θ * is the best hyperparameter combination, arg min is to find the parameter value that minimizes the subsequent expression, θ is the current hyperparameter combination, Θ grid is a predefined hyperparameter grid, k is the number of cross-validation folds, is the loss function, is the output value predicted on the j-fold validation set using the hyperparameter combination θ, y j is the true label on the j-th fold validation set.

9. The wind turbine fault monitoring method based on operation data according to claim 1, characterized in that: The real-time monitoring and fault warning in S6 include: S61, real-time signal monitoring: during the operation of the wind turbine, real-time monitoring of the strain signals and deformation data of the rotating components; S62, comparative analysis of data and model prediction results: input the real-time collected strain signals and deformation data into the trained wind turbine fault pre-detection model, and output the current fault state or health risk probability; S63, fault warning generation and information output: When the real-time prediction result shows the existence of a potential fault, a fault warning signal is immediately generated, including the type of fault, the location where the fault occurs, and the development trend of the fault.

10. A wind turbine fault monitoring device based on operating data, used to implement a wind turbine fault monitoring method based on operating data as claimed in any one of claims 1 to 9, characterized in that: include: Sensor arrangement and installation unit: magnetostrictive sensors are arranged on the rotating parts of the wind turbine to detect the tiny strains and deformations of the rotating parts during operation; Data acquisition unit: collects strain signals and deformation data of rotating parts of wind turbines in real time through magnetostrictive sensors; Data preprocessing unit: preprocess the collected strain signals and deformation data, including signal noise reduction, outlier removal and data smoothing, and optimize the original strain signals and deformation data using adaptive filtering algorithms; Fault feature extraction unit: extracts fault feature parameters based on the preprocessed strain signal and deformation data, including frequency domain features, time domain features and time-frequency domain features, and generates feature vectors for model input; Model building and training unit: Build a wind turbine fault pre-detection model based on the constructed feature vector, and train the model in combination with the wind turbine historical strain signal and deformation data and the fault case library to identify early fault signs and predict the probability of fault occurrence; Real-time monitoring and fault warning unit: During the operation of the wind turbine, the strain signals and deformation data of the rotating parts are monitored in real time, and predictions are made through the trained wind turbine fault pre-detection model. When signs of potential faults are detected, a fault warning signal is generated, and the fault type, location and development trend are output.

Citation Information

Patent Citations

  • Wind turbine generator component fault prediction method

    CN112733283A

  • Fan main bearing fault early warning method based on XGBoost-KDE

    CN115687864A

  • Wind turbine generator fault prediction method and device and electronic equipment

    CN117034001A

  • System and method for monitoring damage of a bearing for a wind turbine

    US20240200535A1

  • Rolling bearing fault diagnosis method based on grcmse and manifold learning

    WO2021135630A1

Cited By

  • Parameter control method of ultrasonic motor

    CN119341433A

  • A parameter control method for ultrasonic motor

    CN119341433B