CN118167569A - Wind turbine generator blade abnormality detection method and device based on vibration - Google Patents
Wind turbine generator blade abnormality detection method and device based on vibration Download PDFInfo
- Publication number
- CN118167569A CN118167569A CN202410564355.7A CN202410564355A CN118167569A CN 118167569 A CN118167569 A CN 118167569A CN 202410564355 A CN202410564355 A CN 202410564355A CN 118167569 A CN118167569 A CN 118167569A
- Authority
- CN
- China
- Prior art keywords
- vibration
- scada
- wind turbine
- data
- blade
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 131
- 230000005856 abnormality Effects 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 claims abstract description 61
- 238000012549 training Methods 0.000 claims abstract description 52
- 238000000605 extraction Methods 0.000 claims abstract description 42
- 238000005516 engineering process Methods 0.000 claims abstract description 29
- 230000002159 abnormal effect Effects 0.000 claims abstract description 26
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 22
- 238000000513 principal component analysis Methods 0.000 claims abstract description 16
- 230000008569 process Effects 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 16
- 238000003860 storage Methods 0.000 claims description 13
- 238000005457 optimization Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 10
- 102000008297 Nuclear Matrix-Associated Proteins Human genes 0.000 claims description 9
- 108010035916 Nuclear Matrix-Associated Proteins Proteins 0.000 claims description 9
- 210000000299 nuclear matrix Anatomy 0.000 claims description 9
- 230000007246 mechanism Effects 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 5
- 230000003190 augmentative effect Effects 0.000 claims description 4
- 238000001228 spectrum Methods 0.000 claims description 4
- 230000001131 transforming effect Effects 0.000 claims description 3
- 239000000306 component Substances 0.000 description 39
- 238000010586 diagram Methods 0.000 description 8
- 230000003044 adaptive effect Effects 0.000 description 6
- 238000004891 communication Methods 0.000 description 6
- 238000010276 construction Methods 0.000 description 5
- 238000002156 mixing Methods 0.000 description 5
- 230000006399 behavior Effects 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 230000002457 bidirectional effect Effects 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 230000001364 causal effect Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 239000008358 core component Substances 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012628 principal component regression Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 238000010408 sweeping Methods 0.000 description 2
- 238000004804 winding Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/70—Wind energy
- Y02E10/72—Wind turbines with rotation axis in wind direction
Landscapes
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The embodiment of the specification discloses a wind turbine generator blade abnormality detection method and device based on vibration. The method comprises the following steps: SCADA data and cabin vibration data of the wind turbine in a target period are obtained; inputting the acquired data into a pre-trained blade abnormality detection model, wherein the blade abnormality detection model comprises a vibration feature extraction layer, an SCADA feature extraction layer and an abnormality detection layer, the vibration feature extraction layer is used for extracting vibration features in the cabin vibration data based on a variation modal decomposition technology and a nuclear principal component analysis technology, the SCADA feature extraction layer is used for extracting SCADA features in the SCADA data according to an ABT network, and the abnormality detection layer is used for outputting a detection result according to a comparison result of a target residual value and a normal residual value range obtained by pre-training; and determining whether the blades of the wind turbine generator are abnormal in the target period according to the detection result output by the blade abnormality identification model.
Description
Technical Field
One or more embodiments of the present disclosure relate to the field of wind power technologies, and in particular, to a method and an apparatus for detecting abnormality of a blade of a wind turbine generator based on vibration.
Background
The wind turbine generator is equipment for converting wind energy into electric energy, and mainly comprises a tower, a cabin, blades, a generator, a control system and the like. The tower is used for supporting the whole wind turbine, the engine room contains core components such as a generator, the blades are key components for capturing wind energy, the power is generated through rotation, the generator is driven to rotate, and finally electric energy is output through the control system.
With the steady promotion of wind power construction, more and more fans are deployed in remote areas such as mountain areas and offshore areas in recent years so as to better utilize wind resources, and meanwhile, wind turbine generators need to face more severe operation and maintenance conditions and more complex weather conditions, so that accidents such as fan blade icing, tower sweeping and blade breaking are more frequent. According to the prior studies, blade failure is mainly caused by the deformation and loosening of the structure of the blade due to stress fatigue load, and finally the blade is broken. The monitoring and data acquisition (SCADA) data of the wind turbine can provide relevant parameters such as wind speed, yaw angle, pitch angle and the like which influence the windward load characteristics of the blade, and deep learning is a main method in a data driving method, and has become more popular in recent years in the aspect of using the SCADA data for detecting the fault of the wind turbine. Therefore, the method and the device can intelligently monitor the blade states of the blades of the wind turbine generator in real time and timely send out abnormal alarms, and have important values in the aspects of reducing operation downtime and reducing operation cost of a wind farm.
In the related art, SCADA data (including parameters such as wind speed, active power, rotor rotation speed, main bearing temperature, gear box oil temperature, gear box bearing temperature, generator front bearing temperature, generator rear bearing temperature, generator winding temperature, generator rotation speed, oil pressure, environmental temperature and the like) are captured through sensors arranged on a wind turbine generator, and whether the blade is abnormal is deduced according to the SCADA data. However, the accuracy of the detection is not high enough, and the phenomena of missed detection and false detection often occur. The related method breaks through the vibration characteristics of the blade faults of the wind turbine unit, lacks of interpretability and is difficult to land in engineering.
Disclosure of Invention
The embodiment of the specification provides a wind turbine generator blade abnormality detection method, device, electronic equipment and storage medium based on vibration, and the technical scheme is as follows:
in a first aspect, an embodiment of the present disclosure provides a method for detecting an abnormality of a blade of a wind turbine generator based on vibration, including:
SCADA data and cabin vibration data of the wind turbine in a target period are obtained;
Inputting the acquired data into a paddle abnormality detection model which is obtained through training in advance, so that the paddle abnormality detection model outputs a detection result; the blade anomaly detection model comprises a vibration feature extraction layer, an SCADA feature extraction layer and an anomaly detection layer, wherein the vibration feature extraction layer is used for extracting vibration features in cabin vibration data based on a variation modal decomposition technology and a nuclear principal component analysis technology, the SCADA feature extraction layer is used for extracting SCADA features in the SCADA data according to an ABT network, the anomaly detection layer is used for outputting detection results according to comparison results of target residual values and a normal residual value range obtained through pre-training, and the target residual values are obtained through calculation according to the extracted vibration features and the SCADA features;
and determining whether the blades of the wind turbine generator are abnormal in the target period according to the detection result output by the blade abnormality identification model.
In a second aspect, an embodiment of the present disclosure provides a training method for a blade anomaly detection model, including:
Obtaining a training dataset, each training sample in the training dataset comprising: SCADA data and cabin vibration data of the wind turbine in the same time period, and real detection results of blades of the wind turbine in the time period;
inputting the training data set into a model to be trained, so that the model to be trained outputs a prediction detection result of the blades of the wind turbine in the period;
and comparing the real detection result with the prediction detection result, and optimizing the model to be trained according to the comparison result.
In a third aspect, embodiments of the present disclosure provide a wind turbine blade anomaly detection device based on vibration, including:
The acquisition unit is used for acquiring SCADA data and cabin vibration data of the wind turbine in a target period;
The input unit is used for inputting the acquired data into a paddle abnormality detection model which is trained in advance, so that the paddle abnormality detection model outputs a detection result; the blade anomaly detection model comprises a vibration feature extraction layer, an SCADA feature extraction layer and an anomaly detection layer, wherein the vibration feature extraction layer is used for extracting vibration features in cabin vibration data based on a variation modal decomposition technology and a nuclear principal component analysis technology, the SCADA feature extraction layer is used for extracting SCADA features in the SCADA data according to an ABT network, the anomaly detection layer is used for outputting detection results according to comparison results of target residual values and a normal residual value range obtained through pre-training, and the target residual values are obtained through calculation according to the extracted vibration features and the SCADA features;
and the determining unit is used for determining whether the blades of the wind turbine generator set are abnormal in the target period according to the detection result output by the blade abnormality identification model.
In a fourth aspect, embodiments of the present disclosure provide a training device for a blade anomaly detection model, including:
An acquisition unit configured to acquire a training data set, where each training sample in the training data set includes: SCADA data and cabin vibration data of the wind turbine in the same time period, and real detection results of blades of the wind turbine in the time period;
The input unit is used for inputting the training data set into a model to be trained so that the model to be trained outputs a prediction detection result of the blades of the wind turbine in the period;
and the optimizing unit is used for comparing the real detection result with the prediction detection result and optimizing the model to be trained according to the comparison result.
In a fifth aspect, embodiments of the present disclosure provide an electronic device comprising a processor and a memory; the processor is connected with the memory; the memory is used for storing executable program codes; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory for performing the steps described in the first aspect or the second aspect of the above embodiment.
In a sixth aspect, embodiments of the present disclosure provide a computer storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the steps of the first or second aspects of the embodiments described above.
The technical scheme provided by some embodiments of the present specification has the following beneficial effects:
On one hand, the method is not limited by single SCADA data any more, but detects the abnormal condition of the blades of the wind turbine based on the SCADA data and the cabin vibration data, so that the accuracy of abnormal detection is improved; on the other hand, the vibration characteristics in the cabin vibration data are extracted based on a variation modal decomposition technology and a nuclear principal component analysis technology, the problem that the vibration signal form is difficult to judge is solved, SCADA characteristics in the SCADA data are extracted according to an ABT network, and the characteristic extraction of the SCADA data with complex, variable and mode mixing is realized.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present description, the drawings that are required in the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present description, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a method for detecting abnormality of a blade of a wind turbine generator based on vibration according to an embodiment of the present disclosure.
Fig. 2 is a schematic diagram of an adaptive bidirectional sequential convolutional network provided in an embodiment of the present disclosure.
Fig. 3 is a schematic diagram of a blade anomaly detection model according to an embodiment of the present disclosure.
Fig. 4 is a flowchart of a training method of a blade anomaly detection model according to an embodiment of the present disclosure.
Fig. 5 is a block diagram of a wind turbine blade abnormality detection device based on vibration according to an embodiment of the present disclosure.
Fig. 6 is a block diagram of a training device for a blade abnormality detection model according to an embodiment of the present disclosure.
Fig. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure.
Detailed Description
The technical solutions in the embodiments of the present specification will be clearly and completely described below with reference to the drawings in the embodiments of the present specification.
The terms first, second and the like in the description and in the claims and in the above-described figures are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order. Furthermore, the term "include" and any variations thereof is intended to cover a non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus.
The method for detecting blade anomalies of a vibration-based wind turbine is described in detail below in connection with one or more embodiments.
The wind turbine generator is equipment for converting wind energy into electric energy, and mainly comprises a tower, a cabin, blades, a generator, a control system and the like. The tower is used for supporting the whole wind turbine, the engine room contains core components such as a generator, the blades are key components for capturing wind energy, the power is generated through rotation, the generator is driven to rotate, and finally electric energy is output through the control system.
Wind turbines are often subjected to severe operating conditions and complex weather, thereby bringing about potential safety hazards. Conventional blade failure is mainly due to stress fatigue loading leading to blade deformation and structural loosening, ultimately leading to blade breakage. In recent years, more and more fans are deployed in remote areas such as mountainous areas and low-temperature offshore areas to utilize better wind resources. Because the wind conditions of such environments are more complex, accidents such as fan blade icing, tower sweeping, blade breaking and the like are more frequent. Therefore, the method and the device can intelligently monitor the blade states of the blades of the wind turbine generator in real time and timely send out abnormal alarms, and have important values in the aspects of reducing operation downtime and reducing operation cost of a wind farm.
In the related art, SCADA data (including parameters such as wind speed, active power, rotor rotation speed, main bearing temperature, gear box oil temperature, gear box bearing temperature, generator front bearing temperature, generator rear bearing temperature, generator winding temperature, generator rotation speed, oil pressure, environmental temperature and the like) are captured through sensors arranged on a wind turbine generator, and whether the blade is abnormal is deduced according to the SCADA data. However, the accuracy of the detection is not high enough, and the phenomena of missed detection and false detection often occur.
In order to solve the problems in the related art, the specification provides a method, a device, electronic equipment and a storage medium for detecting the abnormality of a blade of a wind turbine generator based on vibration.
Referring to fig. 1, fig. 1 shows a flow chart of a vibration-based wind turbine blade abnormality detection method provided in the embodiments of the present disclosure, and as shown in fig. 1, the vibration-based wind turbine blade abnormality detection method may at least include the following steps:
Step 102, SCADA data and cabin vibration data of the wind turbine in a target period are obtained.
The SCADA (Supervisory Control and Data Acquisition ) system is a computer system for industrial control that monitors and controls connected devices remotely. SCADA systems are commonly used in various types of industrial processes including the oil and gas industry, water treatment facilities, electrical power systems, traffic control, manufacturing, and the like.
The main functions of the SCADA system include: collecting data from field devices and sensors in real time, such as temperature, pressure, flow, fluid level, etc.; according to a preset program or an operator command, the field device is remotely controlled, such as opening or closing a valve, starting or stopping a pump and the like; recording and storing a large amount of process data for historical analysis, trend prediction, fault diagnosis and the like; alarm and event management: abnormal conditions such as equipment failure, process parameters exceeding a predetermined range, etc., are monitored and an alarm is triggered to notify an operator.
SCADA data typically contains information from various sensors and control units, which may be transmitted to a central control room via a wired or wireless communication network. The SCADA system can monitor and collect parameters of main bearing temperature, gear box oil temperature, generator bearing temperature, oil pressure, generator rotation speed, torque and other operation working condition parameters such as wind speed, wind direction, environmental temperature and the like related to the operation state of the transmission system through different types of sensors.
Nacelle vibration data may be collected by a condition monitoring system (Condition Monitoring System, CMS) that primarily collects vibration signals to monitor the operational condition of critical components such as the main bearings, gearbox, generator, etc.
104, Inputting the acquired data into a paddle abnormality detection model which is trained in advance, so that the paddle abnormality detection model outputs a detection result; the blade anomaly detection model comprises a vibration feature extraction layer, an SCADA feature extraction layer and an anomaly detection layer, wherein the vibration feature extraction layer is used for extracting vibration features in cabin vibration data based on a variation modal decomposition technology and a nuclear principal component analysis technology, the SCADA feature extraction layer is used for extracting SCADA features in the SCADA data according to an ABT network, the anomaly detection layer is used for outputting detection results according to comparison results of target residual values and a normal residual value range obtained through training in advance, and the target residual values are obtained through calculation according to the extracted vibration features and the SCADA features.
The process of the vibration characteristics extracted by the vibration characteristic extraction layer comprises the following steps: aiming at each channel vibration signal in the cabin vibration data, decomposing a single channel vibration signal into a plurality of modal components, and establishing a constraint optimization problem according to a component narrowband condition; the constraint optimization problem is equivalent to an unconstrained optimization problem based on an augmented lagrange function, and the plurality of modal components and the center frequencies corresponding to the plurality of modal components are calculated based on an alternate direction multiplier method; performing Fourier transform on each modal component, and extracting principal components of a frequency spectrum obtained by transforming each modal component based on a kernel principal component analysis technology; performing centering treatment on the extracted nuclear matrix to obtain a standardized nuclear matrix, and performing eigenvalue decomposition on the standardized nuclear matrix; and determining the vibration characteristics corresponding to each modal variable according to the maximum characteristic value obtained by decomposition and the characteristic vector corresponding to the maximum characteristic value.
The process of extracting the vibration characteristics by the vibration characteristic extraction layer is described in detail below in connection with the calculation formula. First assume that: in an input window of length w, the multichannel vibration signal (i.e. nacelle vibration data) isThe SCADA timing variable multiple timing input (i.e., SCADA data) related to the vibration signal is/>Wherein C is the number of vibration signal channels, N is the dimension of the SCADA related variable, and N is the total number of samples.
The single-channel vibration signal isThe vibration signal can be decomposed into pending/>Each modal component. In the decomposition, it can be assumed that all modal components are concentrated at the respective center frequency/>Nearby narrowband signals, therefore, variational Modal Decomposition (VMD) creates constraint optimization problems based on component narrowband conditions, as follows:
；
after which by an augmented lagrangian function The equation constraint optimization problem is equivalent to the unconstrained optimization problem, and the equation is as follows:
；
And solving pairs by the alternate direction multiplier method ADMM And/>And (3) iteration solution:
Wherein the method comprises the steps of And/>Representing the n+1st iteration result for the c-th vibration channel signal, respectively, r represents the learning step size. Here, the final converged modal decomposition result is expressed as/>。
In extracting c channelsAfter the modal components, carrying out Fourier transformation on each modal component to obtain the frequency spectrum/>, of each modal componentAnd extracting principal components by adopting a nuclear principal component analysis technology (kPCA) method.
Secondly, the core matrix is subjected to centering treatment to obtain a standardized core matrix：
；
Wherein,Is one/>Of which all elements are/>。
Next, for a normalized kernel matrixDecomposing the characteristic values, selecting the largest characteristic value and the corresponding characteristic vector to construct the principal component characteristic after dimension reduction, and/>, of the principal component characteristic of the kth mode of the c vibration channel of the ith sampleCan be expressed as:
Wherein the method comprises the steps of Respectively represent the maximum eigenvalue and the corresponding eigenvector obtained by eigenvalue decomposition,/>Representing a high-dimensional mapping function in the kernel function, a gaussian kernel function that can be used in the present model.
In an embodiment, the ABT network is built based on a dynamic convolution kernel mechanism.
The dynamic convolution kernel construction mechanism is a strategy for adaptive learning features in a Convolutional Neural Network (CNN) that allows the network to dynamically adjust the size or shape of the convolution kernel during training. The mechanism can make the network more flexible, adapt to different types of data and tasks and improve the generalization capability of the model.
ABT (Adaptive Bidirectional Temporal) an adaptive bi-directional sequential convolutional network is a deep learning architecture for processing sequential data. Such network architecture is designed to better capture time dynamics and sequence features in the time series data. The ABT network introduces an adaptive mechanism based on a conventional bidirectional sequential convolutional network (BTCN) to improve the flexibility and performance of the model. ABT networks employ a "time-reversal" technique in the original TCN architecture, i.e., a bi-directional convolution pattern that combines time forward and time backward convolution. Subsequently, the timing characteristics from both directions can be combined using the fully connected layer to maintain the overall dimensions of the timing characteristics.
Because of the complex time-varying nature of SCADA data, it is difficult for a convolution kernel with fixed parameters to meet the requirements of feature extraction. In this embodiment, the ABT network employs a dynamic convolution kernel construction mechanism to replace the constant convolution kernel in the hole causal convolution.
First, the present specification replaces the constant convolution kernel in the hole causal convolution by an adaptive convolution kernel, oriented to the complex time-varying characteristics of SCADA data. As shown in fig. 2, the input X may be twice subjected to i-average pooling of feature dimensions and time dimensions to obtain a stitched 1-dimensional feature vector. The feature vector covers the timing information and feature dimension information of the whole input sample. And sending the complex vector into nonlinear transformation consisting of a forward network and a sigmoid activation function, and constructing a convolution kernel parameter which is finally adopted in convolution calculation. The benefit of using such a nonlinear transformation is that limiting the convolution kernel parameters to between-1 and 1 limits the dramatic changes in parameters in the network, thus suppressing the gradient vanishing problem. The mathematical expression of the above convolution kernel construction process is as follows:
；
Wherein, For network input,/>To get the convolution kernel parameters,/>And/>The weights and biases are to be optimized.
It can be seen that the convolution kernel parameters constructed by the above method are changed along with the change of the input samples, wherein the parameters are dynamically and adaptively changed along with the input, so that the complex timing characteristics can be extracted more effectively.
Secondly, the model aims at integrating complete time sequence information of an input sample and returning to global vibration principal component characteristics. Thus, a bi-directional convolution pattern of time forward and backward convolutions can be employed to extract bi-directional timing features. The timing features from both directions are then fused using a fully connected network to maintain the overall dimensions of the network. The mathematical expression of the above model structure is as follows:
Wherein, Forward and backward dynamic convolution kernel parameters, respectively,/>Is a timing characteristic of the network output.
In the embodiment, the SCADA characteristics in the SCADA data are extracted through the ABT network constructed based on the dynamic convolution kernel mechanism, so that the characteristic extraction of the SCADA data with complex and changeable patterns is realized.
In an embodiment, self-coding normal behavior modeling is performed on other SCADA variables in the wind turbine that are not directly related to blade behavior, to exclude effects of other components outside the blade on nacelle acceleration.
In a normal behavior model of the wind turbine, a multi-layer gating circulating unit (Gated Recurrent Unit, GRU) is adopted to encode other related SCADA operation variables. For input data samplesAt each time t, the GRU can generate a hidden state/>By updating gate/>And reset gate/>Control, which is calculated as follows:
；
Wherein the method comprises the steps of Is a parameter that needs to be learned. hs is the size of the hidden state and m is the number of SCADA variables.
The input of each GRU is from the hidden characteristic h of the previous GRU and the hidden characteristic of the output of the last GRUIs fed into a decoder of fully connected layer construction to obtain reconstructed values for the original input SCADA related variables, namely:
；
Wherein, Is the parameter that needs to be learned in a fully connected decoder. The self-encoder training is trained using the MSE average squared error of the reconstructed values and the original input samples as residuals.
In one embodiment, the calculating process of the target residual value includes: and calculating the difference value of the vibration characteristic and the SCADA characteristic in the same time dimension, and constructing a Markov distance space to quantify the difference value so as to obtain a target residual value. Further, the process of outputting the detection result by the anomaly detection layer according to the comparison result includes: if the number of time points, of which the target residual error value exceeds the normal residual error value range, reaches a preset threshold value, the detection result is abnormal; if the number of time points, of which the target residual value exceeds the normal residual value range, is smaller than a preset threshold value, the detection result is normal.
Since the vibration data is multi-channel, the residual values may also be referred to as residual sequences. Residual sequenceCan be expressed as follows:
；
Wherein, Representing the true value of the principal component feature of the vibration (namely the vibration feature extracted by the vibration feature extraction layer), and/>Refers to the estimated value of the vibration principal component feature (namely, the SCADA feature extracted by the SCADA feature extraction layer).
The memory containsPersonal/>Residual sequence of dimensional feature samples/>The residual of the kth training sample in (1) is/>Its mahalanobis distance/>Can be defined as:
；
Wherein, Respectively mean and covariance matrix of training set residual sequences,/>Representing the residual dimension. The above Markov distance calculation method can be used to obtain the sum/>And/>Constructed vibration principal component residual error distance sequence/>And reconstructing residual distance sequence/>。
According to the definition of the mahalanobis distance, the constructed distance sequence is a complete positive number sequence, and the distribution formed by the sequences belongs to single-side unimodal distribution. Thus, the 99 quantiles of the two distance sequences are calculatedAnd treat them as overrun thresholds. This means that it is considered that when the vibration principal component residual distance is greater than/>When the current sample belongs to a cabin vibration suspicious state; when the reconstructed residual distance is greater than/>When the current sample belongs to the running "suspicious" state.
Because of noise interference of SCADA data of the wind turbine, false alarm is reduced in practical application, and each suspicious state cannot be directly reported as an abnormality. Thus, statistics is performed on the principal component feature residual distance sequence in the training setContinuously exceeding a threshold/>, in the time dimensionAnd is described as/>. In the practical application stage, for a real-time vibration related SCADA variable sample/>, within the same time length wMultichannel cabin acceleration data sample/>And other SCADA related variable samples/>The real-time vibration principal component estimated value and the real value are obtained through the principal component regression model and the vibration principal component extraction module which are completed through training, and the error/> iscalculatedAnd converted to a mahalanobis distance/> according to the reconstruction values described above; Obtaining reconstruction errors through a normal behavior model completed through training and marking the reconstruction errors as/>And converted to a mahalanobis distance/> according to the reconstruction values described above。
Based on the above conditions, when continuousIndividual real-time data samples satisfy/>And/>When the blade is abnormal, the strategy reports out a blade abnormality warning. At this time, the fan state represented by the sample is in abnormal cabin vibration, and the rest parts which are not related to the blades are in normal states, so that the abnormal cabin vibration can be considered to be caused by one side of the blades, namely, an alarm is given.
And step 106, determining whether the blades of the wind turbine generator set are abnormal in the target period according to the detection result output by the blade abnormality recognition model.
The overall architecture of the blade anomaly detection model is shown in fig. 3, and four continuous ABT layers are used as vibration principal component regressions (used for extracting SCADA features, namely vibration principal component estimated values), so that the network gradually extracts time sequence features of local and global multiple time scales from shallow to deep. In the multi-scale time sequence feature fusion, a cross-layer connection mode is used for fusing local and global features so as to accelerate the model convergence speed and solve the gradient disappearance problem of a depth network. On the other hand, the vibration principal component characteristics after the VMD-kPCA are used as target variables for regression of the self-supervision model. In the training process, a training residual is composed of a target estimated value output by a feature extractor composed of ABT and a real target variable, and an average square error is adopted as a loss function. And finally, outputting residual sequences of vibration principal component characteristics of all time sequence samples, and determining whether the blade is abnormal according to the residual sequences.
In the embodiment, on one hand, the method is not limited to single SCADA data, but detects the abnormal blade of the wind turbine based on the SCADA data and the nacelle vibration data, so that the accuracy of abnormality detection is improved; on the other hand, the vibration characteristics in the cabin vibration data are extracted based on a variation modal decomposition technology and a nuclear principal component analysis technology, the problem that the vibration signal form is difficult to judge is solved, SCADA characteristics in the SCADA data are extracted according to an ABT network, and the characteristic extraction of the SCADA data with complex, variable and mode mixing is realized.
The specification also provides a training method of the blade abnormality detection model.
Referring to fig. 4, fig. 4 is a flow chart illustrating a training method of a blade abnormality detection model according to an embodiment of the present disclosure, and as shown in fig. 4, the training method of the blade abnormality detection model may at least include the following steps:
Step 402, acquiring a training data set, wherein each training sample in the training data set comprises: SCADA data and cabin vibration data of the wind turbine in the same time period, and real detection results of blades of the wind turbine in the time period;
step 404, inputting the training data set to a model to be trained, so that the model to be trained outputs a predicted detection result of the blades of the wind turbine in the period;
Step 406, comparing the real detection result with the prediction detection result, and optimizing the model to be trained according to the comparison result.
In the embodiment, on one hand, the method is not limited to single SCADA data, but detects the abnormal blade of the wind turbine based on the SCADA data and the nacelle vibration data, so that the accuracy of abnormality detection is improved; on the other hand, the vibration characteristics in the cabin vibration data are extracted based on a variation modal decomposition technology and a nuclear principal component analysis technology, the problem that the vibration signal form is difficult to judge is solved, SCADA characteristics in the SCADA data are extracted according to an ABT network, and the characteristic extraction of the SCADA data with complex, variable and mode mixing is realized.
Referring to fig. 5, fig. 5 is a block diagram of a wind turbine blade abnormality detection device based on vibration according to an embodiment of the present disclosure. The device comprises:
an acquiring unit 502, configured to acquire SCADA data and nacelle vibration data of a wind turbine generator in a target period;
An input unit 504, configured to input the acquired data into a paddle abnormality detection model that is trained in advance, so that the paddle abnormality detection model outputs a detection result; the blade anomaly detection model comprises a vibration feature extraction layer, an SCADA feature extraction layer and an anomaly detection layer, wherein the vibration feature extraction layer is used for extracting vibration features in cabin vibration data based on a variation modal decomposition technology and a nuclear principal component analysis technology, the SCADA feature extraction layer is used for extracting SCADA features in the SCADA data according to an ABT network, the anomaly detection layer is used for outputting detection results according to comparison results of target residual values and a normal residual value range obtained through pre-training, and the target residual values are obtained through calculation according to the extracted vibration features and the SCADA features;
And the determining unit 506 is configured to determine whether the blade of the wind turbine generator is abnormal within the target period according to the detection result output by the blade abnormality recognition model.
Optionally, the process of the vibration feature extracted by the vibration feature extraction layer includes:
Aiming at each channel vibration signal in the cabin vibration data, decomposing a single channel vibration signal into a plurality of modal components, and establishing a constraint optimization problem according to a component narrowband condition;
The constraint optimization problem is equivalent to an unconstrained optimization problem based on an augmented lagrange function, and the plurality of modal components and the center frequencies corresponding to the plurality of modal components are calculated based on an alternate direction multiplier method;
Performing Fourier transform on each modal component, and extracting principal components of a frequency spectrum obtained by transforming each modal component based on a kernel principal component analysis technology;
Performing centering treatment on the extracted nuclear matrix to obtain a standardized nuclear matrix, and performing eigenvalue decomposition on the standardized nuclear matrix;
And determining the vibration characteristics corresponding to each modal variable according to the maximum characteristic value obtained by decomposition and the characteristic vector corresponding to the maximum characteristic value.
Optionally, the ABT network is constructed based on a dynamic convolution kernel mechanism.
Optionally, the calculating process of the target residual value includes:
and calculating the difference value of the vibration characteristic and the SCADA characteristic in the same time dimension, and constructing a Markov distance space to quantify the difference value so as to obtain a target residual value.
Optionally, the process of outputting the detection result by the anomaly detection layer according to the comparison result includes:
If the number of time points, of which the target residual error value exceeds the normal residual error value range, reaches a preset threshold value, the detection result is abnormal;
If the number of time points, of which the target residual value exceeds the normal residual value range, is smaller than a preset threshold value, the detection result is normal.
In the embodiment, on one hand, the method is not limited to single SCADA data, but detects the abnormal blade of the wind turbine based on the SCADA data and the nacelle vibration data, so that the accuracy of abnormality detection is improved; on the other hand, the vibration characteristics in the cabin vibration data are extracted based on a variation modal decomposition technology and a nuclear principal component analysis technology, the problem that the vibration signal form is difficult to judge is solved, SCADA characteristics in the SCADA data are extracted according to an ABT network, and the characteristic extraction of the SCADA data with complex, variable and mode mixing is realized.
Referring to fig. 6, fig. 6 is a block diagram of a training device for a blade anomaly detection model according to an embodiment of the present disclosure. The device comprises:
An obtaining unit 602, configured to obtain a training data set, where each training sample in the training data set includes: SCADA data and cabin vibration data of the wind turbine in the same time period, and real detection results of blades of the wind turbine in the time period;
the input unit 604 is configured to input the training data set to a model to be trained, so that the model to be trained outputs a predicted detection result of the blades of the wind turbine within the period;
and the optimizing unit 606 is configured to compare the real detection result with the predicted detection result, and optimize the model to be trained according to the comparison result.
In the embodiment, on one hand, the method is not limited to single SCADA data, but detects the abnormal blade of the wind turbine based on the SCADA data and the nacelle vibration data, so that the accuracy of abnormality detection is improved; on the other hand, the vibration characteristics in the cabin vibration data are extracted based on a variation modal decomposition technology and a nuclear principal component analysis technology, the problem that the vibration signal form is difficult to judge is solved, SCADA characteristics in the SCADA data are extracted according to an ABT network, and the characteristic extraction of the SCADA data with complex, variable and mode mixing is realized.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are mutually referred to, and each embodiment mainly describes differences from other embodiments. In particular, for the embodiment of the vibration-based wind turbine blade abnormality detection device, since the embodiment is substantially similar to the embodiment of the vibration-based wind turbine blade abnormality detection method, the description is relatively simple, and the relevant points are referred to in the description of the method embodiment.
Please refer to fig. 7, which illustrates a schematic structural diagram of an electronic device according to an embodiment of the present disclosure.
As shown in fig. 7, the electronic device 700 may include: at least one processor 701, at least one network interface 704, a user interface 703, memory 705, and at least one communication bus 702.
Wherein the communication bus 702 may be used to facilitate communications among the various components described above.
The user interface 703 may include keys, and the optional user interface may also include a standard wired interface, a wireless interface, among others.
The network interface 704 may include, but is not limited to, a bluetooth module, an NFC module, a Wi-Fi module, and the like.
Wherein the processor 701 may include one or more processing cores. The processor 701 utilizes various interfaces and lines to connect various portions of the overall electronic device 700, perform various functions of the electronic device 700, and process data by executing or executing instructions, programs, code sets, or instruction sets stored in the memory 705, and invoking data stored in the memory 705. Alternatively, the processor 701 may be implemented in at least one hardware form of DSP, FPGA, PLA. The processor 701 may integrate one or a combination of several of a CPU, GPU, modem, etc. The CPU mainly processes an operating system, a user interface, an application program and the like; the GPU is used for rendering and drawing the content required to be displayed by the display screen; the modem is used to handle wireless communications. It will be appreciated that the modem may not be integrated into the processor 701 and may be implemented by a single chip.
The memory 705 may include RAM or ROM. Optionally, the memory 705 comprises a non-transitory computer readable medium. Memory 705 may be used to store instructions, programs, code, sets of codes, or instruction sets. The memory 705 may include a stored program area and a stored data area, wherein the stored program area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing the various method embodiments described above, etc.; the storage data area may store data or the like referred to in the above respective method embodiments. The memory 705 may also optionally be at least one storage device located remotely from the processor 701. An operating system, a network communication module, a user interface module, and a vibration-based wind turbine blade anomaly detection application may be included in memory 705 as a computer storage medium. The processor 701 may be configured to invoke the vibration-based wind turbine blade anomaly detection application stored in the memory 705 and perform the steps of vibration-based wind turbine blade anomaly detection mentioned in the previous embodiments.
Embodiments of the present disclosure also provide a computer-readable storage medium having instructions stored therein, which when executed on a computer or processor, cause the computer or processor to perform the steps of one or more of the embodiments shown in fig. 2-4 above. The above-described constituent modules of the electronic apparatus may be stored in the computer-readable storage medium if implemented in the form of software functional units and sold or used as independent products.
In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, produces a flow or function in accordance with embodiments of the present description, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in or transmitted across a computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital subscriber line (Digital Subscriber Line, DSL)), or wireless (e.g., infrared, wireless, microwave, etc.). The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a digital versatile disk (DIGITAL VERSATILE DISC, DVD)), or a semiconductor medium (e.g., a Solid state disk (Solid STATE DISK, SSD)), or the like.
Those skilled in the art will appreciate that implementing all or part of the above-described embodiment methods may be accomplished by way of a computer program, which may be stored in a computer-readable storage medium, instructing relevant hardware, and which, when executed, may comprise the embodiment methods as described above. And the aforementioned storage medium includes: various media capable of storing program code, such as ROM, RAM, magnetic or optical disks. The technical features in the present examples and embodiments may be arbitrarily combined without conflict.
The above-described embodiments are merely preferred embodiments of the present disclosure, and do not limit the scope of the disclosure, and various modifications and improvements made by those skilled in the art to the technical solutions of the disclosure should fall within the protection scope defined by the claims of the disclosure without departing from the design spirit of the disclosure.
Claims (10)
1. A wind turbine blade abnormality detection method based on vibration comprises the following steps:
SCADA data and cabin vibration data of the wind turbine in a target period are obtained;
Inputting the acquired data into a paddle abnormality detection model which is obtained through training in advance, so that the paddle abnormality detection model outputs a detection result; the blade anomaly detection model comprises a vibration feature extraction layer, an SCADA feature extraction layer and an anomaly detection layer, wherein the vibration feature extraction layer is used for extracting vibration features in cabin vibration data based on a variation modal decomposition technology and a nuclear principal component analysis technology, the SCADA feature extraction layer is used for extracting SCADA features in the SCADA data according to an ABT network, the anomaly detection layer is used for outputting detection results according to comparison results of target residual values and a normal residual value range obtained through pre-training, and the target residual values are obtained through calculation according to the extracted vibration features and the SCADA features;
and determining whether the blades of the wind turbine generator are abnormal in the target period according to the detection result output by the blade abnormality identification model.
2. The method for detecting the abnormal condition of the blade of the wind turbine generator based on the vibration according to claim 1, wherein the process of the vibration characteristics extracted by the vibration characteristic extraction layer comprises the following steps:
Aiming at each channel vibration signal in the cabin vibration data, decomposing a single channel vibration signal into a plurality of modal components, and establishing a constraint optimization problem according to a component narrowband condition;
The constraint optimization problem is equivalent to an unconstrained optimization problem based on an augmented lagrange function, and the plurality of modal components and the center frequencies corresponding to the plurality of modal components are calculated based on an alternate direction multiplier method;
Performing Fourier transform on each modal component, and extracting principal components of a frequency spectrum obtained by transforming each modal component based on a kernel principal component analysis technology;
Performing centering treatment on the extracted nuclear matrix to obtain a standardized nuclear matrix, and performing eigenvalue decomposition on the standardized nuclear matrix;
And determining the vibration characteristics corresponding to each modal variable according to the maximum characteristic value obtained by decomposition and the characteristic vector corresponding to the maximum characteristic value.
3. The vibration-based wind turbine blade anomaly detection method according to claim 1, wherein the ABT network is constructed based on a dynamic convolution kernel mechanism.
4. The vibration-based wind turbine blade anomaly detection method according to claim 1, wherein the calculation process of the target residual value comprises the following steps:
and calculating the difference value of the vibration characteristic and the SCADA characteristic in the same time dimension, and constructing a Markov distance space to quantify the difference value so as to obtain a target residual value.
5. The method for detecting the blade abnormality of the wind turbine generator based on the vibration according to claim 1, wherein the process of outputting the detection result by the abnormality detection layer according to the comparison result comprises the following steps:
If the number of time points, of which the target residual error value exceeds the normal residual error value range, reaches a preset threshold value, the detection result is abnormal;
If the number of time points, of which the target residual value exceeds the normal residual value range, is smaller than a preset threshold value, the detection result is normal.
6. A training method of a blade abnormality detection model comprises the following steps:
Obtaining a training dataset, each training sample in the training dataset comprising: SCADA data and cabin vibration data of the wind turbine in the same time period, and real detection results of blades of the wind turbine in the time period;
inputting the training data set into a model to be trained, so that the model to be trained outputs a prediction detection result of the blades of the wind turbine in the period;
and comparing the real detection result with the prediction detection result, and optimizing the model to be trained according to the comparison result.
7. Wind turbine generator system paddle anomaly detection device based on vibration includes:
The acquisition unit is used for acquiring SCADA data and cabin vibration data of the wind turbine in a target period;
The input unit is used for inputting the acquired data into a paddle abnormality detection model which is trained in advance, so that the paddle abnormality detection model outputs a detection result; the blade anomaly detection model comprises a vibration feature extraction layer, an SCADA feature extraction layer and an anomaly detection layer, wherein the vibration feature extraction layer is used for extracting vibration features in cabin vibration data based on a variation modal decomposition technology and a nuclear principal component analysis technology, the SCADA feature extraction layer is used for extracting SCADA features in the SCADA data according to an ABT network, the anomaly detection layer is used for outputting detection results according to comparison results of target residual values and a normal residual value range obtained through pre-training, and the target residual values are obtained through calculation according to the extracted vibration features and the SCADA features;
and the determining unit is used for determining whether the blades of the wind turbine generator set are abnormal in the target period according to the detection result output by the blade abnormality identification model.
8. A training device for a blade anomaly detection model, comprising:
An acquisition unit configured to acquire a training data set, where each training sample in the training data set includes: SCADA data and cabin vibration data of the wind turbine in the same time period, and real detection results of blades of the wind turbine in the time period;
The input unit is used for inputting the training data set into a model to be trained so that the model to be trained outputs a prediction detection result of the blades of the wind turbine in the period;
and the optimizing unit is used for comparing the real detection result with the prediction detection result and optimizing the model to be trained according to the comparison result.
9. An electronic device includes a processor and a memory;
the processor is connected with the memory;
the memory is used for storing executable program codes;
The processor executes a program corresponding to the executable program code by reading the executable program code stored in the memory, for performing the method according to any one of claims 1 to 6.
10. A computer readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, implements the method according to any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410564355.7A CN118167569B (en) | 2024-05-09 | Wind turbine generator blade abnormality detection method and device based on vibration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410564355.7A CN118167569B (en) | 2024-05-09 | Wind turbine generator blade abnormality detection method and device based on vibration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118167569A true CN118167569A (en) | 2024-06-11 |
| CN118167569B CN118167569B (en) | 2024-07-26 |
Family
ID=
Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120133135A1 (en) * | 2010-11-25 | 2012-05-31 | Sinovel Wind Group Co., Ltd. | Wind turbine generator system vibration detecting device and method |
| KR20130046858A (en) * | 2011-10-28 | 2013-05-08 | 엘에스전선 주식회사 | Monitoring system for wind turbine blade and monitoring method using the same |
| JP2019027324A (en) * | 2017-07-27 | 2019-02-21 | 日本精工株式会社 | Abnormality detection system and abnormality detection method of wind power generator |
| US20190072082A1 (en) * | 2016-03-14 | 2019-03-07 | Ventus Engineering GmbH | Method of condition monitoring one or more wind turbines and parts thereof and performing instant alarm when needed |
| CN111310589A (en) * | 2020-01-20 | 2020-06-19 | 河北科技大学 | Fault diagnosis method and device for mechanical system and terminal |
| CN113153659A (en) * | 2021-06-16 | 2021-07-23 | 浙江诺瓦智能科技有限公司 | Health monitoring method and system for blades of wind driven generator |
| CN113719425A (en) * | 2021-05-14 | 2021-11-30 | 北京智慧空间科技有限责任公司 | Fan blade fault prediction method and system and storage medium |
| CN115095487A (en) * | 2022-06-30 | 2022-09-23 | 宁波大学 | Wind turbine state monitoring method based on multi-source heterogeneous SCADA data |
| CN115859148A (en) * | 2022-11-24 | 2023-03-28 | 华能陇东能源有限责任公司 | Fan blade vibration alarm method and device |
| CN116624343A (en) * | 2023-06-14 | 2023-08-22 | 华北电力大学 | Wind turbine generator tower abnormal vibration monitoring and health degree evaluation method and system |
| CN117028164A (en) * | 2023-07-20 | 2023-11-10 | 灵武市隆桥光伏新能源有限公司 | Vibration detection method and equipment for wind turbine generator |
| CN117189505A (en) * | 2022-05-31 | 2023-12-08 | 北京金风慧能技术有限公司 | Abnormal vibration monitoring method and device for wind turbine generator |
| CN117235476A (en) * | 2023-11-14 | 2023-12-15 | 利维智能（深圳）有限公司 | Fault detection method, device, computer equipment and storage medium |
| CN117365869A (en) * | 2023-11-16 | 2024-01-09 | 浙江浙能嘉兴海上风力发电有限公司 | Self-adaptive early warning strategy design method for wind turbine blade tower sweeping faults |
| CN117514644A (en) * | 2023-11-03 | 2024-02-06 | 国能思达科技有限公司 | Method and system for monitoring vibration of fan cabin |
| CN117760734A (en) * | 2023-12-27 | 2024-03-26 | 东北电力大学 | Wind generating set gear box bearing fault diagnosis method and system |
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120133135A1 (en) * | 2010-11-25 | 2012-05-31 | Sinovel Wind Group Co., Ltd. | Wind turbine generator system vibration detecting device and method |
| KR20130046858A (en) * | 2011-10-28 | 2013-05-08 | 엘에스전선 주식회사 | Monitoring system for wind turbine blade and monitoring method using the same |
| US20190072082A1 (en) * | 2016-03-14 | 2019-03-07 | Ventus Engineering GmbH | Method of condition monitoring one or more wind turbines and parts thereof and performing instant alarm when needed |
| JP2019027324A (en) * | 2017-07-27 | 2019-02-21 | 日本精工株式会社 | Abnormality detection system and abnormality detection method of wind power generator |
| CN111310589A (en) * | 2020-01-20 | 2020-06-19 | 河北科技大学 | Fault diagnosis method and device for mechanical system and terminal |
| CN113719425A (en) * | 2021-05-14 | 2021-11-30 | 北京智慧空间科技有限责任公司 | Fan blade fault prediction method and system and storage medium |
| CN113153659A (en) * | 2021-06-16 | 2021-07-23 | 浙江诺瓦智能科技有限公司 | Health monitoring method and system for blades of wind driven generator |
| CN117189505A (en) * | 2022-05-31 | 2023-12-08 | 北京金风慧能技术有限公司 | Abnormal vibration monitoring method and device for wind turbine generator |
| CN115095487A (en) * | 2022-06-30 | 2022-09-23 | 宁波大学 | Wind turbine state monitoring method based on multi-source heterogeneous SCADA data |
| CN115859148A (en) * | 2022-11-24 | 2023-03-28 | 华能陇东能源有限责任公司 | Fan blade vibration alarm method and device |
| CN116624343A (en) * | 2023-06-14 | 2023-08-22 | 华北电力大学 | Wind turbine generator tower abnormal vibration monitoring and health degree evaluation method and system |
| CN117028164A (en) * | 2023-07-20 | 2023-11-10 | 灵武市隆桥光伏新能源有限公司 | Vibration detection method and equipment for wind turbine generator |
| CN117514644A (en) * | 2023-11-03 | 2024-02-06 | 国能思达科技有限公司 | Method and system for monitoring vibration of fan cabin |
| CN117235476A (en) * | 2023-11-14 | 2023-12-15 | 利维智能（深圳）有限公司 | Fault detection method, device, computer equipment and storage medium |
| CN117365869A (en) * | 2023-11-16 | 2024-01-09 | 浙江浙能嘉兴海上风力发电有限公司 | Self-adaptive early warning strategy design method for wind turbine blade tower sweeping faults |
| CN117760734A (en) * | 2023-12-27 | 2024-03-26 | 东北电力大学 | Wind generating set gear box bearing fault diagnosis method and system |
Non-Patent Citations (1)
| Title |
|---|
| 陈雪峰 等: "风电装备故障诊断与健康监测研究综述", 《中国机械工程》, vol. 31, no. 02, 31 January 2020 (2020-01-31), pages 175 - 189 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chatterjee et al. | Scientometric review of artificial intelligence for operations & maintenance of wind turbines: The past, present and future | |
| Cho et al. | Fault detection and diagnosis of a blade pitch system in a floating wind turbine based on Kalman filters and artificial neural networks | |
| Zhang et al. | A review on deep learning applications in prognostics and health management | |
| Zhao et al. | Challenges and opportunities of AI-enabled monitoring, diagnosis & prognosis: A review | |
| Helbing et al. | Deep Learning for fault detection in wind turbines | |
| Wu et al. | A multi-level-denoising autoencoder approach for wind turbine fault detection | |
| Song et al. | Wind turbine health state monitoring based on a Bayesian data-driven approach | |
| Zaher et al. | A multi-agent fault detection system for wind turbine defect recognition and diagnosis | |
| CN113469060A (en) | Multi-sensor fusion convolution neural network aeroengine bearing fault diagnosis method | |
| US20130060524A1 (en) | Machine Anomaly Detection and Diagnosis Incorporating Operational Data | |
| Zhan et al. | Abnormal vibration detection of wind turbine based on temporal convolution network and multivariate coefficient of variation | |
| Zhang | Automatic fault prediction of wind turbine main bearing based on SCADA data and artificial neural network | |
| CN111103137A (en) | Wind turbine gearbox fault diagnosis method based on deep neural network | |
| CN111415070A (en) | Wind turbine generator gearbox oil temperature over-temperature fault early warning method based on SCADA data | |
| Zhang et al. | A novel fault diagnosis method for wind turbine based on adaptive multivariate time-series convolutional network using SCADA data | |
| CN111709190A (en) | Wind turbine generator operation data image identification method and device | |
| US20220345468A1 (en) | System and method for cyberattack detection in a wind turbine control system | |
| CN111198098A (en) | Wind power generator bearing fault prediction method based on artificial neural network | |
| Mohammadi et al. | DeepWind: an accurate wind turbine condition monitoring framework via deep learning on embedded platforms | |
| Praveen et al. | Fingerprinting based data abstraction technique for remaining useful life estimation in a multi-stage gearbox | |
| CN116821730B (en) | Fan fault detection method, control device and storage medium | |
| CN118167569B (en) | Wind turbine generator blade abnormality detection method and device based on vibration | |
| Jiang et al. | An orbit-based encoder–forecaster deep learning method for condition monitoring of large turbomachines | |
| CN118167569A (en) | Wind turbine generator blade abnormality detection method and device based on vibration | |
| Ning et al. | An intelligent device fault diagnosis method in industrial internet of things |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |