CN114298219A - Wind driven generator fault diagnosis method based on deep space-time feature extraction - Google Patents
Wind driven generator fault diagnosis method based on deep space-time feature extraction Download PDFInfo
- Publication number
- CN114298219A CN114298219A CN202111620688.XA CN202111620688A CN114298219A CN 114298219 A CN114298219 A CN 114298219A CN 202111620688 A CN202111620688 A CN 202111620688A CN 114298219 A CN114298219 A CN 114298219A
- Authority
- CN
- China
- Prior art keywords
- time sequence
- fault
- convolution
- time
- driven generator
- 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.)
- Pending
Links
- 238000000605 extraction Methods 0.000 title claims abstract description 38
- 238000003745 diagnosis Methods 0.000 title claims abstract description 27
- 238000000034 method Methods 0.000 title claims abstract description 24
- 230000004913 activation Effects 0.000 claims description 25
- 230000006870 function Effects 0.000 claims description 25
- 238000011176 pooling Methods 0.000 claims description 18
- 238000012544 monitoring process Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 7
- 239000011800 void material Substances 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 claims description 5
- 230000002776 aggregation Effects 0.000 claims description 3
- 238000004220 aggregation Methods 0.000 claims description 3
- 230000036541 health Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 2
- 238000011161 development Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
Images
Abstract
The invention provides a wind driven generator fault diagnosis method based on deep space-time feature extraction, aiming at the characteristic of strong space-time correlation of SCADA multivariable time sequence data of a wind driven generator, time and space feature extraction networks are respectively designed, a time sequence convolution attention module is utilized to screen and extract time sequence fault features, correlation information among variables is simultaneously excavated through a cavity convolution module, then the time sequence features and the space features are combined and finally input into a fault classifier to obtain a final fault diagnosis result, the technical means utilizes the deep fault feature extraction of time and space dimensions to deeply capture fault information of the wind driven generator, improves the fault diagnosis precision of the wind driven generator, thereby timely obtaining fault state information of the wind driven generator, processing and maintaining the fault state information and avoiding deep damage of wind driven generator components, the healthy and stable operation of the wind driven generator is guaranteed.
Description
Technical Field
The invention relates to the technical field of wind driven generator fault diagnosis, in particular to a wind driven generator fault diagnosis method based on deep space-time feature extraction.
Background
With the rapid development of social civilization, the demand of human beings on energy is gradually increased, and wind energy as a clean and renewable energy source has been developed and utilized on a large scale worldwide in recent years. However, the wind turbine generator is extremely prone to failure due to the fact that the wind turbine generator is harsh in operating environment and is affected by various complex effects for a long time, and even the whole wind turbine generator is shut down in severe cases, so that severe economic loss is caused.
At present, a fault diagnosis method based on a traditional physical model is limited by changeable working condition environments and complex internal structures of the wind driven generator, a large number of simplifying assumptions are needed, an accurate model cannot be obtained, and the fault diagnosis precision of the wind driven generator is greatly limited. With the development of sensor technology and wind power Data monitoring systems, a wind power generation set has been widely installed with a Data Acquisition and monitoring Control System (SCADA) to monitor and evaluate the operation state of the set. The SCADA system is integrated in the wind turbine generator, an additional monitoring system does not need to be installed, and massive monitoring data can be generated along with the operation of the generator. Under the background of rapid development of big data technology, the SCADA data is used for modeling and analyzing to become a new hotspot for monitoring the state of the wind turbine generator. Compared with the traditional method based on a physical model, the method based on SCADA data driving only depends on monitoring data, does not need prior knowledge such as a system model or a failure mechanism and has strong applicability and expandability. The wind driven generator fault diagnosis method based on deep learning utilizes SCADA state monitoring data to construct a deep diagnosis model and extract high-dimensional fault features, and becomes the research focus of the current industrial and academic fields.
The existing wind driven generator fault diagnosis research based on deep learning usually only aims at time dimension to extract fault characteristics, and does not consider that SCADA data is multivariable time sequence data, and strong correlation exists among variables. And the time-space correlation characteristics are learned, so that the fault information captured from the SCADA data can be further enriched, and the diagnosis precision of the model is further improved.
Disclosure of Invention
The invention solves the problems: the method for diagnosing the fault of the wind driven generator overcomes the defects of the prior art, and provides the method for diagnosing the fault of the wind driven generator based on deep space-time feature extraction, so that the method for diagnosing the fault of the wind driven generator can effectively improve the precision of diagnosing the fault of the wind driven generator, identifies the fault type of the wind driven generator, and is a fault diagnosis method with engineering practical value.
The technical scheme of the invention is as follows: a wind driven generator fault diagnosis method based on deep space-time feature extraction comprises the following steps:
step S1: acquiring original multivariable time sequence data of a wind power data acquisition and monitoring control system; preprocessing SCADA multivariable time sequence data according to the health state of the wind generating set to obtain standardized two-dimensional multivariable time sequence data;
step S2: inputting the two-dimensional multivariable time sequence data obtained in the step 1 into a time convolution attention module, and extracting the fault time sequence characteristics of the wind driven generator;
step S3: inputting the two-dimensional multivariable time sequence input data obtained in the step (1) into a spatial convolution feature extraction module, and learning spatial features among different variables in the SCADA data to obtain spatial features;
step S4: and combining the time sequence characteristics and the space characteristics obtained in the step 2 and the step 3, inputting the time sequence characteristics and the space characteristics into a fault classifier for classification, and obtaining a final fault type result.
The further step S1 includes the following specific steps:
s11, acquiring original multivariate time sequence data of the wind power data acquisition and monitoring control system, wherein the number of the variables is S;
step S12 is to normalize the obtained raw multivariate time-series data by the maximum-minimum normalization method, and the calculation formula is as follows:
wherein, yijIs the ith value of the variable j in the normalized multivariate time series, xij is the ith value of the variable j in the original multivariate time series, min (x)j) And max (x)j) The minimum and maximum values of the variable j, respectively.
And step S13, dividing the normalized multivariate time sequence data into a plurality of two-dimensional subsequences with the length of N according to sampling time by a non-overlapping sliding window technology to obtain a two-dimensional multivariate time sequence input matrix, wherein the size of the matrix is NxS.
Further, in step S2, the method for extracting the time series fault feature through the time series convolution attention module on the preprocessed two-dimensional input matrix obtained in step 1 specifically includes the following steps:
and step S21, designing a time sequence feature extraction convolution submodule which is mainly formed by stacking a convolution layer, a pooling layer and an activation function layer. Wherein, pooling layer adopts the maximum pooling mode, and activation layer adopts Relu activation function, and the formula is as follows: (x) max (0, x), where x represents the input to the activation layer and max (0, x) represents the output to take the maximum of 0 and x as the Relu activation function;
step S22, designing a time attention submodule, obtaining time sequence weight by using the output of the previous layer as input in a mode of connecting two layers of cavity convolution layers and a Sigmoid activation function layer in series, and obtaining screened time sequence feature mapping after the input weight;
step S23, inputting the multivariable time sequence data obtained in the step 1 into a time sequence convolution attention module obtained by stacking a convolution submodule and a time attention submodule to obtain a time sequence fault characteristic sequence;
further, step S3 needs to further learn multivariate spatial correlation fault characteristics, and specifically includes the following steps:
s31, designing a spatial convolution feature extraction module, wherein the specific structure is formed by stacking a cavity convolution layer and a pooling layer, and the pooling layer performs dimension reduction and information aggregation on data output by the convolution layer in a maximum pooling mode; the void rate adopted by the void convolution layer is dr, the convolution receptive field is enlarged, and the Relu activation function is adopted as the activation function;
and step S32, inputting the multivariate time sequence data obtained in the step 1 into a spatial convolution feature extraction module, carrying out convolution operation on the convolution layer in the variable dimension of the SCADA data to obtain correlation information among the SCADA variables, and learning spatial correlation features among the variables to obtain a spatial feature sequence.
Step S4: combining the time sequence characteristics and the spatial characteristics obtained in the step 2 and the step 3, inputting the time sequence characteristics and the spatial characteristics into a fault classifier for classification to generate a final fault diagnosis result, and specifically comprising the following steps:
step S41, defining the classification task of the wind driven generator fault as a multi-classification problem according to the fault type of the wind driven generator;
step S42, cascading the time sequence characteristics and the space characteristics acquired in the step 2 and the step 3 on characteristic dimensions, inputting the time sequence characteristics and the space characteristics into a Softmax classifier with a cross entropy as a loss function, acquiring output probability values at all given fault labels and under a healthy state, and obtaining the maximum value of the output probability values as a fault classification result through comparison; the calculation formula of the cross entropy loss function is as follows:
where i represents the ith sample output, p (i) represents the true distribution, and q (i) represents the predicted distribution.
Compared with the prior art, the invention has the advantages that: the invention provides a wind driven generator fault diagnosis method based on deep space-time feature extraction, aiming at the characteristic of strong space-time correlation of SCADA multivariable time sequence data of a wind driven generator, time and space feature extraction networks are respectively designed, a time sequence convolution attention module is utilized to screen and extract time sequence fault features, correlation information among variables is simultaneously excavated through a cavity convolution module, then the time sequence features and the space features are combined and finally input into a fault classifier to obtain a final fault diagnosis result, the technical means utilizes the deep fault feature extraction of time and space dimensions to deeply capture fault information of the wind driven generator, improves the fault diagnosis precision of the wind driven generator, thereby timely obtaining fault state information of the wind driven generator, processing and maintaining the fault state information and avoiding deep damage of wind driven generator components, the healthy and stable operation of the wind driven generator is guaranteed.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention;
FIG. 2 is a schematic diagram of a fault feature extraction network according to the present invention;
FIG. 3 is a schematic diagram of a time-series convolution attention module according to the present invention;
FIG. 4 is a schematic diagram of the spatial convolution feature extraction module according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following examples:
as shown in fig. 1, the wind turbine generator fault diagnosis method based on deep space-time feature extraction of the present invention includes the following steps:
step S1: s11, acquiring original multivariate time sequence data of the wind power data acquisition and monitoring control system, wherein the number of the variables is S;
step S12 is to normalize the obtained raw multivariate time-series data by the maximum-minimum normalization method, and the calculation formula is as follows:
wherein, yijIs the ith value, x, of the variable j in the normalized multivariate time seriesijIs the ith value, min (x), of variable j in the original multivariate time seriesj) And max (x)j) The minimum and maximum values of the variable j, respectively.
And step S13, dividing the normalized multivariate time sequence data into a plurality of two-dimensional subsequences with the length of N according to sampling time by a non-overlapping sliding window technology to obtain a two-dimensional multivariate time sequence input matrix, wherein the size of the matrix is NxS.
As shown in fig. 2, the fault feature extraction is performed from the time sequence dimension and the variable space dimension, and specifically includes the following two steps:
first, as shown in fig. 3, for the timing dimension, the following steps are specifically performed: and step S21, designing a time sequence feature extraction convolution submodule which is mainly formed by stacking a convolution layer, a pooling layer and an activation function layer. Wherein, pooling layer adopts the maximum pooling mode, and activation layer adopts Relu activation function, and the formula is as follows: (x) max (0, x), where x represents the input to the activation layer and max (0, x) represents the output to take the maximum of 0 and x as the Relu activation function;
step S22, designing a time attention submodule, obtaining time sequence weight by using the output of the previous layer as input in a mode of connecting two layers of cavity convolution layers and a Sigmoid activation function layer in series, and obtaining screened time sequence feature mapping after the input weight;
step S23, inputting the multivariable time sequence data obtained in the step 1 into a time sequence convolution attention module obtained by stacking a convolution submodule and a time attention submodule to obtain a time sequence fault characteristic sequence;
next, as shown in fig. 4, considering the correlation between variables, performing feature extraction from a spatial dimension, specifically including the following steps:
s31, designing a spatial convolution feature extraction module, wherein the specific structure is formed by stacking a cavity convolution layer and a pooling layer, and the pooling layer performs dimension reduction and information aggregation on data output by the convolution layer in a maximum pooling mode; the void rate adopted by the void convolution layer is dr, the convolution receptive field is enlarged, and the Relu activation function is adopted as the activation function;
and step S32, inputting the multivariate time sequence data obtained in the step 1 into a spatial convolution feature extraction module, carrying out convolution operation on the convolution layer in the variable dimension of the SCADA data to obtain correlation information among the SCADA variables, and learning spatial correlation features among the variables to obtain a spatial feature sequence.
Step S4: and (3) combining the time sequence characteristics and the space characteristics obtained in the steps (2) and (3) to obtain space-time fault characteristics, and inputting the space-time fault characteristics into a classifier for fault classification, wherein the specific steps are as follows:
step S41, defining the classification task of the wind driven generator fault as a multi-classification problem according to the fault type of the wind driven generator;
step S42, cascading the time sequence characteristics and the space characteristics acquired in the step 2 and the step 3 on characteristic dimensions, inputting the time sequence characteristics and the space characteristics into a Softmax classifier with a cross entropy as a loss function, acquiring output probability values at all given fault labels and under a healthy state, and obtaining the maximum value of the output probability values as a fault classification result through comparison; the calculation formula of the cross entropy loss function is as follows:
where i represents the ith sample output, p (i) represents the true distribution, and q (i) represents the predicted distribution.
From the analysis, the embodiment of the invention provides an effective wind driven generator fault diagnosis method, aiming at the characteristics of time-space correlation of SCADA data variables of a wind driven generator, the SCADA multivariable time sequence data is collected, a two-dimensional multivariable time sequence input matrix is obtained through data preprocessing, then time and space convolution feature extraction networks are respectively designed for fault feature extraction, a time attention mechanism is added into a time sequence module in consideration of noise and interference influence, time sequence fault information is screened out, and the fault classification precision of the wind driven generator is effectively improved by combining the correlation characteristics among the variables, so that the fault of key components is timely processed and maintained, and the deep damage of the wind driven generator is avoided, and the serious economic loss is caused.
The principle of the invention is as follows: the method is based on time and space two-dimensional fault feature extraction of SCADA data, self-adaptively screens out time sequence feature information with high fault correlation, and combines correlation space features among variables, so that fault diagnosis is performed on the wind driven generator. Firstly, SCADA multivariable time sequence data is obtained and is preprocessed. Then, fault feature extraction modules are respectively designed from the time dimension and the space dimension, and the space-time fault features are automatically learned. Meanwhile, a time attention mechanism is embedded into the time sequence feature extraction module, and the time sequence features with high fault correlation are screened out and combined with the variable correlation space features. And finally, outputting the obtained fault classification result through a Softmax classifier.
The above-mentioned embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solution of the present invention by those skilled in the art should fall within the protection scope defined by the claims of the present invention without departing from the spirit of the present invention.
Claims (5)
1. A wind driven generator fault diagnosis method based on deep space-time feature extraction is characterized by comprising the following steps:
step S1: acquiring original multivariable time sequence data of a wind power data acquisition and monitoring control system; preprocessing SCADA multivariable time sequence data according to the health state of the wind generating set to obtain standardized two-dimensional multivariable time sequence data;
step S2: inputting the two-dimensional multivariable time sequence data obtained in the step 1 into a time convolution attention module, and extracting the fault time sequence characteristics of the wind driven generator;
step S3: inputting the two-dimensional multivariable time sequence input data obtained in the step (1) into a spatial convolution feature extraction module, and learning spatial features among different variables in the SCADA data to obtain spatial features;
step S4: and combining the time sequence characteristics and the space characteristics obtained in the step 2 and the step 3, inputting the time sequence characteristics and the space characteristics into a fault classifier for classification, and obtaining a final fault type result.
2. The wind turbine generator fault diagnosis method based on deep spatiotemporal feature extraction according to claim 1, characterized in that: the step 1 comprises the following specific steps:
s11, acquiring original multivariate time sequence data of the wind power data acquisition and monitoring control system, wherein the number of the variables is S;
step S12 is to normalize the obtained raw multivariate time series data by the maximum-minimum normalization method, and the calculation formula is as follows:
wherein, yijIs after standardization treatment
Is given by the ith value, x of the variable j in the multivariate time seriesijIs the ith value, min (x), of variable j in the original multivariate time seriesj) And max (x)j) The minimum and maximum values of the variable j, respectively.
And step S13, dividing the normalized multivariable time sequence data into a plurality of two-dimensional subsequences with the length of N according to sampling time by a non-overlapping sliding window technology to obtain two-dimensional multivariable time sequence data, wherein the size of the matrix is NxS.
3. The wind turbine generator fault diagnosis method based on deep spatiotemporal feature extraction according to claim 1, characterized in that: in the step 2, the time convolution attention module includes a time sequence feature extraction convolution submodule and a time attention submodule, and is specifically implemented as follows:
step S21, the time sequence feature extraction convolution submodule is formed by stacking a convolution layer, a pooling layer and an activation function layer; the pooling layer adopts a maximum pooling mode, the activation layer adopts a Relu activation function, and the calculation formula of the activation function is as follows: (x) max (0, x), where x represents the input to the activation layer and max (0, x) represents the output to take the maximum of 0 and x as the Relu activation function;
step S22, designing a time attention submodule, obtaining time sequence weight by using the output of the previous layer as input in a mode of connecting two layers of cavity convolution layers and a Sigmoid activation function layer in series, and obtaining screened time sequence feature mapping after the input weight;
and step S23, inputting the multivariable time sequence data obtained in the step 1 into a time sequence convolution attention module obtained by stacking a convolution submodule and a time attention submodule, and extracting the fault time sequence characteristics of the wind driven generator.
4. The wind turbine generator fault diagnosis method based on deep spatiotemporal feature extraction according to claim 1, characterized in that: the step 3 comprises the following specific steps:
s31, designing a spatial convolution feature extraction module, wherein the specific structure is formed by stacking a cavity convolution layer and a pooling layer, and the pooling layer performs dimension reduction and information aggregation on data output by the convolution layer in a maximum pooling mode; the void rate adopted by the void convolution layer is dr, the convolution receptive field is enlarged, and the Relu activation function is adopted as the activation function;
and step S32, inputting the multivariate time sequence data obtained in the step 1 into a spatial convolution feature extraction module, carrying out convolution operation on the convolution layer in the variable dimension of the SCADA data to obtain correlation information among the SCADA variables, and learning spatial correlation features among the variables to obtain a spatial feature sequence.
5. The wind turbine generator fault diagnosis method based on deep spatiotemporal feature extraction according to claim 1, characterized in that: the step 4 comprises the following specific steps:
step S41, defining a classification task of the wind driven generator fault as a multi-classification problem according to the fault type of the wind driven generator, wherein the fault type of the wind driven generator comprises health and specific fault positions;
step S42, cascading the time sequence characteristics and the space characteristics acquired in the step 2 and the step 3 on characteristic dimensions, inputting the time sequence characteristics and the space characteristics into a Softmax classifier with a cross entropy as a loss function, acquiring output probability values at all given fault labels and under a healthy state, and obtaining the maximum value of the output probability values as a fault classification result through comparison; the calculation formula of the cross entropy loss function is as follows:
where i represents the ith sample output, p (i) represents the true distribution, and q (i) represents the predicted distribution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111620688.XA CN114298219A (en) | 2021-12-27 | 2021-12-27 | Wind driven generator fault diagnosis method based on deep space-time feature extraction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111620688.XA CN114298219A (en) | 2021-12-27 | 2021-12-27 | Wind driven generator fault diagnosis method based on deep space-time feature extraction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114298219A true CN114298219A (en) | 2022-04-08 |
Family
ID=80970113
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111620688.XA Pending CN114298219A (en) | 2021-12-27 | 2021-12-27 | Wind driven generator fault diagnosis method based on deep space-time feature extraction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114298219A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114997333A (en) * | 2022-06-29 | 2022-09-02 | 清华大学 | Fault diagnosis method and device for wind driven generator |
| CN114997333B (en) * | 2022-06-29 | 2024-04-23 | 清华大学 | Fault diagnosis method and device for wind driven generator |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9189215B1 (en) * | 2014-08-26 | 2015-11-17 | National Instruments Corporation | Convergence analysis of program variables |
| US20180284737A1 (en) * | 2016-05-09 | 2018-10-04 | StrongForce IoT Portfolio 2016, LLC | Methods and systems for detection in an industrial internet of things data collection environment with large data sets |
| CN111426950A (en) * | 2020-03-19 | 2020-07-17 | 燕山大学 | Wind driven generator fault diagnosis method of multi-scale space-time convolution depth belief network |
| US20210203157A1 (en) * | 2019-12-30 | 2021-07-01 | Utopus Insights, Inc. | Scalable systems and methods for assessing healthy condition scores in renewable asset management |
| CN113283580A (en) * | 2021-04-30 | 2021-08-20 | 太原理工大学 | Automatic fault detection method for solar cell panel |
-
2021
- 2021-12-27 CN CN202111620688.XA patent/CN114298219A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9189215B1 (en) * | 2014-08-26 | 2015-11-17 | National Instruments Corporation | Convergence analysis of program variables |
| US20180284737A1 (en) * | 2016-05-09 | 2018-10-04 | StrongForce IoT Portfolio 2016, LLC | Methods and systems for detection in an industrial internet of things data collection environment with large data sets |
| US20210203157A1 (en) * | 2019-12-30 | 2021-07-01 | Utopus Insights, Inc. | Scalable systems and methods for assessing healthy condition scores in renewable asset management |
| CN111426950A (en) * | 2020-03-19 | 2020-07-17 | 燕山大学 | Wind driven generator fault diagnosis method of multi-scale space-time convolution depth belief network |
| CN113283580A (en) * | 2021-04-30 | 2021-08-20 | 太原理工大学 | Automatic fault detection method for solar cell panel |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114997333A (en) * | 2022-06-29 | 2022-09-02 | 清华大学 | Fault diagnosis method and device for wind driven generator |
| CN114997333B (en) * | 2022-06-29 | 2024-04-23 | 清华大学 | Fault diagnosis method and device for wind driven generator |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pang et al. | Spatio-temporal fusion neural network for multi-class fault diagnosis of wind turbines based on SCADA data | |
| CN111426950B (en) | Wind driven generator fault diagnosis method of multi-scale space-time convolution depth belief network | |
| Zhao et al. | Intelligent fault diagnosis of multichannel motor–rotor system based on multimanifold deep extreme learning machine | |
| He et al. | A deep multi-signal fusion adversarial model based transfer learning and residual network for axial piston pump fault diagnosis | |
| CN111352977B (en) | Time sequence data monitoring method based on self-attention bidirectional long-short term memory network | |
| CN113639993B (en) | Gearbox fault diagnosis method of multi-mode multi-task convolutional neural network | |
| CN111562105B (en) | Wind turbine generator gearbox fault diagnosis method based on wavelet packet decomposition and convolutional neural network | |
| CN113255209B (en) | Method for predicting residual life of bearing of gearbox | |
| CN113255210B (en) | Method and system for diagnosing federal fault of wind turbine generator | |
| CN116307212B (en) | Air quality prediction method and system | |
| CN114295368A (en) | Multi-channel fused wind power planetary gear box fault diagnosis method | |
| Zheng et al. | Semi-supervised multivariate time series anomaly detection for wind turbines using generator SCADA data | |
| Zhang et al. | MS-TCN: A multiscale temporal convolutional network for fault diagnosis in industrial processes | |
| Wang et al. | A novel competitive temporal convolutional network for remaining useful life prediction of rolling bearings | |
| CN112508278A (en) | Multi-connected system load prediction method based on evidence regression multi-model | |
| Raihan et al. | A Bi-LSTM Autoencoder Framework for Anomaly Detection-A Case Study of a Wind Power Dataset | |
| CN114298219A (en) | Wind driven generator fault diagnosis method based on deep space-time feature extraction | |
| Zhang et al. | Fault diagnosis of energy networks: A graph embedding learning approach | |
| Yu et al. | Time series cross-correlation network for wind power prediction | |
| Li et al. | Comparison and application potential analysis of autoencoder-based electricity pattern mining algorithms for large-scale demand response | |
| Guo et al. | Rolling bearing fault diagnosis based on the combination of improved deep convolution network and gated recurrent unit | |
| Zhou et al. | Transient identification for nuclear power plants based on the similarity of matrices | |
| CN116826727B (en) | Ultra-short-term wind power prediction method and prediction system based on time sequence representation and multistage attention | |
| CN115750226A (en) | Wind turbine generator fault diagnosis method based on local-global hierarchical feature modeling | |
| Kabilan et al. | A LSTM Deep Learning method with Attention Dense Mechanism for AQI prediction |
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 |