CN113449618A - Method for carrying out deep learning rolling bearing fault diagnosis based on feature fusion and mixed enhancement - Google Patents
Method for carrying out deep learning rolling bearing fault diagnosis based on feature fusion and mixed enhancement Download PDFInfo
- Publication number
- CN113449618A CN113449618A CN202110670565.0A CN202110670565A CN113449618A CN 113449618 A CN113449618 A CN 113449618A CN 202110670565 A CN202110670565 A CN 202110670565A CN 113449618 A CN113449618 A CN 113449618A
- Authority
- CN
- China
- Prior art keywords
- fault diagnosis
- rolling bearing
- dimensional
- enhancement
- mixed
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 37
- 238000005096 rolling process Methods 0.000 title claims abstract description 28
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000013135 deep learning Methods 0.000 title claims abstract description 16
- 230000004927 fusion Effects 0.000 title claims abstract description 16
- 238000012549 training Methods 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 11
- 230000003595 spectral effect Effects 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 2
- 238000000605 extraction Methods 0.000 abstract description 4
- 238000013527 convolutional neural network Methods 0.000 abstract description 2
- 239000013598 vector Substances 0.000 description 9
- 230000015556 catabolic process Effects 0.000 description 5
- 238000006731 degradation reaction Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000004141 dimensional analysis Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Signal Processing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a rolling bearing fault diagnosis method based on feature fusion and mixed enhancement, which adds feature engineering on the basis of deep learning, comprehensively considers time domain features, frequency domain features, working condition features and time difference features of original signals, fuses the features and the original signals into a new one-dimensional signal and converts the new one-dimensional signal into a two-dimensional image format, constructs a virtual sample and a label in a linear interpolation mode through mixed enhancement, and inputs the virtual sample and the label into a ResNet18 network for training by utilizing the powerful feature extraction capability of a two-dimensional convolutional neural network. The method comprehensively considers the potential characteristics of the original data, disturbs the distribution of the data and improves the generalization capability of the model. The method not only improves the precision of fault diagnosis of the rolling bearing, but also has good domain adaptability and is suitable for fault diagnosis under various working conditions.
Description
Technical Field
The invention relates to the technical field of deep learning, in particular to a method for diagnosing faults of a rolling bearing based on feature fusion and mixed enhancement.
Background
In recent years, with the upgrading of national industry and the energy conservation and emission reduction, real-time fault diagnosis is required to be realized for large mechanical systems such as aviation generators, rail transit equipment and agricultural equipment, and an advanced health monitoring system with management functions such as health monitoring, fault diagnosis detection and diagnosis and residual life prediction is developed. Rolling bearings are important parts in rotating machines. Among the failures of the rotating machine, the failure caused by the damage of the bearing accounts for about 30%. Therefore, the fault diagnosis of the rolling bearing is of great significance to the condition monitoring and maintenance of the rotary mechanical equipment.
In a practical industrial environment, although the types of faults are the same, their signals may show large differences under different operating conditions. Many rolling bearing fault diagnosis researches are focused on the high precision of models under the same working condition, the important influence of different operating conditions on data internal representation is neglected, the processing and selection of data characteristics are single, and the comprehensiveness is lacked.
On the other hand, although a very strong feature extraction capability can be trained through the deep neural network, it is difficult to learn general knowledge in the data, so that the finally trained model is only suitable for processing data similar to the distribution of training samples, and the model does not perform well on test data outside the distribution of the training samples. Therefore, in a practical application scenario, a "model degradation" phenomenon occurs: models trained at one stage test for performance degradation on data acquired at another stage.
The deep learning rolling bearing fault diagnosis method has the following problems.
First, the domain adaptation problem. Although the deep neural network can train very strong feature extraction capability, it is difficult to learn general knowledge in the data, so that the finally trained model is only suitable for processing data similar to the distribution of training samples, and the performance on test data except the distribution of the training samples is poor. Therefore, in a practical application scenario, a "model degradation" phenomenon occurs: models trained at one stage test for performance degradation on data acquired at another stage. This is the domain adaptation problem for deep learning.
Second, the feature selection problem. In a practical industrial environment, although the types of faults are the same, their signals may show large differences under different operating conditions. Many researches on fault diagnosis focus on the high precision of models under the same working conditions, the important influence of different operating conditions on data internal representation is ignored, and the processing and selection of data characteristics are single and lack of comprehensiveness.
And thirdly, data enhancement problem. In the field of computer vision, a traditional data enhancement mode generates a new training sample by performing geometric transformation such as cropping, rotating and scaling on an image, and an image label is kept unchanged. However, the data enhancement mode has the following limitations:
1. the generated new samples belong to the same category;
2. the relationships between different samples of different classes are not modeled;
3. an overfitting phenomenon occurs. New data enhancement approaches need to be designed.
Disclosure of Invention
In view of the above, the present invention provides a method for diagnosing rolling bearing faults through deep learning based on feature fusion and mixed enhancement, so as to solve the technical problems in the background art. The method comprehensively considers time domain characteristics, frequency domain characteristics, working condition characteristics and time difference characteristics of original vibration signals, fuses the characteristics and the original signals together to form a new characteristic vector with more comprehensive information, disturbs the distribution of the characteristic vector through a mixed enhancement strategy, inhibits overfitting of a model, and finally improves the domain adaptability and the generalization capability of the rolling bearing fault diagnosis model.
In order to achieve the purpose, the invention adopts the following technical scheme:
a method for carrying out deep learning rolling bearing fault diagnosis based on feature fusion and mixed enhancement comprises the following steps:
step S1, extracting time domain characteristics, frequency domain characteristics, working condition characteristics and time difference characteristics in the original one-dimensional vibration signal, and combining the time domain characteristics, the frequency domain characteristics, the working condition characteristics and the time difference characteristics with the original one-dimensional vibration signal to form a new one-dimensional signal sample;
step S2, converting the new one-dimensional signal sample obtained in the step S1 into a two-dimensional image format;
step S3, performing mixed enhancement processing on the one-dimensional signal sample converted in the step S2 to construct a virtual sample and a virtual label;
step S4, inputting the virtual sample and the virtual label obtained in the step S3 as a training set into a ResNet18 network for training to obtain a fault diagnosis model;
and step S5, using the untrained and un-mixed sample and label as the test set, and testing on the fault diagnosis model obtained in step S4 to obtain the fault diagnosis classification result of the rolling bearing.
Further, in the step S1, the original one-dimensional vibration signal includes bearing experimental data of the university of kaiser, usa, which is obtained from an accelerometer using drive-end bearing data at a sampling frequency of 12K.
Further, in step S1, the original one-dimensional vibration signal is preprocessed to extract a time domain feature, a frequency domain feature, a working condition feature and a time difference feature, and the four extracted features and the original one-dimensional vibration signal are combined, where the combination is sequential splicing; the time domain features include: maximum value, mean value, variance, root mean square value, skewness and kurtosis on a certain section of signal; the frequency domain features include: wavelet coefficient, spectral kurtosis, spectral skewness;
the operating condition characteristics include: one-hot encoding of the working conditions; the time difference characteristic: amplitude of change characteristic within each time window.
Further, in step S2, a sequential splitting is adopted for the conversion of the new one-dimensional signal sample.
Further, in the step S3, the aliasing enhancement processing includes linear interpolation, and the linear interpolation is performed on the converted one-dimensional signal samples to construct new data samples and corresponding one-hot labels.
Further, the fault diagnosis model has a trained model weight file.
The invention has the beneficial effects that:
the method comprehensively considers different working conditions of fault detection, fuses the fault detection with time domain characteristics and original signals, comprehensively extracts performance indexes related to faults, improves the domain adaptability of a fault diagnosis model through a new data enhancement method, enhances the generalization capability of the model, and is suitable for fault diagnosis of the rolling bearing under different working conditions.
Drawings
Fig. 1 is a schematic flowchart of a method for deep learning rolling bearing fault diagnosis based on feature fusion and mixed class enhancement provided in embodiment 1.
Fig. 2 is a schematic diagram of the construction of a new feature vector in example 1.
Fig. 3 is a schematic diagram of a residual network block structure in embodiment 1.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1
Referring to fig. 1 to 3, embodiment 1 provides a method for deep learning rolling bearing fault diagnosis based on feature fusion and mixed class enhancement, which includes the following steps:
step S1, extracting time domain characteristics, frequency domain characteristics, working condition characteristics and time difference characteristics in the original one-dimensional vibration signal, and combining the time domain characteristics, the frequency domain characteristics, the working condition characteristics and the time difference characteristics with the original one-dimensional vibration signal to form a new one-dimensional signal sample;
specifically, in this embodiment, the raw one-dimensional vibration signal was obtained from an accelerometer using bearing experimental data from the university of kaiser university laboratory, usa, and the experimental data was taken using drive-end bearing data at a sampling frequency of 12K. In this experimental data, there are four failure types of rolling bearings: inner ring fault, rolling body fault and outer ring fault are normal. The failure diameters for each failure type are 0.007 inches, 0.014 inches and 0.021 inches, respectively, and thus there are ten failure labels.
More specifically, in the embodiment, in addition to extracting the original features, time domain features, frequency domain features, operating condition features and time difference features are also extracted, so as to increase the diversity of the features.
The above features specifically include:
time domain characteristics: maximum, mean, variance, root mean square, skewness, kurtosis, etc. over a segment of the signal.
Frequency domain characteristics: wavelet coefficients, spectral kurtosis, spectral skewness, etc.
The working condition characteristics are as follows: for one-hot coding under working conditions, taking 4 working conditions as an example, the feature vectors corresponding to the one-hot coding are [1,0,0,0], [0,1,0,0], [0,0,1,0], [0,0,0,1], respectively.
Time difference characteristics: the amplitude of change characteristic in each time window, in particular, defines a time sequence in a time window as X ═ X (X)1,x2,…,xm) Starting from a second time point in the time window, the variation amplitude, i.e. ax, of the second time point relative to the previous time point is determinedi＝(xi+1-xi)/xiEventually, a new time series can be formed
More specifically, in this embodiment, the combination is specifically: and after extracting features from multiple angles, sequentially splicing the extracted features and the original one-dimensional signals to construct a new one-dimensional feature vector.
Step S2, converting the new one-dimensional signal sample obtained in the step S1 into a two-dimensional image format;
specifically, in this embodiment, the step adopts sequential splitting for the conversion of the new one-dimensional signal samples, and the converted samples have the representation format of the two-dimensional image and satisfy the input format of the training model.
More specifically, conventional one-dimensional analysis often faces the difficulty of capturing the natural patterns of machine fault conditions. Compared with a one-dimensional signal, an image is a two-dimensional data matrix, can carry stronger information, and represents more complex structural distribution. In order to capture the characteristics of the periodic signal and utilize the strong characteristic extraction capability of the two-dimensional convolutional neural network on image classification, the one-dimensional signal after characteristic fusion is converted into a two-dimensional image format to be used as the input of the deep neural network.
The conversion process is as follows:
firstly, dividing an original one-dimensional vibration signal into n equal parts;
each section is then arranged in the line order of the signal image:
in the formula, I represents the converted signal image, and x (t) is the vibration signal at time t. The converted image format is a single channel, in order to meet the requirement of selecting the input format of the neural network model, the single channel is expanded into three channels, and each channel has the same information.
Step S3, performing mixed enhancement processing on the one-dimensional signal sample converted in the step S2 to construct a virtual sample and a virtual label;
specifically, in this embodiment, the mixture enhancement processing in this step adopts a linear interpolation mode, and this mode can construct a batch of new data samples and corresponding one-hot labels.
More specifically, the mixed enhancement is an unconventional data enhancement mode, and the method obtains new extension data through a linear interpolation method: virtual training samples and labels. In essence, the mixed class enhancement trains the neural network on a convex combination of training samples and their labels, normalizing the linear expression of the model. Furthermore, the mixed class enhancement is simple and effective to implement, consistent with the Oncam razor principle.
The processing procedure of the mixed enhancement on the sample is as follows:
in the formula (x)i,yi)，(xj,yj) Are two samples, x, randomly drawn from the training dataiAnd xjRepresenting the original input sample vector, yiAnd yjRepresenting the one-hot tag code. lambda-Beta [ alpha, alpha ]]And λ ∈ [0, 1]]. α is a hyper-parameter.
Step S4, inputting the virtual sample and the virtual label obtained in the step S3 as a training set into a ResNet18 network for training to obtain a fault diagnosis model;
specifically, in the present embodiment, the ResNet18 network is used as an existing classical network model, and when the network input is matched, the virtual samples are converted from a single-channel format to a three-channel format, and each channel has the same information.
More specifically, the ResNet network is constructed by residual modules, and fig. 2 shows the structure of the residual network blocks.
The ResNet network can effectively relieve the degradation problem caused by the number of layers increased by the deep neural network model. In the residual block, the input can propagate forward through the data lines across the layers more quickly without adding additional parameters and computation to the network.
The industrial floor environment requires that the neural network model has the characteristics of high speed and small size, and the ResNet18 network is lighter than other ResNet50, ResNet101 and other networks, so the ResNet18 network is selected as the training model in the embodiment, and the network has 17 convolutional layers and 1 full-connection layer.
After a series of preprocessing of the samples, the input of the ResNet18 residual network is the preprocessed two-dimensional image format samples and the corresponding one-hot labels during training, and the weights of the network are continuously updated through iteration and an Adam optimization algorithm in the training process.
And step S5, using the untrained and un-mixed sample and label as the test set, and testing on the fault diagnosis model obtained in step S4 to obtain the fault diagnosis classification result of the rolling bearing.
Specifically, in this embodiment, the fault diagnosis model is trained on the basis of the original ResNet18 network, and has a trained model weight file, and the test set is a two-dimensional image format sample without performing aliasing enhancement and a one-hot label corresponding to the two-dimensional image format sample.
In summary, the present invention mainly includes three stages, i.e., feature fusion, format conversion and mixed class enhancement.
Firstly, extracting time domain characteristics, frequency domain characteristics, working condition characteristics and time difference characteristics of a vibration signal of the rolling bearing, and sequentially splicing the characteristics and an original signal to form a new characteristic vector, wherein the characteristic vector has richer characteristic information.
And secondly, converting the new one-dimensional feature vector into a two-dimensional image format.
And finally, constructing virtual samples and labels through mixed enhancement, and inputting the virtual samples and labels into a ResNet18 network for training. The trained model can be tested on a two-dimensional image sample which is not subjected to mixed enhancement.
The invention has the following advantages: on one hand, different working condition conditions of fault detection are comprehensively considered, the fault detection, time domain characteristics and original signals are fused, performance indexes related to faults are comprehensively extracted, on the other hand, the domain adaptability of a fault diagnosis model is improved through a new data enhancement method, the generalization capability of the model is enhanced, and the method is suitable for fault diagnosis of the rolling bearing under different working condition conditions.
The invention is not described in detail, but is well known to those skilled in the art.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (6)
1. A method for carrying out deep learning rolling bearing fault diagnosis based on feature fusion and mixed enhancement is characterized by comprising the following steps:
step S1, extracting time domain characteristics, frequency domain characteristics, working condition characteristics and time difference characteristics in the original one-dimensional vibration signal, and combining the time domain characteristics, the frequency domain characteristics, the working condition characteristics and the time difference characteristics with the original one-dimensional vibration signal to form a new one-dimensional signal sample;
step S2, converting the new one-dimensional signal sample obtained in the step S1 into a two-dimensional image format;
step S3, performing mixed enhancement processing on the one-dimensional signal sample converted in the step S2 to construct a virtual sample and a virtual label;
step S4, inputting the virtual sample and the virtual label obtained in the step S3 as a training set into a ResNet18 network for training to obtain a fault diagnosis model;
and step S5, using the untrained and un-mixed sample and label as the test set, and testing on the fault diagnosis model obtained in step S4 to obtain the fault diagnosis classification result of the rolling bearing.
2. The method for deep learning rolling bearing fault diagnosis based on feature fusion and mixed class enhancement as claimed in claim 1, wherein in the step S1, the raw one-dimensional vibration signal includes bearing experimental data of the university of kaiser storage laboratory, usa, which is obtained from an accelerometer using drive-end bearing data at a sampling frequency of 12K.
3. The method for diagnosing the rolling bearing fault based on the deep learning of the feature fusion and the mixed enhancement as claimed in claim 1, wherein in the step S1, the time domain feature, the frequency domain feature, the operating condition feature and the time difference feature are extracted by preprocessing the original one-dimensional vibration signal, and the four extracted features are combined with the original one-dimensional vibration signal, and the combination form is sequential splicing; the time domain features include: maximum value, mean value, variance, root mean square value, skewness and kurtosis on a certain section of signal; the frequency domain features include: wavelet coefficient, spectral kurtosis, spectral skewness;
the operating condition characteristics include: one-hot encoding of the working conditions; the time difference characteristic: amplitude of change characteristic within each time window.
4. The method for deep learning rolling bearing fault diagnosis based on feature fusion and mixed class enhancement as claimed in claim 1, wherein in the step S2, sequential splitting is adopted for the conversion of the new one-dimensional signal samples.
5. The method for deep learning rolling bearing fault diagnosis based on feature fusion and mixed enhancement as claimed in claim 1, wherein in the step S3, the mixed enhancement processing includes linear interpolation, and the linear interpolation is performed on the converted one-dimensional signal samples to construct new data samples and corresponding one-hot tags.
6. The method for deep learning rolling bearing fault diagnosis based on feature fusion and mixed class enhancement as claimed in claim 1, wherein the fault diagnosis model has a trained model weight file.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110670565.0A CN113449618A (en) | 2021-06-17 | 2021-06-17 | Method for carrying out deep learning rolling bearing fault diagnosis based on feature fusion and mixed enhancement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110670565.0A CN113449618A (en) | 2021-06-17 | 2021-06-17 | Method for carrying out deep learning rolling bearing fault diagnosis based on feature fusion and mixed enhancement |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113449618A true CN113449618A (en) | 2021-09-28 |
Family
ID=77811751
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110670565.0A Pending CN113449618A (en) | 2021-06-17 | 2021-06-17 | Method for carrying out deep learning rolling bearing fault diagnosis based on feature fusion and mixed enhancement |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113449618A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114647909A (en) * | 2022-04-02 | 2022-06-21 | 江苏科技大学 | Method for determining rolling bearing degradation point based on maximum value of spectral kurtosis characteristic |
| CN117951529A (en) * | 2024-03-26 | 2024-04-30 | 济南浪潮数据技术有限公司 | Sample acquisition method, device and equipment for hard disk data fault prediction |
| CN117951529B (en) * | 2024-03-26 | 2024-06-21 | 济南浪潮数据技术有限公司 | Sample acquisition method, device and equipment for hard disk data fault prediction |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111122161A (en) * | 2019-12-24 | 2020-05-08 | 北京交通大学 | Variable working condition bearing fault diagnosis method based on fast kurtogram and deep residual learning |
| CN112304614A (en) * | 2020-11-27 | 2021-02-02 | 中国人民解放军海军工程大学 | End-to-end rolling bearing intelligent fault diagnosis method adopting multi-attention machine system |
| CN112541389A (en) * | 2020-09-29 | 2021-03-23 | 西安交通大学 | Power transmission line fault detection method based on EfficientDet network |
-
2021
- 2021-06-17 CN CN202110670565.0A patent/CN113449618A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111122161A (en) * | 2019-12-24 | 2020-05-08 | 北京交通大学 | Variable working condition bearing fault diagnosis method based on fast kurtogram and deep residual learning |
| CN112541389A (en) * | 2020-09-29 | 2021-03-23 | 西安交通大学 | Power transmission line fault detection method based on EfficientDet network |
| CN112304614A (en) * | 2020-11-27 | 2021-02-02 | 中国人民解放军海军工程大学 | End-to-end rolling bearing intelligent fault diagnosis method adopting multi-attention machine system |
Non-Patent Citations (2)
| Title |
|---|
| 吴定会 等: "《基于多特征融合与 XGBoost 的风机轴承故障诊断》", 《传感器与微系统》, vol. 39, no. 07, 2 July 2020 (2020-07-02), pages 146 - 147 * |
| 孟祥峰: "《基于深度学习的滚动轴承的故障诊断及预测》", 《中国优秀硕士学位论文全文数据库》, no. 08, 15 August 2020 (2020-08-15), pages 6 - 16 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114647909A (en) * | 2022-04-02 | 2022-06-21 | 江苏科技大学 | Method for determining rolling bearing degradation point based on maximum value of spectral kurtosis characteristic |
| CN117951529A (en) * | 2024-03-26 | 2024-04-30 | 济南浪潮数据技术有限公司 | Sample acquisition method, device and equipment for hard disk data fault prediction |
| CN117951529B (en) * | 2024-03-26 | 2024-06-21 | 济南浪潮数据技术有限公司 | Sample acquisition method, device and equipment for hard disk data fault prediction |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110361176B (en) | Intelligent fault diagnosis method based on multitask feature sharing neural network | |
| CN110647830B (en) | Bearing fault diagnosis method based on convolutional neural network and Gaussian mixture model | |
| CN113792758B (en) | Rolling bearing fault diagnosis method based on self-supervision learning and clustering | |
| Lu et al. | A novel feature extraction method using deep neural network for rolling bearing fault diagnosis | |
| CN111931625B (en) | Product key part residual life prediction method based on asymmetric loss neural network | |
| CN114004252A (en) | Bearing fault diagnosis method, device and equipment | |
| Chen et al. | Deep residual network for identifying bearing fault location and fault severity concurrently | |
| Wang et al. | A light weight multisensory fusion model for induction motor fault diagnosis | |
| CN114091504A (en) | Rotary machine small sample fault diagnosis method based on generation countermeasure network | |
| Yang et al. | Few-shot learning for rolling bearing fault diagnosis via siamese two-dimensional convolutional neural network | |
| CN116011507A (en) | Rare fault diagnosis method for fusion element learning and graph neural network | |
| CN116842379A (en) | Mechanical bearing residual service life prediction method based on DRSN-CS and BiGRU+MLP models | |
| CN115587290A (en) | Aero-engine fault diagnosis method based on variational self-coding generation countermeasure network | |
| Liu et al. | Multi-Scale Fusion Attention Convolutional Neural Network for Fault Diagnosis of Aero-Engine Rolling Bearing | |
| CN113449618A (en) | Method for carrying out deep learning rolling bearing fault diagnosis based on feature fusion and mixed enhancement | |
| CN111428772A (en) | Photovoltaic system depth anomaly detection method based on k-nearest neighbor adaptive voting | |
| CN108898157B (en) | Classification method for radar chart representation of numerical data based on convolutional neural network | |
| CN116484258A (en) | Elevator traction machine bearing fault diagnosis method | |
| CN116541771A (en) | Unbalanced sample bearing fault diagnosis method based on multi-scale feature fusion | |
| CN116007937A (en) | Intelligent fault diagnosis method and device for mechanical equipment transmission part | |
| CN111523557A (en) | Wind power intelligent fault diagnosis method based on big data | |
| CN114383845A (en) | Bearing composite fault diagnosis method based on embedded zero sample learning model | |
| CN115825239A (en) | Generator stator bar insulation defect identification method | |
| CN112257773B (en) | Mechanical equipment fault diagnosis method based on multiple measurement vectors of wireless sensor network | |
| CN113177584A (en) | Zero sample learning-based composite fault diagnosis method |
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 |