CN108764341A - A kind of adaptive deep neural network model of operating mode and variable working condition method for diagnosing faults - Google Patents
A kind of adaptive deep neural network model of operating mode and variable working condition method for diagnosing faults Download PDFInfo
- Publication number
- CN108764341A CN108764341A CN201810530399.2A CN201810530399A CN108764341A CN 108764341 A CN108764341 A CN 108764341A CN 201810530399 A CN201810530399 A CN 201810530399A CN 108764341 A CN108764341 A CN 108764341A
- Authority
- CN
- China
- Prior art keywords
- module
- characteristic extracting
- fault
- source domain
- extracting module
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Acoustics & Sound (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The edge distribution that the present invention is directed to failure under different working conditions is identical, but it is distributed in the characteristics of changing on scale and position per the condition of class fault sample, a kind of adaptive deep neural network model of operating mode and variable working condition method for diagnosing faults are provided, the adaptive deep neural network model of the operating mode is made of five parts, including source domain characteristic extracting module, fault grader, target domain characteristic extracting module, Location Scale conversion module and field difference regularization module.For the fault sample of source domain after source domain characteristic extracting module and the processing of Location Scale conversion module, the condition distribution that condition is distributed fault sample similar with target domain is similar.The present invention overcomes the differences that sensing data under the conditions of variable working condition is distributed, operating mode influence can be eliminated by providing one kind, and the method for obtaining the information for only reflecting rolling bearing fault or performance degradation has high promotional value to keep the diagnosis of rolling bearing fault more accurate.
Description
Technical field
The present invention relates to Fault Diagnosis of Roller Bearings, the adaptive deep neural network model of especially a kind of operating mode and
Variable working condition method for diagnosing faults belongs to mechanical fault diagnosis field.
Background technology
Rolling bearing is in electric power, petrochemical industry, metallurgy, machinery, aerospace and some war industry departments using most wide
General machine components, and most easy damaged one of component.It is with efficient, frictional resistance is small, easy to assembly, lubrication is easily real
The advantages that existing, using very universal on rotating machinery, and plays key effect.Many failures of rotating machinery all with rolling
Dynamic bearing has close association.According to relevant statistics, the 70% of mechanical breakdown is vibration fault, and is had in vibration fault
30% is caused by rolling bearing.This is because rolling bearing plays the work for bearing load and transmitting load in mechanical equipment
With, and operating condition is more severe, and long continuous operation is easy to be damaged and break down under top load, high rotating speed.
Direct result caused by rolling bearing fault gently then reduces and loses certain functions of system, heavy then cause serious even calamity
The accident of difficulty.Therefore, the method for diagnosing faults of rolling bearing, be always the technology given priority in mechanical fault diagnosis it
One, there is important social and economic significance.
Often changeable (load, rotating speed etc. continuously or intermittently become operating condition rolling bearing in mechanical equipment
Change).There are direct correlation relationships with operating mode for collected transducing signal.When system variable parameter operation, new data continues to bring out, former
It is first available to there are tag sensor data to produce distributional difference with the test sample under new working condition.Existing training sample
It has been not enough to training and has obtained a reliable fault diagnosis model.Meanwhile the failure under the new working condition of a batch is marked again
Sample is not only time-consuming and laborious but also very expensive.
This just causes a major issue of rolling bearing fault diagnosis, that is, how using on a small quantity under the conditions of variable working condition
Have label training sample or source domain data, establish a reliable model to new working condition or target domain data into
Row prediction (source domain data and target domain data can not have identical data distribution).
Invention content
Technical problem to be solved by the present invention lies in the difference that sensing data is distributed under the conditions of variable working condition is overcome, provide
One kind can eliminate operating mode influence, and the method for obtaining the information for only reflecting rolling bearing fault or performance degradation, and accurately sentence
The diagnostic method of disconnected rolling bearing fault.
In order to solve the above-mentioned technical problem, the present invention devises the rolling bearing fault diagnosis side under the conditions of a kind of variable working condition
Method, this method be based on a kind of adaptive deep neural network model of operating mode, the adaptive deep neural network model of the operating mode it is defeated
Enter be vibration signal x the ∈ X, feature space X of different working condition lower bearings can be original vibration signal by quick Fu
The spectral vectors obtained after leaf transformation export as the operating mode type belonging to fault type label y ∈ Y={ 1,2 ..., K } and sample
Label d ∈ { 0,1 }.Assuming thatWithIt indicates the distribution situation of fault sample under different operating modes, is denoted as source domain distribution respectively
It is distributed with target domain.WithIndicate the edge distribution situation of vibration signal under different operating modes.If vibration signal xiIt comes from
Source domain, i.e.,So di=0.If vibration signal xiFrom target domain, i.e.,So di=1.
Experiment shows under different working conditions that source domain is identical with the edge distribution of target domain failure, but fault sample
Condition distribution it is variant, that is,By further studying, it has been found that the original of fault sample
Beginning vibration signal is only the change on scale and position after characteristic extracting module, between the condition distribution per class fault sample
Change.Therefore, for failure yi∈ Y, it is (W that we, which can find a parameter,i,bi) linear transformation, make the failure sample of source domain
This is after the linear change, condition distributionThe condition of fault sample similar with target domain point
ClothIt is similar.
Since the sample of target domain lacks faulty tag, we can not directly obtain target domain failure yiThe condition of sample
DistributionBut it considersAccording to Bayesian formula, if
SoThat is all kinds of fault samples of source domain respectively after linear transformation,
New sample has edge distribution identical with target domain sample.Therefore, we can by minimizing following distributional difference,
Obtain the corresponding linear change parameter of all kinds of failures
Wherein MMD indicates Maximum Mean Discrepancy, is distributional difference between a kind of common measurement sample
Method.
The adaptive deep neural network model of operating mode proposed by the present invention includes 5 parts:
1, source domain characteristic extracting module MS:MSIncluding 5 layer 1 dimension convolutional neural networks layer (Conv1~Conv5) and 2 entirely
Articulamentum (FC1, FC2);The source domain vibration signal of inputFirst pass around Fast Fourier Transform (FFT) (Fast Fourier
Transform, abbreviation FFT) processing, then input first layer convolutional neural networks layer (Conv1);The last one full articulamentum
(FC2) include K neuron of quantity identical with fault type;Source domain samplePass through MS5 layer 1 dimension convolutional neural networks
Layer is mapped as feature vector with 2 layers of full articulamentum
2, fault grader C:Fault diagnosis essence is a multicategory classification problem, we use Soft-max regression models
Estimate the source domain vibration signal of inputBelong to the probability of each fault category, that is,
Wherein,It isJ-th of element value;It, can be with for given source domain sample set
Estimate source domain characteristic extracting module M by maximizing following cost functionSParameter,
Wherein,It isJ-th of element value, i.e.,Belong to the probability of each fault category j;
3, target domain characteristic extracting module MT:MTWith MSWith identical network structure;The target domain of input shakes
Dynamic signalIt is mapped as feature vector by 5 layer of 1 dimension convolutional neural networks layer and 2 layers of full articulamentum
In view of the sample of target domain does not have the label of fault category, so using MSIn 5 layer of 1 dimension convolutional neural networks layer and 2 layers
The parameter of full articulamentum initializes MT;
4, Location Scale conversion module LS:K class failures need K Location Scale conversion moduleLSiParameter
It is denoted as (Wi,bi), output is expressed asFault sample i.e. in source domain after linear change, wherein yiIndicate failure classes
Type;
5, field difference regularization term module:Use the distributional difference of sample between field after MMD expression linear transformations.Pass through
The distributional difference is minimized, estimates model parameter, including target domain characteristic extracting module MTIn 5 layer 1 dimension convolutional neural networks
The parameter of layer and the parameter and K Location Scale conversion module of 2 layers of full articulamentum
The present invention proposes the Fault Diagnosis of Roller Bearings under the conditions of a kind of variable working condition, including:
Step 1:Training source domain characteristic extracting module MS:Input source domain has exemplar, using back-propagation algorithm,
Minimize cost function LclsEstimate source domain characteristic extracting module MSIn 5 layer of 1 dimension convolutional neural networks layer and 2 layers of full articulamentum
Parameter:
Wherein,It isJ-th of element value, i.e.,Belong to the general of each fault category j
Rate;
Step 2:Use MSIn the parameter of 5 layer of 1 dimension convolutional neural networks layer and 2 layers of full articulamentum initialize MT;
Step 3:Training objective domain features extraction module MT, Location Scale conversion module LS:Input source domain has label
Sample and target domain unlabeled exemplars minimize following distributional difference using back-propagation algorithm:
Estimate target domain characteristic extracting module MTIn 5 layer of 1 dimension convolutional neural networks layer and 2 layers of full articulamentum parameter and
The parameter of K Location Scale conversion module;
Step 4:Fault diagnosis:Target domain fault sampleBy target domain characteristic extracting module MTIt is defeated after processing
Go out forIt is calculated by the following formula sampleFault typeThat is the corresponding failure of most probable value
Type:
Wherein,It isJ-th of element value.
Advantageous effect:The edge distribution that the present invention is directed to failure under different working conditions is identical, but per class fault sample
Condition is distributed in the characteristics of changing on scale and position, devises the adaptive deep neural network model of operating mode and corresponding
Variable working condition method for diagnosing faults overcomes the difference of sensing data distribution under the conditions of variable working condition, and work can be eliminated by providing one kind
Condition influences, and the method for obtaining the information for only reflecting rolling bearing fault or performance degradation, to make examining for rolling bearing fault
It is disconnected more accurate, there is high promotional value.
Description of the drawings
Fig. 1 is the adaptive deep neural network model structural schematic diagram of operating mode of the present invention;
Fig. 2 is the source domain characteristic extracting module M of the present inventionSStructural schematic diagram;
Fig. 3 is the flow diagram of the variable working condition Fault Diagnosis of Roller Bearings of the present invention.
Specific implementation mode
It elaborates to the present invention below in conjunction with attached drawing.
As shown in Figs. 1-2, the adaptive deep neural network model of operating mode provided by the invention, including, 1 source domain feature
Extraction module MS, 1 fault grader C, 1 target domain characteristic extracting module MT, 4 Location Scale conversion module LS:Respectively
Corresponding four class failures, i.e. inner ring failure (IF), outer ring failure (OF), rolling element failure (BF) and normal condition (NO);1 field
Difference regularization term module;Dotted portion indicates to need the parameter using back-propagation algorithm estimation response, bold portion in figure
Statement network parameter has determined.
As shown in Fig. 2, source domain characteristic extracting module MSAs a part for the adaptive deep neural network model of operating mode,
Including 5 convolutional neural networks layers, 2 full articulamentums;The effect of the module is to obtain the nerve for having identification to failure
Network parameter;Dotted portion expression needs to estimate corresponding parameter using back-propagation algorithm in figure;Input source domain has label
Sample minimizes cost function L using back-propagation algorithmclsEstimate source domain characteristic extracting module MSThe parameter of each layer.
As shown in figure 3, the Fault Diagnosis of Roller Bearings under the conditions of a kind of variable working condition includes the following steps:
Step 1:Training source domain characteristic extracting module MS:Input source domain has exemplar, using back-propagation algorithm,
Minimize cost function LclsEstimate source domain characteristic extracting module MSIn 5 layer of 1 dimension convolutional neural networks layer and 2 layers of full articulamentum
Parameter:
Wherein,It isJ-th of element value, i.e.,Belong to the general of each fault category j
Rate;
Step 2:Use MSIn the parameter of 5 layer of 1 dimension convolutional neural networks layer and 2 layers of full articulamentum initialize MT;
Step 3:Training objective domain features extraction module MT, Location Scale conversion module LS:Input source domain has label
Sample and target domain unlabeled exemplars minimize following distributional difference using back-propagation algorithm:
Estimate target domain characteristic extracting module MTIn 5 layer of 1 dimension convolutional neural networks layer and 2 layers of full articulamentum parameter and
The parameter of K Location Scale conversion module;
Step 4:Fault diagnosis:Target domain fault sampleBy target domain characteristic extracting module MTIt is defeated after processing
Go out forIt is calculated by the following formula sampleFault typeThat is the corresponding failure of most probable value
Type:
Wherein,It isJ-th of element value.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, several improvement can also be made under the premise of not departing from inventive principle, including increase fault type, these improvement
It should be regarded as protection scope of the present invention.
Claims (3)
1. the Fault Diagnosis of Roller Bearings under the conditions of a kind of variable working condition, it is characterised in that:This method is based on a kind of operating mode certainly
Deep neural network model is adapted to, which includes source domain characteristic extracting module MS, fault grader C, target domain feature
Extraction module MT, Location Scale conversion module LS and field difference regularization module;The source domain characteristic extracting module MSIncluding
5 layer of 1 dimension convolutional neural networks layer (Conv1~Conv5) and 2 full articulamentums (FC1, FC2), target domain characteristic extracting module
MTWith source domain characteristic extracting module MSNetwork structure it is identical;The fault sample of source domain passes through target domain feature extraction mould
Block MTAfter the LS processing of Location Scale conversion module, the condition that condition is distributed fault sample similar with target domain is distributed phase
Seemingly;Include the following steps:
Step 1:Training source domain characteristic extracting module MS:Input source domain has exemplar, minimum using back-propagation algorithm
Change cost function LclsEstimate source domain characteristic extracting module MSThe parameter of each layer:
Wherein,It isJ-th of element value, i.e.,Belong to the probability of each fault category j;
Step 2:Use source domain characteristic extracting module MSIn the parameter of each layer carry out initialized target domain features extraction module MT;
Step 3:Training objective domain features extraction module MT, Location Scale conversion module LS:Input source domain have exemplar and
Target domain unlabeled exemplars minimize following distributional difference using back-propagation algorithm:
Estimate target domain characteristic extracting module MTThe parameter of the parameter of each layer and K Location Scale conversion module LS;
Step 4:Fault diagnosis:Target domain fault sampleBy target domain characteristic extracting module MTAfter processing, exports and beIt is calculated by the following formula sampleFault typeThat is the corresponding fault type of most probable value:
Wherein,It isJ-th of element value.
2. the Fault Diagnosis of Roller Bearings under the conditions of variable working condition according to claim 1, it is characterised in that:Institute's rheme
The quantity for setting change of scale module LS is consistent with the quantity of fault type.
3. a kind of adaptive deep neural network model of operating mode, it is characterised in that:Including source domain characteristic extracting module MS, failure
Grader C, target domain characteristic extracting module MT, Location Scale conversion module LS and field difference regularization module;The source
Domain features extraction module MSIncluding 5 layer of 1 dimension convolutional neural networks layer (Conv1~Conv5) and 2 full articulamentums (FC1,
FC2), target domain characteristic extracting module MTWith source domain characteristic extracting module MSNetwork structure having the same.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810530399.2A CN108764341B (en) | 2018-05-29 | 2018-05-29 | A kind of Fault Diagnosis of Roller Bearings under the conditions of variable working condition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810530399.2A CN108764341B (en) | 2018-05-29 | 2018-05-29 | A kind of Fault Diagnosis of Roller Bearings under the conditions of variable working condition |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108764341A true CN108764341A (en) | 2018-11-06 |
CN108764341B CN108764341B (en) | 2019-07-19 |
Family
ID=64003317
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810530399.2A Active CN108764341B (en) | 2018-05-29 | 2018-05-29 | A kind of Fault Diagnosis of Roller Bearings under the conditions of variable working condition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108764341B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109580215A (en) * | 2018-11-30 | 2019-04-05 | 湖南科技大学 | A kind of wind-powered electricity generation driving unit fault diagnostic method generating confrontation network based on depth |
CN110031227A (en) * | 2019-05-23 | 2019-07-19 | 桂林电子科技大学 | A kind of Rolling Bearing Status diagnostic method based on binary channels convolutional neural networks |
CN110334478A (en) * | 2019-07-22 | 2019-10-15 | 山东浪潮人工智能研究院有限公司 | Machinery equipment abnormality detection model building method, detection method and model |
CN111458142A (en) * | 2020-04-02 | 2020-07-28 | 苏州智传新自动化科技有限公司 | Sliding bearing fault diagnosis method based on generation of countermeasure network and convolutional neural network |
CN111538947A (en) * | 2020-05-18 | 2020-08-14 | 中车永济电机有限公司 | Method for constructing wind power generator bearing fault classification model |
CN112629863A (en) * | 2020-12-31 | 2021-04-09 | 苏州大学 | Bearing fault diagnosis method for dynamic joint distribution alignment network under variable working conditions |
CN115389247A (en) * | 2022-11-01 | 2022-11-25 | 青岛睿发工程咨询服务合伙企业(有限合伙) | Rotating machinery fault monitoring method based on speed self-adaptive encoder |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101067930A (en) * | 2007-06-07 | 2007-11-07 | 深圳先进技术研究院 | Intelligent audio frequency identifying system and identifying method |
US20130304683A1 (en) * | 2010-01-19 | 2013-11-14 | James Ting-Ho Lo | Artificial Neural Networks based on a Low-Order Model of Biological Neural Networks |
CN105467975A (en) * | 2015-12-29 | 2016-04-06 | 山东鲁能软件技术有限公司 | Equipment fault diagnosis method |
CN106991999A (en) * | 2017-03-29 | 2017-07-28 | 北京小米移动软件有限公司 | Audio recognition method and device |
-
2018
- 2018-05-29 CN CN201810530399.2A patent/CN108764341B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101067930A (en) * | 2007-06-07 | 2007-11-07 | 深圳先进技术研究院 | Intelligent audio frequency identifying system and identifying method |
US20130304683A1 (en) * | 2010-01-19 | 2013-11-14 | James Ting-Ho Lo | Artificial Neural Networks based on a Low-Order Model of Biological Neural Networks |
CN105467975A (en) * | 2015-12-29 | 2016-04-06 | 山东鲁能软件技术有限公司 | Equipment fault diagnosis method |
CN106991999A (en) * | 2017-03-29 | 2017-07-28 | 北京小米移动软件有限公司 | Audio recognition method and device |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109580215A (en) * | 2018-11-30 | 2019-04-05 | 湖南科技大学 | A kind of wind-powered electricity generation driving unit fault diagnostic method generating confrontation network based on depth |
CN109580215B (en) * | 2018-11-30 | 2020-09-29 | 湖南科技大学 | Wind power transmission system fault diagnosis method based on deep generation countermeasure network |
CN110031227A (en) * | 2019-05-23 | 2019-07-19 | 桂林电子科技大学 | A kind of Rolling Bearing Status diagnostic method based on binary channels convolutional neural networks |
CN110334478A (en) * | 2019-07-22 | 2019-10-15 | 山东浪潮人工智能研究院有限公司 | Machinery equipment abnormality detection model building method, detection method and model |
CN110334478B (en) * | 2019-07-22 | 2023-07-25 | 山东浪潮科学研究院有限公司 | Machine equipment abnormality detection model construction method, detection method and model |
CN111458142A (en) * | 2020-04-02 | 2020-07-28 | 苏州智传新自动化科技有限公司 | Sliding bearing fault diagnosis method based on generation of countermeasure network and convolutional neural network |
CN111458142B (en) * | 2020-04-02 | 2022-08-23 | 苏州新传品智能科技有限公司 | Sliding bearing fault diagnosis method based on generation of countermeasure network and convolutional neural network |
CN111538947A (en) * | 2020-05-18 | 2020-08-14 | 中车永济电机有限公司 | Method for constructing wind power generator bearing fault classification model |
CN111538947B (en) * | 2020-05-18 | 2022-06-14 | 中车永济电机有限公司 | Method for constructing wind power generator bearing fault classification model |
CN112629863A (en) * | 2020-12-31 | 2021-04-09 | 苏州大学 | Bearing fault diagnosis method for dynamic joint distribution alignment network under variable working conditions |
CN115389247A (en) * | 2022-11-01 | 2022-11-25 | 青岛睿发工程咨询服务合伙企业(有限合伙) | Rotating machinery fault monitoring method based on speed self-adaptive encoder |
Also Published As
Publication number | Publication date |
---|---|
CN108764341B (en) | 2019-07-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108764341B (en) | A kind of Fault Diagnosis of Roller Bearings under the conditions of variable working condition | |
CN110555273B (en) | Bearing life prediction method based on hidden Markov model and transfer learning | |
Wei et al. | A novel intelligent method for bearing fault diagnosis based on affinity propagation clustering and adaptive feature selection | |
Li et al. | Fault diagnosis of rotating machinery based on combination of deep belief network and one-dimensional convolutional neural network | |
Wong et al. | Modified self-organising map for automated novelty detection applied to vibration signal monitoring | |
CN106874957A (en) | A kind of Fault Diagnosis of Roller Bearings | |
Zhang et al. | Intelligent fault diagnosis of mechanical equipment under varying working condition via iterative matching network augmented with selective Signal reuse strategy | |
Liang et al. | A deep learning method for motor fault diagnosis based on a capsule network with gate-structure dilated convolutions | |
CN113741394B (en) | Industrial equipment fault diagnosis system based on semi-supervised incremental learning | |
CN112487890B (en) | Bearing acoustic signal fault diagnosis method based on parallel sparse filtering | |
CN110008898A (en) | Industrial equipment data edges processing method based on symbol and convolutional neural networks | |
CN107370617A (en) | Cellular network fault diagnosis system based on SVM | |
CN114358124B (en) | New fault diagnosis method for rotary machinery based on deep countermeasure convolutional neural network | |
CN114564987A (en) | Rotary machine fault diagnosis method and system based on graph data | |
Islam et al. | Motor bearing fault diagnosis using deep convolutional neural networks with 2d analysis of vibration signal | |
CN112364706A (en) | Small sample bearing fault diagnosis method based on class imbalance | |
Appana et al. | Reliable fault diagnosis of bearings using distance and density similarity on an enhanced k-NN | |
CN112860183A (en) | Multisource distillation-migration mechanical fault intelligent diagnosis method based on high-order moment matching | |
WO2019178930A1 (en) | Fault diagnosis method for mechanical device | |
Wang et al. | Rotating machine fault detection based on HOS and artificial neural networks | |
CN113505639B (en) | Rotary machine multi-parameter health state assessment method based on TPE-XGBoost | |
TWI780434B (en) | Abnormal diagnosis device and method | |
CN116383739B (en) | Intelligent fault diagnosis method based on domain self-adaption multi-mode data fusion | |
CN116894215A (en) | Gear box fault diagnosis method based on semi-supervised dynamic graph attention | |
Peng et al. | IEPE accelerometer fault diagnosis for maintenance management system information integration in a heavy industry |
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 | ||
GR01 | Patent grant |