CN107451624A - The mechanized equipment health status recognition methods of network is locally connected based on orthogonalization - Google Patents
The mechanized equipment health status recognition methods of network is locally connected based on orthogonalization Download PDFInfo
- Publication number
- CN107451624A CN107451624A CN201710784356.2A CN201710784356A CN107451624A CN 107451624 A CN107451624 A CN 107451624A CN 201710784356 A CN201710784356 A CN 201710784356A CN 107451624 A CN107451624 A CN 107451624A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- sample
- msubsup
- munderover
- 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
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2136—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on sparsity criteria, e.g. with an overcomplete basis
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Acoustics & Sound (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
Abstract
The mechanized equipment health status recognition methods of network is locally connected based on orthogonalization, obtains the vibration signal sample of mechanized equipment monitoring first, and passes through non-overlapping segmentation, obtains sample fragment collection;Then fractional sample fragment collection is randomly selected, after whitening processing, trains orthogonal sparse autoencoder network;The orthogonal sparse autoencoder network trained is locally connected with sample fragment collection again, the local feature of sample fragment is extracted, the feature of vibration signal sample is calculated by local feature arithmetic average;Finally using the feature of vibration signal sample as input, Softmax graders are trained, recycles the Softmax graders trained to export diagnostic result, realizes the Intelligent Recognition of machine performance health status;The present invention can directly identify the health status of mechanized equipment by original vibration signal, avoid the human intervention in identification process;The orthogonality of sparse autoencoder network is ensure that, the sample characteristics for promoting model learning to arrive are various, and the accuracy of diagnosis is higher.
Description
Technical field
The invention belongs to mechanized equipment fault diagnosis technology field, and in particular to one kind locally connects network based on orthogonalization
Mechanized equipment health status recognition methods.
Background technology
With industrial automation and the continuous improvement of intelligent level, the safety military service of mechanized equipment is most important, and such as
What identifies that the health status of mechanized equipment is to ensure its key being on active service safely using method for diagnosing faults.The health of mechanized equipment
State is resided in the Monitoring Data acquired in sensor network, however, with the increase of sensor network scale, data biography per second
Throughput rate improves, and the Monitoring Data for causing sensor network to obtain is growing day by day, and machine can not increasingly be ensured by only relying on artificial experience
The ageing and reliability of tool equipment failure diagnosis.Therefore, how from the monitoring big data of mechanized equipment health status is excavated
Information, ensure that it is on active service safely, turn into the focus studied both at home and abroad and difficult point.
Intelligent trouble diagnosis is capable of the health status of automatic identification mechanized equipment by machine learning method, reduces tradition
To the undue dependence of diagnostic experiences and professional knowledge in failure diagnostic process, obvious accuracy and high efficiency are showed, is big
Mechanized equipment health status identification under data background provides method support.Traditional intelligent trouble diagnosis process is mainly by two
Part forms:1) fault signature extraction and preferably:Based on signal processing method or data statistics means, mechanical monitoring of equipment is extracted
The fault signature of signal, preferred Fault-Sensitive feature is concentrated from multiple features in conjunction with technologies such as distance estimations, principal component analysis;2)
Fault type recognition:With the data such as preferable Fault-Sensitive features training artificial neural network, SVMs or decision tree point
Class model, realize the automatic identification of fault type.Although traditional intelligence method for diagnosing faults can realize mechanized equipment health shape
The automatic identification of state, but its fault signature extracts and the preferably stage still needs to rely on artificial priori, constrains mechanized equipment intelligence
The popularization and application of energy method for diagnosing faults.
Machine learning method of the sparse autoencoder network as forward position, the failure that can be characterized in machinery monitoring signal are special
Sign, supported to solve the problems, such as that the human intervention of traditional intelligence failure diagnostic process provides method.Based on sparse autoencoder network
Intelligent failure diagnosis method by sample training model, can establish between machinery monitoring signal and health status non-linear reflects
Penetrate, reach the purpose for automatically extracting mechanical fault signature, identifying health status.But the sample currently used for network training is frequency domain
Or time-frequency domain sample, do not utilize original vibration signal directly, cause intelligent trouble diagnosis process can not fully achieve automation, study carefully
Its reason, on the one hand, original vibration signal has time shift characteristic, and this is directly to train sparse autoencoder network to bring with the signal
Difficulty;On the other hand, sparse regularization constraint is difficult to ensure that the orthogonal property of autoencoder network, the sample for causing model learning to arrive
Feature is similar, influences the accuracy of Fault Identification.
The content of the invention
The shortcomings that in order to overcome above-mentioned prior art, it is an object of the invention to provide locally connect network based on orthogonalization
Mechanized equipment health status recognition methods, be health that is accurate, efficiently and automatically identifying mechanized equipment under big data background
State provides effective method and supported.
To reach above-mentioned purpose, the technical scheme that the present invention takes is:
The mechanized equipment health status recognition methods of network is locally connected based on orthogonalization, is comprised the following steps:
Step 1:Obtain vibration signal sample set during mechanized equipment R kind health statusWherein,For m-th of health status sample, it is made up of N number of vibration number strong point, its sample label is ym∈{1,2,3,...R};
Step 2:Determine the input dimension N of orthogonal sparse autoencoder networkinWith output dimension Nout, according to input dimension
Nin, by health status sample xmIt is divided into J non-overlapping sample fragment, and J=N/Nin, form sample fragment collection
Wherein,It is j-th of sample fragment, by NinIndividual data point composition;
Step 3:The sample fragment collection obtained at random from step 2Middle selection NsIndividual sample fragmentAnd
Form fractional sample fragment collectionAfter whitening processing, albefaction fractional sample fragment collection is obtainedWhereinRecycle the orthogonal sparse autoencoder network of albefaction fractional sample fragment collection training, the partial weight of calculating network
MatrixMinimize object function:
In formula, σr() is ReLU activation primitives;λ is orthogonalization constraint factor;For partial weight matrix
WlocRow k row vector;
Step 4:The partial weight matrix W that calculation procedure 3 obtainslocThe sample fragment obtained with step 2Inner product, obtain
To sample fragmentLocal featureI.e.:
By each local featureArithmetic average, that is, the state sample that secures good health xmSample characteristics
Step 5:Utilize the sample characteristics { f of tape labelm,ymTraining Sotfmax graders, obtain Sotfmax graders
Weight matrixMinimize object function:
In formula,Respectively grader weight matrix WclassR rows, l every trades vector;
By the weight matrix W for calculating graderclassWith sample characteristics fmInner product, output sample characteristics correspond to each sample
The probability distribution of label, the sample label corresponding to maximum probability is taken as sample xmHealth status, complete mechanized equipment be good for
The Intelligent Recognition of health state.
Beneficial effects of the present invention are:The present invention overcomes the time shift characteristic of original vibration signal using local attachment structure
Caused influence;The sample characteristics for promoting sparse autoencoder network to learn by orthogonalization constraint are various;Connect with reference to local
Binding structure and the advantage of orthogonal sparse autoencoder network, can extract failure spy directly from the original vibration signal of mechanized equipment
Sign, automatic identification health status, the health status accuracy of identification of intelligent Fault Diagnosis Model is improved, is under big data background,
Accurately, efficiently and automatically identify that the health status of mechanized equipment provides effective method.
Brief description of the drawings
Fig. 1 is the flow chart of the present invention.
Fig. 2 is the schematic diagram that orthogonalization locally connects network.
Embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
As shown in figure 1, the mechanized equipment health status recognition methods of network is locally connected based on orthogonalization, including following step
Suddenly:
Step 1:Obtain vibration signal sample set during mechanized equipment R kind health statusWherein,For m-th of health status sample, it is made up of N number of vibration number strong point, its sample label is ym∈{1,2,3,...R};
Step 2:As shown in Fig. 2 determine the input dimension N of orthogonal sparse autoencoder networkinWith output dimension Nout, according to
Input dimension Nin, by health status sample xmJ non-overlapping sample fragment is divided into, forms sample fragment collectionI.e.:
In formula, J=N/Nin, It is j-th of sample fragment, includes NinIndividual data point composition;
Step 3:The sample fragment collection obtained at random from step 2Middle selection NsIndividual sample fragmentAnd
Form fractional sample fragment collectionTo reduce the intersegmental similitude of fractional sample piece, accelerating orthogonal sparse autoencoder network instruction
Experienced convergence rate, to local sample chips section collectionWhitening processing is carried out, obtains the fractional sample fragment collection after albefactionWhereinAs shown in Fig. 2 being assembled for training using the sample fragment practices orthogonal sparse autoencoder network, calculating network
Partial weight matrixMinimize object function:
In formula, σr() is Relu activation primitives;λ is orthogonalization constraint factor;For partial weight matrix
WlocRow k row vector;
Step 4:As shown in Fig. 2 the partial weight matrix W that calculation procedure 3 obtainslocThe sample fragment obtained with step 2
Inner product, sample fragment can be obtainedLocal featureI.e.:
By the J sample fragment split in step 2 respectively with partial weight matrix WlocCarry out inner product calculating, you can obtain
Failure sample xmLocal feature collectionAgain by each local featureArithmetic average, you can the state sample that secures good health xm
Feature
Step 5:Utilize the sample characteristics { f of tape labelm,ymTraining Sotfmax graders, obtain Sotfmax graders
Weight matrixMinimize object function:
In formula,Respectively grader weight matrix WclassR rows, l every trades vector;
By the weight matrix W for calculating graderclassWith sample characteristics fmInner product, output sample characteristics correspond to each sample
The probability distribution of label, the sample label corresponding to maximum probability is taken as sample xmHealth status, complete mechanized equipment be good for
The Intelligent Recognition of health state.
Embodiment:By taking the health status identification of two-stage planetary gear as an example, the feasibility of the present invention is verified.
It is as shown in table 1 to obtain the vibration signal sample of two-stage planetary gear, includes 7 kinds of health status:First order planet
The peeling of wheel bearing needle roller, first order sun gear rippling, first order planetary gear cracks in tooth surface, first order planet wheel bearing inner ring
Failure, second level sun gear hypodontia, second level sun gear peel off, are normal.Vibration signal sample is in 4 kinds of different motor speeds
Gathered under (2100r/min, 2400r/min, 2700r/min and 3000r/min) and 2 kinds of load behaviors (unloaded, heavy duty), except the
A kind of fault sample of health status is included outside 6 kinds of operating modes, and the fault sample of other health status forms by 8 kinds of operating modes.Shake
The sample frequency of dynamic signal is 5120Hz, and each health status contains 335 samples under single operating mode, and each sample contains 1800 numbers
Strong point.
The two-stage planetary gear vibration signal sample of table 1
Health status sample based on above-mentioned two-stage planetary gear, the sample for randomly selecting 25% are used for model training,
Remaining is used for model measurement.In view of each health status sample includes 1800 data points, fault diagnosis mould proposed by the present invention
Shape parameter sets as follows:Input dimension NinWith output dimension NoutIt is 100, orthogonalization constraint factor λ is 0.25.By this hair
The bright health status to two-stage planetary gear is identified, and training precision reaches 100%, and measuring accuracy reaches 99.43%.
Selection traditional intelligence method for diagnosing faults-SVMs and the diagnosis effect of the present invention are contrasted.Artificially carry
The diagnostic characteristic of each health status sample is taken, including:FM0, FM4, energy ratio, Sideband index, Sideband
The conventional Gear Faults Diagnostic feature of 10 kinds of level factor, root-mean-square valve, energy operator, the degree of bias, kurtosis and peak value etc. and 4 kinds
Wavelet energy ratio, then using the training set that this 14 kinds of diagnostic characteristics are formed as input, two-stage planet is identified using SVMs
The health status of gear-box.As shown in table 2, the training precision of this method reaches 97.14%, and measuring accuracy reaches 95.54%, low
In the measuring accuracy of the present invention, the human intervention during illustrating present invention, avoiding intelligent diagnostics, and improve planetary gear
The health status accuracy of identification of case.
Selection stacks sparse autoencoder network and the diagnosis effect of the present invention is contrasted.By stacking sparse own coding net
Network directly extracts the diagnostic characteristic of sample data, then using the diagnostic characteristic of extraction as input, two are identified using Softmax graders
The health status of level epicyclic gearbox.As shown in table 2, because this method can not overcome the shadow of original vibration signal time shift characteristic
Ring, its training precision is 48.63%, measuring accuracy 34.75%, hence it is evident that less than the measuring accuracy of the present invention, illustrates the present invention
With reference to local attachment structure and orthogonal sparse autoencoder network, the time shift characteristic for overcoming original vibration signal is known to health status
The influence of other precision, and the orthogonality of sparse autoencoder network is ensure that, the sample characteristics for promoting model learning to arrive are various, effectively
Improve the health status accuracy of identification of epicyclic gearbox.
2 three kinds of method diagnosis effect contrast tables of table
By contrasting the present invention and traditional intelligence method for diagnosing faults and the diagnosis effect of the sparse autoencoder network of stacking, table
The bright present invention can directly extract diagnostic characteristic from the epicyclic gearbox vibration signal of monitoring, and then identify epicyclic gearbox
Health status, the human intervention during intelligent diagnostics is avoided, effectively increase the accuracy of identification of health status.
Claims (1)
1. the mechanized equipment health status recognition methods of network is locally connected based on orthogonalization, it is characterised in that including following step
Suddenly:
Step 1:Obtain vibration signal sample set during mechanized equipment R kind health statusWherein,For
M-th of health status sample, it is made up of N number of vibration number strong point, its sample label is ym∈{1,2,3,...R};
Step 2:Determine the input dimension N of orthogonal sparse autoencoder networkinWith output dimension Nout, according to input dimension Nin, will be strong
Health state sample xmIt is divided into J non-overlapping sample fragment, and J=N/Nin, form sample fragment collectionWherein,It is j-th of sample fragment, by NinIndividual data point composition;
Step 3:The sample fragment collection obtained at random from step 2Middle selection NsIndividual sample fragmentAnd form
Fractional sample fragment collectionAfter whitening processing, albefaction fractional sample fragment collection is obtainedWherein
Recycle the orthogonal sparse autoencoder network of albefaction fractional sample fragment collection training, the partial weight matrix of calculating networkMinimize object function:
<mrow>
<munder>
<mrow>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
</mrow>
<msub>
<mi>W</mi>
<mrow>
<mi>l</mi>
<mi>o</mi>
<mi>c</mi>
</mrow>
</msub>
</munder>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>s</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>N</mi>
<mi>s</mi>
</msub>
</munderover>
<mo>|</mo>
<mo>|</mo>
<msubsup>
<mi>W</mi>
<mrow>
<mi>l</mi>
<mi>o</mi>
<mi>c</mi>
</mrow>
<mi>T</mi>
</msubsup>
<msub>
<mi>&sigma;</mi>
<mi>r</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>W</mi>
<mrow>
<mi>l</mi>
<mi>o</mi>
<mi>c</mi>
</mrow>
</msub>
<msubsup>
<mi>x</mi>
<mi>w</mi>
<mi>s</mi>
</msubsup>
<mo>)</mo>
</mrow>
<mo>-</mo>
<msub>
<mi>x</mi>
<mi>s</mi>
</msub>
<mo>|</mo>
<msubsup>
<mo>|</mo>
<mn>2</mn>
<mn>2</mn>
</msubsup>
<mo>+</mo>
<mi>&lambda;</mi>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>s</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>N</mi>
<mi>s</mi>
</msub>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>N</mi>
<mrow>
<mi>o</mi>
<mi>u</mi>
<mi>t</mi>
</mrow>
</msub>
</munderover>
<mo>|</mo>
<msub>
<mi>&sigma;</mi>
<mi>r</mi>
</msub>
<mrow>
<mo>(</mo>
<msubsup>
<mi>W</mi>
<mrow>
<mi>l</mi>
<mi>o</mi>
<mi>c</mi>
</mrow>
<mi>k</mi>
</msubsup>
<msubsup>
<mi>x</mi>
<mi>w</mi>
<mi>s</mi>
</msubsup>
<mo>)</mo>
</mrow>
<mo>|</mo>
</mrow>
In formula, σr() is ReLU activation primitives;λ is orthogonalization constraint factor;For partial weight matrix WlocKth
Every trade vector;
Step 4:The partial weight matrix W that calculation procedure 3 obtainslocThe sample fragment obtained with step 2Inner product, obtain sample
This fragmentLocal featureI.e.:
<mrow>
<msubsup>
<mi>f</mi>
<mrow>
<mi>l</mi>
<mi>o</mi>
<mi>c</mi>
</mrow>
<mrow>
<mi>m</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>=</mo>
<msub>
<mi>&sigma;</mi>
<mi>r</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>W</mi>
<mrow>
<mi>l</mi>
<mi>o</mi>
<mi>c</mi>
</mrow>
</msub>
<mo>&CenterDot;</mo>
<msubsup>
<mi>x</mi>
<mi>m</mi>
<mi>j</mi>
</msubsup>
<mo>)</mo>
</mrow>
</mrow>
By each local featureArithmetic average, that is, the state sample that secures good health xmSample characteristics
<mrow>
<msub>
<mi>f</mi>
<mi>m</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>J</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>J</mi>
</munderover>
<msubsup>
<mi>f</mi>
<mrow>
<mi>l</mi>
<mi>o</mi>
<mi>c</mi>
</mrow>
<mrow>
<mi>m</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
</mrow>
Step 5:Utilize the sample characteristics { f of tape labelm,ymTraining Sotfmax graders, the weight of acquisition Sotfmax graders
MatrixMinimize object function:
<mrow>
<mi>J</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>W</mi>
<mrow>
<mi>c</mi>
<mi>l</mi>
<mi>a</mi>
<mi>s</mi>
<mi>s</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mo>-</mo>
<mfrac>
<mn>1</mn>
<mi>M</mi>
</mfrac>
<mo>&lsqb;</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>m</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>r</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>R</mi>
</munderover>
<mn>1</mn>
<mo>{</mo>
<msub>
<mi>y</mi>
<mi>m</mi>
</msub>
<mo>=</mo>
<mi>r</mi>
<mo>}</mo>
<mi>l</mi>
<mi>o</mi>
<mi>g</mi>
<mfrac>
<msup>
<mi>e</mi>
<mrow>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>W</mi>
<mrow>
<mi>c</mi>
<mi>l</mi>
<mi>a</mi>
<mi>s</mi>
<mi>s</mi>
</mrow>
<mi>r</mi>
</msubsup>
<mo>)</mo>
</mrow>
<mi>T</mi>
</msup>
<msub>
<mi>f</mi>
<mi>m</mi>
</msub>
</mrow>
</msup>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>l</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>R</mi>
</munderover>
<msup>
<mi>e</mi>
<mrow>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>W</mi>
<mrow>
<mi>c</mi>
<mi>l</mi>
<mi>a</mi>
<mi>s</mi>
<mi>s</mi>
</mrow>
<mi>l</mi>
</msubsup>
<mo>)</mo>
</mrow>
<mi>T</mi>
</msup>
<msub>
<mi>f</mi>
<mi>m</mi>
</msub>
</mrow>
</msup>
</mrow>
</mfrac>
<mo>&rsqb;</mo>
</mrow>
In formula,Respectively grader weight matrix WclassR rows, l every trades vector;
By the weight matrix W for calculating graderclassWith sample characteristics fmInner product, output sample characteristics correspond to each sample label
Probability distribution, take the sample label corresponding to maximum probability as sample xmHealth status, complete mechanized equipment health shape
The Intelligent Recognition of state.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710784356.2A CN107451624B (en) | 2017-09-04 | 2017-09-04 | Mechanical equipment health state identification method based on orthogonalization local connection network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710784356.2A CN107451624B (en) | 2017-09-04 | 2017-09-04 | Mechanical equipment health state identification method based on orthogonalization local connection network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107451624A true CN107451624A (en) | 2017-12-08 |
CN107451624B CN107451624B (en) | 2020-03-24 |
Family
ID=60494763
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710784356.2A Active CN107451624B (en) | 2017-09-04 | 2017-09-04 | Mechanical equipment health state identification method based on orthogonalization local connection network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107451624B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108225750A (en) * | 2017-12-12 | 2018-06-29 | 南京航空航天大学 | A kind of rotary machinery fault diagnosis method based on the more correlations of fault signature |
CN107451624B (en) * | 2017-09-04 | 2020-03-24 | 西安交通大学 | Mechanical equipment health state identification method based on orthogonalization local connection network |
CN113177484A (en) * | 2021-04-30 | 2021-07-27 | 洛阳师范学院 | Intelligent mechanical fault diagnosis method based on LOF self-encoding |
CN114897154A (en) * | 2022-03-23 | 2022-08-12 | 阿里巴巴(中国)有限公司 | Data prediction method and device |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130242977A1 (en) * | 2012-03-19 | 2013-09-19 | Fujitsu Limited | Apparatus and method for transferring a data signal by bypassing a power-saving segment in a communication network |
CN103995237A (en) * | 2014-05-09 | 2014-08-20 | 南京航空航天大学 | Satellite power supply system online fault diagnosis method |
CN106017876A (en) * | 2016-05-11 | 2016-10-12 | 西安交通大学 | Wheel set bearing fault diagnosis method based on equally-weighted local feature sparse filter network |
CN106441897A (en) * | 2016-10-27 | 2017-02-22 | 石家庄铁道大学 | Morphological component bearing failure diagnosis method based on dictionary study |
CN106555788A (en) * | 2016-11-11 | 2017-04-05 | 河北工业大学 | Application of the deep learning based on Fuzzy Processing in hydraulic equipment fault diagnosis |
CN107024271A (en) * | 2017-03-29 | 2017-08-08 | 兰州理工大学 | Mechanical oscillation signal compression reconfiguration method and system |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107451624B (en) * | 2017-09-04 | 2020-03-24 | 西安交通大学 | Mechanical equipment health state identification method based on orthogonalization local connection network |
-
2017
- 2017-09-04 CN CN201710784356.2A patent/CN107451624B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130242977A1 (en) * | 2012-03-19 | 2013-09-19 | Fujitsu Limited | Apparatus and method for transferring a data signal by bypassing a power-saving segment in a communication network |
CN103995237A (en) * | 2014-05-09 | 2014-08-20 | 南京航空航天大学 | Satellite power supply system online fault diagnosis method |
CN106017876A (en) * | 2016-05-11 | 2016-10-12 | 西安交通大学 | Wheel set bearing fault diagnosis method based on equally-weighted local feature sparse filter network |
CN106441897A (en) * | 2016-10-27 | 2017-02-22 | 石家庄铁道大学 | Morphological component bearing failure diagnosis method based on dictionary study |
CN106555788A (en) * | 2016-11-11 | 2017-04-05 | 河北工业大学 | Application of the deep learning based on Fuzzy Processing in hydraulic equipment fault diagnosis |
CN107024271A (en) * | 2017-03-29 | 2017-08-08 | 兰州理工大学 | Mechanical oscillation signal compression reconfiguration method and system |
Non-Patent Citations (1)
Title |
---|
YAGUO LEI 等: "Reconstruction Independent Component Analysis-Based Methods for Intelligent Fault Diagnosis", 《PROCEEDINGS OF THE 2016 IEEE 20TH INTERNATIONAL CONFERENCE ON COMPUTER SUPPORTED COOPERATIVE WORK IN DESIGN》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107451624B (en) * | 2017-09-04 | 2020-03-24 | 西安交通大学 | Mechanical equipment health state identification method based on orthogonalization local connection network |
CN108225750A (en) * | 2017-12-12 | 2018-06-29 | 南京航空航天大学 | A kind of rotary machinery fault diagnosis method based on the more correlations of fault signature |
CN113177484A (en) * | 2021-04-30 | 2021-07-27 | 洛阳师范学院 | Intelligent mechanical fault diagnosis method based on LOF self-encoding |
CN113177484B (en) * | 2021-04-30 | 2023-02-28 | 洛阳师范学院 | Intelligent mechanical fault diagnosis method based on LOF self-encoding |
CN114897154A (en) * | 2022-03-23 | 2022-08-12 | 阿里巴巴(中国)有限公司 | Data prediction method and device |
Also Published As
Publication number | Publication date |
---|---|
CN107451624B (en) | 2020-03-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107451624A (en) | The mechanized equipment health status recognition methods of network is locally connected based on orthogonalization | |
CN106017876A (en) | Wheel set bearing fault diagnosis method based on equally-weighted local feature sparse filter network | |
CN100485342C (en) | Integrated supporting vector machine mixed intelligent diagnosing method for mechanical fault | |
CN111089720B (en) | Regularization sparse filtering method suitable for gear fault diagnosis under variable rotating speed | |
CN107688856B (en) | Indoor robot scene active identification method based on deep reinforcement learning | |
CN112304614B (en) | Intelligent fault diagnosis method for end-to-end rolling bearing by adopting multi-attention mechanism | |
CN109858352A (en) | A kind of method for diagnosing faults based on compressed sensing and the multiple dimensioned network of improvement | |
CN106168539A (en) | Fault Diagnosis of Roller Bearings based on convolutional neural networks and system | |
CN109489946A (en) | A kind of fault diagnosis method and system of rotating machinery | |
CN112541233A (en) | Rotary machine fault diagnosis method based on improved convolutional neural network | |
CN108827605A (en) | A kind of mechanical breakdown characteristic automatic extraction method based on improvement sparseness filtering | |
CN107491606A (en) | Variable working condition epicyclic gearbox sun gear method for diagnosing faults based on more attribute convolutional neural networks | |
CN107451760B (en) | Rolling bearing fault diagnosis method based on time window slip limited Boltzmann machine | |
CN109946080B (en) | Mechanical equipment health state identification method based on embedded circulation network | |
CN110487547B (en) | Rolling bearing fault diagnosis method under variable working conditions based on vibration diagram and transfer learning | |
CN111444871A (en) | Fault diagnosis method for multi-scale deep convolution neural network planetary gearbox | |
CN109100648A (en) | Ocean current generator impeller based on CNN-ARMA-Softmax winds failure fusion diagnosis method | |
CN112305388B (en) | On-line monitoring and diagnosing method for insulation partial discharge faults of generator stator winding | |
CN113158984A (en) | Bearing fault diagnosis method based on complex Morlet wavelet and lightweight convolution network | |
CN111855202A (en) | Gear box fault diagnosis method and system | |
CN112766200A (en) | Planetary gear box small sample fault diagnosis method based on one-dimensional convolutional neural network | |
CN107016440B (en) | The multiresolution deep neural network intelligent diagnosing method of machine driving failure | |
CN110231165B (en) | Mechanical equipment fault diagnosis method based on expectation difference constraint confidence network | |
CN116821762A (en) | Mechanical fault diagnosis method based on multi-scale graph attention fusion network | |
CN111523557A (en) | Wind power intelligent fault diagnosis method based on big data |
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 |