CN109655259A - Combined failure diagnostic method and device based on depth decoupling convolutional neural networks - Google Patents
Combined failure diagnostic method and device based on depth decoupling convolutional neural networks Download PDFInfo
- Publication number
- CN109655259A CN109655259A CN201811406372.9A CN201811406372A CN109655259A CN 109655259 A CN109655259 A CN 109655259A CN 201811406372 A CN201811406372 A CN 201811406372A CN 109655259 A CN109655259 A CN 109655259A
- Authority
- CN
- China
- Prior art keywords
- decoupling
- depth
- failure
- neural networks
- convolutional neural
- 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
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 50
- 238000002405 diagnostic procedure Methods 0.000 title claims abstract description 15
- 238000000034 method Methods 0.000 claims abstract description 41
- 238000012549 training Methods 0.000 claims abstract description 30
- 238000003745 diagnosis Methods 0.000 claims abstract description 16
- 230000001133 acceleration Effects 0.000 claims abstract description 11
- 238000012360 testing method Methods 0.000 claims abstract description 11
- 238000002474 experimental method Methods 0.000 claims abstract description 10
- 239000000284 extract Substances 0.000 claims abstract description 4
- 210000002569 neuron Anatomy 0.000 claims description 32
- 230000006870 function Effects 0.000 claims description 25
- 239000011159 matrix material Substances 0.000 claims description 13
- 230000002195 synergetic effect Effects 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 6
- 239000000203 mixture Substances 0.000 claims description 6
- 235000013399 edible fruits Nutrition 0.000 claims description 5
- 230000008878 coupling Effects 0.000 claims description 4
- 238000010168 coupling process Methods 0.000 claims description 4
- 238000005859 coupling reaction Methods 0.000 claims description 4
- 238000003860 storage Methods 0.000 claims description 4
- 235000009854 Cucurbita moschata Nutrition 0.000 claims description 3
- 240000001980 Cucurbita pepo Species 0.000 claims description 3
- 235000009852 Cucurbita pepo Nutrition 0.000 claims description 3
- 150000001875 compounds Chemical class 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 230000005484 gravity Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 235000020354 squash Nutrition 0.000 claims description 3
- 230000008859 change Effects 0.000 claims 1
- 230000005540 biological transmission Effects 0.000 description 4
- 238000013461 design Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000009429 distress Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000001965 increasing effect Effects 0.000 description 3
- 210000005036 nerve Anatomy 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000003475 lamination Methods 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 210000004205 output neuron Anatomy 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000007634 remodeling Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000002834 transmittance Methods 0.000 description 1
Classifications
-
- 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
- G01M13/02—Gearings; Transmission mechanisms
- G01M13/021—Gearings
-
- 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
- G01M13/02—Gearings; Transmission mechanisms
- G01M13/028—Acoustic or vibration analysis
-
- 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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses it is a kind of based on depth decoupling convolutional neural networks combined failure diagnostic method and device wherein method comprising steps of step 1: the acquisition and calibration of data set, vibration acceleration signal of the acquisition under rotating machinery is single and combined failure experiment condition obtains several samples and separately constitutes training set and test set respectively, and class label is set separately；Step 2: one-dimensional depth convolutional neural networks model is built；Step 3: building has the decoupling classifier of multi-tag output characteristics under keras frame；Step 4: the training of network model finally obtains optimal depth decoupling convolutional neural networks model；Step 5: the intelligent diagnostics of combined failure and classification, output obtain the real-time diagnosis result of combined failure.The present invention extracts each single failure feature in combined failure signal under the premise of only using single failure signal to train depth Decoupling network model, and the decoupling and classification of combined failure are realized by decoupling classifier.
Description
Technical field
The invention belongs to machinery manufacturing technology fields, are related to a kind of technology for mechanical fault diagnosis, and in particular to one kind is based on
The combined failure diagnostic method of depth decoupling convolutional neural networks.
Background technique
Bearing and gear rotary member are a kind of essential spare parts in common use in mechanical equipment, in modern industrial equipment
In play important role.Fault diagnosis to rotating machinery is the important component of preventative maintenance system, to extension
The service life of mechanical equipment, the safety for reducing maintenance cost and increasing equipment operation are of great significance.
Method for diagnosing faults based on artificial intelligence has been widely used in the fault diagnosis of rotating machinery and acquirement
Preferable effect.General rotary machinery fault diagnosis can be completed by the acquisition, feature extraction, classification of vibration signal, and
Assorting process can be realized by machine learning algorithms such as reverse transmittance nerve network, support vector machines.Convolutional neural networks are with it
Efficient ability in feature extraction and the translation invariance of input is received significant attention, is especially obtained in fields such as image recognitions
Application rapidly.In recent years, scholars begin one's study traditional convolutional neural networks one-dimensional, and apply to one-dimensional sequence data
Classification in.
It is big at present since the coupled mode of a variety of single failures is excessively complicated for the combined failure of rotating machinery
Most intelligent diagnosing methods single-row identify its for a kind of fault mode.For a deep learning diagnostic model, if
It will be for possible components combined failure modelings a variety of in some system, it will so that model complexity improves, model parameter
Quantity is significantly increased, and also results in the problems such as certain combined failure data volumes are insufficient.Currently, the decoupling of mechanical combined failure
Analysis is still the challenge that intelligent diagnostics face, it is also of great significance to the intelligent maintenance of mechanical equipment.
Thus existing technology is not met by the demand of practical application in industry, it is also necessary to improve, develop one kind
The intelligent diagnostics algorithm of combined failure become there is an urgent need to.
Summary of the invention
In order to solve the problems of prior art, the present invention designs a kind of based on depth decoupling convolutional neural networks
Combined failure diagnostic method and device are deeply excavated each in combined failure by the feature learning ability of depth convolutional neural networks
The single failure feature of components, and the characteristic of decoupling classifier multi-tag output is combined, so that constructed depth decouples net
Network model also can be realized the decoupling and classification to combined failure in the case where only single failure signal makees training set.It is logical
The present invention is crossed, accuracy, intelligence and the use of fault diagnosis system being monitored to equipment state with prediction can be improved
Convenience.
The present invention is realized using following technical scheme:
A kind of combined failure diagnostic method based on depth decoupling convolutional neural networks, comprising steps of
Step 1: the acquisition and calibration of data set are acquired respectively under rotating machinery is single and combined failure experiment condition
Vibration acceleration signal is intercepted to obtain several samples, using the method for overlap sampling come to data by certain sample point length
The sample of collection is enhanced, and sample length is 8 periods, Duplication 0.5；Single failure sample is formed into training set, and
All combined failure samples form test set, and class label is set separately；
Step 2: building depth distinguishability Feature Selection Model is stacked by multilayer sequence and is rolled up under keras frame
Lamination and pond layer build one-dimensional depth convolutional neural networks model；
Step 3: building decoupling classifier is input with the feature that step 2 is extracted, passes through stacking under keras frame
Multiple decouplings classification layers, construct the decoupling classifier with multi-tag output characteristics；
Step 4: the training of network model is damaged using the cross entropy that marginal loss function replaces existing intelligent diagnostics algorithm
Function is lost, network model is trained under the premise of only using single failure data in combination with routing synergetic,
And corresponding hyper parameter is chosen using Searching algorithm, finally obtain optimal depth decoupling convolutional neural networks model；
Step 5: the intelligent diagnostics of combined failure and classification have trained the input of collected combined failure signal
Depth decoupling convolutional neural networks model in tested, the output by decoupling classifier obtains the real-time of combined failure and examines
Disconnected result.
The used marginal loss function of this programme effectively can increase between class distance as new objective function and reduce
Inter- object distance, so that the depth characteristic extracted has more distinguishability, this will make the neural network for training multi-tag to export
Become to be more easier；Meanwhile in the training process, single failure data are only used only, network model is trained, is not depended on
In combined failure data and its markup information, so that it may realize the decoupling and diagnosis to combined failure, and use web search
Algorithm chooses corresponding hyper parameter, avoids the need for the process of artificial parameter tuning, may be implemented to automatically obtain optimal network
Model.Under conditions of device hardware is supported, on-line monitoring and diagnostic system can be applied to, complex rotation mechanical equipment is carried out
Real-time condition monitoring and fault diagnosis.
Further, in step 1, the sample is directly intercepted by vibration acceleration signal and is got, and is dividing sample set
When, only include single failure sample in training set, then simultaneously includes single failure sample and combined failure sample in test set, with
This verifies the powerful decoupling ability of mentioned network.
Further,
When intercepted samples direct by vibration acceleration signal, carried out using the method for overlap sampling come the sample to data set
Enhancing, sample length are 8 periods, Duplication 0.5.
Further, pass through multiple convolutional layers and pond based on one-dimensional depth convolutional neural networks in the step 2
The sequence for changing layer stacks building depth distinguishability Feature Selection Model, and structure mainly includes input layer, convolutional layer, Chi Hua
Layer；The depth distinguishability Feature Selection Model is obtained by carrying out layer-by-layer feature extraction to signal with distinguishability
Depth characteristic.
Further, the step 3 constructs decoupling in such a way that vector neuron replaces traditional classifier scalar neuron
Classifier；The decoupling classifier is stacked by multiple decouplings classification layer heap, two neighboring to optimize by routing synergetic
Mapping and coupled relation, construction step between decoupling classification layer specifically may be expressed as:
(1) the multichannel one-dimensional characteristic of the most after-bay layer output obtained by step 2 passes through dimension transformation composition one
Matrix, the matrix are the input of initial decoupling classification layer, are usedIt indicates, wherein KlTable
The number for showing vector neuron, the convolution kernel number being also equal to before most after-bay layer, each vector neuron store a kind of spy
Reference breath, RlIndicate the dimension of each initial vector neuron；
(2) number for assuming the vector neuron of second decoupling classification layer is C, i.e. the classification number of single failure, is usedIndicate its output, then entire calculating process are as follows:
Wherein: i=1,2 ..., Kl, j=1,2 ..., C,For weight matrix,It is intermediate special
Sign, cijFor the coefficient of coup, squash function is the activation primitive of decoupling classification layer；
(3) coefficient of coup c in step (2)ijUpdate is iterated by routing synergetic, gives initial value bij, meter
Calculating formula is
Size by calculating the routing collaboration between adjacent layer vector neuron updates adjacent layer vector nerve to calculate
Coefficient of coup c between memberij, it is represented by
cij=softmax (bij)
(4) output v that the last one decoupling classification layer can be obtained with update iteration is successively calculated, by asking the L2 norm of v can
The length of each output vector neuron is obtained, this length represents probability existing for respective classes；
(5) finally by the next threshold value of settingIf the length of corresponding vector neuron obtained in step (4) is more than this
The classification of threshold value then exports corresponding single failure label, when detecting to combined failure, because the present invention can detect simultaneously
Combined failure is decoupled, because the label of multiple single failures of decoupling classifier output composition combined failure, realizes combined failure
Decoupling and identification.
Further, in step (3), initial value b is givenijWhen, zero initialization is selected, first to guarantee upper one layer of vector nerve
The probability that member is directed toward next layer of vector neuron is all identical.
Further, in the step 4, when carrying out network model training, used objective function is marginal loss letter
Number expression are as follows:
Wherein, M and C respectively indicates number of samples and classification number；Tc is the indicative function of classification, for m-th of sample
This, if its real conditions includes c class failureOtherwise Indicate the corresponding c of m-th of sample
The activation value of a output neuron；P+ expression is worked asWhen i.e. m-th of sample really includes c class failure, when correctly predictedLower limit value；P- expression is worked asThat is m-th of sample is practical when not including c class failure, when correctly predictedIt is upper
Limit value takes 0.1 herein；λ is used to adjust the specific gravity of two phase plus items.
A kind of combined failure diagnostic device based on depth decoupling convolutional neural networks, comprising:
The acquisition of data set and demarcating module, for being acquired under rotating machinery is single and combined failure experiment condition respectively
Vibration acceleration signal is intercepted to obtain several samples by certain sample point length, and single failure sample is formed training set, with
And all combined failure samples form test set, and class label is set separately；
Depth distinguishability Feature Selection Model constructs module, under keras frame, stacks convolutional layer by multilayer sequence
With pond layer, one-dimensional depth convolutional neural networks model is built；
It decouples classifier and constructs module, under keras frame, with the feature of depth distinguishability Feature Selection Model extraction
The decoupling classifier with multi-tag output characteristics is constructed by stacking multiple decouplings classification layers for input；
The training module of network model, for replacing the cross entropy damage of existing intelligent diagnostics algorithm using marginal loss function
Function is lost, network model is trained under the premise of only using single failure data in combination with routing synergetic,
And corresponding hyper parameter is chosen using Searching algorithm, finally obtain optimal depth decoupling convolutional neural networks model；
Collected combined failure signal is inputted trained depth by the intelligent diagnostics and categorization module of combined failure
It is tested in degree decoupling convolutional neural networks model, the output by decoupling classifier obtains the real-time diagnosis knot of combined failure
Fruit.
A kind of electronic equipment including memory, processor, stores the calculating that can be run on a memory and on a processor
Machine program when the processor runs described program, realizes the method.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
The method is realized when row.
Compared with prior art, the beneficial effects of the present invention are:
1, the present invention can distinguish Feature Selection Model comprising depth, have powerful feature learning ability, can be from original
Deeply excavate the single failure feature of each components in combined failure in vibration signal, these features not only have excellent similar
Similitude and foreign peoples's otherness, and more structured messages are remained than the feature of traditional scalar form, it can be preferably
Decoupling of the network to combined failure is helped, to improve the accuracy of identification of network.
2, the present invention has decoupling classifier structure, has the characteristic for exporting multiple labels, so that output knot of the invention
Fruit can more intuitively reflect rotating machinery fault state.By the decoupling and classification to combined failure, it is more advantageous to target
Equipment fault carries out quickly positioning and maintenance, reduces the production line stop time, greatly improves the production effect of target device
Rate reduces production cost, and can avoid major accident caused by occurring due to combined failure to a certain extent, brings
Casualties and economic loss.
3, the depth Decoupling network model constructed by the present invention, fully considers the actual conditions of industrial application, it is intended to only
In the case that normal and single failure signal makees training set, to extract the depth characteristic with distinguishability, and realization pair
The decoupling and classification of combined failure.It is completely independent of the priori knowledges such as combined failure data and its markup information, thus nothing
The combined failure that need to carry out high cost is tested to acquire and mark a variety of combined failure data, and economy is largely saved
Pay wages and liberated experiment labour.
4, present invention employs Searching algorithms to choose corresponding hyper parameter, avoid the process of artificial parameter tuning,
It may be implemented to automatically obtain optimal network model.
Detailed description of the invention
Fig. 1 is the algorithm flow chart of the method for the present invention；
Fig. 2 is the schematic network structure of the method for the present invention；
Fig. 3 is the schematic illustration of the decoupling classifier of inventive network；
Fig. 4 is sensor position view on Five block transmission in example；
Fig. 5 is the recognition result confusion matrix of traditional network in instances；
Fig. 6 is the recognition result confusion matrix of the one-dimensional convolutional neural networks of the present invention in instances；
Fig. 7 is the reality output label form of inventive network in instances.
Specific embodiment
The present invention is described in detail with specific implementation step with reference to the accompanying drawing, but embodiments of the present invention are unlimited
In this.
Embodiment
As shown in Figure 1, a kind of combined failure diagnostic method based on depth decoupling convolutional neural networks, the method pass through
The feature learning ability of depth convolutional neural networks deeply excavates the single failure feature of each components in combined failure, and combines
The characteristic for decoupling the output of classifier multi-tag, so that constructed depth Decoupling network model is instructed in only single failure signal
In the case where practicing collection, the decoupling and classification to combined failure also can be realized, depth decouples the structure of one-dimensional convolutional neural networks
Schematic diagram is as shown in Fig. 2, the method comprising the steps of:
Step 1: the acquisition and calibration of data set, design rotating machinery is single and combined failure is tested, and acquires these respectively
Vibration acceleration signal under operating condition is intercepted to obtain great amount of samples, single failure sample is formed by certain sample point length
Training set and all combined failure samples form test set, and class label is set separately；
Step 2: building depth distinguishability Feature Selection Model is stacked by multilayer sequence and is rolled up under keras frame
Lamination and pond layer build one-dimensional depth convolutional neural networks model；
Step 3: design decoupling classifier is input with the feature that step 2 is extracted, passes through stacking under keras frame
Multiple decouplings classification layers, build the decoupling classifier with multi-tag output characteristics；
Step 4: the training of network model, the present invention replace traditional intelligence diagnosis algorithm using marginal loss function
Cross entropy loss function, in combination with routing synergetic, under the premise of only using single failure data, to network model into
Row training, and corresponding hyper parameter is chosen using Searching algorithm, finally obtain optimal network model；
Step 5: the intelligent diagnostics of combined failure and classification have trained the input of collected combined failure signal
Depth decoupling convolutional neural networks model in tested, the output by decoupling classifier obtains the real-time of combined failure and examines
Disconnected result.
Further, step 3 specifically: step 3 replaces the side of traditional classifier scalar neuron with vector neuron
Formula, to construct decoupling classifier；Decoupling classifier is stacked by multiple decouplings classification layer heap, is optimized by routing synergetic
Mapping and coupled relation, construction step between two neighboring decoupling classification layer specifically may be expressed as:
(1) the multichannel one-dimensional characteristic of the most after-bay layer output obtained by step 2 passes through dimension transformation composition one
Matrix, this matrix are the input of initial decoupling classification layer, are usedIt indicates, wherein KlTable
The number for showing vector neuron, the convolution kernel number being also equal to before most after-bay layer, each vector neuron store a kind of spy
Reference breath, RlIndicate the dimension of each initial vector neuron；
(2) number for assuming the vector neuron of second decoupling classification layer is C (i.e. the classification number of single failure), defeated
It can be used outIt indicates, then entire calculating process are as follows:
Wherein: i=1,2 ..., Kl, j=1,2 ..., C,For weight matrix,It is intermediate special
Sign, cijFor the coefficient of coup, squash function is the activation primitive of decoupling classification layer；
(3) coefficient of coup c in step (2)ijIt can be iterated update by routing synergetic, give initial value bijIt is (logical
Zero initialization is often first selected, all identical with the probability for guaranteeing that upper one layer of vector neuron is directed toward next layer of vector neuron), meter
Calculating formula is
Size by calculating the routing collaboration between adjacent layer vector neuron (is defined as)
To calculate the coefficient of coup c updated between adjacent layer vector neuronij, it is represented by
cij=softmax (bij)
(4) output v that the last one decoupling classification layer can be obtained with update iteration is successively calculated, by asking the L2 norm of v can
The length of each output vector neuron is obtained, this length represents probability existing for respective classes；
(5) finally by the next threshold value of settingThe length of corresponding vector neuron obtained in step (4) is more than this threshold
The classification of value then exports corresponding single failure label, when detecting to combined failure, because the present invention can be detected and be solved
Coupling combined failure, because decoupling classifier can export the label of multiple single failures of composition combined failure, to reach compound event
The decoupling and identification of barrier.Classifier structural schematic diagram is decoupled as shown in figure 3, the process of entire algorithm is as shown in table 1.
Table 1.Dynamic Routing algorithm flow
Further, when step 4 carries out the training of depth network model, used objective function is marginal loss function,
Its expression is
Wherein M and C respectively indicates number of samples and classification number；Tc is the indicative function of classification, for m-th of sample
This, if its real conditions includes c class failureOtherwise Indicate the corresponding c of m-th of sample
The activation value of a output neuron；P+ expression is worked asIt is correctly predicted when (i.e. m-th of sample really includes c class failure)
WhenLower limit value；P- expression is worked asWhen (i.e. m-th of sample is practical not to include c class failure), when correctly predicted
Upper limit value, take 0.1 herein；λ is used to adjust the specific gravity of two phase plus items.
The depth network model training method can effectively be increased using marginal loss function as new objective function
Big between class distance and reduce inter- object distance so that the depth characteristic extracted has more distinguishability, this will be so that the more marks of training
The neural network of label output becomes to be more easier；Meanwhile in the training process, single failure data are only used only, to network mould
Type is trained, independent of combined failure data and its markup information, so that it may realize the decoupling and diagnosis to combined failure, and
And Searching algorithm is used to choose corresponding hyper parameter, the process of artificial parameter tuning is avoided the need for, may be implemented certainly
It is dynamic to obtain optimal network model.
Combined failure intelligent diagnosing method based on depth decoupling convolutional neural networks described in step 5 is, it can be achieved that compound
The intelligent diagnostics of failure and classification, by the input of collected combined failure signal, trained depth decouples convolutional Neural net
It is tested in network model, the output by decoupling classifier obtains the real-time diagnosis of combined failure as a result, this algorithm exists simultaneously
Under conditions of device hardware is supported, on-line monitoring and diagnostic system can be applied to, complex rotation mechanical equipment is carried out real-time
Condition monitoring and fault diagnosis.
Experiment case study:
Bearing and gear are two critical components of gear-box, are the important components of rotating machinery.In order to verify this
The validity for inventing proposed method devises the identification reality of bearing and gear distress using certain automobile Five block transmission as research object
It tests.
1, experimental data
The Five block transmission schematic diagram studied is as shown in figure 4, five gear gears engagement of setting, output shaft revolving speed are when testing
1100r/m, load torque 50Nm.By the accelerometer being mounted on automotive transmission output shaft bearing seat, with 24kHz's
Sample frequency acquires original vibration signal.Sample length is set as 8192 points, Duplication is 0.5 progress sample division.It is related
The details of gear-box data set are listed in Table 2 below.
2. gear-box data set information of table
2, the determination of network parameter
By the original vibration data in above-mentioned training set be input to depth decoupling convolutional neural networks in, network first therefrom from
Dynamic study and extract further feature, these features subsequently input decoupling classifier, calculate cost function and guide network weight with
The adjustment of biasing.
The hyper parameter of network is to carry out exhaustion by the Grid Search method in Scikit-Learn frame to select.Most
Determining network includes 2 groups of convolution-pond layer, 1 remodeling layer, 2 decouplings classification layers eventually.The detail parameters information of network is shown in Table
3。
The network optimized parameter that table 3. obtains
Experiment realizes that maximum number of iterations is set as 20, and data batch are sized to 64, training using the tool box Keras
Optimizer selects Adam Optimizer.
3, experimental result and analysis
In this experiment, by one-dimensional convolutional neural networks model and traditional one-dimensional convolutional Neural net based on decoupling classifier
Network model compares.In order to evaluate the superiority of the proposed method of the present invention, two methods are evaluated using confusion matrix
Classification accuracy and misclassification rate.What Fig. 5 and Fig. 6 reflected conventional model and the mentioned method recognition result of the present invention obscures square
Battle array.From confusion matrix as can be seen that all combined failures have all mistakenly been divided into normally by traditional one-dimensional convolutional neural networks
Class, and the mentioned method of the present invention achieves preferable classifying quality, it was demonstrated that the validity and robustness of mentioned method.In addition,
Fig. 6 is also shown that mentioned method can accurately export the label of single or multiple samples by decoupling classifier, meanwhile, it is not
Single fault condition can be identified as to the combining case being made of normal condition and this failure.
Mentioned method provides proposed method and carries out failure to the decoupling ability of combined failure in order to further illustrate the present invention
The physical tags dished out during identification, as shown in Figure 7.It can be seen from the figure that the mentioned method energy of the present invention will be by bearing
The full decoupled combined failure that failure and gear distress couple is two single failures.
In conclusion the invention discloses a kind of combined failure diagnostic method based on depth decoupling convolutional neural networks,
Firstly, utilizing the feature of depth decoupling convolutional neural networks study and extraction vibration original signal；Secondly, utilizing decoupling classifier
Decoupling identification is carried out to combined failure；Finally, using normal condition, bearing fault, gear distress and by two failure couplings is contained
The gear-box data set of the combined failure of conjunction verifies this method.It is demonstrated experimentally that coming only by single failure signal
Training depth Decoupled Model, the present invention can extract each single failure feature in combined failure signal, and real by decoupling classifier
The decoupling and classification of existing combined failure.
In order to realize above-described embodiment, the embodiment of the invention also provides a kind of based on depth decoupling convolutional neural networks
Combined failure diagnostic device, comprising:
The acquisition of data set and demarcating module, for being acquired under rotating machinery is single and combined failure experiment condition respectively
Vibration acceleration signal is intercepted to obtain several samples by certain sample point length, and single failure sample is formed training set, with
And all combined failure samples form test set, and class label is set separately；
Depth distinguishability Feature Selection Model constructs module, under keras frame, stacks convolutional layer by multilayer sequence
With pond layer, one-dimensional depth convolutional neural networks model is built；
It decouples classifier and constructs module, under keras frame, with the feature of depth distinguishability Feature Selection Model extraction
The decoupling classifier with multi-tag output characteristics is constructed by stacking multiple decouplings classification layers for input；
The training module of network model, in combination with routing synergetic, is being only used for using marginal loss function
Under the premise of single failure data, network model is trained, and chooses corresponding hyper parameter using Searching algorithm, most
Optimal depth is obtained eventually decouples convolutional neural networks model；
Collected combined failure signal is inputted trained depth by the intelligent diagnostics and categorization module of combined failure
It is tested in degree decoupling convolutional neural networks model, the output by decoupling classifier obtains the real-time diagnosis knot of combined failure
Fruit.
In order to realize above-described embodiment, the embodiment of the invention also provides a kind of electronic equipment, including memory, processing
Device stores the computer program that can be run on a memory and on a processor, when the processor runs described program, realizes
Method described in previous embodiment.
In order to realize above-described embodiment, the embodiment of the invention also provides a kind of computer readable storage mediums, deposit thereon
Computer program is contained, method described in previous embodiment is realized when the computer program is executed by processor.
Finally, it should be noted that above embodiment is only described to made by a preferred embodiment of the invention, and
The non-restriction that the scope of the present invention is carried out, without departing from the spirit of the design of the present invention, those skilled in the art couple
Various equivalent variations, modification and the improvement that technical solution of the present invention is made, should all be included in scope of the present invention patent
It is interior.
Claims (10)
1. a kind of combined failure diagnostic method based on depth decoupling convolutional neural networks, which is characterized in that comprising steps of
Step 1: the acquisition and calibration of data set acquire the vibration under rotating machinery is single and combined failure experiment condition respectively
Acceleration signal is intercepted to obtain several samples by certain sample point length, and single failure sample is formed training set, Yi Jisuo
Test set is formed by combined failure sample, and class label is set separately；
Step 2: building depth distinguishability Feature Selection Model stacks convolutional layer by multilayer sequence under keras frame
With pond layer, one-dimensional depth convolutional neural networks model is built；
Step 3: building decoupling classifier is input with the feature that step 2 is extracted under keras frame, multiple by stacking
Decoupling classification layer, constructs the decoupling classifier with multi-tag output characteristics；
Step 4: the training of network model, it is single only using in combination with routing synergetic using marginal loss function
Under the premise of fault data, network model is trained, and chooses corresponding hyper parameter using Searching algorithm, is finally obtained
It obtains optimal depth and decouples convolutional neural networks model；
Step 5: collected combined failure signal is inputted trained depth by the intelligent diagnostics of combined failure and classification
It is tested in degree decoupling convolutional neural networks model, the output by decoupling classifier obtains the real-time diagnosis knot of combined failure
Fruit.
2. the combined failure diagnostic method according to claim 1 based on depth decoupling convolutional neural networks, feature exist
In in step 1, the sample is directly intercepted by vibration acceleration signal and got, and when dividing sample set, is only wrapped in training set
Sample containing single failure then includes single failure sample and combined failure sample in test set simultaneously.
3. the combined failure diagnostic method according to claim 2 based on depth decoupling convolutional neural networks, feature exist
When, intercepted samples direct by vibration acceleration signal, the sample of data set is enhanced using the method for overlap sampling,
Sample length is 8 periods, Duplication 0.5.
4. the combined failure diagnostic method according to claim 1 based on depth decoupling convolutional neural networks, feature exist
In, in the step 2 based on one-dimensional depth convolutional neural networks, pass through the sequence of multiple convolutional layers and pond layer stack
Depth distinguishability Feature Selection Model is constructed, structure mainly includes input layer, convolutional layer, pond layer；The depth is distinguishable
Other property Feature Selection Model obtains the depth characteristic with distinguishability by carrying out layer-by-layer feature extraction to signal.
5. the combined failure diagnostic method according to claim 1 based on depth decoupling convolutional neural networks, feature exist
In the step 3 constructs decoupling classifier in such a way that vector neuron replaces traditional classifier scalar neuron；The solution
Coupling classifier is stacked by multiple decouplings classification layer heap, is optimized between two neighboring decoupling classification layer by routing synergetic
Mapping and coupled relation, construction step specifically may be expressed as:
(1) the multichannel one-dimensional characteristic of the most after-bay layer output obtained by step 2 converts one matrix of composition by dimension,
The matrix is the input of initial decoupling classification layer, is usedIt indicates, wherein KlIndicate vector
The number of neuron, the convolution kernel number being also equal to before most after-bay layer, each vector neuron store a kind of characteristic information,
RlIndicate the dimension of each initial vector neuron；
(2) number for assuming the vector neuron of second decoupling classification layer is C, i.e. the classification number of single failure, is usedIndicate its output, then entire calculating process are as follows:
Wherein: i=1,2 ..., Kl, j=1,2 ..., C,For weight matrix,For intermediate features,
cijFor the coefficient of coup, squash function is the activation primitive of decoupling classification layer；
(3) coefficient of coup c in step (2)ijUpdate is iterated by routing synergetic, gives initial value bij, calculate public
Formula is
By calculate adjacent layer vector neuron between routing collaboration size come calculate update adjacent layer vector neuron it
Between coefficient of coup cij, it is represented by
cij=softmax (bij)；
(4) output v that the last one decoupling classification layer can be obtained with update iteration is successively calculated, by asking the L2 norm of v that can obtain often
The length of one output vector neuron, this length represent probability existing for respective classes；
(5) finally by the next threshold value of settingIf the length of corresponding vector neuron obtained in step (4) is more than this threshold value
Classification then export corresponding single failure label, when being detected to combined failure, decoupling classifier output composition it is compound
The label of multiple single failures of failure realizes the decoupling and identification of combined failure.
6. the combined failure diagnostic method according to claim 5 based on depth decoupling convolutional neural networks, feature exist
In, in step (3), given initial value bijWhen, first select zero initialization.
7. the combined failure diagnostic method according to claim 1 based on depth decoupling convolutional neural networks, feature exist
In in the step 4, when carrying out network model training, used objective function is marginal loss function expression
Are as follows:
Wherein, M and C respectively indicates number of samples and classification number；Tc is the indicative function of classification, for m-th of sample, such as
Its real conditions of fruit include c class failure thenOtherwise Indicate corresponding c-th of the output of m-th of sample
The activation value of neuron；P+ expression is worked asWhen i.e. m-th of sample really includes c class failure, when correctly predicted's
Lower limit value；P- expression is worked asThat is m-th of sample is practical when not including c class failure, when correctly predictedUpper limit value,
0.1 is taken herein；λ is used to adjust the specific gravity of two phase plus items.
8. a kind of combined failure diagnostic device based on depth decoupling convolutional neural networks characterized by comprising
The acquisition of data set and demarcating module, for acquiring the vibration under rotating machinery is single and combined failure experiment condition respectively
Acceleration signal is intercepted to obtain several samples by certain sample point length, and single failure sample is formed training set, Yi Jisuo
Test set is formed by combined failure sample, and class label is set separately；
Depth distinguishability Feature Selection Model constructs module, under keras frame, stacks convolutional layer and pond by multilayer sequence
Change layer, builds one-dimensional depth convolutional neural networks model；
It decouples classifier and constructs module, be defeated with the feature that depth distinguishability Feature Selection Model extracts under keras frame
Enter, by stacking multiple decouplings classification layers, constructs the decoupling classifier with multi-tag output characteristics；
The training module of network model is single only using in combination with routing synergetic for using marginal loss function
Under the premise of fault data, network model is trained, and chooses corresponding hyper parameter using Searching algorithm, is finally obtained
It obtains optimal depth and decouples convolutional neural networks model；
Collected combined failure signal is inputted trained depth solution by the intelligent diagnostics and categorization module of combined failure
It is tested in coupling convolutional neural networks model, the output by decoupling classifier obtains the real-time diagnosis result of combined failure.
9. a kind of electronic equipment, it is characterised in that: on a memory and can be on a processor including memory, processor, storage
The computer program of operation when the processor runs described program, realizes the side as described in any one of claims 1 to 7
Method.
10. a kind of computer readable storage medium, is stored thereon with computer program, it is characterised in that: the computer program
The method as described in any one of claims 1 to 7 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811406372.9A CN109655259B (en) | 2018-11-23 | 2018-11-23 | Compound fault diagnosis method and device based on deep decoupling convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811406372.9A CN109655259B (en) | 2018-11-23 | 2018-11-23 | Compound fault diagnosis method and device based on deep decoupling convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109655259A true CN109655259A (en) | 2019-04-19 |
| CN109655259B CN109655259B (en) | 2021-02-19 |
Family
ID=66111512
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811406372.9A Active CN109655259B (en) | 2018-11-23 | 2018-11-23 | Compound fault diagnosis method and device based on deep decoupling convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109655259B (en) |
Cited By (37)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110361176A (en) * | 2019-06-05 | 2019-10-22 | 华南理工大学 | A kind of intelligent failure diagnosis method for sharing neural network based on multitask feature |
| CN110398369A (en) * | 2019-08-15 | 2019-11-01 | 贵州大学 | A kind of Fault Diagnosis of Roller Bearings merged based on 1-DCNN and LSTM |
| CN110443488A (en) * | 2019-07-31 | 2019-11-12 | 中国石油大学(华东) | The recognition methods of drilling well spill hazard, system and equipment based on convolutional neural networks |
| CN110443174A (en) * | 2019-07-26 | 2019-11-12 | 浙江大学 | A kind of pedestrian's recognition methods again based on decoupling self-adaptive identification feature learning |
| CN110619386A (en) * | 2019-09-09 | 2019-12-27 | 国家电网有限公司 | TMR operation monitoring and fault intelligent research and judgment method and system |
| CN110647923A (en) * | 2019-09-04 | 2020-01-03 | 西安交通大学 | Variable working condition mechanical fault intelligent diagnosis method based on self-learning under small sample |
| CN110647830A (en) * | 2019-09-12 | 2020-01-03 | 华中科技大学 | Bearing fault diagnosis method based on convolutional neural network and Gaussian mixture model |
| CN110749447A (en) * | 2019-11-27 | 2020-02-04 | 淮安信息职业技术学院 | Loader engine fault diagnosis method using big data |
| CN111353587A (en) * | 2020-03-10 | 2020-06-30 | 中国科学院计算技术研究所厦门数据智能研究院 | Interpretable generation method of deep neural network |
| CN111505424A (en) * | 2020-05-06 | 2020-08-07 | 哈尔滨工业大学 | Large experimental device power equipment fault diagnosis method based on deep convolutional neural network |
| CN111860775A (en) * | 2020-07-03 | 2020-10-30 | 南京航空航天大学 | Ship fault real-time diagnosis method based on CNN and RNN fusion |
| CN111899905A (en) * | 2020-08-05 | 2020-11-06 | 哈尔滨工程大学 | Fault diagnosis method and system based on nuclear power device |
| CN112067916A (en) * | 2019-09-20 | 2020-12-11 | 武汉理工大学 | Time series data intelligent fault diagnosis method based on deep learning |
| CN112132102A (en) * | 2020-09-28 | 2020-12-25 | 天津工业大学 | Intelligent fault diagnosis method combining deep neural network with artificial bee colony optimization |
| CN112163630A (en) * | 2020-10-12 | 2021-01-01 | 清华大学 | Compound fault diagnosis method and device based on unbalanced learning |
| CN112284735A (en) * | 2020-10-21 | 2021-01-29 | 兰州理工大学 | Multi-sensor rolling bearing fault diagnosis based on one-dimensional convolution and dynamic routing |
| CN112329650A (en) * | 2020-11-09 | 2021-02-05 | 中国兵器工业集团第二一四研究所苏州研发中心 | Accelerometer fault diagnosis method based on convolutional neural network |
| CN112326210A (en) * | 2019-07-17 | 2021-02-05 | 华北电力大学（保定） | Large motor fault diagnosis method combining sound vibration signals with 1D-CNN |
| CN112395968A (en) * | 2020-11-11 | 2021-02-23 | 北京京航计算通讯研究所 | Mechanical rotating part fault diagnosis method and device based on neural network |
| CN112417766A (en) * | 2020-12-03 | 2021-02-26 | 深制科技(苏州)有限公司 | Fault diagnosis method mainly based on fault-free data |
| CN112633362A (en) * | 2020-12-21 | 2021-04-09 | 华中科技大学 | Self-adaptive fault diagnosis method, device, equipment and medium for rotary machine |
| CN112836577A (en) * | 2020-12-30 | 2021-05-25 | 中南大学 | Intelligent traffic unmanned vehicle fault gene diagnosis method and system |
| CN112834194A (en) * | 2021-01-22 | 2021-05-25 | 西安交通大学 | Fault intelligent detection method based on soft target measurement under fault-free sample |
| CN113063595A (en) * | 2021-03-27 | 2021-07-02 | 温州大学 | Full convolution neural network rolling bearing fault identification method based on one-dimensional convolution |
| CN113240016A (en) * | 2021-05-18 | 2021-08-10 | 哈尔滨理工大学 | Rolling bearing fault diagnosis method based on deep learning |
| CN113705695A (en) * | 2021-08-31 | 2021-11-26 | 重庆邮电大学 | Power distribution network fault data identification method based on convolutional neural network |
| CN114091553A (en) * | 2020-08-06 | 2022-02-25 | 长沙理工大学 | Diagnosis method for rolling bearing fault |
| CN114296005A (en) * | 2021-12-09 | 2022-04-08 | 黄冈师范学院 | Modular multilevel converter submodule composite fault diagnosis method |
| CN114342003A (en) * | 2019-09-03 | 2022-04-12 | 奥格瑞系统有限公司 | Sensor-independent machine fault detection |
| CN114326638A (en) * | 2021-12-30 | 2022-04-12 | 杭州电子科技大学 | Multi-scale convolution and depth bidirectional GRU network fusion performance evaluation method |
| CN114354185A (en) * | 2021-12-29 | 2022-04-15 | 重庆邮电大学 | Gear box fault diagnosis method based on decoupling network |
| CN114383846A (en) * | 2022-01-06 | 2022-04-22 | 合肥工业大学 | Bearing composite fault diagnosis method based on fault label information vector |
| CN114707632A (en) * | 2022-03-03 | 2022-07-05 | 长沙理工大学 | Sensor network sensor fault positioning method, system, equipment and medium |
| CN115560966A (en) * | 2022-09-30 | 2023-01-03 | 新黎明科技股份有限公司 | Weak fault diagnosis method and system for key components of enhanced sparse filtering fusion motor |
| CN112347917B (en) * | 2020-11-05 | 2023-05-16 | 北京信息科技大学 | Gas turbine fault diagnosis method, system, equipment and storage medium |
| CN117499199A (en) * | 2023-08-30 | 2024-02-02 | 长沙理工大学 | VAE-based information enhanced decoupling network fault diagnosis method and system |
| CN117874665A (en) * | 2024-03-13 | 2024-04-12 | 西北工业大学宁波研究院 | SOFC system multi-fault diagnosis method and system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104390780A (en) * | 2014-11-25 | 2015-03-04 | 沈阳化工大学 | Gear case fault diagnosis method based on blind source separation |
| CN107179194A (en) * | 2017-06-30 | 2017-09-19 | 安徽工业大学 | Rotating machinery fault etiologic diagnosis method based on convolutional neural networks |
| CN107451340A (en) * | 2017-07-17 | 2017-12-08 | 安徽工业大学 | Rotating machinery fault quantitative Diagnosis method based on more attribute convolutional neural networks |
| CN108445861A (en) * | 2018-02-05 | 2018-08-24 | 华南理工大学 | A kind of goat fault detection method and system based on convolutional neural networks algorithm |
| CN108613802A (en) * | 2018-05-10 | 2018-10-02 | 重庆大学 | A kind of mechanical failure diagnostic method based on depth mixed network structure |
| US20180284735A1 (en) * | 2016-05-09 | 2018-10-04 | StrongForce IoT Portfolio 2016, LLC | Methods and systems for industrial internet of things data collection in a network sensitive upstream oil and gas environment |
-
2018
- 2018-11-23 CN CN201811406372.9A patent/CN109655259B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104390780A (en) * | 2014-11-25 | 2015-03-04 | 沈阳化工大学 | Gear case fault diagnosis method based on blind source separation |
| US20180284735A1 (en) * | 2016-05-09 | 2018-10-04 | StrongForce IoT Portfolio 2016, LLC | Methods and systems for industrial internet of things data collection in a network sensitive upstream oil and gas environment |
| CN107179194A (en) * | 2017-06-30 | 2017-09-19 | 安徽工业大学 | Rotating machinery fault etiologic diagnosis method based on convolutional neural networks |
| CN107451340A (en) * | 2017-07-17 | 2017-12-08 | 安徽工业大学 | Rotating machinery fault quantitative Diagnosis method based on more attribute convolutional neural networks |
| CN108445861A (en) * | 2018-02-05 | 2018-08-24 | 华南理工大学 | A kind of goat fault detection method and system based on convolutional neural networks algorithm |
| CN108613802A (en) * | 2018-05-10 | 2018-10-02 | 重庆大学 | A kind of mechanical failure diagnostic method based on depth mixed network structure |
Non-Patent Citations (1)
| Title |
|---|
| 韩涛 等: "基于MWT和CNN的滚动轴承智能复合故障诊断方法", 《机械传动》 * |
Cited By (54)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110361176A (en) * | 2019-06-05 | 2019-10-22 | 华南理工大学 | A kind of intelligent failure diagnosis method for sharing neural network based on multitask feature |
| CN110361176B (en) * | 2019-06-05 | 2021-11-19 | 华南理工大学 | Intelligent fault diagnosis method based on multitask feature sharing neural network |
| US11640521B2 (en) | 2019-06-05 | 2023-05-02 | South China University Of Technology | Intelligent fault diagnosis method based on multi-task feature sharing neural network |
| CN112326210A (en) * | 2019-07-17 | 2021-02-05 | 华北电力大学（保定） | Large motor fault diagnosis method combining sound vibration signals with 1D-CNN |
| CN110443174A (en) * | 2019-07-26 | 2019-11-12 | 浙江大学 | A kind of pedestrian's recognition methods again based on decoupling self-adaptive identification feature learning |
| CN110443174B (en) * | 2019-07-26 | 2021-08-10 | 浙江大学 | Pedestrian re-identification method based on decoupling self-adaptive discriminant feature learning |
| CN110443488A (en) * | 2019-07-31 | 2019-11-12 | 中国石油大学(华东) | The recognition methods of drilling well spill hazard, system and equipment based on convolutional neural networks |
| CN110398369A (en) * | 2019-08-15 | 2019-11-01 | 贵州大学 | A kind of Fault Diagnosis of Roller Bearings merged based on 1-DCNN and LSTM |
| CN114342003A (en) * | 2019-09-03 | 2022-04-12 | 奥格瑞系统有限公司 | Sensor-independent machine fault detection |
| CN110647923A (en) * | 2019-09-04 | 2020-01-03 | 西安交通大学 | Variable working condition mechanical fault intelligent diagnosis method based on self-learning under small sample |
| CN110619386A (en) * | 2019-09-09 | 2019-12-27 | 国家电网有限公司 | TMR operation monitoring and fault intelligent research and judgment method and system |
| CN110619386B (en) * | 2019-09-09 | 2022-12-16 | 国家电网有限公司 | TMR operation monitoring and fault intelligent research and judgment method and system |
| CN110647830A (en) * | 2019-09-12 | 2020-01-03 | 华中科技大学 | Bearing fault diagnosis method based on convolutional neural network and Gaussian mixture model |
| CN110647830B (en) * | 2019-09-12 | 2021-12-03 | 华中科技大学 | Bearing fault diagnosis method based on convolutional neural network and Gaussian mixture model |
| CN112067916A (en) * | 2019-09-20 | 2020-12-11 | 武汉理工大学 | Time series data intelligent fault diagnosis method based on deep learning |
| CN110749447B (en) * | 2019-11-27 | 2021-04-16 | 淮安信息职业技术学院 | Loader engine fault diagnosis method using big data |
| CN110749447A (en) * | 2019-11-27 | 2020-02-04 | 淮安信息职业技术学院 | Loader engine fault diagnosis method using big data |
| CN111353587B (en) * | 2020-03-10 | 2024-01-12 | 中科(厦门)数据智能研究院 | Interpretable generation method of deep neural network |
| CN111353587A (en) * | 2020-03-10 | 2020-06-30 | 中国科学院计算技术研究所厦门数据智能研究院 | Interpretable generation method of deep neural network |
| CN111505424A (en) * | 2020-05-06 | 2020-08-07 | 哈尔滨工业大学 | Large experimental device power equipment fault diagnosis method based on deep convolutional neural network |
| CN111860775B (en) * | 2020-07-03 | 2024-05-03 | 南京航空航天大学 | Ship fault real-time diagnosis method based on CNN and RNN fusion |
| CN111860775A (en) * | 2020-07-03 | 2020-10-30 | 南京航空航天大学 | Ship fault real-time diagnosis method based on CNN and RNN fusion |
| CN111899905A (en) * | 2020-08-05 | 2020-11-06 | 哈尔滨工程大学 | Fault diagnosis method and system based on nuclear power device |
| CN111899905B (en) * | 2020-08-05 | 2022-11-01 | 哈尔滨工程大学 | Fault diagnosis method and system based on nuclear power device |
| CN114091553A (en) * | 2020-08-06 | 2022-02-25 | 长沙理工大学 | Diagnosis method for rolling bearing fault |
| CN112132102A (en) * | 2020-09-28 | 2020-12-25 | 天津工业大学 | Intelligent fault diagnosis method combining deep neural network with artificial bee colony optimization |
| CN112132102B (en) * | 2020-09-28 | 2024-04-19 | 天津工业大学 | Intelligent fault diagnosis method combining deep neural network with artificial bee colony optimization |
| CN112163630A (en) * | 2020-10-12 | 2021-01-01 | 清华大学 | Compound fault diagnosis method and device based on unbalanced learning |
| CN112284735A (en) * | 2020-10-21 | 2021-01-29 | 兰州理工大学 | Multi-sensor rolling bearing fault diagnosis based on one-dimensional convolution and dynamic routing |
| CN112347917B (en) * | 2020-11-05 | 2023-05-16 | 北京信息科技大学 | Gas turbine fault diagnosis method, system, equipment and storage medium |
| CN112329650A (en) * | 2020-11-09 | 2021-02-05 | 中国兵器工业集团第二一四研究所苏州研发中心 | Accelerometer fault diagnosis method based on convolutional neural network |
| CN112395968B (en) * | 2020-11-11 | 2021-08-27 | 北京京航计算通讯研究所 | Mechanical rotating part fault diagnosis method and device based on neural network |
| CN112395968A (en) * | 2020-11-11 | 2021-02-23 | 北京京航计算通讯研究所 | Mechanical rotating part fault diagnosis method and device based on neural network |
| CN112417766A (en) * | 2020-12-03 | 2021-02-26 | 深制科技(苏州)有限公司 | Fault diagnosis method mainly based on fault-free data |
| CN112417766B (en) * | 2020-12-03 | 2024-05-14 | 深制科技(苏州)有限公司 | Fault diagnosis method based on fault-free data |
| CN112633362B (en) * | 2020-12-21 | 2024-03-29 | 华中科技大学 | Rotary machine self-adaptive fault diagnosis method, device, equipment and medium |
| CN112633362A (en) * | 2020-12-21 | 2021-04-09 | 华中科技大学 | Self-adaptive fault diagnosis method, device, equipment and medium for rotary machine |
| CN112836577B (en) * | 2020-12-30 | 2024-02-20 | 中南大学 | Intelligent traffic unmanned vehicle fault gene diagnosis method and system |
| CN112836577A (en) * | 2020-12-30 | 2021-05-25 | 中南大学 | Intelligent traffic unmanned vehicle fault gene diagnosis method and system |
| CN112834194A (en) * | 2021-01-22 | 2021-05-25 | 西安交通大学 | Fault intelligent detection method based on soft target measurement under fault-free sample |
| CN113063595A (en) * | 2021-03-27 | 2021-07-02 | 温州大学 | Full convolution neural network rolling bearing fault identification method based on one-dimensional convolution |
| CN113240016A (en) * | 2021-05-18 | 2021-08-10 | 哈尔滨理工大学 | Rolling bearing fault diagnosis method based on deep learning |
| CN113705695A (en) * | 2021-08-31 | 2021-11-26 | 重庆邮电大学 | Power distribution network fault data identification method based on convolutional neural network |
| CN114296005A (en) * | 2021-12-09 | 2022-04-08 | 黄冈师范学院 | Modular multilevel converter submodule composite fault diagnosis method |
| CN114354185A (en) * | 2021-12-29 | 2022-04-15 | 重庆邮电大学 | Gear box fault diagnosis method based on decoupling network |
| CN114326638A (en) * | 2021-12-30 | 2022-04-12 | 杭州电子科技大学 | Multi-scale convolution and depth bidirectional GRU network fusion performance evaluation method |
| CN114383846B (en) * | 2022-01-06 | 2023-06-30 | 合肥工业大学 | Bearing composite fault diagnosis method based on fault label information vector |
| CN114383846A (en) * | 2022-01-06 | 2022-04-22 | 合肥工业大学 | Bearing composite fault diagnosis method based on fault label information vector |
| CN114707632A (en) * | 2022-03-03 | 2022-07-05 | 长沙理工大学 | Sensor network sensor fault positioning method, system, equipment and medium |
| CN115560966B (en) * | 2022-09-30 | 2023-09-15 | 新黎明科技股份有限公司 | Weak fault diagnosis method and system for key components of reinforced sparse filtering fusion motor |
| CN115560966A (en) * | 2022-09-30 | 2023-01-03 | 新黎明科技股份有限公司 | Weak fault diagnosis method and system for key components of enhanced sparse filtering fusion motor |
| CN117499199A (en) * | 2023-08-30 | 2024-02-02 | 长沙理工大学 | VAE-based information enhanced decoupling network fault diagnosis method and system |
| CN117874665A (en) * | 2024-03-13 | 2024-04-12 | 西北工业大学宁波研究院 | SOFC system multi-fault diagnosis method and system |
| CN117874665B (en) * | 2024-03-13 | 2024-05-10 | 西北工业大学宁波研究院 | SOFC system multi-fault diagnosis method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109655259B (en) | 2021-02-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109655259A (en) | Combined failure diagnostic method and device based on depth decoupling convolutional neural networks | |
| Tama et al. | Recent advances in the application of deep learning for fault diagnosis of rotating machinery using vibration signals | |
| CN109635677B (en) | Compound fault diagnosis method and device based on multi-label classification convolutional neural network | |
| CN106682814B (en) | Wind turbine generator fault intelligent diagnosis method based on fault knowledge base | |
| CN109782603A (en) | The detection method and monitoring system of rotating machinery coupling fault | |
| CN111046945B (en) | Fault type and damage degree diagnosis method based on combined convolutional neural network | |
| CN110110768B (en) | Rolling bearing fault diagnosis method based on parallel feature learning and multiple classifiers | |
| CN110334764A (en) | Rotating machinery intelligent failure diagnosis method based on integrated depth self-encoding encoder | |
| CN105372087A (en) | Fault diagnosis method based on multi-sensor signal analysis | |
| CN114492533B (en) | Construction method and application of variable-working-condition bearing fault diagnosis model | |
| CN112257530A (en) | Rolling bearing fault diagnosis method based on blind signal separation and support vector machine | |
| CN112257767B (en) | Product key part state classification method aiming at class imbalance data | |
| CN116894187A (en) | Gear box fault diagnosis method based on deep migration learning | |
| CN112729834B (en) | Bearing fault diagnosis method, device and system | |
| CN113485302A (en) | Vehicle operation process fault diagnosis method and system based on multivariate time sequence data | |
| CN112305388B (en) | On-line monitoring and diagnosing method for insulation partial discharge faults of generator stator winding | |
| CN116593157A (en) | Complex working condition gear fault diagnosis method based on matching element learning under small sample | |
| CN114462480B (en) | Multi-source sensor rolling mill fault diagnosis method based on unbalanced data set | |
| CN117786507B (en) | Rolling bearing unknown fault detection method based on global and local feature coupling guidance | |
| CN118011994A (en) | Cross-modal industrial large model-based fault diagnosis system and method | |
| CN113551904B (en) | Gear box multi-type concurrent fault diagnosis method based on hierarchical machine learning | |
| CN116720109B (en) | FPGA-based improved local linear embedded fan bearing fault diagnosis method | |
| CN116541771A (en) | Unbalanced sample bearing fault diagnosis method based on multi-scale feature fusion | |
| CN113465924A (en) | Bearing fault diagnosis method and system based on improved BP neural network | |
| Wang et al. | Adversarial based unsupervised domain adaptation for bearing fault diagnosis |
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 | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20190419 Assignee: GUANGZHOU HANGXIN AVIATION TECHNOLOGY Co.,Ltd. Assignor: SOUTH CHINA University OF TECHNOLOGY Contract record no.: X2024980011496 Denomination of invention: Composite fault diagnosis method and device based on deep decoupled convolutional neural network Granted publication date: 20210219 License type: Common License Record date: 20240808 Application publication date: 20190419 Assignee: Zhongxin Ocean Helicopter Co.,Ltd. Assignor: SOUTH CHINA University OF TECHNOLOGY Contract record no.: X2024980011494 Denomination of invention: Composite fault diagnosis method and device based on deep decoupled convolutional neural network Granted publication date: 20210219 License type: Common License Record date: 20240808 |
|
| EE01 | Entry into force of recordation of patent licensing contract |