CN106980761A - A kind of rolling bearing running status degradation trend Forecasting Methodology - Google Patents

A kind of rolling bearing running status degradation trend Forecasting Methodology Download PDF

Info

Publication number
CN106980761A
CN106980761A CN201710195149.3A CN201710195149A CN106980761A CN 106980761 A CN106980761 A CN 106980761A CN 201710195149 A CN201710195149 A CN 201710195149A CN 106980761 A CN106980761 A CN 106980761A
Authority
CN
China
Prior art keywords
rolling bearing
running status
degradation trend
bearing running
forecasting methodology
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201710195149.3A
Other languages
Chinese (zh)
Inventor
唐樟春
夏艳君
刘杰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN201710195149.3A priority Critical patent/CN106980761A/en
Publication of CN106980761A publication Critical patent/CN106980761A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16ZINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
    • G16Z99/00Subject matter not provided for in other main groups of this subclass

Abstract

The invention discloses rolling bearing running status degradation trend Forecasting Methodology, it is related to trend prediction field.The present invention provides a kind of rolling bearing running status degradation trend Forecasting Methodology, including:Step 1: gathering and obtaining characterizing the original vibration signal of rolling bearing running status;Step 2: the original vibration signal obtained to the step one carries out denoising Processing, obtain being capable of the useful signal of embodiments rolling bearing running status;Step 3: extracting the temporal signatures and frequency domain character for the useful signal that the step 2 is obtained;Step 4: carrying out Fusion Features to the temporal signatures and the frequency domain character, obtain characterizing the characteristic index of rolling bearing operation trend;Step 5: building degradation trend forecast model, degradation trend prediction is carried out to the characteristic index.The rolling bearing running status degradation trend Forecasting Methodology that the present invention is provided can carry out good prediction to the running status of rolling bearing.

Description

A kind of rolling bearing running status degradation trend Forecasting Methodology
Technical field
The present invention relates to performance trend field, in particular to rolling bearing running status degradation trend Forecasting Methodology.
Background technology
Plant equipment is applied to the every aspect of human lives, work and production, and plays wherein very important Role.At present, plant equipment just develops towards maximization, high speed, precise treatment, systematization, serialization and automation direction, machine The running environment of tool equipment becomes increasingly complex changeable, and the health control even more for equipment proposes new challenge.With equipment Can operation, the problems such as part aging, reliability reduction, residual life are reduced, continue safely and efficiently to work, safeguard with equipment Can timely and effectively it perform, it would be highly desirable to which the mankind solve.Bearing is the important element of plant equipment, is played in mechanical system Highly important effect.The performance degradation trend and life-span prediction method of bearing are always what plant equipment health control was studied Emphasis.In order to effectively prevent equipment operating accuracy from declining, the plant equipment key zero using bearing as representative is maximally utilised The ability to work of part, saves material spending, reduces accident, is increasingly necessary to track the running of parts.It is right Plant equipment key components and parts carry out degradation trend and life search have been turned into extremely important one in modern comfort health control Ring.
The content of the invention
It is an object of the invention to provide a kind of rolling bearing running status degradation trend Forecasting Methodology, it, which is significantly reduced, sets Standby maintenance cost, the global reliability for improving equipment.
The present invention provides a kind of technical scheme:
A kind of rolling bearing running status degradation trend Forecasting Methodology, including:Step 1: gathering and obtaining characterizing the axis of rolling Hold the original vibration signal of running status;Step 2: the original vibration signal obtained to the step one is carried out at de-noising Reason, obtains being capable of the useful signal of embodiments rolling bearing running status;Step 3: that extracts that the step 2 obtains is described The temporal signatures and frequency domain character of useful signal;Step 4: Fusion Features are carried out to the temporal signatures and the frequency domain character, Obtain characterizing the characteristic index of rolling bearing operation trend;Step 5: degradation trend forecast model is built, to the characteristic index Carry out degradation trend prediction.
Further, the degradation trend model includes least square method supporting vector machine, particle cluster algorithm and deviation accumulation And method, the most young waiter in a wineshop or an inn is into SVMs for building forecast model, and the particle cluster algorithm is for the prediction mould The parameter of type is optimized, and the deviation accumulation and method are used to be controlled the predicated error of the forecast model.
Further, the method for the Fusion Features is PCA, the temporal signatures and the frequency domain character Multidimensional characteristic matrix is constituted, the multidimensional characteristic matrix obtains the characteristic index by the PCA.
Further, the step 2 carries out de-noising using Methods for Wavelet Denoising Used to the original vibration signal, and obtains The useful signal.
The beneficial effects of the present invention are:The vibration signal of the rolling bearing gathered by online monitoring vibration sensor, Obtain useful information and extract the state feature of information in domains such as time domain, frequency domain, time-frequency domains, obtain that bearing running status can be characterized Multidimensional characteristic collection, and carry out the performance degradation trend prediction and life prediction of rolling bearing running status, be that bearing is based on regarding Feelings maintenance, the important foundation for improving reliability.But multidimensional characteristic concentrates the two or more spies that there may be mutual redundancy Levy, it is also possible to there is the feature completely irrelevant with bearing runnability state, simultaneously because rolling bearing structure, working environment etc. Complexity, cause single simple feature index be difficult intactly performance of the reflection rolling bearing under arms in periodic process move back Change trend, it is difficult to determine its residual life, it is very unfavorable for bearing maintenance, maintenance and raising reliability.And the present invention is carried Go out a kind of rolling bearing running status performance degradation trend and method for predicting residual useful life by online monitoring vibration signal, it is first First pass through wavelet noise to reject the noise in collection signal and retain useful information, then respectively in time domain, frequency domain, time-frequency domain etc. Extract the state feature of information and form multidimensional characteristic collection in domain.With reference to PCA, multidimensional characteristic collection is merged, On the basis of not reducing the information content that legacy data included and rejecting redundancy and invalid components in multidimensional characteristic information, The optimal characteristic index of statistical significance upside deviation is obtained, complete reflection bearing under arms in the cycle comprehensively is then based on The characteristic index of runnability degradation trend and using least square method supporting vector machine as trend prediction model, uses cross validation Method, and the comprehensive new method for using particle cluster algorithm Optimal Parameters, obtaining being capable of degree of precision prediction bearing performance degradation trend. The present invention can obtain the new method of an energy full forecast rolling bearing performance degradation trend, can tie up further to carry out regarding feelings Solid foundation is established in maintenance strategy research, so as to significantly reduce expense, improve the global reliability of equipment.
Brief description of the drawings
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be attached to what is used required in embodiment Figure is briefly described.It should be appreciated that the following drawings illustrate only certain embodiments of the present invention, therefore it is not construed as pair The restriction of scope.For those of ordinary skill in the art, on the premise of not paying creative work, can also be according to this A little accompanying drawings obtain other related accompanying drawings.
The flow signal for the rolling bearing running status degradation trend Forecasting Methodology that Fig. 1 provides for embodiments of the invention Figure;
Deviation accumulation and the schematic flow sheet of method that Fig. 2 provides for embodiments of the invention.
Embodiment
Below in conjunction with the accompanying drawings, the embodiment to the present invention is described in detail.
Referring to Fig. 1, present embodiments providing a kind of rolling bearing running status degradation trend Forecasting Methodology.Below to this The specific steps of method are illustrated.
Step (1):On-line monitoring vibrating sensor collection bearing vibration signal (includes reflection bearing runnability The useful signal and noise signal of state).
Step (2):Wavelet noise function (the Main Basiss Matlab algorithms, wherein base letter carried using MATLAB softwares It is crucial that number, Decomposition order, threshold method and threshold value etc., which are chosen) most of noise jamming in vibration signal is eliminated, and retain The useful signal of bearing runnability state can be reflected.
Step (3):Extract temporal signatures collection (including dimension feature and dimensionless feature, 1, table 2 is shown in Table in detail), frequency domain special Collect (being shown in Table 3 in detail), time and frequency domain characteristics collection (empirical mode decomposition method (Empirical Model Decomposition, EMD non-linear, the non-stationary signal feature in bearing condition monitoring signal) can be efficiently extracted, and possesses higher time-frequency and is differentiated Rate, time and frequency domain characteristics collection herein is the nonlinear characteristic in the bearing condition monitoring signal that EMD is extracted) and Weibull parameter (form parameter and scale parameter of Weibull distribution can be used as the characteristic quantity for reflecting bearing running status to feature set.Due to Xi Er Bert becomes transducing and preferably extracts bearing initial failure information, therefore status monitoring signal first passes through Hilbert transform and obtained newly Data, then using maximum likelihood method be the form parameter and scale parameter that can obtain Weibull distribution, i.e., Weibull join Number feature set), obtain that the multidimensional characteristic collection of bearing runnability state can be characterized, and build multidimensional characteristic collection matrix.
The index of temporal signatures containing dimension of table 1
The non_dimensional time domain characteristic index of table 2
The frequency domain character of table 3
Non-linear, non-stationary information, empirical modal are generally comprised in the status monitoring signal of plant equipment key components and parts Decomposition method (Empirical Mode Decomposition, EMD) is a kind of non-linear, Non-stationary Signal Analysis method, it By signal decomposition into a series of intrinsic mode functions (Intrinsic Mode Function, IMF) sum.For nonlinear spy Levy, widely used method also has wavelet method.Compared to wavelet method, empirical mode decomposition method is not only able to effectively Extract signal in Weak characteristic, and without choose basic function, therefore, adaptivity is stronger, be especially suitable for handle non-thread Property signal.Influence of the human factor to result, decomposition result dependency analysis signal sheet are eliminated by empirical mode decomposition method Body, can efficiently extract non-linear, non-stationary signal Weak characteristic, and available higher time frequency resolution, with good Time-frequency locality.So the present embodiment extracts the non-linear spy in parts monitoring signals using empirical modal analysis method Levy.
If by empirical mode decomposition, signal obtains multiple modal components fi(t) with remainder rn(t), rn(t) letter is regarded as Number (n+1)th component fn+1(t), then i-th (i=1,2 ..., n+1) individual component fi(t) energy is:
In formula, N is IMF components fi(t) data length.
Rule of thumb mode decomposition Complete Orthogonal, it can be deduced that:
E [x (t)]=E [f1(t)]+E[f2(t)]+...+E[fn+1(t)]
Weibull distribution is one of conventional statistical distribution pattern in fail-safe analysis, can be divided into two parameter Weibull point Cloth and three-parameter Weibull distribution.The form parameter and scale parameter of Weibull distribution can also be used as reflection parts running status Characteristic quantity.The present embodiment considers two parameter Weibull distribution, and its probability density function is:
β is form parameter in formula, and η is scale parameter.
Method for parameter estimation is generally divided into diagram method and the major class of analytic method two, diagram method includes experience distribution map method, prestige Boolean's probability graph method and relative risk statistic map method etc.;Analytic method includes Maximum Likelihood Estimation Method and regression estimates method etc..This implementation Example solves parameter by Maximum Likelihood Estimation Method.Maximum-likelihood estimation function is:
In formula, N is data length.Because the extraction of Hilbert transform pairs parts initial failure information has good effect Really, xiThe data for being state detection signal after Hilbert transform.
On above formula, local derviation is sought β and η respectively, and makes it be zero, is obtained:
Above formula is Nonlinear System of Equations, is solved using Newton iteration method and obtains parameter beta and η estimate.Due to passing through pole The estimate that the maximum-likelihood estimation technique is obtained usually there will be larger error, and amendment form parameter β can reduce evaluated error, its Correction formula is:
In formula, N is data length, βUIt is correction value, β is estimate.
Step (4):Using PCA PCA (Principal Components Analysis) to rolling bearing The multidimensional characteristic of vibration signal is merged after de-noising, is not being reduced on the basis of legacy data includes the information content, is being obtained The characteristic index of rolling bearing running status performance degradation trend can be characterized comprehensively.
The performance degradation characteristic index in parts running is obtained using principal component analysis progress multidimensional characteristic fusion It is, by certain Linear Mapping, to form the new multidimensional characteristic collection less than original dimension by original multidimensional characteristic collection.New It is mutually orthogonal two-by-two between feature set, and vector sorted from big to small according to the otherness between feature, sequence first vector reflection Maximum difference between feature.The new vector that maximum difference amount is constituted so between multidimensional characteristic can not only be covered in feature Useful information, and the redundancy section of original multidimensional characteristic collection is eliminated, realize with the mode of principal component to describe high dimension According to feature.
If having n data in certain data collected, each data have p characteristic variable, then constitute a n × p The multidimensional characteristic matrix of rank:
X in formulai=(x1i,x2i,...,xni)T, xijIt is j-th of characteristic variable of i-th of data.
Obtained by principal component transform by x1,x2,...,xpThe linear combination of expression:
If coefficient lijMeet:1)li1 2+li2 2+...+lip 2=1, i=1,2 .., p;2) coefficient lijMake linear combination yi With yj(i ≠ j) is independent of each other;3)yiAccording to x1,x2,...,xpThe sequence of all linear combination variance sizes, then former multidimensional characteristic Integrate first, second, pth principal component is y1,y2,...,yp
Principal component analysis process represented by above formula is exactly to ask characteristic value and spy by the covariance matrix to multidimensional characteristic Levy vector, and the characteristic vector corresponding to the characteristic value after order by size is arranged is used as the coefficient of linear combination.Utilize master Component analyzing method carries out merging the step of obtaining parts performance degradation characteristic index to multidimensional characteristic X:
(1) parts multidimensional characteristic matrix X mean vector is calculated:
By multidimensional characteristic centralization:
(2) covariance matrix of the special matrix of multidimensional is calculated:
The eigen vector of multidimensional characteristic matrix is calculated, using the characteristic vector corresponding to eigenvalue of maximum as right The coefficient of multidimensional characteristic matrix linear change, obtains parts performance degradation characteristic index.
Step (5):The conclusion obtained according to step (4), design forecast model and parameter optimization.With least square support to Amount machine uses cross-validation method as forecast model in optimizing to parameter model, and synthesis uses particle swarm optimization algorithm, to carry Accuracy rate and accuracy that height is trained every time.And using deviation accumulation and method (Cumulative Sum, CUSUM) to current mould The predictive ability of type is tested, and further improves the accuracy of prediction.
Obtained forecast model is optimized by particle cluster algorithm, although parameter has reached in some sense optimal, but It is but to lack further to examine the predictive ability of model, the present embodiment, which is utilized, uses error accumulation and method (Cumulative Sum, CUSUM) tests to the predictive ability of "current" model, further improves the accuracy of prediction.
Referring to Fig. 2, the specific steps of deviation accumulation given below and method.
Assuming that the error Z=y of forecast modeli-yiIt is to obey average for μ, variance is σ2Normal distribution, to obey normal state The array Z of distribution is obtained after being standardized:
Accumulation and method examine the predictive ability of "current" model using two indices:
UBi=max [0, (di-m)+UBi-1]
LBi=max [0, (- di-m)+LBi-1]
In above formula, UBiRepresent the overgauge of error, LBiRepresent the minus deviation of error.Original state UB0And LB0All it is 0, it is generally the case that parameter m takes 0.5, and threshold value h takes 3.
What the present embodiment was proposed is entered based on Least Square Support Vector Regression, particle cluster algorithm and error accumulation and method The research step of row plant equipment key components and parts trend prediction is as follows:
(1) running status to plant equipment key components and parts is monitored on-line, and the vibration signal of acquisition is filtered Ripple handles (Methods for Wavelet Denoising Used), to ensure that the characteristic information that vibration data is included is not flooded by noise;
(2) the multidimensional characteristic information of Monitoring Data is extracted, is obtained using the mode (principal component analytical method) of Fusion Features The characteristic index of plant equipment key components and parts running status performance degradation trend;
(3) the trend prediction mould based on Least Square Support Vector Regression, particle cluster algorithm and accumulation and method is set up Type;
(4) by characteristic index input prediction model, the degradation trend to plant equipment key components and parts running status is realized Predict, and output predicts the outcome.
Step (6):Trend prediction, using the characteristic index obtained in step 4 as trend prediction model input data, and Determine training set and test set;Rolling bearing running status performance degradation trend is predicted.
In the present embodiment, training set length is 100, and institute's extracting method of the present invention and control methods carry out Single-step Prediction, its Middle cross-validation method selects 5-Fold;The major parameter of particle cluster algorithm is set:Maximum evolution quantity is 200, planted Group's maximum quantity is that 20, global and local search capability is respectively set to 1.7 and 1.5;The major parameter of genetic algorithm is set to: Maximum evolutionary generation is 200, and population maximum quantity is 20.

Claims (4)

1. a kind of rolling bearing running status degradation trend Forecasting Methodology, it is characterised in that including:
Step 1: gathering and obtaining characterizing the original vibration signal of rolling bearing running status;
Step 2: the original vibration signal obtained to the step one carries out denoising Processing, obtaining being capable of embodiments rolling The useful signal of dynamic bearing running status;
Step 3: extracting the temporal signatures and frequency domain character for the useful signal that the step 2 is obtained;
Step 4: carrying out Fusion Features to the temporal signatures and the frequency domain character, obtain characterizing rolling bearing operation trend Characteristic index;
Step 5: building degradation trend forecast model, degradation trend prediction is carried out to the characteristic index.
2. rolling bearing running status degradation trend Forecasting Methodology according to claim 1, it is characterised in that the degeneration Trend prediction model includes least square method supporting vector machine, particle cluster algorithm and deviation accumulation and method, and the most young waiter in a wineshop or an inn is into branch Holding vector machine is used to build forecast model, and the particle cluster algorithm is used to optimize the parameter of the forecast model, described Deviation accumulation and method are used to be controlled the predicated error of the forecast model.
3. rolling bearing running status degradation trend Forecasting Methodology according to claim 1, it is characterised in that the feature The method of fusion is PCA, and the temporal signatures constitute multidimensional characteristic matrix, the multidimensional with the frequency domain character Eigenmatrix obtains the characteristic index by the PCA.
4. rolling bearing running status degradation trend Forecasting Methodology according to claim 1, it is characterised in that the step Two carry out de-noising using Methods for Wavelet Denoising Used to the original vibration signal, and obtain the useful signal.
CN201710195149.3A 2017-03-29 2017-03-29 A kind of rolling bearing running status degradation trend Forecasting Methodology Pending CN106980761A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710195149.3A CN106980761A (en) 2017-03-29 2017-03-29 A kind of rolling bearing running status degradation trend Forecasting Methodology

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710195149.3A CN106980761A (en) 2017-03-29 2017-03-29 A kind of rolling bearing running status degradation trend Forecasting Methodology

Publications (1)

Publication Number Publication Date
CN106980761A true CN106980761A (en) 2017-07-25

Family

ID=59339235

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710195149.3A Pending CN106980761A (en) 2017-03-29 2017-03-29 A kind of rolling bearing running status degradation trend Forecasting Methodology

Country Status (1)

Country Link
CN (1) CN106980761A (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108760256A (en) * 2018-03-23 2018-11-06 佛山科学技术学院 A kind of prediction technique of axle box remaining life
CN108898050A (en) * 2018-05-17 2018-11-27 广东工业大学 A kind of flexible material process equipment roll shaft performance index calculation method
CN109214097A (en) * 2018-09-14 2019-01-15 上海工程技术大学 A kind of long related failure trend prediction method of dimensionless group rolling bearing
CN109359791A (en) * 2018-12-26 2019-02-19 湖南科技大学 A kind of mechanical system degradation trend prediction technique and system
CN109827777A (en) * 2019-04-01 2019-05-31 哈尔滨理工大学 Rolling bearing fault prediction technique based on Partial Least Squares extreme learning machine
CN109855875A (en) * 2019-01-15 2019-06-07 沈阳化工大学 A kind of rolling bearing operational reliability prediction technique

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103954450A (en) * 2014-05-19 2014-07-30 重庆交通大学 Bearing life degradation performance evaluation index construction method based on main component analysis
CN104655423A (en) * 2013-11-19 2015-05-27 北京交通大学 Rolling bearing fault diagnosis method based on time-frequency domain multidimensional vibration feature fusion

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104655423A (en) * 2013-11-19 2015-05-27 北京交通大学 Rolling bearing fault diagnosis method based on time-frequency domain multidimensional vibration feature fusion
CN103954450A (en) * 2014-05-19 2014-07-30 重庆交通大学 Bearing life degradation performance evaluation index construction method based on main component analysis

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
杨鹏: "基于在线监测信息的机械设备关键零部件寿命预测方法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 *

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108760256A (en) * 2018-03-23 2018-11-06 佛山科学技术学院 A kind of prediction technique of axle box remaining life
CN108898050A (en) * 2018-05-17 2018-11-27 广东工业大学 A kind of flexible material process equipment roll shaft performance index calculation method
CN109214097A (en) * 2018-09-14 2019-01-15 上海工程技术大学 A kind of long related failure trend prediction method of dimensionless group rolling bearing
CN109359791A (en) * 2018-12-26 2019-02-19 湖南科技大学 A kind of mechanical system degradation trend prediction technique and system
CN109359791B (en) * 2018-12-26 2020-06-05 湖南科技大学 Mechanical system degradation trend prediction method and system
CN109855875A (en) * 2019-01-15 2019-06-07 沈阳化工大学 A kind of rolling bearing operational reliability prediction technique
CN109827777A (en) * 2019-04-01 2019-05-31 哈尔滨理工大学 Rolling bearing fault prediction technique based on Partial Least Squares extreme learning machine
CN109827777B (en) * 2019-04-01 2020-12-18 哈尔滨理工大学 Rolling bearing fault prediction method based on partial least square method extreme learning machine

Similar Documents

Publication Publication Date Title
Zheng et al. Generalized composite multiscale permutation entropy and Laplacian score based rolling bearing fault diagnosis
Haidong et al. Intelligent fault diagnosis of rolling bearing using deep wavelet auto-encoder with extreme learning machine
Zhao et al. Critical-to-fault-degradation variable analysis and direction extraction for online fault prognostic
Gama et al. A survey on concept drift adaptation
Chen et al. Multi-fault diagnosis study on roller bearing based on multi-kernel support vector machine with chaotic particle swarm optimization
Hong et al. A hybrid deep learning-based neural network for 24-h ahead wind power forecasting
Xu et al. PHM-oriented integrated fusion prognostics for aircraft engines based on sensor data
Zheng et al. A rolling bearing fault diagnosis method based on multi-scale fuzzy entropy and variable predictive model-based class discrimination
Shao et al. Rolling bearing fault detection using continuous deep belief network with locally linear embedding
Angelov et al. A new type of simplified fuzzy rule-based system
Jia et al. An optimized classification algorithm by BP neural network based on PLS and HCA
Lu et al. Dominant feature selection for the fault diagnosis of rotary machines using modified genetic algorithm and empirical mode decomposition
Saxena et al. Evolving an artificial neural network classifier for condition monitoring of rotating mechanical systems
Jia et al. An optimized RBF neural network algorithm based on partial least squares and genetic algorithm for classification of small sample
Fazzolari et al. A review of the application of multiobjective evolutionary fuzzy systems: Current status and further directions
Gholizadeh et al. Optimal design of structures subjected to time history loading by swarm intelligence and an advanced metamodel
JP5363927B2 (en) Abnormality detection / diagnosis method, abnormality detection / diagnosis system, and abnormality detection / diagnosis program
Wang et al. Deep separable convolutional network for remaining useful life prediction of machinery
CN108009674A (en) Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks
CN108344564B (en) A kind of state recognition of main shaft features Testbed and prediction technique based on deep learning
Giraud-Carrier et al. Introduction to the special issue on meta-learning
Hachicha et al. A survey of control-chart pattern-recognition literature (1991–2010) based on a new conceptual classification scheme
Karakasis et al. On the use of metamodel-assisted, multi-objective evolutionary algorithms
Li et al. A directed acyclic graph network combined with CNN and LSTM for remaining useful life prediction
Yang et al. A hybrid feature selection scheme for unsupervised learning and its application in 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
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20170725