CN109060350A - A kind of Rolling Bearing Fault Character extracting method dictionary-based learning - Google Patents
A kind of Rolling Bearing Fault Character extracting method dictionary-based learning Download PDFInfo
- Publication number
- CN109060350A CN109060350A CN201811027613.9A CN201811027613A CN109060350A CN 109060350 A CN109060350 A CN 109060350A CN 201811027613 A CN201811027613 A CN 201811027613A CN 109060350 A CN109060350 A CN 109060350A
- Authority
- CN
- China
- Prior art keywords
- signal
- dictionary
- fault
- rolling bearing
- bearing fault
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a kind of Rolling Bearing Fault Character extracting methods dictionary-based learning, are field of signal processing.Include: to obtain vibration signal from the sensor being mounted on bearings, be expressed as s, note sample frequency is fs;Using harmonic wave separation filter, the harmonic component that other mechanism vibrations contained in s are generated is removed, and obtains processing signal y;Shift-Invariant K-SVD dictionary learning is carried out to signal y, obtains rarefaction representation and best dictionary, is rebuild according to rarefaction representation and dictionary and restores pure fault-signal x.Envelope spectral transformation is carried out to signal x, obtains the frequecy characteristic of bearing fault vibration signal.This method, which uses sparse representation method, can promptly and accurately extract the Weak fault feature of early stage, have better robustness and accuracy compared to traditional Signal De-noising Method, provide strong support to rolling bearing fault judgement and maintenance decision.
Description
Technical field
The present invention relates to rotating machinery and vibration signal processing field, in particular to a kind of axis of rolling dictionary-based learning
Hold fault signature extracting method.
Background technique
Rolling bearing is the critical component of motor and rotating machinery, such as car transmissions, wind turbine, aeroplane engine
Machine, lathe etc..Since many failures in these devices may be since rolling bearing fault, it is therefore necessary to diagnosis event as early as possible
Barrier is to prevent serious or catastrophic failure.These equipment are usually under high speed, heavy duty, high/low temperature and the poor working conditions of pollution
Operation, therefore be easily damaged, lead to serious mechanical breakdown.Heavy to closing is become to bearing failure diagnosis in industrial manufacturing process
It wants.Due to the dynamic characteristic of the direct representing fault bearing of vibration signal, this leads to their sensibility to failure, therefore their quilts
It is widely used in detecting bearing fault.For the mechanical fault diagnosis based on vibration signal, total there are two key steps, i.e. feature
Extraction and pattern-recognition.Therefore need to carry out efficient, accurate feature extraction.
In order to extract representative fault signature from complicated nonstationary noise vibration signal, many signals have been developed
Processing method is used for rolling bearing fault diagnosis, such as statisticallys analyze, Fourier transform, wavelet transformation, Hilbert-Huang
Transform and empirical mode decomposition.
Different from traditional diagnostic method, rarefaction representation is intended to find the most sparse or approximate most sparse expression of signal,
It captures the higher level feature in the data in redundant dictionary.The main thought of sparse representation theory is by the energy of characteristic information
Amount focuses in a small number of elements, and by projecting to different fault-signals in different rarefaction representation spaces, provides one kind
The feasible method of multiple characteristic informations is identified simultaneously.Due to the ability with powerful extraction feature, sparse representation method is fast
Speed exploitation is used for fault diagnosis, including match tracing (MP), base tracking denoising and the sparse reconstruction of manifold.Although effectively, in work
A large amount of priori knowledge is needed using them in industry system.The selection of dictionary directly affects the validity of expression, due to operating ring
The variation in border, the complexity and unstable condition of component are difficult to carry out fault diagnosis to generator.Dictionary learning is a kind of effective
Data-driven version, for constructing the empirical learning dictionary for being used for rarefaction representation, wherein generating atom comes from basic experience number
According to rather than from certain theoretical models.Most common dictionary learning method is that K-SVD passes through the training on alternately current dictionary
Signal sparse approximate and dictionary is constructed to the optimization of dictionary according to the decomposition of calculating.However, when handling long signal, dictionary
Learning method needs to divide the signal into segment, or even the same clip with out of phase will lead to different atoms.As a result, learning
The dictionary of habit will have bulk redundancy information.Due to the characteristic of mechanical signal, such as strong periodically, changeability and phase problem,
It may be useful for learning invariant shift dictionary.
Therefore, the characteristics of being based on mechanical oscillation signal, proposes a kind of the sparse of combination invariant shift dictionary learning technology
Time-frequency representation method, the fault signature for rolling bearing extract, and overcome the office of traditional rarefaction representation and dictionary learning method
It is sex-limited.All pulse characteristics at different location with same characteristic features can be by the method that is proposed only by a basic function
It indicates, without long signal is divided into small frame.Therefore, study dictionary is smaller compared with K-SVD and is more suitable for industry
Vibration signal in system, with faster convergence rate.
Summary of the invention
The purpose of the present invention is to provide a kind of Rolling Bearing Fault Character extracting methods dictionary-based learning, effectively solve
Certainly existing fault signature extracts inaccuracy, is difficult to separation problem to fault-signal under strong noise environment, improves failure knowledge
Other accuracy rate.In order to achieve the above objectives, the present invention provides the following technical solutions, comprising the following steps:
Step 1: obtaining vibration signal from the sensor being mounted on bearings, and is expressed as s, and note sample frequency is fs;
Step 2: using harmonic wave separation filter, and the harmonic component that other mechanism vibrations contained in s are generated removes,
Obtain processing signal y;
Step 3: Shift-Invariant K-SVD dictionary learning is carried out to signal y, obtains rarefaction representation and best word
Allusion quotation rebuilds according to rarefaction representation and dictionary and restores pure fault-signal x.
Step 4: carrying out envelope spectral transformation to signal x, obtains the frequecy characteristic of bearing fault vibration signal and existing
Rolling bearing parameter can provide strong support compared to Dui for fault identification and maintenance decision.
Further, specifically includes the following steps: 11 in step 1: establishing coordinate system: establishing space coordinates XYZ, X-axis
It is directed toward bearing, Y-axis is directed toward vertical direction.12: unidirectional acceleration transducer being installed as Z-direction using bearing direction and computer connects
Connect acquisition data.13: the data of record acceleration transducer acquisition are denoted as vibration acceleration time-domain signal S, and note sample frequency is
fs.14: being intercepted from signal S often as tsVibration acceleration signal analyzed, be denoted as s.
Further, specifically includes the following steps: 21 in step 2: input signal s, initiation parameter threshold value are ε, iteration
Number k, creation Fourier Dictionary D.22: solving rarefaction representation, objective function are as follows:It is obtained by orthogonal matching pursuit (OMP) derivation algorithm by k iteration
Rarefaction representation α, reconstruction signal y, as isolates harmonic component signal.
Further, specifically includes the following steps: 31 in step 3: input parameter, including: signal y, degree of rarefication L,
Atomic length l, atomic quantity K.32: the objective function that original signal is restored are as follows:It is chased after with matching
Track (MP) algorithm calculates rarefaction representation:According to formulaDictionary atom is updated, according to formulaMore
New dictionary.34: repeat step 32,33 untilIt is sufficiently small, export dictionary atom M and rarefaction representation
αk,τ.35: according to formulaReconstruct fault-signal x.
Further, specifically includes the following steps: 41 in step 4: doing Hilbert transformation to fault-signal x, variation is public
Formula is42: Fast Fourier is done to obtained Hilbert transformation H (x) (t)
Transform obtains the envelope spectrum of signal.43: common rolling bearing fault position includes: roller failure fBS, inner ring failure fiWith
Outer ring failure fo, can be with the different faults standard frequency of current rotating speed by bearing parameter.44: the failure-frequency feature that will be obtained
It compares with standard frequency, provides strong support for accident analysis and maintenance decision.
Detailed description of the invention
In order to keep the purpose of the present invention, technical scheme and beneficial effects clearer, the present invention provides following attached drawing and carries out
Illustrate:
Fig. 1 is the flow chart of the method for the invention;
Fig. 2 is the bearing vibration signal time domain waveform that accelerometer obtains;
Fig. 3 is the primary fault time domain plethysmographic signal figure obtained using dictionary learning;
Fig. 4 is that the frequency domain figure for carrying out Trouble Match is converted by Hilbert;
Specific embodiment
Below in conjunction with attached drawing, a preferred embodiment of the present invention will be described in detail.
Fig. 1 be the method for the invention flow chart, this method the following steps are included:
Step S1:
S11: it establishes coordinate system: establishing space coordinates XYZ, X-axis is directed toward bearing, and Y-axis is directed toward vertical direction.
S12: unidirectional acceleration transducer is installed as Z-direction using bearing direction and connects acquisition data with computer.
S13: the data of record acceleration transducer acquisition are denoted as vibration acceleration time-domain signal S, and note sample frequency is
fs。
S14: it is intercepted from signal S often as tsVibration acceleration signal analyzed, be denoted as s.
Step S2:
S21: input signal s, initiation parameter threshold value is ε, the number of iterations k, creates Fourier Dictionary D.
S22: rarefaction representation, objective function are solved are as follows:Pass through orthogonal
Rarefaction representation is obtained by k iteration with tracking (OMP) derivation algorithm, reconstruction signal y as isolates harmonic component letter
Number.
Step S3:
S31: input parameter, including: signal y, degree of rarefication L, atomic length l, atomic quantity K.
S32: the objective function that original signal is restored are as follows:With match tracing (MP) algorithm,
Calculate rarefaction representation:
S33: according to formulaDictionary atom is updated, according to formulaUpdate dictionary.
S34: repeat step 32,33 untilIt is sufficiently small, export dictionary atom M and rarefaction representation
αk,τ。
S35: according to formulaReconstruct fault-signal x.
Step S4:
S41: Hilbert transformation is done to fault-signal x, variation formula is
S42: Fast Fourier Transform is done to obtained Hilbert transformation H (x) (t) and obtains the envelope of signal
Spectrum.
S43: common rolling bearing fault position includes: roller failure fBS, inner ring failure fiWith outer ring failure fo, pass through axis
It holds parameter and current rotating speed calculates different faults standard frequency.
S44: obtained failure-frequency feature and standard frequency are compared, provide strong branch for accident analysis and maintenance decision
Support.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1, the present invention obtains the excessively complete word of down pulse atomic structre by dictionary learning using fault-signal as training set
Allusion quotation, resulting dictionary more can true and accurate expression fault-signals compared to other predefined dictionaries.
2, meter can be substantially reduced using the method for moving constant dictionary learning according to the shifting invariant feature of periodic pulse signal
Evaluation time improves signaling protein14-3-3 precision.
3, it is suitable for the various faults types such as the fluctuation of speed, rolling element sliding, and there is preferable noiseproof feature, faint
Pulse signal, which extracts aspect, has good effect.
Claims (5)
1. a kind of Rolling Bearing Fault Character extracting method dictionary-based learning, comprising:
S1: obtaining vibration signal from the sensor being mounted on bearings, and is expressed as s, and note sample frequency is fs;
S2: using harmonic wave separation filter, and the harmonic component that other mechanism vibrations contained in s are generated removes, and is handled
Signal y;
S3: Shift-Invariant K-SVD dictionary learning is carried out to signal y, rarefaction representation and best dictionary is obtained, finally weighs
It builds and restores pure fault-signal x.
S4: envelope spectral transformation is carried out to signal x, obtains the frequecy characteristic and existing rolling bearing of bearing fault vibration signal
Parameter can provide strong support compared to Dui for fault identification and maintenance decision.
2. Rolling Bearing Fault Character extracting method dictionary-based learning according to claim 1, which is characterized in that institute
It is as follows to state vibration signal obtaining step in step S1:
S11: it establishes coordinate system: establishing space coordinates XYZ, X-axis is directed toward bearing, and Y-axis is directed toward vertical direction.
S12: unidirectional acceleration transducer is installed as Z-direction using bearing direction and connects acquisition data with computer.
S13: the data of record acceleration transducer acquisition are denoted as vibration acceleration time-domain signal S, and note sample frequency is fs。
S14: it is intercepted from signal S often as tsVibration acceleration signal analyzed, be denoted as s.
3. Rolling Bearing Fault Character extracting method dictionary-based learning according to claim 1, which is characterized in that institute
It is as follows to state harmonic component separating step in step S2:
S21: input signal s, initiation parameter threshold value is ε, the number of iterations k, creates Fourier Dictionary D.
S22: rarefaction representation, objective function are solved are as follows:s.t.y-D1∝1< ∈, passes through orthogonal matching pursuit
(OMP) derivation algorithm obtains rarefaction representation by k iteration, and reconstruction signal y as isolates harmonic component signal.
4. Rolling Bearing Fault Character extracting method dictionary-based learning according to claim 1, which is characterized in that institute
Stating dictionary learning and fault-signal reconstruct in step S3, steps are as follows:
S31: input parameter, including: signal y, degree of rarefication L, atomic length l, atomic quantity K.
S32: the objective function that original signal is restored are as follows:With match tracing (MP) algorithm, calculate
Rarefaction representation:s.t.||α||0≤L。
S33: according to formulaDictionary atom is updated, according to formulaUpdate dictionary.
S34: repeat step S32, S33 untilIt is sufficiently small, export dictionary atom M and rarefaction representation
αk,τ。
S35: according to formulaReconstruct fault-signal x.
5. Rolling Bearing Fault Character extracting method dictionary-based learning according to claim 1, which is characterized in that institute
It is as follows to state envelope spectral transformation and fault-signal frequency analysis step in step S4:
S41: Hilbert transformation is done to fault-signal x, variation formula is
S42: Fast Fourier Transform is done to obtained Hilbert transformation H (x) (t) and obtains the envelope spectrum of signal.
S43: common rolling bearing fault position includes: roller failure fBS, inner ring failure fiWith outer ring failure fo, joined by bearing
Several and current rotating speed calculates different faults standard frequency.
S44: obtained failure-frequency feature and standard frequency are compared, and are provided strong support for accident analysis and maintenance decision.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811027613.9A CN109060350A (en) | 2018-09-04 | 2018-09-04 | A kind of Rolling Bearing Fault Character extracting method dictionary-based learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811027613.9A CN109060350A (en) | 2018-09-04 | 2018-09-04 | A kind of Rolling Bearing Fault Character extracting method dictionary-based learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109060350A true CN109060350A (en) | 2018-12-21 |
Family
ID=64759473
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811027613.9A Pending CN109060350A (en) | 2018-09-04 | 2018-09-04 | A kind of Rolling Bearing Fault Character extracting method dictionary-based learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109060350A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110160790A (en) * | 2019-05-14 | 2019-08-23 | 中国地质大学(武汉) | A kind of rolling bearing fault impact signal extracting method and system based on improvement K-SVD |
CN111307455A (en) * | 2020-03-06 | 2020-06-19 | 西南交通大学 | Train bogie bearing fault monitoring method and system based on dictionary learning |
CN111382792A (en) * | 2020-03-09 | 2020-07-07 | 兰州理工大学 | Rolling bearing fault diagnosis method based on double-sparse dictionary sparse representation |
CN112988918A (en) * | 2021-04-06 | 2021-06-18 | 中车青岛四方机车车辆股份有限公司 | Bearing fault dictionary construction method, analysis method and system |
CN113203565A (en) * | 2021-03-25 | 2021-08-03 | 长江大学 | Bearing fault identification method and system based on EEMD sparse decomposition |
CN113295410A (en) * | 2021-05-14 | 2021-08-24 | 上海交通大学 | Bearing fault diagnosis method under variable rotating speed working condition |
CN113468760A (en) * | 2021-07-21 | 2021-10-01 | 中南大学 | Motor weak fault detection method and system based on dictionary learning |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06186136A (en) * | 1992-12-18 | 1994-07-08 | Ono Sokki Co Ltd | Fault diagnostic apparatus for ball-and-roller bearing |
CN105241666A (en) * | 2015-09-21 | 2016-01-13 | 华南理工大学 | Rolling bearing fault feature extraction method based on signal sparse representation theory |
CN105424359A (en) * | 2015-11-25 | 2016-03-23 | 华南理工大学 | Sparse-decomposition-based hybrid fault feature extraction method of gear wheel and bearing |
CN106441897A (en) * | 2016-10-27 | 2017-02-22 | 石家庄铁道大学 | Morphological component bearing failure diagnosis method based on dictionary study |
CN107543722A (en) * | 2017-08-18 | 2018-01-05 | 西安交通大学 | The Rolling Bearing Fault Character extracting method of dictionary learning is stacked based on depth |
CN108388692A (en) * | 2018-01-17 | 2018-08-10 | 西安交通大学 | Rolling Bearing Fault Character extracting method based on layering sparse coding |
-
2018
- 2018-09-04 CN CN201811027613.9A patent/CN109060350A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06186136A (en) * | 1992-12-18 | 1994-07-08 | Ono Sokki Co Ltd | Fault diagnostic apparatus for ball-and-roller bearing |
CN105241666A (en) * | 2015-09-21 | 2016-01-13 | 华南理工大学 | Rolling bearing fault feature extraction method based on signal sparse representation theory |
CN105424359A (en) * | 2015-11-25 | 2016-03-23 | 华南理工大学 | Sparse-decomposition-based hybrid fault feature extraction method of gear wheel and bearing |
CN106441897A (en) * | 2016-10-27 | 2017-02-22 | 石家庄铁道大学 | Morphological component bearing failure diagnosis method based on dictionary study |
CN107543722A (en) * | 2017-08-18 | 2018-01-05 | 西安交通大学 | The Rolling Bearing Fault Character extracting method of dictionary learning is stacked based on depth |
CN108388692A (en) * | 2018-01-17 | 2018-08-10 | 西安交通大学 | Rolling Bearing Fault Character extracting method based on layering sparse coding |
Non-Patent Citations (1)
Title |
---|
苗中华等: "基于稀疏编码的振动信号特征提取算法与实验研究", 《振动与冲击》 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110160790A (en) * | 2019-05-14 | 2019-08-23 | 中国地质大学(武汉) | A kind of rolling bearing fault impact signal extracting method and system based on improvement K-SVD |
CN110160790B (en) * | 2019-05-14 | 2020-08-07 | 中国地质大学(武汉) | Rolling bearing fault impact signal extraction method and system based on improved K-SVD |
CN111307455A (en) * | 2020-03-06 | 2020-06-19 | 西南交通大学 | Train bogie bearing fault monitoring method and system based on dictionary learning |
CN111382792A (en) * | 2020-03-09 | 2020-07-07 | 兰州理工大学 | Rolling bearing fault diagnosis method based on double-sparse dictionary sparse representation |
CN111382792B (en) * | 2020-03-09 | 2022-06-14 | 兰州理工大学 | Rolling bearing fault diagnosis method based on double-sparse dictionary sparse representation |
CN113203565A (en) * | 2021-03-25 | 2021-08-03 | 长江大学 | Bearing fault identification method and system based on EEMD sparse decomposition |
CN112988918A (en) * | 2021-04-06 | 2021-06-18 | 中车青岛四方机车车辆股份有限公司 | Bearing fault dictionary construction method, analysis method and system |
CN112988918B (en) * | 2021-04-06 | 2022-10-21 | 中车青岛四方机车车辆股份有限公司 | Bearing fault dictionary construction method, analysis method and system |
CN113295410A (en) * | 2021-05-14 | 2021-08-24 | 上海交通大学 | Bearing fault diagnosis method under variable rotating speed working condition |
CN113468760A (en) * | 2021-07-21 | 2021-10-01 | 中南大学 | Motor weak fault detection method and system based on dictionary learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109060350A (en) | A kind of Rolling Bearing Fault Character extracting method dictionary-based learning | |
Grezmak et al. | Interpretable convolutional neural network through layer-wise relevance propagation for machine fault diagnosis | |
Islam et al. | Automated bearing fault diagnosis scheme using 2D representation of wavelet packet transform and deep convolutional neural network | |
Zheng et al. | Sparse elitist group lasso denoising in frequency domain for bearing fault diagnosis | |
Zhao et al. | Sparse classification based on dictionary learning for planet bearing fault identification | |
Wu et al. | Rub-impact fault diagnosis of rotating machinery based on 1-D convolutional neural networks | |
CN108982107B (en) | It is a kind of based on morphology and it is multiple dimensioned arrangement entropy mean value bearing fault quantify trend diagnosis method | |
Zhang et al. | A new feature extraction approach using improved symbolic aggregate approximation for machinery intelligent diagnosis | |
CN105373700A (en) | Method for mechanical fault diagnosis based on information entropies and evidence theory | |
Han et al. | Fault diagnosis of rolling element bearings based on multiscale dynamic time warping | |
Tang et al. | Feature extraction of multi-sensors for early bearing fault diagnosis using deep learning based on minimum unscented kalman filter | |
Liang et al. | An information-based K-singular-value decomposition method for rolling element bearing diagnosis | |
CN107451515A (en) | A kind of rotating machinery fault recognition method and system | |
CN111678699A (en) | Early fault monitoring and diagnosing method and system for rolling bearing | |
Sun et al. | A novel rolling bearing vibration impulsive signals detection approach based on dictionary learning | |
Li et al. | Intelligent fault diagnosis of aeroengine sensors using improved pattern gradient spectrum entropy | |
Lu et al. | Early fault warning and identification in condition monitoring of bearing via wavelet packet decomposition coupled with graph | |
Deng et al. | Impulse feature extraction method for machinery fault detection using fusion sparse coding and online dictionary learning | |
Tian et al. | Approach for hydraulic pump fault diagnosis based on wpt-svd and svm | |
Yu et al. | Sparse time–frequency representation for the transient signal based on low-rank and sparse decomposition | |
TW201633025A (en) | Diagnostic method for malfunction mode of machine tool main shaft and system thereof | |
Zhang et al. | Review on fault diagnosis on the rolling bearing | |
Zhi et al. | Anti-noise motor fault diagnosis method based on decision tree and the feature extraction methods in the time domain and frequency domain | |
Long et al. | Applications of fractional lower order frequency spectrum technologies to bearing fault analysis | |
Hao et al. | New fusion features convolutional neural network with high generalization ability on rolling 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: 20181221 |