CN106595850A - Mechanical oscillation signal fault analysis method - Google Patents
Mechanical oscillation signal fault analysis method Download PDFInfo
- Publication number
- CN106595850A CN106595850A CN201611188946.0A CN201611188946A CN106595850A CN 106595850 A CN106595850 A CN 106595850A CN 201611188946 A CN201611188946 A CN 201611188946A CN 106595850 A CN106595850 A CN 106595850A
- Authority
- CN
- China
- Prior art keywords
- fault
- mechanical
- signal
- oscillation signal
- frequency
- 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
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a mechanical oscillation signal fault analysis method, and the method comprises the steps: signal preprocessing, signal formal processing, signal feature parameter extraction, and signal fault recognition. The beneficial effects of the invention are that a sample entropy can serve as a mechanical fault feature parameter for extraction through the change of a sample entropy value of a vibration signal; the method carries out the classification and analysis of the frequency of the mechanical fault vibration signal, carries out the fault recognition through employing a support vector machine, and simplifies the recognition process of the mechanical fault.
Description
Technical field
The present invention relates to a kind of fault diagnosis field of mechanical oscillation signal, is related to a kind of failure of mechanical oscillation signal point
Analysis method.
Background technology
The vibration and its characteristic information that plant equipment is produced in motor process is reflection plant equipment and its running status
The main signal of change, obtains, records and analyze these Dynamic Signals by various dynamic testers, is to carry out plant equipment
The main path of condition monitoring and fault diagnosis.Key technology therein is to extract machinery by the analyzing and processing to vibration signal
Fault characteristic information.Therefore, by being analyzed to mechanical oscillation signal, the malfunction of plant equipment can be obtained, and is recognized
Failure therein.
The content of the invention
In view of this, the technical problem to be solved in the present invention is to provide a kind of failure analysis methods of mechanical oscillation signal,
For solving insurmountable problem set forth above.
To reach the effect of above-mentioned technical proposal, the technical scheme is that：A kind of failure of mechanical oscillation signal point
Analysis method, including formal process, the characteristic parameter extraction of signal, the Fault Identification of signal of Signal Pretreatment, signal；
Signal Pretreatment passes through emulation to mechanical oscillation signal contrast experiment, using the noise-reduction method of empirical mode decomposition,
The improvement of the anti-acoustic capability of the noise-reduction method of empirical mode decomposition is illustrated, finally uses it for surveying the drop of mechanical oscillation signal
Make an uproar；
The formal process of signal after Signal Pretreatment, is obtaining the noise reduction form of mechanical oscillation signal, in mistake
Mechanical oscillation signal fault signature extracting method is employed in journey, and the relative wavelet energy fusion of mechanical oscillation signal is existed
It is interior, for the few problem of fault sample number in diagnosis, using support vector machine as Fault Identification grader, and by failure
Diagnostic test, demonstrates the effectiveness of mechanical oscillation signal fault signature extracting method, obtains the event of the mechanical oscillation signal
Barrier feature；
The characteristic parameter extraction of signal, further extracts Sample Entropy, together first from the fault signature of mechanical oscillation signal
Shi Liyong Sample Entropies are analyzed to the mechanical oscillation signal of different faults type, Injured level, as a result show Sample Entropy
The matrix of Fault characteristic parameters as Fault characteristic parameters, can be set up, and carries out the singular value of the matrix of Fault characteristic parameters
Decompose, while carrying out mechanical breakdown characteristic parameter extraction using Fault Identification grader；
The Fault Identification of signal, to classifying to Mechanical Fault Vibration Signals by the size of frequency of vibration, is divided into frequency
The low Mechanical Fault Vibration Signals of high Mechanical Fault Vibration Signals and frequency, the mechanical breakdown vibration letter high to frequency
Number Mechanical Fault Vibration Signals are analyzed using independent component analysis, statistically separate component are obtained, by machinery
Vibration signal is expressed as the linear combination of separate component, wherein the coefficient conduct of the linear combination of separate component
Characteristic vector, and mechanical breakdown identification is carried out with reference to nearest neighbor algorithm, obtain the failure of the high Mechanical Fault Vibration Signals of frequency
Type；The Mechanical Fault Vibration Signals low to frequency carry out independent component analysis, obtain the independence point of different working condition signals
Amount, then carries out mechanical breakdown knowledge as feature using the absolute value of the isolated component of different working condition signals using support vector machine
Do not obtain the fault signature of the low Mechanical Fault Vibration Signals of frequency.
The useful achievement of the present invention is that, by the change of the sample entropy of vibration signal, Sample Entropy can be as mechanical event
The characteristic parameter of barrier is extracted, and to the frequency of Mechanical Fault Vibration Signals classification analysises are carried out, and is entered using support vector machine
The identification of row failure, simplifies the identification process of mechanical breakdown.
Specific embodiment
In order that the technical problem to be solved, technical scheme and beneficial effect become more apparent, below tie
Embodiment is closed, the present invention will be described in detail.It should be noted that specific embodiment described herein is only to explain
The present invention, is not intended to limit the present invention, and the product that can realize said function belongs to equivalent and improvement, is all contained in this
Within bright protection domain.Concrete grammar is as follows：
Embodiment one：Plant equipment due to being affected by factors such as impacts that rotating speed, load and failure are produced, its vibration
Signal often shows strong non-stationary.For the non-stationary signal of vibration signal only understands signal in time domain or frequency domain
Global property is inadequate, it is also desirable to obtain the time dependent situation of signal spectrum of vibration signal.Time-frequency analysis technology is
Convert the signal into and be analyzed in two-dimentional time-frequency domain, be the effective means for analyzing non-stationary signal.Passed through based on vibration signal
In the method for diagnosing faults of different frequency bands temporal signatures after decomposition, the frequency domain character of vibration signal is not accounted for.Based on time-frequency domain
The mechanical oscillation signal method for diagnosing faults of feature can be used for singular value decomposition method in the feature extraction of time-frequency matrix.Mesh
Before, time-frequency analysis technology has been widely used for mechanical fault diagnosis field, comprising Short Time Fourier Transform, wavelet transformation.It is short
When Fourier transformation there is a problem of window function fix, distribution there is a problem of cross term interference, then there is energy in wavelet transformation
The problem that leakage and wavelet basis function are selected.The conversion that it is proposed be it is a kind of new with adaptive Time-Frequency Analysis Method,
Adaptive Time-frequency Decomposition can be carried out according to the local time varying characteristic of signal, be especially suitable for carrying out non-stationary signal
Analysis.Conversion is made up of two parts：If any complicated signal can be decomposed in thousand by empirical mode decomposition method
Accumulate the addition of mode function, and instantaneous frequency is all defined on each.Conversion has become mechanical fault diagnosis field and grinds
The focus studied carefully.At present, based on conversion fault diagnosis major part using decompose obtain in accumulate mode function, internally accumulate mode
Function is analyzed the mechanical fault signature of extraction, using to bearing divided oscillation signal solution, the high frequency that local damage bearing is produced
Amplitude-modulated signal composition is separated as the interior mode function that accumulates, and then obtains its envelope signal with conversion, is extracted by envelope spectrum
Bearing fault characteristics frequency.Accumulate the Energy-Entropy of mode function in obtaining by the use of after Gear Vibration Signal Analysis as feature, it is right
Gear carries out crackle and broken teeth fault diagnosis.Based on the interior fault diagnosis side for accumulateing mode function singular value decomposition and support vector machine
Method, using the interior mode function that accumulates vector matrix is formed, and then carries out singular value decomposition to the matrix, extracts its singular value as event
Barrier characteristic vector, then carries out fault diagnosis with support vector machine.
Embodiment two：Based on the support vector machine of Statistical Learning Theory be at present solve small sample classification problem most
Best method.Support vector machine are that pattern vector is mapped to high dimensional feature based on a kind of very simple but highly effective thought one
Space, constructs the Optimal Separating Hyperplane of " optimum " in this space, as maximum class interval hyperplane.
Traditional support vector machine are commonly used to solve two class classification problems, and there are various faults in mechanical fault diagnosis needs
Differentiate, accordingly, it would be desirable to two class classification problems are generalized to into multicategory classification problem.At present, the structure of multi-class classification support vector machine
Making method mainly has following several：
1) one-to-many svm classifier
Its main thought is that some two classification devices of joint constitute a multi classifier.For N class classification problems, the party
Method needs the sub-classifier SVM for constructing N number of two classes classification.Wherein, all sample marks of i-th sub-classifier i-th classification
A class is designated as, the sample labeling of other all categories is another kind of.Maximum corresponding that classification of output valve is sentenced in N number of grader
The classification broken belonging to sample to be identified.
2) one-to-one svm classifier
This sorting technique main thought：For N class samples, one sub-classifier SVM of training will between every two classes sample
This two classes sample separates, and each sub-classifier SVM is only trained with two class samples, so needs to construct N (N-1)/2 altogether
Sub-classifier.When the classification belonging to new samples is predicted, needs are compared using paired sub-classifier, and one is produced every time
Winner can obtain a classification, finally with the method for ballot determining last output, will sample be input into certain
Grader, if the result of the grader is belonging to jth class, just adds a ticket to jth class, when all graders enter to recognizing sample
After row classification, which kind of other poll at most, decides which kind of new samples belong to.
3) directed acyclic graph classification
The training side that directed acyclic graph sorting technique is combined in the training stage using any two classes sample of One-against-one
Formula, needs construction N (N-1)/2 sub-classifier altogether, but except for the difference that in categorizing process, directed acyclic graph sorting technique
Sub-classifier used is configured to into a directed acyclic diagram form, including an i.e. node and a leaf, each node is one
The individual sub-classifier for carrying out two class classification.When being classified to the sample identified, from the beginning of root node, complete by only need to walking
Classification.Comparative study has been done with to multi-category support vector machines, it is indicated that " one-to-one " sorting technique is better than in actual applications it
Its method.
The when one frequency plane of spectrum is divided into into the time frequency block of an area equation, to carrying out energy normalized, Ran Houfang per block
Time-frequency entropy is defined according to the mode of comentropy.For the matrix of any one row or column linear correlation, by left and right being multiplied respectively to its
One orthogonal matrix enters line translation, original matrix can be converted into into a diagonal matrix, and the singular value number for obtaining is reflected
The number of independent row (column) vector in original matrix.Singular value decomposition has good stability, can preferably portray matrix character
Advantage, has become the important tool of signal processing and analysis of statistical data.At present, singularity value decomposition is obtained widely
Using, such as data compression, signal de-noising, machine state monitoring etc..Based on the time spectrum frequency feature extracting method of singular value decomposition,
Mechanical oscillation signal is decomposed into into several component sums first with method, then line translation is entered to each component and is obtained instantaneously
Frequency and instantaneous amplitude, so as to obtaining the spectrum of bearing vibration signal, the spectral representation frequency distribution of time one of signal integrity.To spectrum
Singular value decomposition is carried out, then the singular value for obtaining is carried out as the characteristic vector of bearing failure diagnosis using support vector machine
Failure modes.
The present invention will be described in detail for above-described embodiment.It should be noted that specific embodiment described herein
Only to explain the present invention, it is not intended to limit the present invention, the product that can realize said function belongs to equivalent and improvement,
It is included within protection scope of the present invention.
The useful achievement of the present invention is that, by the change of the sample entropy of vibration signal, Sample Entropy can be as mechanical event
The characteristic parameter of barrier is extracted, and to the frequency of Mechanical Fault Vibration Signals classification analysises are carried out, and is entered using support vector machine
The identification of row failure, simplifies the identification process of mechanical breakdown.
Claims (1)
1. a kind of failure analysis methods of mechanical oscillation signal, it is characterised in that including Signal Pretreatment, the formal place of signal
Reason, the characteristic parameter extraction of signal, the Fault Identification of signal；
The Signal Pretreatment passes through emulation to the mechanical oscillation signal contrast experiment, using the noise reduction side of empirical mode decomposition
Method, wherein being improved the anti-acoustic capability of the noise-reduction method of the empirical mode decomposition, finally uses it for actual measurement described
The noise reduction of mechanical oscillation signal；
The formal process of the signal after the Signal Pretreatment, is obtaining the noise reduction shape of the mechanical oscillation signal
Formula, employs mechanical oscillation signal fault signature extracting method during this, and by the relatively small of the mechanical oscillation signal
Including wave energy fusion, for the few problem of fault sample number, carried out using support vector machine as Fault Identification grader
Fault diagnosis, obtains the fault signature of the mechanical oscillation signal；
The characteristic parameter extraction of the signal, further extracts sample first from the fault signature of the mechanical oscillation signal
Entropy, while being analyzed to the mechanical oscillation signal of different faults type, Injured level using the Sample Entropy, is tied
Fruit shows that the Sample Entropy can repeatedly be extracted the Sample Entropy to set up the Fault characteristic parameters as Fault characteristic parameters
Matrix, and the matrix singular value decomposition of the Fault characteristic parameters is carried out, while entering using the Fault Identification grader
Row mechanical breakdown characteristic parameter extraction；
The Mechanical Fault Vibration Signals are classified by the Fault Identification of the signal to the size by frequency of vibration, are divided into
The low Mechanical Fault Vibration Signals of the high Mechanical Fault Vibration Signals of frequency and frequency, the institute high to the frequency
State Mechanical Fault Vibration Signals to be analyzed the Mechanical Fault Vibration Signals using independent component analysis, obtain statistically phase
Mutually independent component, is expressed as the linear combination of the separate component, wherein the phase by the mechanical oscillation signal
Mutually the coefficient of the linear combination of independent component brings into the mechanical breakdown characteristic parameter as characteristic vector, and combines most
Nearest neighbor algorithm carries out mechanical breakdown identification, obtains the fault type of the high Mechanical Fault Vibration Signals of the frequency；To institute
Stating the low Mechanical Fault Vibration Signals of frequency carries out independent component analysis, obtains the isolated component of different working condition signals, so
Afterwards using the absolute value of the isolated component of the different working condition signals as feature, with the mechanical breakdown characteristic parameter as reference,
Reusing support vector machine carries out the failure that mechanical breakdown identification obtains the low Mechanical Fault Vibration Signals of the frequency
Type.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611188946.0A CN106595850A (en) | 2016-12-21 | 2016-12-21 | Mechanical oscillation signal fault analysis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611188946.0A CN106595850A (en) | 2016-12-21 | 2016-12-21 | Mechanical oscillation signal fault analysis method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106595850A true CN106595850A (en) | 2017-04-26 |
Family
ID=58600590
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611188946.0A Pending CN106595850A (en) | 2016-12-21 | 2016-12-21 | Mechanical oscillation signal fault analysis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106595850A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106840379A (en) * | 2017-03-08 | 2017-06-13 | 潘小胜 | A kind of failure analysis methods of mechanical oscillation signal |
| CN108957262A (en) * | 2018-07-26 | 2018-12-07 | 深圳友讯达科技股份有限公司 | Signal de-noising method and device |
| CN113295416A (en) * | 2021-05-21 | 2021-08-24 | 中国人民解放军国防科技大学 | Bearing fault classification method and system based on frequency spectrum |
| CN113383215A (en) * | 2018-04-30 | 2021-09-10 | 通用电气公司 | System and process for mode-matched bearing vibration diagnostics |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102539192A (en) * | 2012-01-20 | 2012-07-04 | 北京信息科技大学 | Failure prediction method based on ICA reconstruction |
| CN103743477A (en) * | 2013-12-27 | 2014-04-23 | 柳州职业技术学院 | Mechanical failure detecting and diagnosing method and apparatus |
| CN105157821A (en) * | 2015-08-30 | 2015-12-16 | 华电电力科学研究院 | Rotary machinery vibration fault diagnosis and quantitative analysis method |
-
2016
- 2016-12-21 CN CN201611188946.0A patent/CN106595850A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102539192A (en) * | 2012-01-20 | 2012-07-04 | 北京信息科技大学 | Failure prediction method based on ICA reconstruction |
| CN103743477A (en) * | 2013-12-27 | 2014-04-23 | 柳州职业技术学院 | Mechanical failure detecting and diagnosing method and apparatus |
| CN105157821A (en) * | 2015-08-30 | 2015-12-16 | 华电电力科学研究院 | Rotary machinery vibration fault diagnosis and quantitative analysis method |
Non-Patent Citations (1)
| Title |
|---|
| 赵志宏: "基于振动信号的机械故障特征提取与诊断研究", 《中国优秀博士学位论文全文数据库工程科技Ⅱ辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106840379A (en) * | 2017-03-08 | 2017-06-13 | 潘小胜 | A kind of failure analysis methods of mechanical oscillation signal |
| CN113383215A (en) * | 2018-04-30 | 2021-09-10 | 通用电气公司 | System and process for mode-matched bearing vibration diagnostics |
| CN108957262A (en) * | 2018-07-26 | 2018-12-07 | 深圳友讯达科技股份有限公司 | Signal de-noising method and device |
| CN113295416A (en) * | 2021-05-21 | 2021-08-24 | 中国人民解放军国防科技大学 | Bearing fault classification method and system based on frequency spectrum |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105275833B (en) | CEEMD (Complementary Empirical Mode Decomposition)-STFT (Short-Time Fourier Transform) time-frequency information entropy and multi-SVM (Support Vector Machine) based fault diagnosis method for centrifugal pump | |
| Yu et al. | Weak fault feature extraction of rolling bearings using local mean decomposition-based multilayer hybrid denoising | |
| CN106595850A (en) | Mechanical oscillation signal fault analysis method | |
| CN102661783B (en) | Characteristic extracting method for prediction of rotating mechanical failure trend | |
| CN102721545B (en) | Rolling bearing failure diagnostic method based on multi-characteristic parameter | |
| CN104748961A (en) | Gear fault diagnosis method based on SVD decomposition and noise reduction and correlation EEMD entropy features | |
| Jiang et al. | Using bispectral distribution as a feature for rotating machinery fault diagnosis | |
| CN105758644A (en) | Rolling bearing fault diagnosis method based on variation mode decomposition and permutation entropy | |
| CN105258789B (en) | A kind of extracting method and device of vibration signal characteristics frequency band | |
| CN105004498A (en) | Vibration fault diagnosis method of hydroelectric generating set | |
| CN101949895B (en) | Method for identifying fine crack impact signal of metal deep drawing part | |
| CN110044623A (en) | The rolling bearing fault intelligent identification Method of empirical mode decomposition residual signal feature | |
| CN104330258A (en) | Method for identifying grey relational degree of rolling bearing fault based on characteristic parameters | |
| CN101145280A (en) | Independent component analysis based automobile sound identification method | |
| CN107478418A (en) | A kind of rotating machinery fault characteristic automatic extraction method | |
| CN106908232A (en) | A kind of vibration signal fault recognition method of rotating machinery | |
| CN102539159A (en) | Fault diagnosis method for valve mechanism of diesel engine | |
| CN102901630A (en) | Adaptive redundant lifting wavelet noise reduction analysis-based bearing failure recognition method | |
| CN106771598A (en) | A kind of Adaptive spectra kurtosis signal processing method | |
| CN106840379A (en) | A kind of failure analysis methods of mechanical oscillation signal | |
| Mengyuan et al. | Research of circuit breaker intelligent fault diagnosis method based on double clustering | |
| Huang et al. | Feature extraction for rolling element bearing faults using resonance sparse signal decomposition | |
| CN103530660A (en) | Early diagnosis method for faults of strip steel tension sensor | |
| Qu et al. | Study on the algorithm of vibration source identification based on the optical fiber vibration pre-warning system | |
| Li et al. | A new method of nonlinear feature extraction for multi-fault diagnosis of rotor systems |
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: 20170426 |