CN108801634B - The method and its application of bearing fault characteristics frequency are extracted based on singular value decomposition and the frequency band entropy of optimization - Google Patents
The method and its application of bearing fault characteristics frequency are extracted based on singular value decomposition and the frequency band entropy of optimization Download PDFInfo
- Publication number
- CN108801634B CN108801634B CN201810233116.8A CN201810233116A CN108801634B CN 108801634 B CN108801634 B CN 108801634B CN 201810233116 A CN201810233116 A CN 201810233116A CN 108801634 B CN108801634 B CN 108801634B
- Authority
- CN
- China
- Prior art keywords
- value
- frequency
- kurtosis
- optimization
- change rate
- 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.)
- Active
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
Landscapes
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- General Physics & Mathematics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The present invention relates to a kind of method and its application that bearing fault characteristics frequency is extracted based on singular value decomposition and the frequency band entropy of optimization, belong to mechanical fault diagnosis and field of signal processing.The present invention is based on kurtosis indexs to propose the concept of unusual kurtosis value relative change rate, and determine that SVD reconstructs order using unusual kurtosis value relative change rate, its principle is simple and compared to other methods, using kurtosis value as theoretical basis, with solid theoretical foundation, and the denoising effect more more excellent than other methods can be obtained.The present invention obtains reconstruction signal after SVD reconstruct order determines, analyzes result to the further noise reduction process of reconstruction signal using the bandpass filter of the frequency band entropy design optimization of optimization and obtains good effect.The present invention can effectively extract bearing fault characteristics frequency, be applied to bearing emulation signal and the analysis of practical bearing signal, have wide practicability.
Description
Technical field
The present invention relates to a kind of methods for extracting bearing fault characteristics frequency based on singular value decomposition and the frequency band entropy of optimization
And its application, belong to mechanical fault diagnosis and field of signal processing.
Background technique
Bearing is the core component of machine driven system, and failure is to cause the one of the major reasons of rotating machinery fault.
It therefore, is always the hot and difficult issue of mechanical fault diagnosis to the condition monitoring and fault diagnosis of bearing.Work as rolling bearing
When breaking down, vibration signal contains a large amount of running state information, shows as non-stationary and multi -components modulation
Signal, especially in failure early stage, since modulation source is weak, fault features are usually very faint, and by surrounding devices, environment
Noise jamming causes fault features frequency to be difficult to extract, identify.
The key of bearing failure diagnosis is extraction bearing fault characteristics signal (bearing fault characteristics frequency) from original signal.
Laub AJ proposes SVD theory, and describes some applications and algorithm realization of SVD.Lv Zhimin etc. is by singularity value decomposition
For in signal de-noising and fault diagnosis.The reconstruct order of SVD is very big to its influential effect, therefore, how to determine that it reconstructs rank
It is secondary, it is to need the problem of studying;Also, expection is often not achieved in the noise reduction effect of single SVD, it is also necessary to reconstruction signal into
The subsequent further noise reduction process of row.
Summary of the invention
The present invention provides a kind of sides that bearing fault characteristics frequency is extracted based on singular value decomposition and the frequency band entropy of optimization
Method and its application, to obtain reconstruction signal, and filter to the band logical that reconstruction signal optimizes for determining the reconstruct order of SVD
Wave, and then realize the feature extraction of bearing fault, for identifying bearing fault.
The technical scheme is that a kind of extract bearing fault characteristics frequency based on singular value decomposition and the frequency band entropy of optimization
The method of rate, the relative change rate for being primarily based on unusual kurtosis value determine the reconstruct order of SVD, and then obtain the reconstruct letter of SVD
Number;Then to the reconstruction signal progress frequency band entropy analysis of SVD, the centre frequency of adaptive determination bandpass filter and to its band
Width optimizes, the adaptive bandpass filter optimized using the bandwidth Design of optimization;Finally utilize the adaptive band logical of optimization
Filter carries out bandpass filtering to the reconstruction signal of SVD, and carries out Envelope Demodulation Analysis to filtering signal, extracts rolling bearing
Fault characteristic frequency.
The relative change rate based on unusual kurtosis value determines the reconstruct order of SVD, and then obtains the reconstruct letter of SVD
Number, specifically:
Singular value decomposition is carried out to bearing original vibration signal, the kurtosis value of each singular value reconstruct is calculated, according to kurtosis
Value calculates each unusual kurtosis value relative change rate of rank, and obtains unusual kurtosis value relative change rate maximum absolute value value: working as acquirement
Unusual kurtosis value relative change rate maximum value by positive value come when, choose before i order component be reconstructed, acquisition
The reconstruction signal of SVD;When the maximum value of the unusual kurtosis value relative change rate of acquirement is from negative value, i+1 before choosing
Order component is reconstructed, and obtains the reconstruction signal of SVD.
It is described that each unusual kurtosis value relative change rate of rank is calculated according to kurtosis value, specifically:
In formula, dk (i) is the kurtosis value of i order reconstruction signal, and k is singular value number, SVKiAs the i-th order reconstruct letter
Number singular value relative change rate.
The bandwidth optimizes, specifically: corresponding each a value in region of search calculates the kurtosis value of filtering signal,
Compare the kurtosis value size under each a value, bandwidth parameter a corresponding to kurtosis maximum value is determined as optimum bandwidth parameter
a*;And then according to Δ f*≈a*·fs/NwObtain the bandwidth deltaf f of optimization*;The lower limit of region of search is amin, the region of search upper limit is amax,
The step-length used when optimizing bandwidth parameter is S1.
The described search domain upper limit is divided to two kinds:
IfThen haveIt obtains
IfThen haveIt obtains
Wherein, fnFor center frequency, fsFor sample frequency, Δ f is the bandwidth of bandpass filter, NwFor the change of Fourier in short-term
The window changed is long.
The aminIt is 0.1 for 0.1, S1 value.
The method bearing for identification of bearing fault characteristics frequency will be extracted based on the frequency band entropy of singular value decomposition and optimization
Failure.
The beneficial effects of the present invention are:
1, the present invention is based on kurtosis indexs to propose the concept of unusual kurtosis value relative change rate, and uses unusual kurtosis value
Relative change rate determines that SVD reconstructs order, and principle is simple and has compared to other methods using kurtosis value as theoretical basis
Solid theoretical foundation, and the denoising effect more more excellent than other methods can be obtained.
2, the present invention obtains reconstruction signal, utilizes the band of the frequency band entropy design optimization of optimization after SVD reconstruct order determines
Bandpass filter analyzes result and obtains good effect to the further noise reduction process of reconstruction signal.
3, the reconstruct order of the SVD of the relative change rate of unusual kurtosis value of the invention determines method, and the frequency with optimization
The extraction for Rolling Bearing Fault Character frequency is combined with entropy, can effectively extract bearing fault characteristics frequency, is applied
Signal and the analysis of practical bearing signal are emulated in bearing, there is wide practicability.
Detailed description of the invention
Fig. 1 is that the reconstruct order of the SVD of the relative change rate provided by the present invention based on unusual kurtosis value determines method,
And the flow chart of the extraction for Rolling Bearing Fault Character frequency is combined with the frequency band entropy of optimization;
Fig. 2, which is that simulation analysis unusual kurtosis value under different state of signal-to-noise is opposite in Application Example 1 of the present invention, to be changed
The comparison diagram of rate and singular value relative change rate, i.e., in the case where different background noise, the SVD that two methods are chosen reconstructs letter
Number kurtosis value compare;SNR ranges are [- 30,0], step-length 1;
Fig. 3 is emulation signal opposite change of unusual kurtosis value under different state of signal-to-noise in Application Example 1 of the present invention
Rate is compared with reconstruct order selected by singular value relative change rate;SNR ranges are [- 30,0], step-length 1;
Fig. 4 is the relational graph of unusual kurtosis value relative change rate and reconstruct order in Application Example 1 of the present invention;For convenient for
It shows, the relational graph of preceding 100 rank is only provided in figure;
Fig. 5 is unusual kurtosis value relative change rate SVK in Application Example 1 of the present inventioniThe reconstruction signal envelope spectrum of selection,
F in figurerTo turn frequency, fiFor inner ring fault characteristic frequency;
Fig. 6 is singular value relative change rate applied to the relational graph in example 1 with reconstruct order;For convenient for display, in figure
Only provide the relational graph of preceding 50 rank;
Fig. 7 is the reconstruction signal envelope spectrogram that singular value relative change rate chooses;Wherein, frTo turn frequency, fiFor inner ring failure
Characteristic frequency;
Fig. 8 is in Application Example 1 of the present invention, and kurtosis maximum principle optimizes the bandwidth parameter a of bandpass filter, kurtosis
The relational graph ([0.1,10] range is only provided in figure) of value and bandwidth parameter a;
Fig. 9 is after the frequency band entropy of optimization is analyzed, to the envelope spectrogram after reconstruction signal bandpass filtering;F in figurerTo turn frequency, fi
For inner ring fault characteristic frequency;
Figure 10 is the relational graph of unusual kurtosis value relative change rate and reconstruct order in Application Example 2 of the present invention;For just
The relational graph of preceding 100 rank is only provided in display, figure;
Figure 11 is reconstruct signal envelope spectrogram, f in figurerTo turn frequency, fiFor inner ring fault characteristic frequency;
Figure 12 is the bandwidth parameter a of kurtosis maximum principle optimization bandpass filter in Application Example 2 of the present invention, high and steep
The relational graph ([0.1,10] range is only provided in figure) of angle value and bandwidth parameter a;
Figure 13 is the envelope spectrogram after carrying out self-adaptive band-pass filter to reconstruction signal after the frequency band entropy analysis of optimization;Figure
Middle frTo turn frequency, fiFor inner ring fault characteristic frequency;
Figure 14 is the relational graph of unusual kurtosis value relative change rate and reconstruct order in Application Example 3 of the present invention;For just
The relational graph of preceding 100 rank is only provided in display, figure;
Figure 15 is the bandwidth parameter a of kurtosis maximum principle optimization bandpass filter in Application Example 3 of the present invention, high and steep
The relational graph ([0.1,20] range is only provided in figure) of angle value and bandwidth parameter a;
Figure 16 is the packet after carrying out bandpass filtering to the reconstruction signal of the reconstruct order of Figure 14 after the frequency band entropy analysis of optimization
Network spectrogram;F in figureoFor inner ring fault characteristic frequency.
Specific embodiment
Embodiment 1: as shown in figs 1-9, a kind of that bearing fault characteristics are extracted based on singular value decomposition and the frequency band entropy of optimization
The method of frequency, specific step is as follows for the method:
(f is analyzed to bearing inner race fault simulation signal according to process described in foregoing inventionsFor 12000Hz, fnFor
3000Hz), and in Matlab software it is handled.
SVD reconstruct order is chosen based on unusual kurtosis value relative change rate and is made comparisons with singular value relative change rate.Such as
Shown in Fig. 2-3.It under different state of signal-to-noise, is analyzed from the kurtosis index of reconstruction signal, effect acquired by unusual kurtosis value
Fruit is better than singular value relative change rate (or being equal to) substantially, and its reconstruct order value is more steady, and there is no singular value phases
To the biggish fluctuation of appearance selected by change rate.Analyze the SNR ranges [- 30,0] provided, step-length 1.
Step1, Signal to Noise Ratio (SNR)=- 8dB is chosen, above-mentioned emulation signal is analyzed.First with unusual kurtosis value phase
Order is reconstructed to change rate Selection Model, as shown in figure 4, giving relational graph, it is seen then that the reconstruct order of selection is 1 (because obtaining
The maximum value of relative change rate of unusual kurtosis value come by positive value, preceding 1 order component of selection is reconstructed).In Fig. 5
Give corresponding reconstruction signal envelope spectrum.Turn frequency f from wherein can more clearly extractrWith fault characteristic frequency fiWith
Two frequency multiplication 2fi。
Step2, as shown in Figure 6,7, utilize singular value relative change rate choose reconstruct order and its reconstruction signal envelope
Spectrum.It will be appreciated from fig. 6 that the reconstruct order chosen is 2.It can also be extracted in its envelope spectrum and turn frequency frWith fault characteristic frequency fiWith two
Frequency multiplication 2fi.But effect does not have that Fig. 5's is good, because sideband amplitude is greater than failure-frequency amplitude, and its passband noise is thicker
It is close.In addition, the kurtosis value of unusual kurtosis value relative change rate reconstruction signal is 3.4710, and singular value from the point of view of kurtosis index
The kurtosis value of relative change rate's reconstruction signal is 3.1756, therefore, also illustrates its superiority.
Step3, the analysis of frequency band entropy, the centre frequency of adaptive determination bandpass filter are carried out to the reconstruction signal of SVD
And its bandwidth is optimized, the adaptive bandpass filter optimized using the bandwidth Design of optimization.It is former using kurtosis maximum value
Then optimize bandwidth parameter a, and then utilizes formula Δ f*≈a*·fs/Nw, calculate optimum bandwidth Δ f*.Initialize a=0.1, counterweight
Structure signal is analyzed, and is designed bandpass filter, is calculated and save the kurtosis value of filtering signal under this value, then, a=a+0.1
Continue the above analysis, the upper limit until getting aUntil.Compare the kurtosis value size under each a value, it will
Bandwidth parameter a decision bits optimum bandwidth parameter a corresponding to kurtosis maximum value*, as shown in Figure 8, as a=5.6, there is kurtosis
Maximum value 6.133.Therefore optimum bandwidth parameter is a*=5.6.Also, frequency band entropy is analyzed, and a length of N of window can be obtainedw=128, and then can obtain
Optimum bandwidth is Δ f*=525Hz.
Step4, using the bandpass filter of above-mentioned optimization, bandpass filtering is carried out to reconstruction signal.And to filtering signal into
Row Envelope Demodulation Analysis (characteristic frequency for extracting bearing inner race fault simulation signal), envelope spectrum is as shown in Figure 9.
A kind of reconstruct order of the SVD of the relative change rate based on unusual kurtosis value determines method, and the frequency band with optimization
Entropy combines the extraction for Rolling Bearing Fault Character frequency, i.e., extracts bearing fault theory characteristic frequency and envelope spectrum
Fault characteristic frequency is compared, to identify that failure has occurred in bearing inner race.
Above-mentioned implementing procedure, simulation result such as Fig. 4-Fig. 9.Fig. 4-5 is unusual kurtosis value relative change rate and reconstruct order
Relational graph, reconstruction signal envelope spectrogram.Fig. 6-7 is the relational graph of singular value relative change rate and reconstruct order, reconstruction signal
Envelope spectrogram.Fig. 8 is the relational graph of kurtosis value and bandwidth parameter a, obtains maximum kurtosis value in a=5.6.So optimum bandwidth
Coefficient is a*=5.6.Fig. 9 is filtering signal envelope spectrum, can significantly extract fault characteristic frequency fi and its frequency multiplication 2fi, 3fi
With turn frequency fr, it was demonstrated that the present invention extract bearing inner race fault simulation signal fault characteristic frequency, realize the validity of fault identification.
Above-mentioned implementation case study the result shows that, the reconstruct of the SVD of the relative change rate proposed by the invention based on unusual kurtosis value
Order determines method, and combines the extraction for Rolling Bearing Fault Character frequency with the frequency band entropy of optimization, can there is effect
Emulation signal for bearing inner race failure is analyzed, and provides support for practical application.
Embodiment 2: a kind of that bearing is extracted based on singular value decomposition and the frequency band entropy of optimization as shown in Fig. 1 and Figure 10-13
The method of fault characteristic frequency, specific step is as follows for the method:
(f is analyzed to practical bearing inner ring fault-signal according to process described in foregoing inventionsFor 12000Hz, fnFor
2830Hz), and Matlab software analysis result is given.
Step1, determine that SVD reconstructs order first with unusual kurtosis value relative change rate.As shown in Figure 10, pass is given
System figure, it is seen then that the reconstruct order of selection be 2 (because the unusual kurtosis value of acquirement relative change rate maximum value by positive value
, choose preceding 2 order component and be reconstructed).Therefore, with this reconstruction signal and its envelope spectrum is sought.It has been presented in Fig. 11 corresponding
Reconstruction signal envelope spectrum.Turn frequency f from wherein can more clearly extractrWith fault characteristic frequency fiWith two frequency multiplication 2fi。
Using the further Reduction Analysis of following steps.
Step2, the analysis of frequency band entropy, the centre frequency of adaptive determination bandpass filter are carried out to the reconstruction signal of SVD
And its bandwidth is optimized, the adaptive bandpass filter optimized using the bandwidth Design of optimization.It is former using kurtosis maximum value
Then optimize bandwidth parameter a, and then utilizes formula Δ f*≈a*·fs/Nw, calculate optimum bandwidth Δ f*.Initialize a=0.1, counterweight
Structure signal is analyzed, and is designed bandpass filter, is calculated and save the kurtosis value of filtering signal under this value, then, a=a+0.1
Continue the above analysis, the upper limit until getting aUntil.Compare the kurtosis value size under each a value,
By bandwidth parameter a decision bits optimum bandwidth parameter a corresponding to kurtosis maximum value*, as shown in Figure 12, as a=3.9, have
Kurtosis maximum value 5.494.Therefore optimum bandwidth parameter is a*=3.9.Also, frequency band entropy is analyzed, and a length of N of window can be obtainedw=64, in turn
It is Δ f that optimum bandwidth, which can be obtained,*=731.25Hz.
Step3, using the adaptive bandpass filter of above-mentioned optimization, self-adaptive band-pass filter is carried out to reconstruction signal.And
Envelope Demodulation Analysis (characteristic frequency for extracting bearing inner race fault-signal), envelope spectrum such as Figure 13 institute are carried out to filtering signal
Show.From the extraction bearing inner race fault characteristic frequency f that can be wherein apparentiAnd frequency multiplication 2fi.By bearing fault theory characteristic frequency
Rate is compared with the fault characteristic frequency that envelope spectrum extracts, and can determine that bearing operating status.
According to above-mentioned implementing procedure, the simulation case result such as Figure 10-Figure 13 can get.Figure 10 is that unusual kurtosis value is opposite
The relational graph of change rate and reconstruct order determines the reconstruct order of SVD for 2 with this.Figure 11 is the envelope spectrogram of reconstruction signal, can
Extract bearing inner race fault characteristic frequency fi, turn frequency fr, but still surrounded by noise.Figure 12 is the relationship of kurtosis value and bandwidth parameter a
Figure obtains maximum kurtosis value in a=3.9.So optimum bandwidth parameter is a*=3.9, with the adaptive band logical of this design optimization
Filter.Figure 13 is envelope spectrum of the reconstruction signal after self-adaptive band-pass filter, and failure spy can be clearly extracted in figure
It levying frequency fi and its frequency multiplication 2fi, 3fi and turns frequency fr, it was demonstrated that the present invention extracts the fault characteristic frequency of bearing outer ring fault-signal,
Realize that bearing fault identification is effective.Above-mentioned implementation case study the result shows that, it is proposed by the invention based on unusual kurtosis
The reconstruct order of the SVD of the relative change rate of value determines method, and combines with the frequency band entropy of optimization for rolling bearing fault
The extraction of characteristic frequency, can be effectively applied to practical bearing inner ring fault signal analysis (it is special to extract bearing inner race failure
Frequency is levied, realizes accurately identifying for bearing fault), there is practical application value.
Embodiment 3: as illustrated in figures 14-16, a kind of that bearing fault spy is extracted based on singular value decomposition and the frequency band entropy of optimization
The method for levying frequency, specific step is as follows for the method:
(f is analyzed to practical bearing outer ring fault-signal according to process described in foregoing inventionsFor 25600Hz, fnFor
8148Hz), and Matlab software analysis result is given.F shown in figurerTurn frequency, f for bearingoFor bearing outer ring fault signature
Frequency.
Step1, determine that SVD reconstructs order first with unusual kurtosis value relative change rate.As shown in figure 14, pass is given
System figure, it is seen then that the reconstruct order of selection be 1 (because the unusual kurtosis value of acquirement relative change rate maximum value by positive value
, choose preceding 1 order component and be reconstructed).
Step2, the analysis of frequency band entropy, the centre frequency of adaptive determination bandpass filter are carried out to the reconstruction signal of SVD
And its bandwidth is optimized, the adaptive bandpass filter optimized using the bandwidth Design of optimization.It is former using kurtosis maximum value
Then optimize bandwidth parameter a, and then utilizes formula Δ f*≈a*·fs/Nw, calculate optimum bandwidth Δ f*.Initialize a=0.1, counterweight
Structure signal is analyzed, and is designed bandpass filter, is calculated and save the kurtosis value of filtering signal under this value, then, a=a+0.1
Continue the above analysis, the upper limit until getting aUntil.Compare the kurtosis value under each a value
Size, by bandwidth parameter a decision bits optimum bandwidth parameter a corresponding to kurtosis maximum value*, as shown in Figure 15, as a=14,
With kurtosis maximum value 29.7011.Therefore optimum bandwidth parameter is a*=14.Also, frequency band entropy is analyzed, and a length of N of window can be obtainedw=64,
And then can obtain optimum bandwidth is Δ f*=5600Hz.
Step3, using the adaptive bandpass filter of above-mentioned optimization, self-adaptive band-pass filter is carried out to reconstruction signal.And
Envelope Demodulation Analysis (characteristic frequency for extracting bearing inner race fault-signal), envelope spectrum such as Figure 16 institute are carried out to filtering signal
Show.The extraction bearing outer ring fault characteristic frequency f that can be therefrom apparentiAnd several frequency multiplication nfi.By bearing fault theory characteristic
Frequency is compared with the fault characteristic frequency that envelope spectrum extracts, and can determine that bearing operating status.
Figure 14 is the relational graph of unusual kurtosis value relative change rate and reconstruct order, and the reconstruct order of SVD is determined with this
It is 1.Figure 15 is the relational graph of kurtosis value Yu bandwidth parameter a, obtains maximum kurtosis value in a=14.So optimum bandwidth parameter
For a*=14, with the adaptive bandpass filter of this design optimization.Figure 16 is reconstruction signal after self-adaptive band-pass filter
Envelope spectrum can clearly extract fault characteristic frequency fi and its frequency multiplication 2fi, 3fi etc. in figure, it was demonstrated that the present invention extracts practical
The fault characteristic frequency of bearing outer ring fault-signal realizes that bearing fault identification is feasible.Above-mentioned implementation case study result
Show that the reconstruct order of the SVD of the relative change rate proposed by the invention based on unusual kurtosis value determines method, and with optimization
Frequency band entropy combine the extraction for Rolling Bearing Fault Character frequency, practical bearing outer ring failure can be effectively applied to
Signal analyzes (can extract bearing outer ring fault characteristic frequency, realize accurately identifying for bearing fault), has practical application value,
And there is engineering adaptability.
Embodiment 4: a method of bearing fault characteristics frequency is extracted based on singular value decomposition and the frequency band entropy of optimization, it is first
First the relative change rate based on unusual kurtosis value determines the reconstruct order of SVD, and then obtains the reconstruction signal of SVD;Then to SVD
Reconstruction signal carry out the analysis of frequency band entropy, the centre frequency of adaptive determination bandpass filter simultaneously optimizes its bandwidth,
The adaptive bandpass filter optimized using the bandwidth Design of optimization;Finally using the adaptive bandpass filter of optimization to SVD
Reconstruction signal carry out bandpass filtering, and to filtering signal carry out Envelope Demodulation Analysis, extract rolling bearing fault signature frequency
Rate.
It is possible to further which the reconstruct order for determining SVD based on the relative change rate of unusual kurtosis value is arranged, in turn
The reconstruction signal of SVD is obtained, specifically:
Singular value decomposition is carried out to bearing original vibration signal, the kurtosis value of each singular value reconstruct is calculated, according to kurtosis
Value calculates each unusual kurtosis value relative change rate of rank, and obtains unusual kurtosis value relative change rate maximum absolute value value: acquirement
When the maximum value of unusual kurtosis value relative change rate is from negative value, i+1 order component is reconstructed before choosing, and obtains SVD
Reconstruction signal.
It is possible to further be arranged it is described each unusual kurtosis value relative change rate of rank is calculated according to kurtosis value, specifically:
In formula, dk (i) is the kurtosis value of i order reconstruction signal, and k is singular value number, SVKiAs the i-th order reconstruct letter
Number singular value relative change rate.
It is optimized it is possible to further which the bandwidth is arranged, specifically: corresponding each a value in region of search calculates
The kurtosis value of filtering signal compares the kurtosis value size under each a value, bandwidth parameter a corresponding to kurtosis maximum value is sentenced
It is set to optimum bandwidth parameter a*;And then according to Δ f*≈a*·fs/NwObtain the bandwidth deltaf f of optimization*;The lower limit of region of search is amin,
The region of search upper limit is amax, the step-length used when optimizing bandwidth parameter is S1.
It is divided to two kinds it is possible to further which the described search domain upper limit is arranged:
IfThen haveIt obtains
IfThen haveIt obtains
Wherein, fnFor center frequency, fsFor sample frequency, Δ f is the bandwidth of bandpass filter, NwFor the change of Fourier in short-term
The window changed is long.
It is possible to further which a is arrangedminIt is 0.1 for 0.1, S1 value.
The method bearing for identification of bearing fault characteristics frequency will be extracted based on the frequency band entropy of singular value decomposition and optimization
Failure.
Above in conjunction with attached drawing, the embodiment of the present invention is explained in detail, but the present invention is not limited to above-mentioned
Embodiment within the knowledge of a person skilled in the art can also be before not departing from present inventive concept
Put that various changes can be made.
Claims (5)
1. the method for extracting bearing fault characteristics frequency based on singular value decomposition and the frequency band entropy of optimization, it is characterised in that: first
The reconstruct order of SVD is determined based on the relative change rate of unusual kurtosis value, and then obtains the reconstruction signal of SVD;Then to SVD's
Reconstruction signal carries out the analysis of frequency band entropy, and the centre frequency of adaptive determination bandpass filter simultaneously optimizes its bandwidth, adopts
The adaptive bandpass filter optimized with the bandwidth Design of optimization;Finally using the adaptive bandpass filter of optimization to SVD's
Reconstruction signal carries out bandpass filtering, and carries out Envelope Demodulation Analysis to filtering signal, extracts the fault characteristic frequency of rolling bearing;
The bandwidth optimizes, specifically: corresponding each a value in region of search calculates the kurtosis value of filtering signal, compares
Bandwidth parameter a corresponding to kurtosis maximum value is determined as optimum bandwidth parameter a by the kurtosis value size under each a value*;Into
And according to Δ f*≈a*·fs/NwObtain the bandwidth deltaf f of optimization*;The lower limit of region of search is amin, the region of search upper limit is amax, optimization
The step-length used when bandwidth parameter is S1;
The described search domain upper limit is divided to two kinds:
IfThen haveIt obtains
IfThen haveIt obtains
Wherein, fnFor center frequency, fsFor sample frequency, Δ f is the bandwidth of bandpass filter, NwFor Short Time Fourier Transform
Window is long.
2. the side according to claim 1 for extracting bearing fault characteristics frequency based on singular value decomposition and the frequency band entropy of optimization
Method, it is characterised in that: the relative change rate based on unusual kurtosis value determines the reconstruct order of SVD, and then obtains the weight of SVD
Structure signal, specifically:
Singular value decomposition is carried out to bearing original vibration signal, the kurtosis value of each singular value reconstruct is calculated, according to kurtosis value meter
The unusual kurtosis value relative change rate of each rank is calculated, and obtains unusual kurtosis value relative change rate maximum absolute value value: when the surprise of acquirement
When the maximum value of the relative change rate of different kurtosis value is from positive value, i order component is reconstructed before choosing, and obtains SVD's
Reconstruction signal;When the maximum value of the unusual kurtosis value relative change rate of acquirement is from negative value, i+1 order component before choosing
It is reconstructed, obtains the reconstruction signal of SVD.
3. the side according to claim 2 for extracting bearing fault characteristics frequency based on singular value decomposition and the frequency band entropy of optimization
Method, it is characterised in that: it is described that each unusual kurtosis value relative change rate of rank is calculated according to kurtosis value, specifically:
In formula, dk (i) is the kurtosis value of i order reconstruction signal, and k is singular value number, SVKiAs the i-th order reconstruction signal
Singular value relative change rate.
4. the side according to claim 1 for extracting bearing fault characteristics frequency based on singular value decomposition and the frequency band entropy of optimization
Method, it is characterised in that: aminIt is 0.1 for 0.1, S1 value.
5. extracting bearing fault characteristics based on singular value decomposition and the frequency band entropy of optimization for of any of claims 1-4
The method of frequency bearing fault for identification.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810233116.8A CN108801634B (en) | 2018-03-21 | 2018-03-21 | The method and its application of bearing fault characteristics frequency are extracted based on singular value decomposition and the frequency band entropy of optimization |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810233116.8A CN108801634B (en) | 2018-03-21 | 2018-03-21 | The method and its application of bearing fault characteristics frequency are extracted based on singular value decomposition and the frequency band entropy of optimization |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108801634A CN108801634A (en) | 2018-11-13 |
CN108801634B true CN108801634B (en) | 2019-12-03 |
Family
ID=64095257
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810233116.8A Active CN108801634B (en) | 2018-03-21 | 2018-03-21 | The method and its application of bearing fault characteristics frequency are extracted based on singular value decomposition and the frequency band entropy of optimization |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108801634B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109813547B (en) * | 2019-01-08 | 2020-09-22 | 华南理工大学 | Rotary machine local fault diagnosis method based on sparse decomposition optimization algorithm |
CN109946081B (en) * | 2019-04-15 | 2020-09-22 | 北京航空航天大学 | Fault diagnosis method for rolling bearing slipping under variable rotating speed |
CN112067297B (en) * | 2020-09-23 | 2021-06-22 | 昆明理工大学 | Bearing fault feature extraction method |
CN114441111B (en) * | 2022-01-29 | 2023-09-29 | 常州大学 | Pipeline leakage infrasonic wave signal analysis method and system |
CN114775382B (en) * | 2022-06-21 | 2022-09-06 | 源利腾达(西安)科技有限公司 | Ultrasonic-based road surface quality detection method for highway traffic engineering |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7756649B2 (en) * | 2007-09-06 | 2010-07-13 | Schenck Rotec Gmbh | Method for fault detection and diagnosis of a rotary machine |
CN103940612B (en) * | 2014-04-10 | 2016-05-25 | 昆明理工大学 | A kind of Rolling Bearing Fault Character extracting method and system |
CN104408302A (en) * | 2014-11-19 | 2015-03-11 | 北京航空航天大学 | Bearing variable-condition fault diagnosis method based on LMD-SVD (Local Mean Decomposition-Singular Value Decomposition) and extreme learning machine |
CN105954031B (en) * | 2016-06-29 | 2018-03-13 | 潍坊学院 | A kind of envelope Analysis Method based on unusual spectral factorization filtering |
-
2018
- 2018-03-21 CN CN201810233116.8A patent/CN108801634B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN108801634A (en) | 2018-11-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108801634B (en) | The method and its application of bearing fault characteristics frequency are extracted based on singular value decomposition and the frequency band entropy of optimization | |
CN109682601B (en) | Early fault identification method for rolling bearing under variable rotating speed working condition | |
Li et al. | An enhanced rolling bearing fault detection method combining sparse code shrinkage denoising with fast spectral correlation | |
Luo et al. | A kurtosis-guided adaptive demodulation technique for bearing fault detection based on tunable-Q wavelet transform | |
CN104236908B (en) | Combined slicing bearing fault diagnosis method on basis of MID (modulation intensity distribution) algorithm | |
CN107991706B (en) | Coal seam hydraulic fracturing microseismic signal combined noise reduction method based on wavelet packet multiple threshold and improved empirical mode decomposition | |
CN109145727A (en) | A kind of bearing fault characteristics extracting method based on VMD parameter optimization | |
Djebala et al. | Rolling bearing fault detection using a hybrid method based on empirical mode decomposition and optimized wavelet multi-resolution analysis | |
Li et al. | Research on test bench bearing fault diagnosis of improved EEMD based on improved adaptive resonance technology | |
Zhang et al. | Energy operator demodulating of optimal resonance components for the compound faults diagnosis of gearboxes | |
CN109781412B (en) | Rolling bearing self-adaptive resonance demodulation method based on EEMD | |
Liu et al. | Flexible generalized demodulation for intelligent bearing fault diagnosis under nonstationary conditions | |
CN107957566A (en) | Magnetic resonance depth measurement method for extracting signal based on frequency selection singular spectrum analysis | |
Han et al. | Roller bearing fault diagnosis based on LMD and multi-scale symbolic dynamic information entropy | |
CN108051857B (en) | A kind of local similarity wavelet threshold denoising method based on empirical mode decomposition | |
Sun et al. | Fault detection of rolling bearing using sparse representation-based adjacent signal difference | |
CN109374298B (en) | Bearing fault diagnosis method based on cross-correlation singular value | |
Yao et al. | Weak fault feature extraction of rotating machinery based on double-window spectrum fusion enhancement | |
CN108051659B (en) | Method for separating and extracting rotor noise | |
Chi et al. | Bearing fault diagnosis through stochastic resonance by full-wave signal construction with half-cycle delay | |
Wu et al. | A study on vibration component separation of a rotor system during startup and its application in fault diagnosis | |
CN112485028A (en) | Vibration signal characteristic frequency spectrum extraction method and mechanical fault diagnosis analysis method | |
Yang et al. | Research on Fault Feature Extraction Method Based on FDM‐RobustICA and MOMEDA | |
CN116840153A (en) | Cultural relic crack detection system with fault diagnosis function | |
Zhao et al. | Variable step adaptive kurtogram method based on empirical wavelet transform for 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |