CN108181098A - A kind of high pedestal jib crane low-speed heave-load unit failure feature extracting method - Google Patents
A kind of high pedestal jib crane low-speed heave-load unit failure feature extracting method Download PDFInfo
- Publication number
- CN108181098A CN108181098A CN201711350955.XA CN201711350955A CN108181098A CN 108181098 A CN108181098 A CN 108181098A CN 201711350955 A CN201711350955 A CN 201711350955A CN 108181098 A CN108181098 A CN 108181098A
- Authority
- CN
- China
- Prior art keywords
- singular value
- speed heave
- low
- dual
- signal
- 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M7/00—Vibration-testing of structures; Shock-testing of structures
- G01M7/02—Vibration-testing by means of a shake table
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The present invention relates to a kind of high pedestal jib crane low-speed heave-load unit failure feature extracting methods.This method is by the way that the vibration signal of the high pedestal jib crane low-speed heave-load component of acquisition is decomposed by dual-tree complex wavelet packet, it is decomposed into the component of different frequency range, obtain the corresponding dual-tree complex wavelet packet decomposition coefficient of different frequency range, singular value decomposition is carried out to the dual-tree complex wavelet packet decomposition coefficient of the frequency band comprising fault characteristic information, seek singular value difference spectral curve, the number of singular value reconstruct component is selected according to its maximum sudden change point, then singular value reconstruct is carried out to it, noise reduction is carried out to component, then dual-tree complex wavelet reconstruct is carried out to it, the Hilbert envelope of reconstruction signal is asked to compose again, judge that the low-speed heave-load component whether there is failure according to envelope spectrum.The present invention solves the problems, such as that existing high pedestal jib crane low-speed heave-load component ultra-long time vibration signal can not carry out noise reduction using singular value decomposition and fault signature extracts.
Description
Technical field
The invention belongs to harbour facilities high pedestal jib crane fault diagnosis technology field, more particularly to a kind of seat type
Crane low-speed heave-load unit failure feature extracting method.
Background technology
In high pedestal jib crane, there is more low-speed heave-load critical component, as hoisting mechanism, jib lubbing mechanism and
Rotating mechanism etc., when initial failure occurs for these low-speed heave-load critical components, the active ingredient in vibration signal is extremely micro-
It is weak, it is often submerged in strong ambient noise and crane irregular oscillation.Therefore, how to be detected under strong noise background
Effective small-signal characteristic information, it has also become one of critical issue of high pedestal jib crane low-speed heave-load component fault diagnosis.
However, the distinctive low turn of frequency characteristic of low-speed heave-load device, more increases the difficulty for having impulse fault feature extraction.
Extraction will be unable to carry out fault diagnosis less than equipment fault characteristic frequency, when major accident is had no to discover temporarily extremely to endanger
Danger.When low-speed heave-load device carries out fault diagnosis, especially in the process that fault diagnosis is carried out to low-speed heavy-loaded gear
In, since the frequency that turns of slow-speed shaft is generally several hertz to several hertz of zero, the meshing frequency of slow speed turbine stage is generally tens hertz,
The interval time for leading to each failure impact is longer, is difficult to detect fault-signal near slow-speed shaft, can even detect
Fault-signal since it is extremely faint, can consumingly be interfered by noise, it is difficult to carry out comprehensive accident analysis to failure.In work
In Cheng Yingyong, the technology and methods for reliably extracting low-speed heave-load device vibratory impulse feature are less, lead to low-speed heave-load device
During failure hidden danger, it is difficult to find, serious accident frequently occurs.
Signal processing method based on singular value decomposition (Singular Value Decomposition, SVD) is a kind of
The non-linear filtering method of classical signal processing method is different from, has extraordinary processing for non-linear and non-stationary signal
Effect carries out singular value decomposition by structural matrix, chooses suitable singular value reconstruct, so that it may de-noising, and nothing are carried out to signal
Phase offset.
But since equipment rotating speed is low, to collect the low-frequency component of the signal, the frequency discrimination of signal to be ensured
Rate, it is necessary to carry out the vibration signals collecting of ultra-long time, and very noisy is all carried in signal.Event is carried out for this signal
Barrier diagnosis when, can not directly carry out noise reduction using traditional singular value decomposition, because of the limitation of calculator memory, signal into
The length of row singular value decomposition is limited, and the length of this signal has been over computer and allows signal progress unusual
It is worth the length decomposed.
Invention content
The purpose of the present invention is to provide a kind of high pedestal jib crane low-speed heave-load unit failure feature extracting methods, should
When method utilizes dual-tree complex wavelet packet decomposition vibration signal, the decomposition coefficient of each frequency range can be passed with the increase of Decomposition order
Subtract, so as to provide possibility to carry out ultra-long time vibration signal noise reduction using singular value decomposition, overcome existing singular value point
Solution method can not detect the defects of Weak characteristic information in ultra-long time signal.
To achieve these goals, the technical solution adopted by the present invention is as follows：
A kind of high pedestal jib crane low-speed heave-load unit failure feature extracting method, includes the following steps：
1) with sample frequency fsData acquisition is carried out to the vibration signal of high pedestal jib crane low-speed heave-load component, will be adopted
Original vibration signal x (t) the input dual-tree complex wavelet packet algorithms collected, carry out n-layer decomposition, by original vibration signal point to it
Solve is 2nA frequency range, the frequency range of each frequency range is respectively [0, fs/2n]、 [fs/2n,2×fs/2n]、[2×fs/2n,3×fs/
2n]、…、[(2n-1)×fs/2n,2n×fs/2n]；
2) the tach signal r of high pedestal jib crane low-speed heave-load component is acquired, according to the number of gear teeth of low-speed heave-load component
Z calculates one times of meshing frequency of low-speed heave-load component
3) judge frequency range residing for one times of meshing frequency, selection includes the corresponding dual-tree complex wavelet of frequency range of the meshing frequency
Packet decomposition coefficient constructs Hankel matrix As, and carries out singular value decomposition, obtains A=UDVT, D is diagonal matrix；
4) the element σ in diagonal matrix D on diagonal is extracted1≥σ2≥......σq>=0, form unusual value sequence；
5) b is asked for respectively to unusual value sequencei=σi-σi+1(i=1,2 ..., q-1), by all biSequence B=
(b1,b2,......,bq-1) it is known as the Difference Spectrum of singular value；
6) the maximum sudden change point in singular value Difference Spectrum is checked, by the corresponding serial number of maximum sudden change point as singular value weight
The component number retained is needed during structure, singular value reconstruct is carried out, obtains updated dual-tree complex wavelet packet coefficient；
7) updated dual-tree complex wavelet packet coefficient is utilized, reconstructs vibration signal, realizes the noise reduction of vibration signal；
8) Hilbert envelope demodulations are carried out to the vibration signal of reconstruct, judges that the low-speed heave-load component is according to envelope spectrum
It is no that there are failures.
The present invention, the singular value decomposition are to carry out singular value point to the dual-tree complex wavelet packet decomposition coefficient of certain frequency range
It solves rather than vibration signal is directly decomposed, compressible calculation amount, vibrated so as to fulfill low-speed heave-load component ultra-long time
The singular value decomposition and noise reduction of signal, extract its fault signature.
When the present invention utilizes dual-tree complex wavelet packet decomposition vibration signal, the decomposition coefficient of each frequency range can be with Decomposition order
Increase and the characteristic successively decreased, ultra-long time vibration signal can be compressed in less decomposition coefficient, so as to for using strange
Different value decomposes progress ultra-long time vibration signal noise reduction and provides possibility, so as to detect high pedestal jib crane low-speed heave-load portion
Part early-stage weak fault characteristic signal, finds potential faults ahead of time, needs to acquire overlength suitable for extraction low-speed heave-load component
The occasion of time vibration signal, and then judge whether low-speed heave-load device breaks down.
Description of the drawings
Fig. 1 is the overall work schematic diagram of the high pedestal jib crane low-speed heave-load unit failure feature extraction of the present invention；
The dual-tree complex wavelet packet that Fig. 2 is the present invention decomposes and the flow chart of reconstruct；
Fig. 3 be the present invention realize constructed with y=[0.3+0.21 × sin (2 π × 0.25t)] × sin (2 π × 50t)
Amplitude-modulated signal time domain waveform and its amplitude spectrum；
Fig. 4 be the present invention realize constructed with y=[0.3+0.21 × sin (2 π × 0.25t)] × sin (2 π × 50t)
Amplitude-modulated signal amplitude spectrum partial enlarged view；
Fig. 5 be the present invention realize constructed with y=[0.3+0.21 × sin (2 π × 0.25t)] × sin (2 π × 50t)
Amplitude-modulated signal addition noise after time domain waveform and its amplitude spectrum；
Fig. 6 is four frequencies that 2 layers of dual-tree complex wavelet packet are carried out to signal shown in Fig. 5 and are decomposed that the present invention realizes
The wavelet coefficient of section；
Fig. 7 carries out singular value decomposition for what the present invention realized to first frequency range dual-tree complex wavelet packet decomposition coefficient in Fig. 6
The singular value Difference Spectrum obtained afterwards；
Fig. 8 is the drop that 6 singular values reconstruct before the singular value Difference Spectrum according to Fig. 7 that the present invention realizes selects
Signal and its amplitude spectrum after making an uproar；
Fig. 9 is the partial enlarged view of Signal Amplitude after noise reduction shown in Fig. 8 of the invention realized；
Figure 10 is the envelope spectrum of signal after noise reduction shown in Fig. 8 of the invention realized；
Figure 11 is that direct decomposed using dual-tree complex wavelet packet that the present invention realizes carries out obtained by noise reduction signal shown in Fig. 5
Time domain waveform and its amplitude spectrum；
Figure 12 is that direct decomposed using dual-tree complex wavelet packet that the present invention realizes carries out obtained by noise reduction signal shown in Fig. 5
The amplitude spectrum partial enlarged view of signal
Figure 13 is that direct decomposed using dual-tree complex wavelet packet that the present invention realizes carries out obtained by noise reduction signal shown in Fig. 5
The envelope spectrum of signal；
Figure 14 is the low-speed heave-load component original vibration signal and its amplitude spectrum of actual measurement that the present invention realizes；
Figure 15 is that 4 layers of the dual-tree complex wavelet packet that carried out to vibration signal shown in Figure 14 that the present invention realizes decomposes, the 1st frequency range
The singular value Difference Spectrum of dual-tree complex wavelet packet decomposition coefficient；
Figure 16 be the time domain waveform obtained using the reconstruction signal of singular value Difference Spectrum shown in Figure 15 realized of the present invention and its
Amplitude spectrum；
Figure 17 is that the amplitude spectrum obtained using the reconstruction signal of singular value Difference Spectrum shown in Figure 15 that the present invention realizes locally is put
Big figure；
Figure 18 is the envelope spectrum obtained using the reconstruction signal of singular value Difference Spectrum shown in Figure 15 that the present invention realizes.
Specific embodiment
In the following, the substantive distinguishing features and advantage of the present invention are further described with reference to example, but not office of the invention
It is limited to listed embodiment.
It is shown in Figure 1, a kind of high pedestal jib crane low-speed heave-load unit failure feature extracting method, including following step
Suddenly：
1) in the present invention, first with piezoelectric type vibration acceleration transducer pickup high pedestal jib crane low-speed heave-load portion
The original vibration signal of part.
The piezoelectric type vibration acceleration transducer may be mounted at the hoisting mechanism of high pedestal jib crane, jib lubbing mechanism
On the critical components such as rotating mechanism, the critical component can be such as the retarder of hoisting mechanism and cylinder support bearing
Seat, the retarder of jib lubbing mechanism and the bearing block of rack-and-pinion, the large-size pivoting support of rotating mechanism and driving pinion etc.
On.
In the vibration signal that high pedestal jib crane is extracted using piezoelectric type vibration acceleration transducer, acquired by data
Instrument will carry out analog-to-digital conversion after signal amplification filtering, so as to obtain digital signal, can store into computer by the present invention
Method and step, handled automatically by computer program.
2) collected original vibration signal x (t) is inputted into dual-tree complex wavelet packet algorithm, n-layer decomposition is carried out to it, it will
Original vibration signal is decomposed into 2nA frequency range, the frequency range of each frequency range is respectively [0, fs/2n]、 [fs/2n,2×fs/2n]、[2
×fs/2n,3×fs/2n]、…、[(2n-1)×fs/2n,2n×fs/2n]。
It should be noted that double-tree complex wavelet package transforms (dual-tree complex wavelet of the present invention
Packet transform, DT-CWPT) be based on developed on the basis of dual-tree complex wavelet transform can simultaneously to low frequency and
The signal processing method that high-frequency signal is finely divided.
First, dual-tree complex wavelet transform is a kind of with approximate translation invariance, good set direction, anti-frequency
The method of the New Wavelet Transforms of a variety of good characteristics such as aliasing characteristic, higher computational efficiency and Accurate Reconstruction.Double trees are multiple
Wavelet transformation is brought using two parallel and using different low pass and high-pass filter discrete wavelet transformers and realizes signal
It decomposes and reconstructs, be referred to as real part tree and imaginary part tree, it is equivalent to a signal while makees two wavelet transforms,
The wavelet filter of the two wavelet transformations is special designing, they form mutually Hilbert transformation pair.
According to the building method of dual-tree complex wavelet transform, Phase information can be expressed as
ψ (t)=ψh(t)+iψg(t) (1)
ψg(t)=H { ψh(t)} (2)
Or
ψ in formulah(t), ψg(t) the real small echo of two approximations Hilbert transformation each other is represented, H [] represents that Hilbert becomes
Conversion；I is complex unit.
Since the real valued wavelet transform of the transformation of Hilbert each other of dual-tree complex wavelet transform by two independences and parallel forms,
In order to be distinguished, it is respectively designated as real part tree (Re) and imaginary part tree (Im).Therefore, according to Wavelet Analysis Theory, double trees are answered small
The wavelet coefficient and scale coefficient of the real part tree wavelet transformation of wave conversion have no difference with traditional real valued wavelet transform, can be by formula (4)
And formula (5) is calculated：
Similarly, the wavelet coefficient of the imaginary part tree wavelet transformation of dual-tree complex wavelet transform and scale coefficient can by formula (6) and
Formula (7) calculates
Therefore, the wavelet coefficient and scale coefficient of DT-CWT can be obtained：
It finally, can be multiple small to the double trees of signal progress by the dual-tree complex wavelet restructing algorithm in formula (10), (11) and (12)
The wavelet coefficient of one or more scales after Wave Decomposition carries out single branch or combined reconstruction, reaches identical with original signal strength
Reconstruction signal
However, dual-tree complex wavelet transform as traditional wavelet transform, does not carry out continuing to divide to high frequency section.
Double-tree complex wavelet package transforms (dual-tree complex wavelet packet transform, DT-CWPT), to double trees
There is no the radio-frequency head divided to be allocated as further decomposing in complex wavelet transform, as traditional wavelet package transforms, double trees are multiple small
Wave packet transform is further expanding for dual-tree complex wavelet transform, can provide higher frequency discrimination for the entire frequency range of signal
Rate and the frequency content that preferably identification and determining signal are included, so as to reduce the loss of useful information amount.
Double-tree complex wavelet package transforms two-layer decomposition and restructuring procedure are as shown in Figure 2.Two parallel discrete wavelet package transforms,
One of discrete analog method is properly termed as real part tree, another wavelet package transforms is properly termed as imaginary part tree, in order to
Reaching the decomposition to signal has approximate translation invariance, reduces the loss of useful information amount, double-tree complex wavelet package transforms are to letter
When number being decomposed and reconstructed, it is desirable that meet decompose and restructuring procedure in keep the positive benefit in sampling location of imaginary part tree always
In the centre position of the sampling location of real part tree, the information of real part tree and imaginary part tree can be formed in the decomposable process of signal
It is complementary.And double-tree complex wavelet package transforms also decompose the high fdrequency component for not having to decompose in dual-tree complex wavelet transform, subtract
The loss of information is lacked.The method of specific implementation is as shown in Fig. 2, wherein first layer decomposes, by first_1 wave filter groups
It is decomposed for real part tree discrete wavelet packet, first_1 wave filters are made of two rows, the f of the second row in first_1 wave filters1-0
For low-pass filter, the f of the first row1-1For high-pass filter；Equally, by first_2 wave filter groups it is that imaginary part tree is discrete
WAVELET PACKET DECOMPOSITION, first_2 wave filters also have two rows to form, the f of the second row in first_2 wave filters2-0For low-pass filtering
Device, the f of the first row2-1For high-pass filter.Decomposition for second layer more than dual-tree complex wavelet packet, in order to ensure DT-CWPT
Real part tree and the summation of delay difference that is generated on this layer and all front layers of imaginary part tree relative to the input of original signal be one
A sampling period, that is to say, that should have half of sampling between the phase-frequency response of real part tree and imaginary part tree respective filter
The group delay in period, and the amplitude frequency response of the two wave filters also should be identical, the real part of double-tree complex wavelet package transforms
Q_shift wave filter group h, the imaginary part tree WAVELET PACKET DECOMPOSITION alternating of double-tree complex wavelet package transforms is used alternatingly in tree WAVELET PACKET DECOMPOSITION
Use Q_shift wave filter groups g.Employed in every layer of DT-CWPT of decomposable process coefficient of wavelet decomposition dichotomy so as to
Extra calculating is eliminated, and then the treatment effeciency of signal is improved, DT-CWPT is double to the process that signal is reconstructed
The inverse process that tree complex wavelet package transforms decompose signal.
3) judge frequency range residing for one times of meshing frequency, selection includes the corresponding dual-tree complex wavelet of frequency range of the meshing frequency
Packet decomposition coefficient constructs Hankel matrix As, and carries out singular value decomposition, obtains A=UDVT。
It should be noted that in the present invention, the object of Main Diagnosis is the event of high pedestal jib crane low-speed heave-load component
Barrier, the especially initial failure of gear meshing section.If acquire the tach signal r of high pedestal jib crane low-speed heave-load component, root
According to the number of gear teeth Z of low-speed heave-load component, then one times of meshing frequency of low-speed heave-load component can be calculated
If as previously mentioned, using dual-tree complex wavelet packet to collected original vibration signal x (t) carry out n-layer decomposition, can
Original vibration signal is decomposed into 2nA frequency range, the frequency range of each frequency range is respectively [0, fs/2n]、 [fs/2n,2×fs/2n]、
[2×fs/2n,3×fs/2n]、…、[(2n-1)×fs/2n,2n×fs/2n].The one times of meshing frequency calculated according to formula (13)
fm, judge the frequency range residing for it, extract the corresponding dual-tree complex wavelet decomposition coefficient w (n) of the frequency band, by formula (14) Suo Shi
Mode constructs Hankel matrix As.
For real matrix A ∈ Rm×n, let it be, and whether ranks are related, certainly exist a pair of orthogonal matrix U=(u1,
u2,......,um)∈Rm×nWith orthogonal matrix V=(v1,v2,......,vm)∈Rm×nSo that
B=UDVT (15)
In formula, D=(diag (σ1,σ2,......σq), 0) or its transposition D=(diag (σ1,σ2,......σq), 0
)T, this will depend on m ＜ n or m ＞ n, wherein, D ∈ Rm×n, 0 represents null matrix, q=min (m, n), and have：σ1≥σ2
≥......σq>=0, they are called the singular value of matrix B.
4) D in step 3) is diagonal matrix, extracts the element σ on diagonal in diagonal matrix D1≥σ2≥......σq
>=0, form unusual value sequence：
S=σ1,σ2,......σq (16)
5) to the unusual value sequence in step 4), b is asked for respectivelyi=σi-σi+1(i=1,2 ..., q-1), it will be all
biSequence B=(b1,b2,......,bq-1) it is known as the Difference Spectrum of singular value；
6) the maximum sudden change point in singular value Difference Spectrum is checked, by the corresponding serial number of maximum sudden change point as singular value weight
The component number retained is needed during structure, singular value reconstruct is carried out, obtains updated dual-tree complex wavelet packet coefficient；
7) updated dual-tree complex wavelet packet coefficient is utilized, reconstructs vibration signal, realizes the noise reduction of vibration signal；
8) Hilbert envelope demodulations are carried out to the vibration signal of reconstruct, judges that the low-speed heave-load component is according to envelope spectrum
It is no that there are failures.
In order to verify the validity of the above method, constructive simulation signal is as follows：
Y=[0.3+0.21 × sin (2 π × 0.25t)] × sin (2 π × 50t) (17)
The sample frequency of the emulation signal be 1024Hz, sampling number 32768, time domain waveform, spectrogram such as Fig. 3
Shown, corresponding frequency spectrum partial enlarged view is as shown in Figure 4.
After the signal plus noise, time domain waveform and spectrogram are as shown in Figure 5.Modulation letter is not seen completely from Fig. 5
Number sideband 0.25Hz, carry out noise reduction using singular value decomposition, traditional singular value decomposition cannot use, because of present invention profit
Computer carries out signal the length of singular value decomposition no more than 8192.
Two layers of dual-tree complex wavelet packet is carried out to the signal using institute's extracting method of the present invention to decompose, obtains four frequency bands, each
Frequency band points are 8192, and decomposition result is as shown in Figure 6.To the dual-tree complex wavelet packet decomposition coefficient of first 0~128Hz of frequency band
Hankel matrixes are constructed, carry out singular value decomposition and acquire unusual value sequence, and then acquire singular value Difference Spectrum, unusual value difference
Open score overshooting become in front section, all going to zero below.In order to clearly observe the situation of Difference Spectrum, by singular value sequence
100 points are painted under a coordinate system before row and Difference Spectrum, as shown in Figure 7.As we can see from the figure at the 2nd point and the 6th point all
Occur compared with macromutation, if maximum sudden change point is happened at the first two point, the 2nd maximum sudden change point is often taken, because unusual
Value component can lose effective information very little；For existing simultaneously multiple larger catastrophe points, maximum sudden change point only need to be selected.
Therefore retaining preceding 6 components of singular value decomposition, remaining component is set to 0, carries out singular value reconstruct, so as to the
The wavelet coefficient of one frequency band has carried out noise reduction, then carries out dual-tree complex wavelet reconstruct to it, obtain as shown in Figure 8 as a result, its
In upper figure be reconstruct time domain plethysmographic signal, figure below be corresponding amplitude spectrum, Fig. 9 be corresponding amplitude spectrum partial enlarged view.
Extraordinary periodic shock is presented in signal, and impulse period is about 4s, and corresponding frequency is 0.25Hz, is exactly modulated signal
Sideband 0.25Hz.Fig. 8 is almost the same with Fig. 3, it is achieved thereby that noise reduction.Reconstruction signal shown in Fig. 8 is wished
That Bert envelope demodulation, obtains envelope spectrum as shown in Figure 10, what can be will be apparent that sees the frequency of 0.25Hz.
In order to more intuitively experience the validity of this method, dual-tree complex wavelet transform processing is directly carried out to emulation signal,
Figure 11 is the time domain waveform and spectrogram after dual-tree complex wavelet reconstruct, and Figure 12 is corresponding frequency spectrum partial enlarged view, can be seen
To side frequency information, although can still have noise with noise reduction, noise does not completely eliminate, it is impossible to completely inhibit noise.
The envelope spectrum of reconstruction signal is as shown in figure 13, although having the ingredient of 50Hz and 0.25Hz, since noise does not disappear completely
It removes, also has other false frequency contents in envelope spectrum, interference is formed to the frequency content in envelope spectrogram, is carried with the present invention
Go out method to compare, effect is very poor.
In order to further verify validity of the institute's extracting method of the present invention in engineering, certain low-speed heave-load device is acquired
Original vibration signal, time domain waveform and its amplitude spectrum are as shown in figure 14.Since the length of signal is 131072, belong to super
Long-time vibration signal can not carry out singular value decomposition noise reduction process to it using conventional computer, need carry out singular value
The points of the frequency band of decomposition are dropped within 8192, and 4 layers of dual-tree complex wavelet point are carried out to the signal so being carried using the present invention
Solution, the points of each frequency band are 8192.
Since the impact near frequency 70.04Hz is bigger, so to the frequency band containing 70.04Hz, i.e. first frequency
Band carries out singular value decomposition to its dual-tree complex wavelet packet decomposition coefficient, and unusual Difference Spectrum is as shown in figure 15, occurs at the 4th point
Compared with macromutation, preceding 4 components of singular value decomposition are subjected to singular value reconstruct, so as to the wavelet coefficient of first frequency band into
It has gone noise reduction, the reconstruct of dual-tree complex wavelet then is carried out to it, as a result as shown in figure 16, extraordinary periodicity is presented in signal,
And there is meshing frequency and its frequency multiplication, the period is about 4.368s, and corresponding frequency is 0.22894Hz, with gear wheel event
Barrier characteristic frequency 0.24Hz is very close, and 70.04 partial enlarged view of meshing frequency is as shown in figure 17, side frequency occurs
70.27Hz, side frequency bandwidth are 0.23Hz, with gear wheel to turn frequency 0.24Hz very close.
Hilbert envelope is carried out to the reconstruction signal shown in Figure 16 again to demodulate to obtain envelope spectrum as shown in figure 18, it can
See 0.2289Hz with what be will be apparent that, 0.4578Hz frequency contents, this and turn frequency 0.24Hz one times of gear wheel and two times
Frequency is very close, can be concluded that this failure as gear wheel failure, and later factory staff checks equipment, finds big
There is concentrated wear failure in gear, this is matched with the result diagnosed using institute's extracting method of the present invention.
Present invention firstly provides n-layer decomposition is carried out to original vibration signal using dual-tree complex wavelet packet decomposition method, utilize
The characteristic that decomposition coefficient successively successively decreases in dual-tree complex wavelet packet decomposable process extracts the frequency for including fault characteristic information in n-th layer
Section by the shorter decomposition coefficient of ultra-long time vibration signal boil down to, recycles singular value decomposition to decompose dual-tree complex wavelet packet
Coefficient carries out noise reduction process, according to Difference Spectrum maximum sudden change point, selects suitable singular value restructuring matrix, then reconstruct and dropped
Vibration signal after making an uproar.The present invention is solved in high pedestal jib crane low-speed heave-load fault diagnosis, traditional singular value decomposition without
Method handles ultra-long time vibration signal, it is difficult to the problem of extracting low-speed heave-load component initial failure Weak characteristic.
The above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications
It should be regarded as protection scope of the present invention.
Claims (2)
1. a kind of high pedestal jib crane low-speed heave-load unit failure feature extracting method, which is characterized in that include the following steps：
1) with sample frequency fsData acquisition is carried out to the vibration signal of high pedestal jib crane low-speed heave-load component, it will be collected
Original vibration signal x (t) inputs dual-tree complex wavelet packet algorithm, carries out n-layer decomposition, original vibration signal is decomposed into 2nA frequency
Section, the frequency range of each frequency range is respectively [0, fs/2n]、[fs/2n,2×fs/2n]、[2×fs/2n,3×fs/2n]、…、[(2n-
1)×fs/2n,2n×fs/2n]；
2) the tach signal r of high pedestal jib crane low-speed heave-load component is acquired, according to the number of gear teeth Z of low-speed heave-load component, meter
Calculate one times of meshing frequency of low-speed heave-load component
3) judge frequency range residing for one times of meshing frequency, the corresponding dual-tree complex wavelet packet of frequency range of the selection comprising the meshing frequency decomposes
Coefficient constructs Hankel matrix As, and carries out singular value decomposition, obtains A=UDVT, D is diagonal matrix；
4) the element σ in diagonal matrix D on diagonal is extracted1≥σ2≥......σq>=0, form unusual value sequence；
5) b is asked for respectively to unusual value sequencei=σi-σi+1(i=1,2 ..., q-1), by all biSequence B=(b1,
b2,......,bq-1) it is known as the Difference Spectrum of singular value；
6) it checks the maximum sudden change point in singular value Difference Spectrum, is needed when the corresponding serial number of maximum sudden change point is reconstructed as singular value
The component number to be retained carries out singular value reconstruct, obtains updated dual-tree complex wavelet packet coefficient；
7) updated dual-tree complex wavelet packet coefficient is utilized, reconstructs vibration signal, realizes the noise reduction of vibration signal；
8) Hilbert envelope demodulations are carried out to the vibration signal of reconstruct, judges whether the low-speed heave-load component is deposited according to envelope spectrum
In failure.
2. high pedestal jib crane low-speed heave-load unit failure feature extracting method according to claim 1, which is characterized in that profit
With the vibration signal of piezoelectric type vibration acceleration transducer pickup high pedestal jib crane low-speed heave-load component.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711350955.XA CN108181098A (en) | 2017-12-15 | 2017-12-15 | A kind of high pedestal jib crane low-speed heave-load unit failure feature extracting method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711350955.XA CN108181098A (en) | 2017-12-15 | 2017-12-15 | A kind of high pedestal jib crane low-speed heave-load unit failure feature extracting method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108181098A true CN108181098A (en) | 2018-06-19 |
Family
ID=62546137
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711350955.XA Pending CN108181098A (en) | 2017-12-15 | 2017-12-15 | A kind of high pedestal jib crane low-speed heave-load unit failure feature extracting method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108181098A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109799718A (en) * | 2019-01-29 | 2019-05-24 | 特斯联（北京）科技有限公司 | A kind of wisdom building control method and system based on Internet of Things sensing |
| CN109813547A (en) * | 2019-01-08 | 2019-05-28 | 华南理工大学 | Rotating machinery local type method for diagnosing faults based on sparse decomposition optimization algorithm |
| CN111458129A (en) * | 2020-04-29 | 2020-07-28 | 江苏省特种设备安全监督检验研究院 | High-precision online detection system for cantilever beam of crane |
| CN112462137A (en) * | 2020-09-18 | 2021-03-09 | 国网辽宁省电力有限公司电力科学研究院 | Equipment fault feature extraction method based on wavelet packet and Hilbert envelope spectrum analysis |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1548285A2 (en) * | 2003-12-23 | 2005-06-29 | General Electric Company | Method and apparatus for detecting compressor stall precursors |
| CN104729853A (en) * | 2015-04-10 | 2015-06-24 | 华东交通大学 | Rolling bearing performance degradation evaluation device and method |
| CN105426822A (en) * | 2015-11-05 | 2016-03-23 | 郑州轻工业学院 | Non-stable signal multi-fractal feature extraction method based on dual-tree complex wavelet transformation |
-
2017
- 2017-12-15 CN CN201711350955.XA patent/CN108181098A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1548285A2 (en) * | 2003-12-23 | 2005-06-29 | General Electric Company | Method and apparatus for detecting compressor stall precursors |
| CN104729853A (en) * | 2015-04-10 | 2015-06-24 | 华东交通大学 | Rolling bearing performance degradation evaluation device and method |
| CN105426822A (en) * | 2015-11-05 | 2016-03-23 | 郑州轻工业学院 | Non-stable signal multi-fractal feature extraction method based on dual-tree complex wavelet transformation |
Non-Patent Citations (2)
| Title |
|---|
| 刘景浩: "齿轮传动故障诊断专家系统的研究与应用", 《中国优秀博硕士学位论文全文数据库 (硕士) 工程科技Ⅱ辑》 * |
| 胥永刚: "双树复小波和奇异差分谱在滚动轴承故障诊断中的应用", 《振动工程学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109813547A (en) * | 2019-01-08 | 2019-05-28 | 华南理工大学 | Rotating machinery local type method for diagnosing faults based on sparse decomposition optimization algorithm |
| 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 |
| CN109799718A (en) * | 2019-01-29 | 2019-05-24 | 特斯联（北京）科技有限公司 | A kind of wisdom building control method and system based on Internet of Things sensing |
| CN111458129A (en) * | 2020-04-29 | 2020-07-28 | 江苏省特种设备安全监督检验研究院 | High-precision online detection system for cantilever beam of crane |
| CN112462137A (en) * | 2020-09-18 | 2021-03-09 | 国网辽宁省电力有限公司电力科学研究院 | Equipment fault feature extraction method based on wavelet packet and Hilbert envelope spectrum analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108181098A (en) | A kind of high pedestal jib crane low-speed heave-load unit failure feature extracting method | |
| Sung et al. | Locating defects of a gear system by the technique of wavelet transform | |
| CN101294845B (en) | Multi-frequency weak signal detecting method for early failure of rotor | |
| Chen et al. | Fault diagnosis of planetary gearbox under variable-speed conditions using an improved adaptive chirp mode decomposition | |
| CN105115594B (en) | Gear-box vibration signal fault signature extracting method based on Wavelet Entropy and information fusion | |
| CN101259016A (en) | Method for real time automatically detecting epileptic character wave | |
| CN108151869B (en) | Mechanical vibration characteristic index extraction method, system and device | |
| CN106409351A (en) | Method and device for processing vibration signal of loose part of nuclear power plant | |
| CN110135293A (en) | A kind of wind-driven generator electrical vibration signal noise-eliminating method and storage medium | |
| CN108647667B (en) | A kind of implementation method of the frequency domain amplitude spectrum kurtosis figure based on signal Time-frequency Decomposition | |
| CN109063668B (en) | Impact signal envelope demodulation method based on peak value retention and down-sampling | |
| CN110806315A (en) | Gearbox composite fault diagnosis method based on inversion editing | |
| CN105266800B (en) | One kind is based on fetus electrocardio blind separation under Low SNR | |
| CN105092241A (en) | Method and system for diagnosing local fault of gear | |
| CN109765052B (en) | Planetary gearbox early fault diagnosis method based on GOA-ASR | |
| Kocur et al. | Order bispectrum: a new tool for reciprocated machine condition monitoring | |
| CN115730199B (en) | Rolling bearing vibration signal noise reduction and fault feature extraction method and system | |
| Wang et al. | Weak fault diagnosis of rolling bearing under variable speed condition using IEWT-based enhanced envelope order spectrum | |
| CN104215456A (en) | Plane clustering and frequency-domain compressed sensing reconstruction based mechanical fault diagnosis method | |
| CN103940597A (en) | Method for detecting mechanical faults based on generalized extremum morphological filtering | |
| Zhang et al. | Improved local cepstrum and its applications for gearbox and rolling bearing fault detection | |
| CN101839922A (en) | Method for eliminating signal online outliers | |
| CN107966287A (en) | A kind of adaptive dynamoelectric equipment Weak fault feature extracting method | |
| KR102598458B1 (en) | Apparatus and method for detecting fault of planetary gearbox using cepstrum-based signal analysis | |
| CN108267311A (en) | A kind of mechanical multidimensional big data processing method based on tensor resolution |
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: 20180619 |