CN110132403A - A kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold - Google Patents
A kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold Download PDFInfo
- Publication number
- CN110132403A CN110132403A CN201910262768.9A CN201910262768A CN110132403A CN 110132403 A CN110132403 A CN 110132403A CN 201910262768 A CN201910262768 A CN 201910262768A CN 110132403 A CN110132403 A CN 110132403A
- Authority
- CN
- China
- Prior art keywords
- signal
- noise
- eemd
- vacuum pump
- wavelet
- 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
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Compressors, Vaccum Pumps And Other Relevant Systems (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Soundproofing, Sound Blocking, And Sound Damping (AREA)
Abstract
The invention discloses a kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold: EEMD decomposition is carried out to original signal first, obtains several IMF components and a remainder；Secondly all IMF components are normalized with the calculating of auto-correlation function, IMF component is divided into the IMF component of signal leading IMF component and noise dominant according to the characteristic of auto-correlation function zero point decaying；Then wavelet soft-threshold noise reduction process is carried out to the IMF component of noise dominant；Finally by by wavelet soft-threshold treated noise dominant IMF component and the leading IMF component and remainder of signal carry out the reconstruct of signal, to obtain the vacuum pump vibration signal after noise reduction.The present invention is decomposed using EEMD, can overcome the problems such as decomposing bring mode mixing and end effect by EMD, and the effective noise signal removed in vacuum pump vibration signal preferably retains more useful signals, improves the signal-to-noise ratio of signal.
Description
Technical field
The present invention relates to vibration signal noise management technique, in particular to a kind of vacuum pump based on EEMD and wavelet threshold
Vibration signal noise-reduction method.
Background technique
In recent years, China has obtained quick development in space technology field, and more batches of satellites of sequential transmissions and space are visited
Measurement equipment.To guarantee even running of the spacecraft equipment under complicated space environment, need to carry out the ground simulation of space environment
Test, wherein ultralow temperature and high vacuum are the Simulation Test Environments of important requirement.Vacuum pump is as ground space environmental simulation
Important equipment, fault occurrence frequency is higher, will lead to decline, low temperature noise exception of vacuum pump performance etc..Therefore, it is necessary to right
Vacuum pump structure carries out on-line fault diagnosis, the performance change of timely discovering device, to guarantee the normal operation and reality of vacuum pump
Test safety.Status signal carrier of the vibration signal as such mechanical equipment carries out it to acquire monitoring in real time, and takes certain
Signal processing method analyze it processing be always related fields research emphasis.Under actual working environment, due to
It is influenced by site environment noise, the fault characteristic information in vibration signal is often submerged in noise signal, is influenced pair
The extraction of fault characteristic value.Therefore, how effectively to carry out noise reduction process to the vibration signal of vacuum pump is to realize that vacuum pump exists
One of the critical issue of line fault diagnosis.
More for the noise-reduction method of the mechanical equipment vibration signal of the classifications such as vacuum pump, early stage is mainly using with small
Noise-reduction method based on Wave Decomposition such as carries out wavelet de-noising processing to signal using improved threshold function table, all achieves certain
Effect, but its noise reduction effect tends to rely on the selection of wavelet basis and threshold value.The empirical mode decomposition (EMD) then proposed is suitable
For non-linear, non-stationary signal processing method, signal decomposition is arranged successively from high frequency to low frequency by this method at several
Intrinsic mode function (IMF) and a remainder (res), have certain actual physical meaning and frequency layering capabilities, be one
Finish the decomposition method of fully adaptive.But if there are abnormal sudden changes in signal, then can generate modal overlap and end effect from
And influence noise reduction effect.A kind of set empirical mode decomposition (EEMD) method then proposed, it is special using the zero-mean of white noise
Property, different white noises is repeatedly added in original signal, a series of IMF and remainder are obtained, can effectively be inhibited in EMD
There are the influences of the brings such as mode mixing and end effect.
Summary of the invention
The purpose of the present invention is overcoming deficiency in the prior art, a kind of vacuum pump based on EEMD and wavelet threshold is provided
Vibration signal noise-reduction method can effectively remove the noise signal in vacuum pump vibration signal, preferable to retain more useful letters
Number, the signal-to-noise ratio of signal is improved, the root-mean-square error of signal is reduced.
The technical scheme adopted by the invention is that: a kind of vacuum pump vibration signal noise reduction side based on EEMD and wavelet threshold
Method, comprising the following steps:
Step 1, EEMD decomposition is carried out to vacuum pump original vibration signal, obtains several IMF components and remainder；
Step 2, auto-correlation function calculating is carried out to all IMF components that step 1 obtains, according to the normalization of IMF component
IMF component, is divided into the IMF component of signal leading IMF component and noise dominant by the characteristic of auto-correlation function zero point decaying；
Step 3, the IMF component of the noise dominant obtained to step 2 carries out wavelet soft-threshold noise reduction process；
Step 4, the IMF component of the noise dominant in step 3 after wavelet soft-threshold noise reduction process, step 2 are obtained
The obtained remainder of the leading IMF component and step 1 of signal be reconstructed, the vacuum pump vibration signal after obtaining noise reduction.
Further, in step 1, the process for carrying out EEMD decomposition to vacuum pump original vibration signal is as follows:
Step 1-1, it is zero that M mean value, which is added, to vacuum pump original vibration signal, and standard deviation k is the Gauss white noise of constant
Sound generates new signal；
Step 1-2 carries out multiple EMD decomposition to the new signal that generates in step 1-1, obtain several IMF components and
Remainder；
Step 1-3, all IMF components that step 1-2 is obtained carry out population mean calculating and decompose to get to by EEMD
The IMF component and remainder of generation.
Wherein, in step 1-1, the value range of number M is decomposed at 100-300 times, additional Gaussian white noise standard deviation k exists
0.01-0.5 times of original vibration signal standard deviation value.
Wherein, in step 2, the IMF component that IMF component is divided into signal leading IMF component and noise dominant
Process it is as follows:
Step 2-1, auto-correlation function is defined as:
Rx(t, t+ τ)=E [x (t) x (t+ τ)] (3)
Wherein, τ is time interval, and x (t) and x (t+ τ) are respectively value of the signal in t moment and t+ τ moment, E [x (t) x
(t+ τ)] indicate mathematic expectaion；
Step 2-2, normalized autocorrelation functions indicate are as follows:
Wherein, Rx(0) auto-correlation function value of signal and the signal under synchronization itself is indicated；
Step 2-3 decays according to the normalized autocorrelation functions of general signal and random noise signal in zero crossings
This characteristic of slow degree carries out category filter to IMF component.
Wherein, it in step 2-3, in the normalized autocorrelation functions of IMF component, is slowly shaken after zero point obtains maximum value
The signal swung is general signal, and general signal is screened to the IMF component dominated for signal；Have at zero point maximum value and
At zero point both sides decay rapidly close to zero signal be random noise signal, random noise signal is screened as noise dominant
IMF component.
Further, in step 3, the IMF component to noise dominant adopt in wavelet soft-threshold noise reduction process
Threshold value is chosen with fixed threshold rule, wavelet coefficient shrink using soft-threshold function or zero setting is handled.
Further, in step 3, involved wavelet basis selection in the wavelet soft-threshold noise reduction process
Sym8 wavelet systems, Decomposition order are set as 4 layers.
The beneficial effects of the present invention are:
(1) present invention, which carries out EEMD decomposition to signal, can overcome by EMD decomposition bring mode mixing and end effect
The problems such as, and combined with wavelet threshold, denoising effect is substantially better than traditional wavelet noise reduction, EEMD forces noise reduction and
EMD Threshold Denoising.
(2) present invention utilizes general signal and the normalized autocorrelation functions of random noise in zero point attenuation characteristic to IMF
Component is classified, and then the IMF component dominated to signal retains, and is handled the IMF component of noise dominant.
(3) present invention can effectively remove the noise signal in vacuum pump vibration signal, preferable to retain more useful letters
Number, the signal-to-noise ratio of signal is improved, the root-mean-square error of signal is reduced.
Detailed description of the invention
Fig. 1 is a kind of algorithm flow of the vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold of the present invention
Figure；
Fig. 2 a is general signal；
Fig. 2 b is the normalized autocorrelation functions of general signal；
Fig. 2 c is random noise signal；
Fig. 2 d is the normalized autocorrelation functions of random noise signal；
Fig. 3 a is original Blocks signal；
Fig. 3 b is to add Blocks signal of making an uproar；
The single order IMF component (IMF1) that Fig. 4 a is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
The second order IMF component (IMF2) that Fig. 4 b is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
The three rank IMF components (IMF3) that Fig. 4 c is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
The quadravalence IMF component (IMF4) that Fig. 4 d is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
The five rank IMF components (IMF5) that Fig. 4 e is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
The six rank IMF components (IMF6) that Fig. 4 f is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
The seven rank IMF components (IMF7) that Fig. 4 g is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
The eight rank IMF components (IMF8) that Fig. 4 h is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
The nine rank IMF components (IMF9) that Fig. 4 i is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
The remainder (res) that Fig. 4 j is plus Blocks signal of making an uproar obtains after EEMD is decomposed；
Fig. 5 a is the auto-correlation function of single order IMF component (IMF1) in Fig. 4 a；
Fig. 5 b is the auto-correlation function of second order IMF component (IMF2) in Fig. 4 b；
Fig. 5 c is the auto-correlation function of three rank IMF components (IMF3) in Fig. 4 c；
Fig. 5 d is the auto-correlation function of quadravalence IMF component (IMF4) in Fig. 4 d；
Fig. 5 e is the auto-correlation function of five rank IMF components (IMF5) in Fig. 4 e；
Fig. 5 f is the auto-correlation function of six rank IMF components (IMF6) in Fig. 4 f；
Fig. 5 g is the auto-correlation function of seven rank IMF components (IMF7) in Fig. 4 g；
Fig. 5 h is the auto-correlation function of eight rank IMF components (IMF8) in Fig. 4 h；
Fig. 5 i is the auto-correlation function of nine rank IMF components (IMF9) in Fig. 4 i；
Fig. 5 j is the auto-correlation function of remainder (res) in Fig. 4 j；
Fig. 6 a be using traditional wavelet noise-reduction method to add make an uproar Blocks signal carry out noise reduction process result；
Fig. 6 b be using EEMD force noise-reduction method to add make an uproar Blocks signal carry out noise reduction process result；
Fig. 6 c be using EMD Threshold Denoising method to add make an uproar Blocks signal carry out noise reduction process result；
Fig. 6 d is to make an uproar using the present invention is based on the vacuum pump vibration signal noise-reduction methods of EEMD and wavelet threshold to adding
Blocks signal carries out noise reduction process result；
Fig. 7 is vacuum pump original vibration signal figure；
Fig. 8 a is the single order IMF component (IMF1) that vacuum pump original vibration signal obtains after EEMD is decomposed；
Fig. 8 b is the second order IMF component (IMF2) that vacuum pump original vibration signal obtains after EEMD is decomposed；
Fig. 8 c is the three rank IMF components (IMF3) that vacuum pump original vibration signal obtains after EEMD is decomposed；
Fig. 8 d is the quadravalence IMF component (IMF4) that vacuum pump original vibration signal obtains after EEMD is decomposed；
Fig. 8 e is the five rank IMF components (IMF5) that vacuum pump original vibration signal obtains after EEMD is decomposed；
Fig. 8 f is the six rank IMF components (IMF6) that vacuum pump original vibration signal obtains after EEMD is decomposed；
Fig. 8 g is the seven rank IMF components (IMF7) that vacuum pump original vibration signal obtains after EEMD is decomposed；
Fig. 8 h is the eight rank IMF components (IMF8) that vacuum pump original vibration signal obtains after EEMD is decomposed；
Fig. 9 a is to carry out noise reduction process result to vacuum pump original vibration signal using traditional wavelet noise-reduction method；
Fig. 9 b is to force noise-reduction method to carry out noise reduction process result to vacuum pump original vibration signal using EEMD；
Fig. 9 c is to carry out noise reduction process result to vacuum pump original vibration signal using EMD Threshold Denoising method；
Fig. 9 d is using the present invention is based on the vacuum pump vibration signal noise-reduction methods of EEMD and wavelet threshold to vacuum pump original
Beginning vibration signal carries out noise reduction process result.
Specific embodiment
In order to further understand the content, features and effects of the present invention, the following examples are hereby given, and cooperate attached drawing
Detailed description are as follows:
As shown in Figure 1, a kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold, specifically includes following
Step:
Step 1, vacuum pump original vibration signal is decomposed using set empirical mode decomposition (EEMD), is obtained several
A intrinsic mode function (IMF) and remainder (res).
Wherein, carrying out EEMD decomposition to original vibration signal is that white noise spectrum uniform properties and zero-mean spy are utilized
Property, by the signal automatic average to different time scales after decomposition and its mean value is zero, so can be with by multiple decomposition
The phenomenon that effectively weakening modal overlap.The parameter set decomposes number M and additional Gaussian white noise standard deviation k as EEMD,
The middle value range for decomposing number M is at 100-300 times, and additional Gaussian white noise standard deviation k is in 0.01-0.5 times of original vibration letter
Number standard deviation value.
EEMD is the innovatory algorithm based on EMD, and the process that EEMD is decomposed is as follows:
1) M white Gaussian noise signal n is added into original vibration signal x (t)i(t) (i=1,2 ..., M) is generated new
Signal xi(t), it may be assumed that
xi(t)=x (t)+ni(t) (1)
2) to xi(t) multiple EMD decomposition is carried out, N number of IMF component is obtained and is denoted as cij(t) (j=1,2 ..., N) and remainder
r(t).Wherein cij(t) indicate that j-th of IMF component that white Gaussian noise obtains is added in i-th.
3) it to offset the influence that M white Gaussian noise signal is added and generates to IMF component, is united using white Gaussian noise signal
The characteristic that characteristic mean is zero is counted, all IMF components are subjected to population mean calculating, obtains decomposing the IMF generated by EEMD
Component cj(t):
Step 2, auto-correlation function calculating is carried out to all IMF components, is decayed according to its normalized autocorrelation functions zero point
Characteristic, IMF component is divided into the IMF component of signal leading IMF component and noise dominant.
Wherein, the process that IMF component is divided into the IMF component of the leading IMF component of signal and noise dominant is as follows:
A) auto-correlation function has reacted same signal in the similarity degree of different moments, is defined as:
Rx(t, t+ τ)=E [x (t) x (t+ τ)] (3)
Wherein τ is time interval, and x (t) and x (t+ τ) are respectively value of the signal in t moment and t+ τ moment, E [x (t) x
(t+ τ)] indicate mathematic expectaion.
B) normalized autocorrelation functions are generallyd use in engineer application to reflect this degree, i.e., there will be dimensional form transformation
For Dimensionless Form, by between the Ratage Coutpressioit of value to -1 to 1, normalized autocorrelation functions be may be expressed as:
Wherein Rx (0) indicates the signal auto-correlation function value under synchronization with itself.Auto-correlation function embodies
The positive and negative interval of signal, so length is the signal of n, auto-correlation function length is then 2n-1, embodies auto-correlation function
Symmetry.
C) general signal is since with preferable relevance, normalized autocorrelation functions are in zero point acquirement maximum value
There is not the phenomenon that decaying rapidly in slowly concussion afterwards.And random noise signal due to the relevance in different moments is weaker, with
Machine is stronger, thus its normalized autocorrelation functions has maximum value at zero point, and then both sides decays connect rapidly at zero point
It is bordering on zero.According to this characteristic, i.e., up to the IMF component that IMF component is divided into signal leading IMF component and noise dominant.
It as shown in Figure 2 a and 2 b, is general signal and its normalized autocorrelation functions comparison diagram: where Fig. 2 a is general signal, figure
2b is the normalized autocorrelation functions of general signal；It is random noise signal and its normalization from phase as shown in Fig. 2 c and Fig. 2 d
Close function comparison diagram: Fig. 2 c is random noise signal, and Fig. 2 d is the normalized autocorrelation functions of random noise signal.
Step 3, the IMF component of the noise dominant obtained to step 2 carries out wavelet soft-threshold noise reduction process.
Threshold value is chosen using fixed threshold regular (Sqtwolog), formula are as follows:
Wherein σ is noise criteria variance, and N is signal length.
Wavelet coefficient shrink using soft-threshold function or zero setting is handled.Wavelet soft-threshold function is to will be above threshold value
Wavelet coefficient carry out a degree of contraction, the wavelet coefficient less than threshold value is set to zero.It is whole that new wavelet coefficient is obtained in this way
Body continuity is preferable, the useful information in the better stick signal of energy.Soft-threshold function expression formula are as follows:
Wherein x is the wavelet coefficient before processing, ω (x, λ) is that treated wavelet coefficient, and λ is threshold size.
Step 4, the IMF component of the noise dominant in step 3 after wavelet soft-threshold noise reduction process, step 2 are obtained
The obtained remainder of the leading IMF component and step 1 of signal be reconstructed, the vacuum pump vibration signal after obtaining noise reduction.
Embodiment 1
Simulation analysis is carried out based on the vacuum pump vibration signal noise-reduction method of EEMD and wavelet threshold to above-mentioned: being established former
Beginning signal and its signals and associated noises mathematical model, using common Blocks wave in the test of MATLAB noise reduction as emulating signal, and
The white Gaussian noise that signal-to-noise ratio is 8.2955dB, signal sampling rate 1Hz, sampled point is added to Blocks wave in simulation process
Number is set as 1024, is emulation Blocks signal and its noisy signal: where Fig. 3 a is original Blocks as best shown in figures 3 a and 3b
Signal, Fig. 3 b are to add Blocks signal of making an uproar.
To adding Blocks signal of making an uproar to carry out EEMD decomposition, (decomposing number M is 150, and white noise standard deviation k is that 0.5), there are
To 9 IMF components and 1 remainder, Blocks signal of making an uproar is added to pass through the result that EEMD is decomposed as Fig. 4 a to 4j is shown.By EEMD
Decompose obtained IMF component and auto-correlation function processing be normalized, as Fig. 5 a to Fig. 5 j show each rank IMF component from phase
Close function.It can be seen that zero point both sides starts to decay rapidly after IMF1, IMF2 and IMF3 obtain maximum value at zero point, i.e., should
There is apparent noisy feature in component.Therefore it chooses first three IMF component and carries out wavelet soft-threshold processing, wavelet basis selection
Sym8 wavelet systems, Decomposition order are set as 4 layers, while retaining other IMF components and carrying out last IMF component reconstruct with remainder.
Using traditional wavelet Method of Noise, EEMD force Method of Noise, EMD Threshold Denoising Method with it is proposed by the invention
The vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold noisy emulation signal is handled after Comparative result
As shown in Fig. 6 a to Fig. 6 d, Fig. 6 a is traditional wavelet noise-reduction method, and Fig. 6 b is that EEMD forces noise-reduction method, and Fig. 6 c is EMD
Threshold Denoising method, Fig. 6 d are noise-reduction methods of the invention.It is Signal to Noise Ratio (SNR) to the performance measure evaluation index after noise reduction
And mean square error RMSE, formula difference are as follows:
Wherein x (i) be noisy original signal, x'(i) be noise reduction after signal.
Wherein signal-to-noise ratio is bigger, shows that useful information ratio is bigger in result, noise reduction effect is better, and root-mean-square error is then
In contrast.
Table 1 is that the performance indicator of a variety of noise-reduction methods compares.
In conclusion as shown in Figure 6 a to 6 d, although traditional wavelet Method of Noise and EEMD are forced at Method of Noise
The result of reason is very smooth, but is lost mass part detailed information.And EMD Threshold Denoising Method and context of methods obtain
Signal although there is small concussion, but most of useful detailed information has been obtained into good reservation, it is similar to original signal
Degree is high.And by the data of table 1 it is found that EEMD proposed by the present invention and Threshold Denoising method are handling noisy emulation letter
Number when, can not only greatly improve the signal-to-noise ratio of denoised signal, moreover it is possible to by mean square error control in minimum, can be maximum
Original signal characteristic is restored, noise reduction effect performance indicator is superior to other methods, has apparent advantage.
Embodiment 2
Instance analysis is carried out based on the vacuum pump vibration signal noise-reduction method of EEMD and wavelet threshold to above-mentioned: this is sent out
Bright proposed noise-reduction method applies the noise reduction process of Mr. Yu's model vacuum pump vibration data signal.Data collection system is by upper
Machine, 6366 type data collecting card of NI-USB, preposition charge amplifier and acoustic emission sensor are constituted.Experiment is with vacuum pump case
On center carry out real-time data acquisition as vibration signals collecting point.By the way that sample rate 100kHz is arranged, when sampling, is a length of
0.5s carries out the data sampling of vacuum pump vibration signal, and vacuum pump original vibration signal is as shown in Figure 7.
Noise reduction process is carried out to noisy vibration signal using the noise-reduction method based on EEMD and wavelet threshold.Signals and associated noises warp
After crossing EEMD decomposition, one is obtained 14 IMF components and remainder, by the way that autocorrelation calculation is normalized to all IMF components
It was found that preceding 8 IMF components contain noise, as shown in Fig. 8 a to Fig. 8 h, therefore chooses preceding 8 IMF components and carry out wavelet soft-threshold
Processing, then all IMF components are reconstructed again.As a result as shown in Fig. 9 a to Fig. 9 d, Fig. 9 a is traditional wavelet noise reduction
Method, Fig. 9 b are that EEMD forces noise-reduction method, and Fig. 9 c is EMD Threshold Denoising method, and Fig. 9 d is noise-reduction method of the invention.
It can be seen that the vibration signal after the method for the present invention noise reduction more intuitive can see what vacuum pump generated during the work time
Impact signal, the good impulse period feature and trend for remaining original signal, greatly eliminates making an uproar in impact signal
Acoustical signal, and effectively filtering out uncorrelated noise signal in the gap of impact signal twice makes signal more steadily and accurately,
Its anti-acoustic capability is better than other several methods.
To sum up, the present invention can be overcome using EEMD decomposition and be asked by EMD decomposition bring mode mixing and end effect etc.
Topic, the effective noise signal removed in vacuum pump vibration signal, preferably retains more useful signals, improves the noise of signal
Than.
Although the preferred embodiment of the present invention is described above in conjunction with attached drawing, the invention is not limited to upper
The specific embodiment stated, the above mentioned embodiment is only schematical, be not it is restrictive, this field it is common
Technical staff under the inspiration of the present invention, without breaking away from the scope protected by the purposes and claims of the present invention, may be used also
By make it is many in the form of, within these are all belonged to the scope of protection of the present invention.
Claims (7)
1. a kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold, which comprises the following steps:
Step 1, EEMD decomposition is carried out to vacuum pump original vibration signal, obtains several IMF components and remainder；
Step 2, all IMF components obtained to step 1 carry out auto-correlation function calculating, according to the normalization of IMF component from phase
IMF component is divided into the IMF component of signal leading IMF component and noise dominant by the characteristic for closing function zero-point decaying；
Step 3, the IMF component of the noise dominant obtained to step 2 carries out wavelet soft-threshold noise reduction process；
Step 4, the letter IMF component of the noise dominant in step 3 after wavelet soft-threshold noise reduction process, step 2 obtained
The remainder that number leading IMF component and step 1 obtains is reconstructed, the vacuum pump vibration signal after obtaining noise reduction.
2. a kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold according to claim 1, special
Sign is, in step 1, the process for carrying out EEMD decomposition to vacuum pump original vibration signal is as follows:
Step 1-1, it is zero that M mean value, which is added, to vacuum pump original vibration signal, and standard deviation k is the white Gaussian noise of constant, raw
The signal of Cheng Xin；
Step 1-2 carries out multiple EMD decomposition to the new signal generated in step 1-1, obtains several IMF components and remainder；
Step 1-3, all IMF components that step 1-2 is obtained carry out population mean and calculate to get to by EEMD decomposition generation
IMF component and remainder.
3. a kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold according to claim 2, special
Sign is, in step 1-1, decomposes the value range of number M at 100-300 times, k is in 0.01- for additional Gaussian white noise standard deviation
0.5 times of original vibration signal standard deviation value.
4. a kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold according to claim 1, special
Sign is, in step 2, the process of the IMF component that IMF component is divided into the leading IMF component of signal and noise dominant
It is as follows:
Step 2-1, auto-correlation function is defined as:
Rx(t, t+ τ)=E [x (t) x (t+ τ)] (3)
Wherein, τ is time interval, and x (t) and x (t+ τ) are respectively value of the signal in t moment and t+ τ moment, E [x (t) x (t+
τ)] indicate mathematic expectaion；
Step 2-2, normalized autocorrelation functions indicate are as follows:
Wherein, Rx(0) auto-correlation function value of signal and the signal under synchronization itself is indicated；
Step 2-3 decays according to the normalized autocorrelation functions of general signal and random noise signal in zero crossings slow
This characteristic of degree carries out category filter to IMF component.
5. a kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold according to claim 4, special
Sign is, in step 2-3, in the normalized autocorrelation functions of IMF component, the letter slowly shaken after maximum value is obtained in zero point
Number it is general signal, general signal screen to the IMF component leading for signal；There is maximum value at zero point and in zero point
Place both sides decay rapidly close to zero signal be random noise signal, random noise signal is screened into the IMF for noise dominant
Component.
6. a kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold according to claim 1, special
Sign is, in step 3, the IMF component to noise dominant is carried out in wavelet soft-threshold noise reduction process, using fixed threshold
Rule chooses threshold value, shrink to wavelet coefficient using soft-threshold function or zero setting is handled.
7. a kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold according to claim 1, special
Sign is, in step 3, involved wavelet basis selection sym8 wavelet systems in the wavelet soft-threshold noise reduction process,
Decomposition order is set as 4 layers.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910262768.9A CN110132403A (en) | 2019-04-02 | 2019-04-02 | A kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910262768.9A CN110132403A (en) | 2019-04-02 | 2019-04-02 | A kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110132403A true CN110132403A (en) | 2019-08-16 |
Family
ID=67569191
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910262768.9A Pending CN110132403A (en) | 2019-04-02 | 2019-04-02 | A kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110132403A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110659620A (en) * | 2019-09-26 | 2020-01-07 | 中国科学院微电子研究所 | Filtering noise reduction method based on fuzzy control |
| CN111582132A (en) * | 2020-04-30 | 2020-08-25 | 南京信息工程大学 | Improved EEMD and PCNN-based gas leakage signal noise reduction method |
| CN112528857A (en) * | 2020-12-10 | 2021-03-19 | 上海海事大学 | EEMD-based noise reduction method for reciprocating friction vibration signal |
| CN113450285A (en) * | 2021-07-19 | 2021-09-28 | 天津大学 | Method for removing anti-disturbance in image |
| CN114154546A (en) * | 2021-12-08 | 2022-03-08 | 东北大学 | Noise reduction method for steel production process data |
| CN114184274A (en) * | 2021-11-19 | 2022-03-15 | 哈尔滨工程大学 | Nonlinear characteristic diagnosis method for vibration signal interfered by high-frequency noise |
| CN114966403A (en) * | 2022-08-01 | 2022-08-30 | 山东博源精密机械有限公司 | New energy automobile motor locked-rotor fault detection method and system |
| CN114994365A (en) * | 2022-04-18 | 2022-09-02 | 北京理工大学 | Accelerometer output signal noise reduction method based on air cannon test |
| CN115255405A (en) * | 2022-09-23 | 2022-11-01 | 相国新材料科技江苏有限公司 | Intelligent control method and system of additive manufacturing equipment |
| CN115659128A (en) * | 2022-12-12 | 2023-01-31 | 浙江工业大学 | Signal noise reduction method based on ensemble empirical mode decomposition method and power spectrum |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103902844A (en) * | 2014-04-24 | 2014-07-02 | 国家电网公司 | Transformer vibration signal de-noising method based on EEMD kurtosis threshold value |
| CN109058089A (en) * | 2018-06-13 | 2018-12-21 | 天津大学 | A method of the vacuum pump overload fault detection based on acoustic emission signal |
| CN109145729A (en) * | 2018-07-13 | 2019-01-04 | 杭州电子科技大学 | Based on the electromyography signal denoising method for improving wavelet threshold and EEMD |
-
2019
- 2019-04-02 CN CN201910262768.9A patent/CN110132403A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103902844A (en) * | 2014-04-24 | 2014-07-02 | 国家电网公司 | Transformer vibration signal de-noising method based on EEMD kurtosis threshold value |
| CN109058089A (en) * | 2018-06-13 | 2018-12-21 | 天津大学 | A method of the vacuum pump overload fault detection based on acoustic emission signal |
| CN109145729A (en) * | 2018-07-13 | 2019-01-04 | 杭州电子科技大学 | Based on the electromyography signal denoising method for improving wavelet threshold and EEMD |
Non-Patent Citations (2)
| Title |
|---|
| 余发军 等: ""基于EEMD 和自相关函数特性的自适应降噪方法"", 《计算机应用研究》 * |
| 邓青林 等: ""基于EEMD 和小波的爆破振动信号去噪"", 《爆破》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110659620B (en) * | 2019-09-26 | 2022-12-27 | 中国科学院微电子研究所 | Filtering noise reduction method based on fuzzy control |
| CN110659620A (en) * | 2019-09-26 | 2020-01-07 | 中国科学院微电子研究所 | Filtering noise reduction method based on fuzzy control |
| CN111582132A (en) * | 2020-04-30 | 2020-08-25 | 南京信息工程大学 | Improved EEMD and PCNN-based gas leakage signal noise reduction method |
| CN111582132B (en) * | 2020-04-30 | 2023-04-18 | 南京信息工程大学 | Improved EEMD and PCNN-based gas leakage signal noise reduction method |
| CN112528857A (en) * | 2020-12-10 | 2021-03-19 | 上海海事大学 | EEMD-based noise reduction method for reciprocating friction vibration signal |
| CN113450285A (en) * | 2021-07-19 | 2021-09-28 | 天津大学 | Method for removing anti-disturbance in image |
| CN114184274A (en) * | 2021-11-19 | 2022-03-15 | 哈尔滨工程大学 | Nonlinear characteristic diagnosis method for vibration signal interfered by high-frequency noise |
| CN114154546A (en) * | 2021-12-08 | 2022-03-08 | 东北大学 | Noise reduction method for steel production process data |
| CN114994365A (en) * | 2022-04-18 | 2022-09-02 | 北京理工大学 | Accelerometer output signal noise reduction method based on air cannon test |
| CN114966403A (en) * | 2022-08-01 | 2022-08-30 | 山东博源精密机械有限公司 | New energy automobile motor locked-rotor fault detection method and system |
| CN114966403B (en) * | 2022-08-01 | 2022-10-25 | 山东博源精密机械有限公司 | New energy automobile motor locked-rotor fault detection method and system |
| CN115255405A (en) * | 2022-09-23 | 2022-11-01 | 相国新材料科技江苏有限公司 | Intelligent control method and system of additive manufacturing equipment |
| CN115659128A (en) * | 2022-12-12 | 2023-01-31 | 浙江工业大学 | Signal noise reduction method based on ensemble empirical mode decomposition method and power spectrum |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110132403A (en) | A kind of vacuum pump vibration signal noise-reduction method based on EEMD and wavelet threshold | |
| Rai et al. | A review on signal processing techniques utilized in the fault diagnosis of rolling element bearings | |
| Wang et al. | A novel feature enhancement method based on improved constraint model of online dictionary learning | |
| Guo et al. | Envelope extraction based dimension reduction for independent component analysis in fault diagnosis of rolling element bearing | |
| CN105588717A (en) | Gearbox fault diagnosis method | |
| CN113420691A (en) | Mixed domain characteristic bearing fault diagnosis method based on Pearson correlation coefficient | |
| CN101900789A (en) | Tolerance analog circuit fault diagnosing method based on wavelet transform and fractal dimension | |
| CN112326245B (en) | Rolling bearing fault diagnosis method based on variational Hilbert-Huang transform | |
| CN101251445B (en) | Method for analysis of fractal characteristic of rotating machinery bump-scrape acoustic emission signal | |
| Cui et al. | Spectrum-based, full-band preprocessing, and two-dimensional separation of bearing and gear compound faults diagnosis | |
| Shi et al. | The VMD-scale space based hoyergram and its application in rolling bearing fault diagnosis | |
| CN110782041B (en) | Structural modal parameter identification method based on machine learning | |
| CN106706122B (en) | Method for denoising bump-scrape acoustic emission signal based on related coefficient and EMD filtering characteristic | |
| Cao et al. | A method for extracting weak impact signal in NPP based on adaptive Morlet wavelet transform and kurtosis | |
| Yang et al. | Fast nonlinear Hoyergram for bearings fault diagnosis under random impact interference | |
| Xu et al. | Rolling bearing fault feature extraction via improved SSD and a singular-value energy autocorrelation coefficient spectrum | |
| CN110057918A (en) | Damage of composite materials quantitative identification method and system under strong noise background | |
| CN111380680A (en) | Check valve fault feature extraction method based on improved permutation entropy | |
| CN114061746B (en) | Repeated transient signal extraction method in rotary machinery fault diagnosis | |
| Dai et al. | Element analysis and its application in rotating machinery fault diagnosis | |
| CN109212609A (en) | Near surface Noise Elimination method based on wave equation continuation | |
| Hambaba et al. | Multiresolution error detection on early fatigue cracks in gears | |
| CN112766044B (en) | Method and device for analyzing longitudinal and transverse wave speeds of loose sample and computer storage medium | |
| CN114200232A (en) | Method and system for detecting fault traveling wave head of power transmission line | |
| CN113567129A (en) | CEEMD-based noise reduction method for train bearing vibration signal |
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: 20190816 |