CN107677362A - A kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis - Google Patents
A kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis Download PDFInfo
- Publication number
- CN107677362A CN107677362A CN201710793554.5A CN201710793554A CN107677362A CN 107677362 A CN107677362 A CN 107677362A CN 201710793554 A CN201710793554 A CN 201710793554A CN 107677362 A CN107677362 A CN 107677362A
- Authority
- CN
- China
- Prior art keywords
- fft
- vibration signal
- real
- real cepstrum
- fast fourier
- 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
Abstract
The present invention discloses a kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis, including：Step 1, characteristic of rotating machines vibration signal is acquired, obtains discrete vibration signal time series；Step 2, real cepstrum processing is carried out to discrete vibration signal time series, obtains the real cepstrum analysis value corresponding to each point；Step 3, the time i.e. frequency reciprocal each put using discrete vibration signal time series, using the real cepstrum processing costs corresponding to each point as ordinate, is drawn as abscissa and obtains the real cepstrum processing figure for vibrating time-domain signal；Step 4, according to obtained real cepstrum processing figure, the basic modulating frequency of rotating machinery is identified.The modulating frequency of rotating machinery can simply, be directly identified in the real cepstrum processing figure of vibration time-domain signal using the present invention, obtained image information is more prominent distinct, and curvilinear motion becomes apparent from, and crest is more prominent.
Description
Technical field
The present invention relates to the frequency content of signal to extract field, more particularly to a kind of improved based on real cepstrum analysis
Rotating machinery modulating frequency extracting method.
Background technology
The frequency content extraction of signal belongs to the content of signal frequency domain analysis field, the existing frequency applied to actual signal
Domain analysis method mainly has Fast Fourier Transform (FFT), Short Time Fourier Transform and wavelet transformation etc..Fast Fourier Transform (FFT) is
Utilize WNThe periodicity and symmetry of the factor, a kind of fast and efficiently algorithm of construction, more can ideally describe steady letter
Number；Short Time Fourier Transform is by signal adding window, and the signal after adding window is carried out into Fast Fourier Transform (FFT), can describe signal
Two-dimentional time-frequency figure；Wavelet transformation is also to carry out adding window analysis to signal, but its window function shape can change, and can ensure signal
Frequency-domain analysis result have higher frequency resolution at low frequency, have higher temporal resolution in high frequency treatment, so as to realize
To the adaptive analysis of signal.
The time-domain information of actual signal can be converted into the information in frequency domain by these methods, but be directed to rotating machinery
Actual vibration signal for, three of the above method do not identify out sometimes in rotating machinery vibrating some modulate substantially
Frequency content, and these basic modulating frequencies include the extremely critical important informations such as rotor shaft frequency, blade and blade frequency, therefore it is non-
Often it is necessary to be identified.
《Scientific and technological visual field》(machinery and electronics) 94-103 pages of o. 11th " application of the cepstrum in vibration-testing " in 2014
Document disclose it is a kind of with cepstrum analysis technology identify frequency domain modulation signal side frequency composition method.Cepstrum has
Spectral line accurate positioning, amplitude protrude, the characteristics of can preferably identifying the side frequency composition of frequency domain modulation signal.Its shortcoming is to use
When cepstrum definition is handled actual signal, acquired image information is not obvious enough, and curvilinear motion is relatively gentle, symbol
The crest for modulating frequency is also not prominent enough, it is difficult to effectively obtains information needed.
The content of the invention
, can the invention provides a kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis
Overcome in rotating machinery real work in most cases, existing modulating frequency unobvious in spectrogram, it is not easy to quilt
The defects of identification, it can simple in the real cepstrum analysis figure of vibration time-domain signal, directly identify the modulation of rotating machinery
Frequency, obtained image information is more prominent distinct, and curvilinear motion becomes apparent from, and crest is more prominent.
A kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis, including：
Step 1, characteristic of rotating machines vibration signal is acquired, obtains discrete vibration signal time series；
Step 2, real cepstrum processing is carried out to discrete vibration signal time series, obtains the reality corresponding to each point and fall
Spectrum analysis value；
Step 3, time for each being put using discrete vibration signal time series i.e. corresponding frequency reciprocal as abscissa, with
Cepstrum processing costs after the corresponding noise reduction of each point is ordinate, draws and obtains the real cepstrum processing for vibrating time-domain signal
Figure；
Step 4, according to obtained real cepstrum processing figure, the basic modulating frequency of rotating machinery is identified.
In step 1, when being acquired to characteristic of rotating machines vibration signal, vibration signal sample frequency and sampling time interval
Relation be：
In formula：
fsFor the sample frequency of vibration signal；tsFor the sampling time interval of vibration signal；
Vibration signal sampling total duration and signal frequency resolution ratio relation be：
In formula：
f0For the frequency resolution of vibration signal；N is that the sampling of vibration signal is always counted；T is that the sampling of vibration signal is total
Duration.
In step 2, on the basis of real cepstrum processing is carried out to discrete vibration signal time series, to real cepstrum
The curve that assay value changes over time carries out noise reduction process, obtains the real scramble after the noise reduction corresponding to each point of time series
Compose processing costs.
Described noise reduction process, the sentence in MATLAB are：
C (t)=wden (c (t), ' sqtwolog ', ' s ', ' mln ', 3, ' db2 ')；
In formula：
C (t) is the real cepstrum processing costs after the noise reduction that each point is corresponding on Vibration Signal Time Series；C (t) is to shake
Taken absolute value again after the corresponding Fast Fourier Transform (FFT) of each point in dynamic signal time sequence, then natural logrithm taken to it,
Then make Fast Fourier Transform (FFT) inverse transformation to it, then it is carried out again to take real part to operate obtained real cepstrum analysis value；
Wden is wavelet de-noising；Sqtwolog is represented and is used fixed threshold principle；Behalf selection of threshold function mode is soft-threshold；mln
Represent and be adjusted according to the noise level estimation of each layer of wavelet decomposition；3 and db2 represents to enter signal using db2 small echo
3 layers of decomposition of row.
In step 2, the specific steps of described real cepstrum processing include：
Step 2-1, Fast Fourier Transform (FFT) is carried out to discrete vibration signal time series；
Step 2-2, on the basis of Fast Fourier Transform (FFT), acquired results are taken absolute value；
Step 2-3, on the basis of taking absolute value, natural logrithm is taken to acquired results；
Step 2-4, on the basis of natural logrithm is taken, inverse fast Fourier transform is carried out to acquired results；
Step 2-5, on the basis of inverse fast Fourier transform, to it take the operation of real part, obtain vibration signal
The corresponding real cepstrum analysis value of each point in time series.
Wherein：In step 2-1, described Fast Fourier Transform (FFT), the sentence in MATLAB is：
Fft-t=fft (t)
In formula：
Fft-t is the Fast Fourier Transform (FFT) result that each point is corresponding on Vibration Signal Time Series；T is discrete original
Beginning vibration signal time series；Fft represents fast Fourier direct transform.
In step 2-2, described takes absolute value, and the sentence in MATLAB is：
Abs-fft-t=abs (fft-t)
In formula：
Abs-fft-t is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes definitely again afterwards
The result of value；Abs represents to take absolute value.
In step 2-3, described takes natural logrithm, and the sentence in MATLAB is：
Ln-abs-fft-t=log (abs-fft-t)
In formula：
Ln-abs-fft-t is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes again afterwards
Absolute value, the result of natural logrithm is then taken to it again；Log represents to take natural logrithm.
In step 2-4, the inverse transformation of described Fast Fourier Transform (FFT), the sentence in MATLAB is：
Ifft-ln-abs-fft-t=ifft (ln-abs-fft-t)
In formula：
After ifft-ln-abs-fft-t is the corresponding Fast Fourier Transform (FFT) of each point on Vibration Signal Time Series
Take absolute value again, then natural logrithm is taken to it, then make the result of Fast Fourier Transform (FFT) inverse transformation to it again；Ifft represents fast
Fast inverse Fourier transform.
In step 2-5, described sentence of the operation of real part in MATLAB that take is：
C (t)=real (ifft-ln-abs-fft-t)
In formula：
C (t) is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes absolute value again afterwards, then
Natural logrithm is taken to it, then makees Fast Fourier Transform (FFT) inverse transformation to it, then it is carried out again to take real part to operate what is obtained
Real cepstrum analysis value；Real represents to plural number take the operation of real part.
Technical scheme provided by the invention uses real scramble spectrum processing method, only remains the spectral amplitude information of signal,
Phase information has been abandoned, has realized effective extraction to the rotor shaft frequency of rotating machinery, by being carried out to real cepstrum processing costs
Appropriate noise reduction process, actual vibration signal can be largely eliminated because the interference that grass comes, solve existing fast
The methods of fast Fourier transformation, Short Time Fourier Transform and wavelet transformation, can not significantly identify one in rotating machinery vibrating
The defects of a little basic modulating frequency compositions.
Using the present invention can in the real cepstrum analysis figure of vibration time-domain signal it is simple, directly identify rotating machinery
Modulating frequency, obtained image information is more prominent distinct, and curvilinear motion becomes apparent from, and crest is more prominent.With reference to rotating machinery
Machine moving law, axle frequency, the Ye Pin of rotating machinery can be effectively further identified, for the performance point of rotating machinery
Analysis and status monitoring are significant.
Brief description of the drawings
Fig. 1 is the real cepstrum processing figure of original emulation signal；
Fig. 2 is the real cepstrum processing figure of improved emulation signal；
The real cepstrum processing figure of the emulation signal for the white noise that Fig. 3 is plus signal to noise ratio is 20dB；
Fig. 4 a are plus the processing of the real cepstrum of 15dB noise Simulation signals is schemed；
Fig. 4 b are plus the processing of the real cepstrum of 12dB noise Simulation signals is schemed；
Fig. 4 c are plus the processing of the real cepstrum of 9dB noise Simulation signals is schemed；
Fig. 4 d are plus the processing of the real cepstrum of 6dB noise Simulation signals is schemed；
Fig. 4 e are plus the processing of the real cepstrum of 3dB noise Simulation signals is schemed；
Fig. 4 f are plus the processing of the real cepstrum of 0dB noise Simulation signals is schemed；
Fig. 5 is the vibration signal time-domain diagram of small blower fan 50HZ normal works；
Fig. 6 is the real cepstrum processing figure of small blower fan 10HZ work；
Fig. 7 is the real cepstrum processing figure of small blower fan 20HZ work；
Fig. 8 is the real cepstrum processing figure of small blower fan 30HZ work；
Fig. 9 is the real cepstrum processing figure of small blower fan 40HZ work；
Figure 10 is the real cepstrum processing figure of small blower fan 50HZ work；
The real cepstrum processing figure that Figure 11 is the attached heavy 10 grams of working frequency 10HZ of small blower fan；
The real cepstrum processing figure that Figure 12 is the attached heavy 10 grams of working frequency 20HZ of small blower fan；
The real cepstrum processing figure that Figure 13 is the attached heavy 20 grams of working frequency 10HZ of small blower fan；
The real cepstrum processing figure that Figure 14 is the attached heavy 20 grams of working frequency 20HZ of small blower fan.
Embodiment
In order that the object, technical solutions and advantages of the present invention are clearer, the present invention is made below in conjunction with accompanying drawing into
One step it is described in detail.
The improved rotating machinery modulating frequency extracting method based on real cepstrum analysis proposed by the present invention, respectively to imitative
True signal and actual signal carry out Treatment Analysis, and specific steps are described as follows：
(1) according to the sample frequency f for being actually needed selected vibration signals, and then determine sampling time interval ts, choose proper
When sampling total duration T, on this condition, the vibration signal of rotating machinery is measured.
Shown in vibration signal sample frequency and the relation of sampling time interval such as formula (1),
In formula：
fsFor the sample frequency of vibration signal；
tsFor the sampling time interval of vibration signal；
The relation such as formula (2) that vibration signal samples total duration and signal frequency resolution ratio is shown,
In formula：
f0For the frequency resolution of vibration signal；
N is that the sampling of vibration signal is always counted；
T is the sampling total duration of vibration signal.
(2) on the basis of discrete vibration signal time series is obtained, Fast Fourier Transform (FFT) is carried out to it, shaken
The corresponding Fast Fourier Transform (FFT) result of each point in dynamic signal time sequence.
MATLAB (is used for the advanced techniques computational language of algorithm development, data visualization, data analysis and numerical computations
And interactive environment) in sentence be：
Fft-t=fft (t)
In formula：
Fft-t is the Fast Fourier Transform (FFT) result that each point is corresponding on Vibration Signal Time Series；
T is discrete original vibration signal time series；Fft represents fast Fourier direct transform.
(3) on Vibration Signal Time Series are obtained on the basis of the corresponding Fast Fourier Transform (FFT) result of each point,
It is taken absolute value, obtains the value corresponding to each point.
Sentence in MATLAB is：
Abs-fft-t=abs (fft-t)
In formula：
Abs-fft-t is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes definitely again afterwards
The result of value；Abs represents to take absolute value.
(4) on the basis of the value that previous step is calculated that each point is corresponding on Vibration Signal Time Series, it is taken
Natural logrithm, obtain the value corresponding to each point.
Sentence in MATLAB is：
Ln-abs-fft-t=log (abs-fft-t)
In formula：
Ln-abs-fft-t is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes again afterwards
Absolute value, the result of natural logrithm is then taken to it again；Log represents to take natural logrithm.
(5) on the basis of the value that previous step is calculated that each point is corresponding on Vibration Signal Time Series, it is entered
The inverse transformation of row Fast Fourier Transform (FFT), obtain the value corresponding to each point.
Sentence in MATLAB is：
Ifft-ln-abs-fft-t=ifft (ln-abs-fft-t)
In formula：
After ifft-ln-abs-fft-t is the corresponding Fast Fourier Transform (FFT) of each point on Vibration Signal Time Series
Take absolute value again, then natural logrithm is taken to it, then make the result of Fast Fourier Transform (FFT) inverse transformation to it again；Ifft represents fast
Fast inverse Fourier transform.
(6) on the basis of the value that previous step is calculated that each point is corresponding on Vibration Signal Time Series, it is entered
Row takes the operation of real part, obtains the real cepstrum analysis value that each point is corresponding on Vibration Signal Time Series.
Sentence in MATLAB is：
C (t)=real (ifft-ln-abs-fft-t)
In formula：
C (t) is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes absolute value again afterwards, then
Natural logrithm is taken to it, then makees Fast Fourier Transform (FFT) inverse transformation to it, then it is carried out again to take real part to operate what is obtained
Real cepstrum analysis value；Real represents to plural number take the operation of real part.
(7) base of the real cepstrum analysis value that each point is corresponding on Vibration Signal Time Series is calculated in previous step
On plinth, appropriate noise reduction process is carried out to it, corresponding real cepstrum processing costs is each put after obtaining noise reduction.
Sentence in MATLAB is：
C (t)=wden (c (t), ' sqtwolog ', ' s ', ' mln ', 3, ' db2 ')；
In formula：
C (t) is the real cepstrum processing costs after the noise reduction that each point is corresponding on Vibration Signal Time Series；Wden is small
Ripple noise reduction, wavelet de-noising is carried out to c (t) and obtains C (t)；Sqtwolog is represented and is used fixed threshold principle；Behalf threshold function table
Selection mode is soft-threshold；Mln represents to be adjusted according to the noise level estimation of each layer of wavelet decomposition；3 and db2 represents profit
3 layers of decomposition are carried out to signal with db2 small echo.
(8) in the real cepstrum processing costs being calculated after corresponding noise reduction is each put on Vibration Signal Time Series
On the basis of, with the inverse of corresponding time for each being put on Vibration Signal Time Series for abscissa, with vibration obtained in the previous step
Real cepstrum processing costs in signal time sequence after the corresponding noise reduction of each point is ordinate, corresponds, is drawn,
The real cepstrum analysis figure (using frequency as abscissa) of vibration signal is obtained, so as to intuitively simply identify, extract and shake
Modulating frequency in dynamic signal.
In order to highlight the superiority of the present invention, this example is by original emulation signal, improved emulation signal, plus noise
Emulation signal and actual signal, noise reduction after actual signal carry out the processing based on real scramble spectral technology, by tune therein
Frequency content processed identifies.
(1) emulation signal is：
Y (t)=(cos (2 π 10t)+cos (2 π 20t)+cos (2 π 30t)) sin (2 π 100t)
This, which is one, modulates the signal that fundamental frequency is 10HZ, while also 10HZ two frequencys multiplication, the modulating frequency of frequency tripling, this
A little frequencies are together modulated to 100HZ frequency, obtain original emulation signal.
After carrying out real cepstrum processing to original emulation signal, Fig. 1 is obtained, it can be found that occurring first at 10HZ
Individual more obvious crest, and 10HZ is exactly the modulation fundamental frequency of original emulation signal.
(2) detection of multinomial modulated signal
Emulation signal in (1) is improved, makes to contain more modulation fundamental frequency in its modulated signal, and in improved emulation
It is 20dB white noise plus a signal to noise ratio on signal, then real cepstrum processing is carried out to it.Emulation signal after improvement is：
Y (t)=(cos (2 π 10t)+cos (2 π 20t)+cos (2 π 30t)+cos (2 π 7t)+cos (2 π
14t)+cos(2π·21t)+cos(2π·28t))·sin(2π·100t)
As can be seen that in contrast to original emulation signal, modulated signal is on the basis of original 10HZ, 20HZ, 30HZ
7HZ, 14HZ, 21HZ, 28HZ frequency content are added, is made up of 10HZ and its harmonic wave and 7HZ and its harmonic wave.To improved imitative
After true signal carries out real cepstrum processing, Fig. 2 is obtained, it is found that occur two most obvious ripples near 7HZ and 10HZ
Peak, this demonstrate that this method can not only identify single modulation fundamental frequency, moreover it is possible to while identify multiple different modulation fundamental frequencies.
(3) in original emulation basis of signals, plus the white noise that a signal to noise ratio is 20dB, the emulation to plus noise
After signal carries out real cepstrum processing, Fig. 3 is obtained, it can be found that obvious crest at 10HZ be present, can effectively identify tune
Fundamental frequency processed.
In order to verify, the signal to noise ratio size of plus noise identifies the influence of modulating frequency for the present invention, in original emulation
In basis of signals, made an uproar respectively plus the white noise that signal to noise ratio is 15dB, 12dB, 9dB, 6dB, 3dB, 0dB adding to these successively
After the emulation signal of sound carries out real cepstrum processing, Fig. 4 a~Fig. 4 f are obtained, it can be found that the presence that remained unchanged at 10HZ is more bright
Aobvious crest, it was demonstrated that the present invention still can effectively identify modulation fundamental frequency, still in the case where certain noise jamming be present
With the reduction of institute's plus noise signal to noise ratio, the crest at basic modulating frequency gradually becomes no longer to be most obvious crest, this
Illustrate to identify that basic modulating frequency becomes more and more difficult, even if but institute's plus noise signal to noise ratio be 0dB in the case of,
Still basic modulating frequency can effectively be identified, illustrate that the present invention can be extracted effectively in Low SNR signal
Modulating frequency.
(4) vibration signal processing of small blower fan normal work
Handled for the actual data analysis of rotating machinery, small blower fan is have chosen as typical rotating machinery, it is determined that adopting
Sample frequency fsFor 5120HZ, sampling total duration T is 15s, to small blower fan with 10HZ, 20HZ, 30HZ, 40HZ, 50HZ frequency just
Often vibration signal during work is measured, and its medium and small blower fan is with actual vibration signal time-domain diagram during 50HZ normal works as schemed
Shown in 5：
The vibration signal to be worked respectively with 10HZ, 20HZ, 30HZ, 40HZ, 50HZ frequency small blower fan carries out real cepstrum
After processing, Fig. 6 is obtained to Figure 10.
It can be found that in figure 6 and figure 7, there is crest the most obvious in 10HZ and 20HZ or so place, and now small wind
The working frequency of machine is exactly 10HZ and 20HZ.In other words, can have been found by finding the crest occurred in real scramble spectrogram
The frequency content of primary modulation effect, and when small blower fan is worked with relatively low working frequency, primary modulation frequency is exactly its turn
Sub- frequency.
Fig. 9 and Figure 10 is observed, it can be found that when small blower fan is worked with 40HZ or 50HZ frequency, corresponding vibration letter
In number scramble spectrogram, most obvious crest be no longer at corresponding at rotor frequency, and occur from the half of rotor frequency
At frequency multiplication, such as in fig.9, more obvious first crest is at abscissa 19.39HZ, precisely working frequency 40HZ
Half or so, in Fig. 10, more obvious first crest is at abscissa 24.98HZ, precisely working frequency
50HZ half or so.This also just illustrates that, when small blower fan is operated with 40HZ, 50HZ etc. of a relatively high working frequency, it shakes
The basic modulating frequency that in dynamic signal major frequency components are played with modulating action is no longer rotor frequency, but rotor frequency
Half frequency multiplication.
Small blower fan is handled with the vibration signal reality cepstrum of 10HZ, 20HZ, 30HZ, 40HZ, 50HZ frequency work respectively
Result figure carries out across comparison, it can be found that：When small blower fan is operated with relatively low working frequency, such as with 10HZ, 20HZ
Working frequency work when, can be by its basic modulating frequency of real scramble spectrum discrimination, now its basic modulating frequency is exactly turn
Sub- frequency.With gradually stepping up for working frequency, basic modulating frequency is little by little become two points of rotor frequency by rotor frequency
One of frequency multiplication, when such as being worked with 40HZ, 50HZ working frequency, the most obvious modulating frequency that cepstrum identifies has become
Into the half frequency multiplication of rotor frequency.And when being worked with intermediate frequency 30HZ, as shown in figure 8, scheming in the processing of real cepstrum
On can either find the crest for representing rotor frequency, also can find crest at the half frequency multiplication of rotor frequency, still
Now two crests are not particularly evident, and in the suitable crest of also some other height and steep, this proof
Now the main component of modulating frequency is more dispersed, both includes rotor frequency, further comprises the half times of rotor frequency
The multi-frequency composition such as frequency and other integer harmonics, fraction frequency multiplication, is not to concentrate on a certain major frequency components.
(5) the unbalanced vibration signal processing of small fan rotor
The essence of rotor unbalance failure is exactly the bias of rotary part quality, because the barycenter of rotary part deviates rotation
Axis, a unbalanced centrifugal intertia force will be produced when the rotor rotates so that plant equipment produce periodically
Vibration.In order to realize the fault condition of rotor unbalance, increase on some blade of blower fan attached heavy known to quality so that
The situation of mass eccentricity occurs for rotor, and attached heavy amount has 10 grams and 20 grams of two kinds of situations, and the working frequency of blower fan then have selected
Two kinds of situations of 10HZ and 20HZ, why without bigger attached heavy amount and broader operating frequency range is selected, be because
Continue in the case of increasing attached heavy amount and heightening working frequency, will occur the attached heavy unsafe condition thrown away, this is due to
It is attached it is heavy be to be built by the magnet that can adsorb the known weight on blade, once the centrifugal force suffered by magnet exceedes blade
To its absorption affinity, magnet will be thrown away.
In the attached heavy rotor unbalance experiment of blade, four groups of experiments have been carried out altogether, have been attached heavy 10 grams of working frequencies respectively
Four kinds of situations such as 10HZ, attached heavy 10 grams of working frequency 20HZ, attached heavy 20 grams of working frequency 10HZ and attached heavy 20 grams of 20HZ, shake to it
Dynamic signal carries out real cepstrum and handles to obtain Figure 11 to Figure 14.
Comparison diagram 11 and Figure 12, Figure 13 and 14 it can be found that attached heavy phase with the case of, as rotor frequency is by 10HZ
Be changed into 20HZ, most obvious crest is also changed into 20HZ from 10HZ in real scramble spectrogram, by real cepstrum still can intuitively, have
Read out to effect basic modulating frequency --- rotor frequency now.But it has also been discovered that with the increase of rotor frequency, turn
Main crest height representated by sub- frequency declines, and the number and amplitude of other secondary peaks all increased, and this exists with small blower fan
Variation tendency when working frequency gradually increases under normal circumstances is also consistent.
Comparison diagram 11 and Figure 13, Figure 12 with Figure 14 it can be found that when frequency is identical, as attached heavy amount is become by 10 grams
20 grams, in real cepstrum the holding that remains unchanged of most obvious crest abscissa positions be basically unchanged, but under the highly significant of main crest
Drop, also become less obvious with the gap of surrounding other secondary peaks, this illustrates the composition of modulating frequency also to become no longer
Focus primarily upon at rotor frequency, but other more modulating frequency compositions occur, this brings with working frequency increase
Influence is relatively.
Summarized according to case above, under the operating mode of rotor unbalance, with the increase of rotating machinery working frequency,
The increase of unbalance mass, can cause modulating frequency from only exist a main modulating frequency composition be changed into existing simultaneously it is a variety of
More significant modulating frequency composition, and by real cepstrum analysis, it is main that these in rotating machinery can be efficiently identified out
Modulating frequency and its change procedure.
The preferred embodiment of the present invention is the foregoing is only, protection scope of the present invention is not limited in above-mentioned embodiment party
Formula, every technical scheme for belonging to the principle of the invention belong to protection scope of the present invention.For those skilled in the art
Speech, some improvements and modifications carried out on the premise of the principle of the present invention is not departed from, these improvements and modifications also should be regarded as this
The protection domain of invention.
Claims (10)
1. a kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis, including：
Step 1, characteristic of rotating machines vibration signal is acquired, obtains discrete vibration signal time series；
Step 2, real cepstrum processing is carried out to discrete vibration signal time series, obtains the real cepstrum corresponding to each point
Assay value；
Step 3, the time i.e. corresponding frequency reciprocal each put using discrete vibration signal time series is abscissa, with each
The corresponding real cepstrum processing costs of point is ordinate, draws and obtains the real cepstrum processing figure for vibrating time-domain signal；
Step 4, according to obtained real cepstrum processing figure, the basic modulating frequency of rotating machinery is identified.
2. the improved rotating machinery modulating frequency extracting method based on real cepstrum analysis according to claim 1, its
It is characterised by, in step 1, when being acquired to characteristic of rotating machines vibration signal, vibration signal sample frequency and between the sampling time
Every relation be：
<mrow>
<msub>
<mi>f</mi>
<mi>s</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>t</mi>
<mi>s</mi>
</msub>
</mfrac>
</mrow>
In formula：
fsFor the sample frequency of vibration signal；tsFor the sampling time interval of vibration signal；
Vibration signal sampling total duration and signal frequency resolution ratio relation be：
<mrow>
<msub>
<mi>f</mi>
<mn>0</mn>
</msub>
<mo>=</mo>
<mfrac>
<msub>
<mi>f</mi>
<mi>s</mi>
</msub>
<mi>N</mi>
</mfrac>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mi>N</mi>
<mo>&CenterDot;</mo>
<msub>
<mi>t</mi>
<mi>s</mi>
</msub>
</mrow>
</mfrac>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>T</mi>
</mfrac>
</mrow>
In formula：
f0For the frequency resolution of vibration signal；N is that the sampling of vibration signal is always counted；T is the sampling total duration of vibration signal.
3. the improved rotating machinery modulating frequency extracting method based on real cepstrum analysis according to claim 1, its
It is characterised by, in step 2, on the basis of real cepstrum processing is carried out to discrete vibration signal time series, to real scramble
The curve that spectrum assay value changes over time carries out noise reduction process, obtains the reality after the noise reduction corresponding to each point of time series and falls
Frequency spectrum processing value.
4. the improved rotating machinery modulating frequency extracting method based on real cepstrum analysis according to claim 3, its
It is characterised by, described noise reduction process, the sentence in MATLAB is：
C (t)=wden (c (t), ' sqtwolog ', ' s ', ' mln ', 3, ' db2 ')；
In formula：
C (t) is the real cepstrum processing costs after the noise reduction that each point is corresponding on Vibration Signal Time Series；C (t) believes for vibration
Taken absolute value again after the corresponding Fast Fourier Transform (FFT) of each point in number time series, then natural logrithm is taken to it, then
Make Fast Fourier Transform (FFT) inverse transformation to it, then it is carried out again to take real part to operate obtained real cepstrum analysis value；wden
For wavelet de-noising；Sqtwolog is represented and is used fixed threshold principle；Behalf selection of threshold function mode is soft-threshold；Mln is represented
It is adjusted according to the noise level estimation of each layer of wavelet decomposition；3 and db2 represents carries out 3 layers using db2 small echo to signal
Decompose.
5. the improved rotating machinery modulating frequency extraction side based on real cepstrum analysis according to claim 1 or 3 or 4
Method, it is characterised in that in step 2, the specific steps of described real cepstrum processing include：
Step 2-1, Fast Fourier Transform (FFT) is carried out to discrete vibration signal time series；
Step 2-2, on the basis of Fast Fourier Transform (FFT), acquired results are taken absolute value；
Step 2-3, on the basis of taking absolute value, natural logrithm is taken to acquired results；
Step 2-4, on the basis of natural logrithm is taken, inverse fast Fourier transform is carried out to acquired results；
Step 2-5, on the basis of inverse fast Fourier transform, to it take the operation of real part, obtain the vibration signal time
The corresponding real cepstrum analysis value of each point in sequence.
6. the improved rotating machinery modulating frequency extracting method based on real cepstrum analysis according to claim 5, its
It is characterised by, in step 2-1, described Fast Fourier Transform (FFT), the sentence in MATLAB is：
Fft-t=fft (t)
In formula：
Fft-t is the Fast Fourier Transform (FFT) result that each point is corresponding on Vibration Signal Time Series；T is discrete original shakes
Dynamic signal sequence sequence；Fft represents fast Fourier direct transform.
7. the improved rotating machinery modulating frequency extracting method based on real cepstrum analysis according to claim 5, its
It is characterised by, in step 2-2, described takes absolute value, and the sentence in MATLAB is：
Abs-fft-t=abs (fft-t)
In formula：
Abs-fft-t is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes absolute value again afterwards
As a result；Abs represents to take absolute value.
8. the improved rotating machinery modulating frequency extracting method based on real cepstrum analysis according to claim 5, its
It is characterised by, in step 2-3, described takes natural logrithm, and the sentence in MATLAB is：
Ln-abs-fft-t=log (abs-fft-t)
In formula：
Ln-abs-fft-t is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes definitely again afterwards
Value, the result of natural logrithm is then taken to it again；Log represents to take natural logrithm.
9. the improved rotating machinery modulating frequency extracting method based on real cepstrum analysis according to claim 5, its
It is characterised by, in step 2-4, the inverse transformation of described Fast Fourier Transform (FFT), the sentence in MATLAB is：
Ifft-ln-abs-fft-t=ifft (ln-abs-fft-t)
In formula：
Ifft-ln-abs-fft-t is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes again afterwards
Absolute value, then natural logrithm is taken to it, then make the result of Fast Fourier Transform (FFT) inverse transformation to it again；Ifft represents quick Fu
In leaf inverse transformation.
10. the improved rotating machinery modulating frequency extracting method based on real cepstrum analysis according to claim 5, its
It is characterised by, in step 2-5, described sentence of the operation of real part in MATLAB that take is：
C (t)=real (ifft-ln-abs-fft-t)
In formula：
C (t) is that the Fast Fourier Transform (FFT) that each point is corresponding on Vibration Signal Time Series takes absolute value again afterwards, then to it
Natural logrithm is taken, then makees Fast Fourier Transform (FFT) inverse transformation to it, then its reality for carrying out taking real part to operate to obtain is fallen again
Spectrum analysis value；Real represents to plural number take the operation of real part.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710793554.5A CN107677362A (en) | 2017-09-06 | 2017-09-06 | A kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710793554.5A CN107677362A (en) | 2017-09-06 | 2017-09-06 | A kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107677362A true CN107677362A (en) | 2018-02-09 |
Family
ID=61135575
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710793554.5A Pending CN107677362A (en) | 2017-09-06 | 2017-09-06 | A kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107677362A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108593296A (en) * | 2018-04-26 | 2018-09-28 | 济南大学 | A kind of bearing Single Point of Faliure diagnostic method based on cepstrum puppet back gauge |
| CN108709635A (en) * | 2018-08-17 | 2018-10-26 | 国家电网有限公司 | Determine the method and device of characteristic of rotating machines vibration signal major frequency components information |
| CN108843611A (en) * | 2018-06-15 | 2018-11-20 | 浙江大学 | A kind of axial flow blower state identification method based on vibration signal combined spectrum correlation analysis |
| CN110232251A (en) * | 2019-06-18 | 2019-09-13 | 西安微电子技术研究所 | A kind of noise Simulation analysis method suitable for sequential switch circuit |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030028332A1 (en) * | 2001-06-19 | 2003-02-06 | The Aerospace Corporation | Rocket engine gear defect monitoring method |
| CN101464212A (en) * | 2007-12-19 | 2009-06-24 | 西门子公司 | Method and apparatus for vibration-based automatic condition monitoring of a wind turbine |
| RU2363936C1 (en) * | 2008-05-27 | 2009-08-10 | Общество с ограниченной ответственностью НПЦ "Динамика"-Научно-производственный центр "Диагностика", надежность машин и комплексная автоматизация" | Method for vibration-based diagnostics of objects |
-
2017
- 2017-09-06 CN CN201710793554.5A patent/CN107677362A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030028332A1 (en) * | 2001-06-19 | 2003-02-06 | The Aerospace Corporation | Rocket engine gear defect monitoring method |
| CN101464212A (en) * | 2007-12-19 | 2009-06-24 | 西门子公司 | Method and apparatus for vibration-based automatic condition monitoring of a wind turbine |
| RU2363936C1 (en) * | 2008-05-27 | 2009-08-10 | Общество с ограниченной ответственностью НПЦ "Динамика"-Научно-производственный центр "Диагностика", надежность машин и комплексная автоматизация" | Method for vibration-based diagnostics of objects |
Non-Patent Citations (2)
| Title |
|---|
| 汤武初 等: ""倒频谱在压缩机故障诊断中的应用"", 《噪声与振动控制》 * |
| 韩捷 等: "《旋转机械故障机理及诊断技术》", 31 August 1997, 机械工业出版社 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108593296A (en) * | 2018-04-26 | 2018-09-28 | 济南大学 | A kind of bearing Single Point of Faliure diagnostic method based on cepstrum puppet back gauge |
| CN108593296B (en) * | 2018-04-26 | 2020-01-31 | 济南大学 | bearing single-point fault diagnosis method based on cepstrum pseudo-edge distance |
| CN108843611A (en) * | 2018-06-15 | 2018-11-20 | 浙江大学 | A kind of axial flow blower state identification method based on vibration signal combined spectrum correlation analysis |
| CN108843611B (en) * | 2018-06-15 | 2019-08-27 | 浙江大学 | A kind of axial flow blower state identification method based on vibration signal combined spectrum correlation analysis |
| CN108709635A (en) * | 2018-08-17 | 2018-10-26 | 国家电网有限公司 | Determine the method and device of characteristic of rotating machines vibration signal major frequency components information |
| CN108709635B (en) * | 2018-08-17 | 2020-07-28 | 国家电网有限公司 | Method and device for determining main frequency component information of rotary mechanical vibration signal |
| CN110232251A (en) * | 2019-06-18 | 2019-09-13 | 西安微电子技术研究所 | A kind of noise Simulation analysis method suitable for sequential switch circuit |
| CN110232251B (en) * | 2019-06-18 | 2022-09-27 | 西安微电子技术研究所 | Noise simulation analysis method suitable for time sequence switch circuit |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107677362A (en) | A kind of improved rotating machinery modulating frequency extracting method based on real cepstrum analysis | |
| Yang et al. | Fault diagnosis for a wind turbine generator bearing via sparse representation and shift-invariant K-SVD | |
| Wang et al. | Rotating speed isolation and its application to rolling element bearing fault diagnosis under large speed variation conditions | |
| Zhang et al. | Bearing fault diagnosis based on variational mode decomposition and total variation denoising | |
| Yang et al. | An ensemble local means decomposition method and its application to local rub-impact fault diagnosis of the rotor systems | |
| Yang et al. | A fault diagnosis approach for roller bearing based on VPMCD under variable speed condition | |
| Meng et al. | Fault diagnosis of rolling bearing based on second generation wavelet denoising and morphological filter | |
| Li et al. | Detection of bearing faults using a novel adaptive morphological update lifting wavelet | |
| Li et al. | An improved Viterbi algorithm for IF extraction of multicomponent signals | |
| CN109029999B (en) | Rolling bearing fault diagnosis method based on enhanced modulation bispectrum analysis | |
| Xiong et al. | A new procedure for extracting fault feature of multi-frequency signal from rotating machinery | |
| Barbini et al. | Phase editing as a signal pre-processing step for automated bearing fault detection | |
| Lopatinskaia et al. | Monitoring varying speed machinery vibrations—II recursive filters and angle domain | |
| Luo et al. | The diagnosis approach for rolling bearing fault based on Kurtosis criterion EMD and Hilbert envelope spectrum | |
| Varanis et al. | Statements on wavelet packet energy–entropy signatures and filter influence in fault diagnosis of induction motor in non-stationary operations | |
| Hua et al. | Matching and reassignment based time-frequency enhancement for rotating machinery fault diagnosis under nonstationary speed operations | |
| CN109900447B (en) | Cyclic impact vibration detection method and system based on harmonic signal decomposition | |
| CN113761466B (en) | Method and device for constructing vibration signal order ratio spectrum of rotary machine | |
| Sharma | Gear fault detection based on instantaneous frequency estimation using variational mode decomposition and permutation entropy under real speed scenarios | |
| Prosvirin et al. | Efficient rub-impact fault diagnosis scheme based on hybrid feature extraction and SVM | |
| Khelfi et al. | Temporal envelope detection by the square root of the three-phase currents for IM rotor fault diagnosis | |
| Garg et al. | Current signature analysis and its application in the condition monitoring of wind turbine for rotor faults | |
| CN108917918A (en) | One kind being directed to bearing vibration signal instantaneous frequency analysis and diagnosis method | |
| Sokołowski et al. | Multiple local damage detection method based on time-frequency representation and agglomerative hierarchical clustering of temporary spectral content | |
| CN103149047A (en) | Cooling tower acoustic diagnosis method based on nonlinear mixed model |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180209 |
|
| RJ01 | Rejection of invention patent application after publication |