CN110926594B - Method for extracting time-varying frequency characteristics of rotary machine signal - Google Patents
Method for extracting time-varying frequency characteristics of rotary machine signal Download PDFInfo
- Publication number
- CN110926594B CN110926594B CN201911155435.2A CN201911155435A CN110926594B CN 110926594 B CN110926594 B CN 110926594B CN 201911155435 A CN201911155435 A CN 201911155435A CN 110926594 B CN110926594 B CN 110926594B
- Authority
- CN
- China
- Prior art keywords
- time
- frequency
- signal
- real
- frequency distribution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
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
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M99/00—Subject matter not provided for in other groups of this subclass
- G01M99/005—Testing of complete machines, e.g. washing-machines or mobile phones
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention provides a method for extracting time-varying frequency characteristics of a rotary machine signal, which can overcome mutual interference of adjacent time-varying frequency components, remove noise components and obtain signal time-frequency distribution with high time-frequency resolution. The method comprises the following steps: acquiring an angular domain signal corresponding to a time domain signal of target rotating mechanical equipment, wherein the time domain signal of the target rotating mechanical equipment is an original signal; performing time-frequency analysis on the angular domain signals, and performing agent test on each row or part of rows of the obtained time-frequency distribution matrix to obtain real frequency components; reconstructing time domain signals of each real frequency component; and determining Hilbert time-frequency distribution of the time-domain signals of each real frequency component obtained by reconstruction, and superposing the Hilbert time-frequency distribution to obtain the time-frequency distribution of the original signal. The present invention relates to the field of condition monitoring and fault diagnosis of rotating mechanical equipment.
Description
Technical Field
The invention relates to the field of state monitoring and fault diagnosis of rotary mechanical equipment, in particular to a time-varying frequency feature extraction method for a rotary mechanical signal.
Background
The signal feature extraction technology is one of key common technologies in the fields of aerospace, energy power and the like. By accurately analyzing the amplitude and frequency structure of signals such as vibration and noise of mechanical equipment, the running characteristics of complex equipment and the health state of internal components can be revealed. In practical engineering application, a plurality of devices operate under time-varying working conditions, and acquired signals have the characteristic of a time-varying frequency structure. Conventional spectral analysis methods do not have the ability to reveal time-varying frequency structures.
The time-frequency analysis method expresses the signal frequency structure changing along with time and the amplitude information of the corresponding frequency components in a time-frequency joint domain. Compared with a spectrum analysis method, the time-frequency analysis method not only can reveal the frequency structure and amplitude intensity characteristics of the signal, but also can express the change rule of the characteristics along with time, and is more suitable for extracting time-varying frequency characteristics. In actual industrial production and scientific research activities, the rotating machinery often needs to operate under time-varying working conditions, and the collected rotating machinery signals often have time-varying characteristics. Therefore, the time-frequency distribution of the rotating mechanical signals is constructed through time-frequency analysis, and the characteristic information in the rotating mechanical signals can be visually expressed, so that the dynamic characteristics and the health state of a mechanical system are reflected, and the method is widely concerned in the field of state monitoring and fault diagnosis of large rotating mechanical equipment.
Chinese patent 201710593238.3 discloses a time-frequency analysis method based on adaptive adjustment of window function length. Firstly, carrying out fast Fourier transform on an acquired signal to obtain a signal frequency spectrum; secondly, determining a control factor according to the frequency spectrum characteristics and the resolution requirement; and finally, substituting the obtained control function into a window function, multiplying the window function by the signal frequency spectrum after dimension expansion after fast Fourier transform, and then obtaining time-frequency distribution through inverse Fourier transform. Chinese patent 201310256814.7 discloses a rotating machine order tracking method based on adaptive STFT. Firstly, determining a scale function of a window function according to the rotating speed trend; secondly, carrying out self-adaptive STFT of a variable window function according to a scale function; and finally, extracting order components from the obtained time-frequency distribution, and performing time-frequency inverse transformation to obtain an independent time-domain waveform, thereby extracting the characteristics of the rotating machinery.
The two methods are characterized in that the length of a window function in time-frequency analysis can be adjusted by utilizing the self frequency spectrum information of the signal or the collected rotating speed information, so that the time-frequency resolution precision can be improved to a certain extent compared with the time-frequency distribution of a fixed window function. However, under the influence of the heisenberg uncertainty, the constructed window function cannot achieve the highest precision in both time and frequency directions; in addition, noise components inevitably exist in the actually measured rotating machine signals, and the time-frequency distribution obtained by the method also inevitably has interference of the noise components. Therefore, for large-scale complex rotating mechanical equipment, when the complex amplitude and frequency modulation action generates densely distributed time-varying frequency components, adjacent time-varying frequency components are easily confused due to low time-frequency resolution, and noise components are not easily distinguished from real frequency components, so that the signal characteristics are wrongly identified, and the identification and judgment of the health state of mechanical equipment are influenced.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a method for extracting time-varying frequency characteristics of a rotating machinery signal, which can overcome mutual interference of adjacent time-varying frequency components, remove noise components, clearly extract densely distributed time-varying frequency components and obtain signal time-frequency distribution with high time-frequency resolution.
In order to solve the above technical problem, an embodiment of the present invention provides a method for extracting time-varying frequency characteristics of a rotating machine signal, including:
acquiring an angular domain signal corresponding to a time domain signal of target rotating mechanical equipment, wherein the time domain signal of the target rotating mechanical equipment is an original signal;
performing time-frequency analysis on the angular domain signals, and performing agent test on each row or part of rows of the obtained time-frequency distribution matrix to obtain real frequency components;
reconstructing time domain signals of each real frequency component;
and determining Hilbert time-frequency distribution of the time-domain signals of each real frequency component obtained by reconstruction, and superposing the Hilbert time-frequency distribution to obtain the time-frequency distribution of the original signal.
Further, the acquiring an angular domain signal corresponding to a time domain signal of the rotating mechanical device includes:
acquiring a time domain signal of target rotating mechanical equipment, wherein the time domain signal comprises: vibration, displacement, sound or electrical signals;
performing time-frequency analysis on the original signal to preliminarily obtain time-frequency distribution of the original signal;
extracting a frequency trend from the time-frequency distribution of the original signal;
and according to the extracted frequency trend, carrying out angular domain resampling on the original signal to obtain an angular domain signal.
Further, the acquiring an angular domain signal corresponding to a time domain signal of the rotating mechanical device includes:
acquiring a time domain signal of target rotating mechanical equipment, and synchronously acquiring a rotating speed signal; wherein the time domain signal comprises: vibration, displacement, sound or electrical signals;
and performing angular domain resampling on the original signal by using the rotating speed signal to obtain an angular domain signal.
Further, the performing time-frequency analysis on the angular domain signal, and performing agent test on each row or part of rows of the obtained time-frequency distribution matrix to obtain real frequency components includes:
performing time-frequency analysis on the angular domain signals to obtain a time-frequency distribution matrix;
independently extracting each line or partial lines of the obtained time-frequency distribution matrix into vectors and using the vectors as candidate components;
and automatically judging whether each candidate component is a real frequency component or not by utilizing a Fourier transform agent test method.
Further, the independently extracting each row of the obtained time-frequency distribution matrix as a vector and using the vector as a candidate component includes:
extracting each row of the obtained time-frequency distribution matrix as a vector to obtain K row vectors y of 1 XNk(N), where K is 1,2, … K and N is 1,2, … N, K representing the number of rows of the time-frequency distribution matrix and N representing the number of columns of the time-frequency distribution matrix;
the automatically judging whether each candidate component is a real frequency component by using the Fourier transform agent test method comprises the following steps:
determining a row vector ykInstantaneous frequency IF of (n)yk(n);
Based on the obtained instantaneous frequency IFyk(n) determining a row vector yk(n) instantaneous frequency entropy;
determining a row vector ykN of (N)sA Fourier transform proxy signal yβk(N), wherein β ═ 1,2, … Ns;
Determining a row vector ykAll proxy signals y of (n)βk(n) instantaneous frequency entropy;
determining a row vector yk(n) all proxy signals having an instantaneous frequency entropy greater than the row vector yk(n) a percentage of instantaneous frequency entropy, if said percentage is greater than a predetermined threshold,then the row vector yk(n) corresponding frequency component fk(t) is the true frequency content; otherwise, the row vector yk(n) corresponding frequency component fk(t) is random noise.
Further, the instantaneous frequency IFyk(n) is represented by:
IFyk(n)=(1/2π)[dφyk(n)/dn]
φyk(n)=arctan{H[yk(n)]/yk(n)},
wherein H (-) represents a Hilbert transform phiyk(n) is a shorthand form of instantaneous phase;
the instantaneous frequency entropy is expressed as:
wherein E isykRepresenting instantaneous frequency entropy, pk(m) represents the instantaneous frequency IFyk(n) a probability distribution at the mth value, M being 1,2, … M, M representing the instantaneous frequency IFyk(n) number of values;
fourier transform proxy signal yβk(n) is represented by:
yβk(n)=∫{Xk(f)exp[iγβ(f)]}exp(i2πfn)df
wherein, Xk(f) Denotes yk(n) Fourier transform, i denotes an imaginary unit, γβ(f) At a frequency value of f]Randomly taken phase within the range.
Further, the reconstructing the time domain signal of the real frequency component includes:
reconstructing an independent time domain waveform of a real frequency component by using a Vold-Kalman filter; or the like, or, alternatively,
extracting the angular domain waveform of the real frequency component by using a band-pass filter, then performing time domain resampling, and reconstructing the independent time domain waveform of the real frequency component; or the like, or, alternatively,
and extracting the amplitude envelope and instantaneous frequency of the real frequency components from the time-frequency distribution of the original signal, and reconstructing the independent time domain waveform of the real frequency components according to the extracted amplitude envelope and instantaneous frequency of the real frequency components.
Further, the determining hilbert time-frequency distribution of the time-domain signal of each real frequency component obtained by reconstruction, and superimposing the hilbert time-frequency distribution to obtain the time-frequency distribution of the original signal includes:
and calculating the amplitude envelope and the instantaneous frequency of the reconstructed real frequency components, independently constructing the Hilbert time-frequency distribution of each real frequency component, and overlapping the Hilbert time-frequency distributions of the real frequency components to obtain the time-frequency distribution of the original signal.
Further, the calculation formula of the hilbert time-frequency distribution is as follows:
TFRz(t,f)=a2(t)δ[f-f(t)]
wherein z (t) represents the reconstructed time-domain signal at time t, a (t) represents the amplitude envelope of the reconstructed time-domain signal, phi (t) represents the instantaneous phase of the reconstructed time-domain signal, f (t) represents the instantaneous frequency of the reconstructed time-domain signal, H (-) represents the Hilbert transform, delta (-) represents the Dirac function, TFR represents the time-domain signal, and f (t) represents the time-domain signalz(t, f) represents the time-frequency distribution matrix of z (t) signals, and f represents the frequency variation.
Further, the hilbert time-frequency distribution of the time-domain signal of each real frequency component obtained by reconstruction is determined, and the hilbert time-frequency distributions are superposed to obtain the time-frequency distribution of the original signal, which can be replaced by:
performing time-frequency analysis on the time domain signals of the real frequency components obtained by reconstruction, and constructing time-frequency distribution of each real frequency component; wherein the time-frequency analysis comprises: short-time fourier transform, continuous wavelet transform or wigner-vila transform;
and overlapping the time-frequency distribution of each real frequency component to obtain the time-frequency distribution of the original signal.
The technical scheme of the invention has the following beneficial effects:
in the scheme, before time-frequency transformation is carried out on a time-domain signal of rotating mechanical equipment, real frequency components are identified in a self-adaptive mode through proxy testing, the real frequency components are reconstructed into independent time-domain signals (the independent time-domain signals are single-component signals), time-frequency distribution of the real frequency components is constructed based on Hilbert transformation, and finally Hilbert time-frequency distribution of each real frequency component is overlapped.
Drawings
Fig. 1 is a schematic flow chart of a method for extracting time-varying frequency characteristics of a rotating machine signal according to an embodiment of the present invention;
fig. 2 is a detailed flowchart of a method for extracting time-varying frequency characteristics of a rotating machine signal according to an embodiment of the present invention;
fig. 3 is a schematic diagram of short-time fourier transform time-frequency distribution of a radial displacement signal of a certain turbine provided in an embodiment of the present invention;
fig. 4 is a schematic time-frequency distribution diagram of a radial displacement signal of a certain turbine rotor according to an embodiment of the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments.
As shown in fig. 1, a method for extracting time-varying frequency characteristics of a rotating machine signal according to an embodiment of the present invention includes:
s101, acquiring an angular domain signal corresponding to a time domain signal of target rotating mechanical equipment, wherein the time domain signal of the target rotating mechanical equipment is an original signal;
s102, carrying out time-frequency analysis on the angle domain signals, and carrying out agent test on each row or partial rows of the obtained time-frequency distribution matrix to obtain real frequency components;
s103, reconstructing time domain signals of each real frequency component;
and S104, determining the Hilbert time-frequency distribution of the time-domain signal of each real frequency component obtained by reconstruction, and superposing the Hilbert time-frequency distributions to obtain the time-frequency distribution of the original signal.
Before time-frequency transformation is carried out on a time-domain signal of rotating mechanical equipment, real frequency components are identified in a self-adaptive mode through proxy testing, the real frequency components are reconstructed into independent time-domain signals (the independent time-domain signals are single-component signals), time-frequency distribution of the real frequency components is constructed based on Hilbert transformation, and finally Hilbert time-frequency distribution of the real frequency components is overlapped.
In order to better understand the method for extracting time-varying frequency characteristics of a rotating machine signal provided by the embodiment of the present invention, as shown in fig. 2, the method for extracting time-varying frequency characteristics of a rotating machine signal is described in detail, and specifically may include the following steps:
s101, acquiring an angular domain signal corresponding to a time domain signal of target rotating mechanical equipment;
in this embodiment, an angular domain signal corresponding to a time domain signal of the target rotating mechanical device may be obtained by any one of the manners described in manner 1 or manner 2, where manner 1 may specifically include the following steps:
h1, acquiring a time domain signal of the target rotating mechanical equipment, wherein the time domain signal comprises: vibration, displacement, sound or electrical signals;
in this embodiment, time domain signals x (t) of the target rotating mechanical device are collected at equal time intervals, where the time domain signals may be vibration, displacement, sound, electrical signals, and the like, and t represents a sampling time.
H2, performing time-frequency analysis on the original signal to preliminarily obtain time-frequency distribution of the original signal, namely traditional time-frequency distribution;
in this embodiment, time-frequency analysis is performed on the time domain signal x (t) by methods such as short-time fourier transform, continuous wavelet transform, or wigner-vela transform, to obtain the time-frequency distribution of the original signal.
In this embodiment, a short-time fourier transform time-frequency distribution graph of a radial displacement signal of a certain turbine rotor is shown in fig. 3.
H3, extracting a frequency trend f (t) from the time-frequency distribution of the original signal;
in the present embodiment, f (t) may represent the rotational speed frequency at time t, for example.
H4, according to the extracted frequency trend f (t), carrying out angular domain resampling on the original signal x (t) to obtain an angular domain signalWherein the content of the first and second substances,indicating the phase of the rotor in the rotating machine,is a variable of the angular domain signal.
The method 2 may specifically include the following steps:
h1, acquiring time domain signals of the target rotating mechanical equipment, and synchronously acquiring rotating speed signals; wherein the time domain signal comprises: vibration, displacement, sound or electrical signals;
in this embodiment, the signal time domain signal x (t), such as vibration, displacement, sound, electrical signal, etc., of the target rotating mechanical device is collected at equal time intervals.
H2, performing angular domain resampling on the original signal x (t) by using the rotation speed signal to obtain an angular domain signal
S102, performing time-frequency analysis on the angle domain signal, and performing a proxy test on each row or a part of rows of the obtained time-frequency distribution matrix to obtain a real frequency component, which may specifically include the following steps:
a1, performing time-frequency analysis on the angle domain signals to obtain a time-frequency distribution matrix;
in this embodiment, diagonal domain signalsPerforming time-frequency analysis to obtain a time-frequency distribution matrix with the size of KxNWhere f represents a frequency variable.
A2, independently extracting each line or partial lines of the obtained time-frequency distribution matrix as vectors and using the vectors as candidate components;
in this embodiment, each behavior example is extracted for explanation:
extracting each row of the obtained time-frequency distribution matrix as a vector to obtain K row vectors y of 1 XNk(N), where K is 1,2, … K and N is 1,2, … N, K representing the number of rows of the time-frequency distribution matrix and N representing the number of columns of the time-frequency distribution matrix.
A3, automatically determining whether each candidate component is a real frequency component by using a fourier transform proxy test method, which may specifically include the following steps:
a31, determining a row vector ykInstantaneous frequency IF of (n)yk(n); wherein the instantaneous frequency IFyk(n) is represented by:
IFyk(n)=(1/2π)[dφyk(n)/dn]
φyk(n)=arctan{H[yk(n)]/yk(n)},
wherein H (-) represents a Hilbert transform phiyk(n) is a shorthand form of instantaneous phase;
a32, according to the obtained instantaneous frequency IFyk(n) determining a row vector yk(n) instantaneous frequency entropy; wherein the instantaneous frequency entropy is represented as:
wherein E isykRepresenting instantaneous frequency entropy, pk(m) represents the instantaneous frequency IFyk(n) a probability distribution at the mth value, M being 1,2, … M, M representing the instantaneous frequency IFyk(n) number of values;
a33, determining a row vector ykN of (N)sA Fourier transform proxy signal yβk(n) wherein the Fourier transform proxy signal yβk(n) is represented by:
yβk(n)=∫{Xk(f)exp[iγβ(f)]}exp(i2πfn)df
wherein, Xk(f) Denotes yk(n) Fourier transform, γβ(f) At a frequency value of f]Random phase in the range of 1,2, … Ns,NsThe number of constructed Fourier transform proxy signals is shown, and i represents an imaginary number unit;
a34, determining the row vector y according to the method of the steps A31 and A32kAll proxy signals y of (n)βk(n) instantaneous frequency entropy;
a35, determining a row vector yk(n) all proxy signals having an instantaneous frequency entropy greater than the row vector yk(n) a percentage of instantaneous frequency entropy, if said percentage is greater than a preset threshold, then the row vector yk(n) corresponding frequency component fk(t) is the true frequency content; otherwise, the row vector yk(n) corresponding frequency component fk(t) is random noise.
In this embodiment, for the row vector yk(N) determining its instantaneous frequency entropy and its NsThe magnitude of the instantaneous frequency entropy of the proxy signal is such that if any proxy signal exceeds a predetermined threshold (e.g., 95%) the ratio y isk(n) larger instantaneous frequency entropy, then judge the line vector yk(n) corresponding frequency component fk(t) true presence, i.e. fk(t) is a real frequency component.
In this embodiment, steps A31-A35 are repeated until all row vectors yk(n) all have been detected, all frequency components f that are actually presentk(t) has already been screened out.
In addition, the following should be noted:
if the original signal x (t) contains not only the harmonic component of the rotational speed frequency but also other frequency trends, in step H3, the corresponding frequency trend needs to be extracted, and step H4 and the following steps are performed based on the frequency trend until all real frequency components in the original signal are effectively screened out.
S103, reconstructing the time domain signal of each real frequency component by any one mode of 1), 2) and 3), wherein,
1) reconstructing independent time domain waveforms of real frequency components using a Vold-Kalman filter
2) Reconstruction of independent time domain waveforms of real frequency content using band pass filters
In the embodiment, the band-pass filter is used for extracting the angular domain waveform of the real frequency component, then time domain resampling is carried out, and the independent time domain waveform of the real frequency component is reconstructed;
3) reconstructing independent time domain waveform of real frequency component by ridge line reconstruction mode
In this embodiment, the amplitude envelope and the instantaneous frequency of the real frequency component are extracted from the time-frequency distribution of the original signal, and the independent time domain waveform of the real frequency component is reconstructed according to the extracted amplitude envelope and instantaneous frequency of the real frequency component.
And S104, determining the Hilbert time-frequency distribution of the time-domain signal of each real frequency component obtained by reconstruction, and superposing the Hilbert time-frequency distributions to obtain the time-frequency distribution of the original signal.
In this embodiment, the amplitude envelope and the instantaneous frequency of the reconstructed real frequency components are calculated, hilbert time-frequency distribution of each real frequency component is independently constructed, and the hilbert time-frequency distributions of the real frequency components are superimposed to obtain time-frequency distribution of the original signal.
In this embodiment, the calculation formula of the hilbert time-frequency distribution is as follows:
TFRz(t,f)=a2(t)δ[f-f(t)]
wherein z (t) represents the reconstructed time-domain signal at time t, a (t) represents the amplitude envelope of the reconstructed time-domain signal, phi (t) represents the instantaneous phase of the reconstructed time-domain signal, f (t) represents the instantaneous frequency of the reconstructed time-domain signal, H (-) represents the Hilbert transform, delta (-) represents the Dirac function, TFR represents the time-domain signal, and f (t) represents the time-domain signalz(t, f) represents the time-frequency distribution matrix of z (t) signals, and f represents the frequency variation.
In this embodiment, a schematic diagram of time-frequency distribution of a radial displacement signal of a certain water turbine rotor is shown in fig. 4, and it can be seen by comparing fig. 3 and fig. 4 that a traditional time-frequency analysis method represented by short-time fourier transform simultaneously expresses real frequency components and noise components on the time-frequency distribution, and because readability is not strong, feature recognition errors are easily caused.
In this embodiment, S104 may be replaced by: performing time-frequency analysis on the time domain signals of the real frequency components obtained by reconstruction, and constructing time-frequency distribution of each real frequency component; wherein the time-frequency analysis comprises: short-time fourier transform, continuous wavelet transform or wigner-vila transform; and overlapping the time-frequency distribution of each real frequency component to obtain the time-frequency distribution of the original signal.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (6)
1. A method for extracting time-varying frequency characteristics of a rotating machinery signal is characterized by comprising the following steps:
acquiring an angular domain signal corresponding to a time domain signal of target rotating mechanical equipment, wherein the time domain signal of the target rotating mechanical equipment is an original signal;
performing time-frequency analysis on the angular domain signals, and performing agent test on each row or part of rows of the obtained time-frequency distribution matrix to obtain real frequency components;
reconstructing time domain signals of each real frequency component;
determining Hilbert time-frequency distribution of the time-domain signals of each real frequency component obtained by reconstruction, and overlapping the Hilbert time-frequency distribution to obtain time-frequency distribution of the original signals;
wherein the time domain signal of the reconstructed real frequency component is: reconstructing an independent time domain waveform of a real frequency component by using a Vold-Kalman filter;
performing time-frequency analysis on the diagonal domain signals, and performing agent test on each row or part of rows of the obtained time-frequency distribution matrix to obtain real frequency components, wherein the step of obtaining the real frequency components comprises the following steps:
performing time-frequency analysis on the angular domain signals to obtain a time-frequency distribution matrix;
independently extracting each line or partial lines of the obtained time-frequency distribution matrix into vectors and using the vectors as candidate components;
automatically judging whether each candidate component is a real frequency component or not by utilizing a Fourier transform agent test method;
each row of the obtained time-frequency distribution matrix is independently extracted as a vector and comprises the following candidate components:
extracting each row of the obtained time-frequency distribution matrix as a vector to obtain K row vectors y of 1 XNk(N), where K is 1, 2.. K and N is 1, 2.. N, K representing the number of rows of the time-frequency distribution matrix and N representing the number of columns of the time-frequency distribution matrix;
the automatically judging whether each candidate component is a real frequency component by using the Fourier transform agent test method comprises the following steps:
determining a row vector ykInstantaneous frequency IF of (n)yk(n);
Based on the obtained instantaneous frequency IFyk(n) determining a row vector yk(n) instantaneous frequency entropy;
determining a row vector ykN of (N)sA Fourier transform proxy signal yβk(N), wherein β ═ 1, 2.. Ns;
Determining a row vector ykAll proxy signals y of (n)βk(n) instantaneous frequency entropy;
determining a row vector yk(n) all proxy signals having an instantaneous frequency entropy greater than the row vector yk(n) a percentage of instantaneous frequency entropy, if said percentage is greater than a preset threshold, then the row vector yk(n) corresponding frequency component fk(t) is the true frequency content; otherwise, the row vector yk(n) corresponding frequency component fk(t) is random noise;
instantaneous frequency IFyk(n) is represented by:
IFyk(n)=(1/2π)[dφyk(n)/dn]
φyk(n)=arctan{H[yk(n)]/yk(n)},
wherein H (-) represents a Hilbert transform phiyk(n) is a shorthand form of instantaneous phase;
the instantaneous frequency entropy is expressed as:
wherein E isykRepresenting instantaneous frequency entropy, pk(m) represents the instantaneous frequency IFyk(n) a probability distribution at the mth value, M ═ 1, 2.. M, M denotes the instantaneous frequency IFyk(n) number of values;
fourier transform proxy signal yβk(n) is represented by:
yβk(n)=∫{Xk(f)exp[iγβ(f)]}exp(i2πfn)df
wherein, Xk(f) Denotes yk(n) Fourier transform, i denotes an imaginary unit, γβ(f) At a frequency value of f]Randomly taken phase within the range.
2. The method according to claim 1, wherein the obtaining an angular domain signal corresponding to a time domain signal of a rotating machine device comprises:
acquiring a time domain signal of target rotating mechanical equipment, wherein the time domain signal comprises: vibration, displacement, sound or electrical signals;
performing time-frequency analysis on the original signal to preliminarily obtain time-frequency distribution of the original signal;
extracting a frequency trend from the time-frequency distribution of the original signal;
and according to the extracted frequency trend, carrying out angular domain resampling on the original signal to obtain an angular domain signal.
3. The method according to claim 1, wherein the obtaining an angular domain signal corresponding to a time domain signal of a rotating machine device comprises:
acquiring a time domain signal of target rotating mechanical equipment, and synchronously acquiring a rotating speed signal; wherein the time domain signal comprises: vibration, displacement, sound or electrical signals;
and performing angular domain resampling on the original signal by using the rotating speed signal to obtain an angular domain signal.
4. The method according to claim 1, wherein the determining hubert time-frequency distributions of the time-domain signals of the real frequency components obtained by reconstruction, and the superimposing the hubert time-frequency distributions to obtain the time-frequency distribution of the original signal comprises:
and calculating the amplitude envelope and the instantaneous frequency of the reconstructed real frequency components, independently constructing the Hilbert time-frequency distribution of each real frequency component, and overlapping the Hilbert time-frequency distributions of the real frequency components to obtain the time-frequency distribution of the original signal.
5. The method according to claim 4, wherein the calculation formula of the Hilbert time-frequency distribution is as follows:
TFRz(t,f)=a2(t)δ[f-f(t)]
wherein z (t) represents the reconstructed time-domain signal at time t, a (t) represents the amplitude envelope of the reconstructed time-domain signal, phi (t) represents the instantaneous phase of the reconstructed time-domain signal, f (t) represents the instantaneous frequency of the reconstructed time-domain signal, H (-) represents the Hilbert transform, delta (-) represents the Dirac function, TFR represents the time-domain signal, and f (t) represents the time-domain signalz(t, f) represents the time-frequency distribution matrix of z (t) signals, and f represents the frequency variation.
6. The method according to claim 1, wherein the determining of the hubert time-frequency distribution of the reconstructed time-domain signal of each real frequency component superimposes the hubert time-frequency distributions to obtain the time-frequency distribution of the original signal, and the time-frequency distribution of the original signal may be replaced by:
performing time-frequency analysis on the time domain signals of the real frequency components obtained by reconstruction, and constructing time-frequency distribution of each real frequency component; wherein the time-frequency analysis comprises: short-time fourier transform, continuous wavelet transform or wigner-vila transform;
and overlapping the time-frequency distribution of each real frequency component to obtain the time-frequency distribution of the original signal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911155435.2A CN110926594B (en) | 2019-11-22 | 2019-11-22 | Method for extracting time-varying frequency characteristics of rotary machine signal |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911155435.2A CN110926594B (en) | 2019-11-22 | 2019-11-22 | Method for extracting time-varying frequency characteristics of rotary machine signal |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110926594A CN110926594A (en) | 2020-03-27 |
CN110926594B true CN110926594B (en) | 2021-04-20 |
Family
ID=69850711
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911155435.2A Active CN110926594B (en) | 2019-11-22 | 2019-11-22 | Method for extracting time-varying frequency characteristics of rotary machine signal |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110926594B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112001314A (en) * | 2020-08-25 | 2020-11-27 | 江苏师范大学 | Early fault detection method for variable speed hoist |
CN112229627B (en) * | 2020-09-29 | 2021-12-03 | 电子科技大学 | Rotating machine instantaneous rotating speed estimation method based on short-time sparse Fourier transform |
CN113222147B (en) * | 2021-05-11 | 2024-02-13 | 北华航天工业学院 | Construction method of conditional double-countermeasure learning reasoning model |
CN113283399B (en) * | 2021-07-16 | 2021-11-05 | 北京科技大学 | Transient time-frequency feature extraction method based on dynamic response of single-degree-of-freedom system |
CN115687862B (en) * | 2022-10-19 | 2023-08-01 | 北京科技大学 | Time-varying filtering-based time-frequency analysis method for rotating mechanical signals |
CN116010806B (en) * | 2023-03-28 | 2023-09-01 | 湖北工业大学 | Time-frequency analysis method for rotary mechanical time-varying multi-component signal |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101709997A (en) * | 2009-12-11 | 2010-05-19 | 南京航空航天大学 | Harmonic window function of vibration signal processing |
CN102519578A (en) * | 2011-12-15 | 2012-06-27 | 广东石油化工学院 | Method for extracting time-frequency domain spectrum of mixed signals of rotating machinery |
CN105717490A (en) * | 2016-02-21 | 2016-06-29 | 上海大学 | LFM signal separating and parameter estimating method based on time frequency analysis |
CN110376437A (en) * | 2019-07-18 | 2019-10-25 | 北京科技大学 | A kind of order analysis method overcoming non-order frequency content interference |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4373350B2 (en) * | 2005-02-15 | 2009-11-25 | 株式会社日立製作所 | Shaft vibration monitoring system |
CN100554917C (en) * | 2007-09-29 | 2009-10-28 | 南京航空航天大学 | Obtain the method for system features function and signal characteristic value |
CN101603854A (en) * | 2009-07-15 | 2009-12-16 | 南京信息工程大学 | The rotating machinery non-stationery vibration signal instantaneous frequency estimation algorithm in start and stop period |
US9267226B2 (en) * | 2012-02-03 | 2016-02-23 | General Electric Company | Dynamic unbalance detection in a washing machine |
-
2019
- 2019-11-22 CN CN201911155435.2A patent/CN110926594B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101709997A (en) * | 2009-12-11 | 2010-05-19 | 南京航空航天大学 | Harmonic window function of vibration signal processing |
CN102519578A (en) * | 2011-12-15 | 2012-06-27 | 广东石油化工学院 | Method for extracting time-frequency domain spectrum of mixed signals of rotating machinery |
CN105717490A (en) * | 2016-02-21 | 2016-06-29 | 上海大学 | LFM signal separating and parameter estimating method based on time frequency analysis |
CN110376437A (en) * | 2019-07-18 | 2019-10-25 | 北京科技大学 | A kind of order analysis method overcoming non-order frequency content interference |
Non-Patent Citations (2)
Title |
---|
《Adaptive iterative generalized demodulation for nonstationary complex signal analysis: Principle and application in rotating machinery fault diagnosis》;ZhipengFeng;《Mechanical Systems and Signal Processing》;20180930;第110卷(第15期);全文 * |
《时变转速工况下行星齿轮箱故障诊断方法研究》;陈小旺;《中国博士学位论文全文数据库工程科技Ⅱ辑》;20180815(第08期);第69页-第76页 * |
Also Published As
Publication number | Publication date |
---|---|
CN110926594A (en) | 2020-03-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110926594B (en) | Method for extracting time-varying frequency characteristics of rotary machine signal | |
McCormick et al. | Cyclostationarity in rotating machine vibrations | |
Wang et al. | Rolling element bearing fault diagnosis via fault characteristic order (FCO) analysis | |
Miao et al. | Improvement of kurtosis-guided-grams via Gini index for bearing fault feature identification | |
CN107356432B (en) | Fault Diagnosis of Roller Bearings based on frequency domain window experience small echo resonance and demodulation | |
Ou et al. | A new rolling bearing fault diagnosis method based on GFT impulse component extraction | |
Zhao et al. | A tacho-less order tracking technique for large speed variations | |
Chen et al. | Fault diagnosis of planetary gearbox under variable-speed conditions using an improved adaptive chirp mode decomposition | |
Qu et al. | A new acoustic emission sensor based gear fault detection approach | |
CN106053080B (en) | Rolling bearing fault feature extraction method based on energy slice wavelet transformation | |
CN104655380A (en) | Method for extracting fault features of rotating mechanical equipment | |
Xiao et al. | A fault frequency bands location method based on improved fast spectral correlation to extract fault features in axial piston pump bearings | |
Chen et al. | Compound fault identification of rolling element bearing based on adaptive resonant frequency band extraction | |
CN109029999B (en) | Rolling bearing fault diagnosis method based on enhanced modulation bispectrum analysis | |
Wang et al. | Sparse and low-rank decomposition of the time–frequency representation for bearing fault diagnosis under variable speed conditions | |
CN112668518A (en) | VMSST time-frequency analysis method for vibration fault signal | |
Wang et al. | Rotating machinery fault diagnosis based on EEMD time-frequency energy and SOM neural network | |
Wang et al. | Weak fault diagnosis of rolling bearing under variable speed condition using IEWT-based enhanced envelope order spectrum | |
CN107831013A (en) | A kind of Method for Bearing Fault Diagnosis for strengthening cyclic bispectrum using probability principal component analysis | |
Lin et al. | A review and strategy for the diagnosis of speed-varying machinery | |
Zhang et al. | Improved local cepstrum and its applications for gearbox and rolling bearing fault detection | |
Zhao et al. | Vibration health monitoring of rolling bearings under variable speed conditions by novel demodulation technique | |
Zhao et al. | Rolling element bearing instantaneous rotational frequency estimation based on EMD soft-thresholding denoising and instantaneous fault characteristic frequency | |
CN112485028B (en) | Feature spectrum extraction method of vibration signal and mechanical fault diagnosis analysis method | |
CN110376437B (en) | Order analysis method for overcoming non-order frequency component interference |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |