CN111665051A - Bearing fault diagnosis method under strong noise variable-speed condition based on energy weight method - Google Patents

Bearing fault diagnosis method under strong noise variable-speed condition based on energy weight method Download PDF

Info

Publication number
CN111665051A
CN111665051A CN202010626838.7A CN202010626838A CN111665051A CN 111665051 A CN111665051 A CN 111665051A CN 202010626838 A CN202010626838 A CN 202010626838A CN 111665051 A CN111665051 A CN 111665051A
Authority
CN
China
Prior art keywords
order
frequency
time
bearing
energy
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
Application number
CN202010626838.7A
Other languages
Chinese (zh)
Inventor
王鹏
王太勇
张兰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin University
Original Assignee
Tianjin University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tianjin University filed Critical Tianjin University
Priority to CN202010626838.7A priority Critical patent/CN111665051A/en
Publication of CN111665051A publication Critical patent/CN111665051A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M13/00—Testing of machine parts
    • G01M13/04—Bearings
    • G01M13/045—Acoustic or vibration analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/17—Mechanical parametric or variational design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Geometry (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Computer Hardware Design (AREA)
  • General Engineering & Computer Science (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)

Abstract

The invention relates to a bearing fault diagnosis method under the condition of strong noise and variable rotating speed based on an energy weight method, which extracts vibration signal orders by utilizing a time-frequency ridge characteristic point linear interpolation and covering algorithm method according to a time-frequency representation graph based on Gabor transformation; carrying out instantaneous frequency estimation and quadratic fitting on the vibration signal by using a local extremum search algorithm and the extracted order; carrying out equal-angle resampling on the vibration signal by using a key phase time scale method according to the fitted instantaneous frequency; carrying out Hilbert-Huang transformation of CEEMDAN on the resampled equal angular domain signal to obtain an order-frequency spectrum of the signal; extracting the occurrence position of impact energy in the order-frequency spectrum, and further binarizing the order-frequency spectrum; and obtaining an energy weight order sequence capable of reflecting the impact through multi-scale binary spectrum analysis, and performing power spectrum analysis on the energy weight order sequence to obtain the impact component related to the fault. The method can eliminate the influence of strong noise and variable rotating speed on vibration signal analysis, and improve the accuracy of fault diagnosis of the rolling bearing.

Description

Bearing fault diagnosis method under strong noise variable-speed condition based on energy weight method
Technical Field
The invention belongs to the technical field of fault detection, and particularly relates to a bearing fault diagnosis method under the condition of strong noise and variable rotating speed based on an energy weight method.
Background
With the introduction of "manufacturing 2025", more and more attention is paid to industrial production. Rolling bearings are important parts of common equipment in industrial production, and the health state of the rolling bearings is concerned with production efficiency, operation cost and production safety. With the diversification, complication and large-scale of modern industrial equipment, the application occasions of the rolling bearing are more and more extensive.
In actual use scenes, the rolling bearing is often complex in environment, and complex vibration interference exists between the rolling bearing and the rolling bearing or between different components of the bearing, so that the acquired vibration signals often contain a large amount of interference noise, and when the fault is not obvious, the fault characteristic vibration signals are often submerged by the noise and are difficult to separate. Meanwhile, the running working condition of the rolling bearing is also changed frequently, the irregular variable-speed running is more common than the stable working condition, and great difficulty is brought to vibration signal analysis, so that the method for diagnosing the fault of the rolling bearing under the condition of strong noise and variable speed has great theoretical significance and practical significance.
Disclosure of Invention
The invention aims to: the bearing fault diagnosis method based on the energy weight method under the strong noise variable-speed condition solves the problem that the traditional fault diagnosis method is difficult to realize accurate judgment of the fault state under the variable-speed condition because vibration signals of a rolling bearing often show non-stable characteristics under the high-noise variable-speed operation condition.
In order to achieve the purpose, the technical scheme of the invention is as follows: the bearing fault diagnosis method under the condition of strong noise and variable rotating speed based on the energy weight method comprises the following steps:
step 1: acquiring a vibration signal of a bearing by using an acceleration sensor;
step 2: carrying out Gabor expansion on the acquired signals to obtain a Gabor time-frequency diagram;
and step 3: selecting a clearest and complete certain-order component in a Gabor time-frequency graph, placing control points on ridge lines of the order component, and connecting the control points by using straight lines to obtain a filtering center frequency line;
and 4, step 4: solving Gabor coefficient of the order component selected in the step 2 through a covering algorithm
Figure BDA0002565100670000021
Obtaining a time-frequency spectrum M of the selected order componentq(t,f)；
And 5: carrying out instantaneous frequency estimation and carrying out quadratic fitting according to a local extremum searching algorithm;
step 6: performing key phase time scale calculation and equal-angle resampling by using the fitted instantaneous frequency function;
and 7: performing Hilbert-Huang transformation order-frequency analysis based on adaptive noise Complete set Empirical Mode Decomposition (CEEMDAN) on the equiangular resampling signals;
and 8: extracting the occurrence position of impact energy in the order-frequency spectrum, and binarizing the order-frequency spectrum;
and step 9: obtaining an energy weight order sequence capable of reflecting impact through multi-scale binary spectrum analysis, summing the energy weight order sequence to obtain an aggregate energy weight time sequence, and performing power spectrum analysis on the aggregate energy weight time sequence;
step 10: searching theoretical fault characteristic orders of the bearing in the obtained power spectrum, and if corresponding order characteristics exist, indicating that corresponding faults exist, so that fault diagnosis of the rolling bearing under the conditions of strong noise and variable rotating speed is realized;
the theoretical fault characteristic order of the bearing comprises the following contents:
theoretical fault characteristic order of bearing outer ring:
Figure BDA0002565100670000031
theoretical fault characteristic order of bearing inner race:
Figure BDA0002565100670000032
theoretical fault characteristic order of bearing rolling body:
Figure BDA0002565100670000033
theoretical fault characteristic order of bearing retainer:
Figure BDA0002565100670000034
wherein n is the number of the balls; d is the diameter of the rolling body; d is the pitch diameter of the bearing; beta is the rolling element contact angle.
Preferably, the Gabor unfolding algorithm in step 2 is:
Figure BDA0002565100670000035
Figure BDA0002565100670000036
wherein x (k) is a signal, L is a period of the signal, M and N are a time domain sample number and a frequency domain sample number, respectively, Cm,nIs a Gabor coefficient, hL(k) And gammaL(k) Is defined as
hL(k)＝h(k-mΔM)e2πnΔNk/L
γL(k)＝γ(k-mΔM)e2πnΔNk/L
hL(k) And gammaL(k) The two are in a biorthogonal relationship of
Figure BDA0002565100670000037
In the formula, Δ M and Δ N are a time sampling interval and a frequency sampling interval, respectively, and Δ M Δ N is L in the case of critical sampling.
Preferably, the masking algorithm in step 4 is: setting one and C according to time-varying filtering neighborhoodm,nBinary mask array phi with same dimensionm,nExtracting
Figure BDA0002565100670000041
Is of the formula
Figure BDA0002565100670000042
Preferably, the filtering neighborhood is calculated by adopting an equal frequency method or an equal order method;
the bandwidth method, i.e. the filter bandwidth, does not change with the change of the center frequency, if the q-th order center frequency is fq(t), if the constant frequency bandwidth delta f is adopted, the filtering neighborhood calculation formula is
Figure BDA0002565100670000043
The bandwidth of the order of magnitude Δ o varies with the center frequency of the filter, which is related to the center frequency fqThe ratio of (t) is constant, and the filtering neighborhood calculation formula of the equal order method is
Figure BDA0002565100670000044
Preferably, the local extremum searching algorithm is formulated as
IFq(t+1)＝Argmax|Mq(t,f)|,f∈[fq(t)-Δf,fq(t)+Δf]
In the formula, IFq(t) is a function of instantaneous frequency, fq(t) is the center frequency, t is time, f is frequency, Arg is the average operator, max is the maximum operator.
Preferably, the equation of quadratic fit in step 5 is IFq(t)＝at2+bt+c
In the formula, IFqAnd (t) is an instantaneous frequency function, t is time, and a, b and c are constants obtained after fitting.
Preferably, the key phase time scale T of the equiangular resamplingnIs given by the formula
Figure BDA0002565100670000045
In the formula T0A, b and c are constants obtained after the fitting in the step 5, Q, which is the time domain sampling starting momentmaxIs the maximum analysis order.
Preferably, the equiangular resampling adopts Lagrange linear interpolation algorithm, and the formula is
Figure BDA0002565100670000051
Where x (T) is a time domain signal, TnFor the key phase time scale for equiangular resampling, t is time.
Preferably, the specific method of step 7 is:
and 7-11: dividing the signal into a plurality of intrinsic mode functions IMF by using a CEEMDAN method;
and 7-12: performing Hilbert transform on each IMF component;
Figure BDA0002565100670000052
where y (t) is the Hilbert transform signal, c (t) is the IMF component, and x is the convolution sign.
Preferably, the algorithm steps of the CEEMDAN are as follows:
and 7-21: adding white noise N to the original signal X (t)i(t) constructing a signal X to be analyzedi(t)＝X(t)+β0Ni(t)，i＝1，2，…，I；
And 7-22: to Xi(t) EMD, defining operator Ek(. h) the k-th IMF is taken and the first IMF of CEEMDAN is calculated:
Figure BDA0002565100670000053
and 7-23: calculate the first residual:
Figure BDA0002565100670000054
and 7-24: noise is added to the first residue and its first IMF is determined, which is defined as the second IMF of CEEMDAN:
Figure BDA0002565100670000055
and 7-25: calculating the kth residue to obtain the kth +1 IMF of CEEMDAN, where K is 1, 2, …, K:
Figure BDA0002565100670000061
Figure BDA0002565100670000062
and 7-26: repeating the step 5 to obtain K IMFs and residues, wherein the original signal x (t) can be expressed as:
Figure BDA0002565100670000063
preferably, the algorithm steps of the EMD are as follows
Step 7-221: let h0(t) finding h0(t) obtaining the upper envelope e described by all the maximum points through cubic spline interpolation for all the maximum points and minimum points+(t) and a lower envelope e characterized by all minima points-(t)；
Step 7-222: calculate the mean of the upper and lower envelopes
Figure BDA0002565100670000064
Step 7-223: average value m of envelope0(t) from h0(t) subtracting to obtain a signal h to be detected1(t)
h1(t)＝h0(t)-m0(t)
Step 7-224: inspection h1(t) whether IMF is satisfied, if not, h1(t) is regarded as h0(t) repeating the above process to obtain h2(t),h3(t),...,hi(t) up to hi(t) satisfies the condition of IMF: the extreme points and the zero-crossing points have the same number or one difference, and the local mean value defined by the upper envelope line and the lower envelope line is 0Time imf1＝hi(t)；
And 7-225: subtracting the first IMF from the original signal to obtain a first residual:
r1(t)＝X(t)-imf1(t)
step 7-226: let h0(t)＝r1(t), repeating the sieving process until two adjacent signals h to be detected are obtained in the IMF sieving processi-1(t) and hi(t) standard deviation less than a set value, i.e.:
Figure BDA0002565100670000065
where T is the time span, usually 0.2-0.3,
and 7-227: finally obtaining n IMFs and residual rn(t), then the original signal can be represented as:
Figure BDA0002565100670000071
preferably, the conditions of the IMF include: (1) the number of extreme points and the number of zero-crossing points must be equal or differ by one at most; (2) at any time, the average of the upper envelope formed by the local maximum point and the lower envelope formed by the local minimum point is equal to zero, that is, the upper and lower envelopes are locally symmetrical with respect to the time axis.
Preferably, the multi-scale binary spectrum comprises the following specific steps:
step 9-1, extracting energy time series X under different frequency intervals from the time-frequency spectrum matrix of M × Nn(t)，n∈(1,N)；
Step 9-2: in the energy time series XnSetting a sliding window with the length of 2d +1 points in (t), when | xn(ti)|＝max{|xn(tk) When i-d is not less than k is not more than i + d, let Bn(ti) 1, otherwise Bn(ti) 0. When the midpoint energy value of the window is a local energy extreme value, the weight is set to be 1, otherwise, the weight is 0;
step 9-3: binarizing the energy time sequences on different frequency intervals, and repeating the step (2) for N times to obtain N binary time sequences, namely a binary matrix B (t, f) with the size of M multiplied by N, wherein the matrix is called a binary spectrum;
step 9-4: setting different window lengths 2d +1, and repeating the steps 9-2 and 9-3 for multiple times to obtain the multi-scale binary spectrum.
Preferably, the formula of d in the point of the length 2d +1 of the sliding window is as follows:
Figure BDA0002565100670000072
in the formula fsAs the sampling rate, c as the number of calculations, ffIn order to be the characteristic frequency of the fault,
Figure BDA0002565100670000073
the operator rounding up the even.
Preferably, the energy weight order sequence formula is
Figure BDA0002565100670000081
Wherein B iscIs a binary spectrum at the c-th scale.
Preferably, the set energy weight order sequence formula is
Figure BDA0002565100670000082
Wherein C is the number of scales in the multi-scale binary spectrum.
Preferably, the power spectrum analysis formula is
Figure BDA0002565100670000083
In the formula FW(omega) is
Figure BDA0002565100670000084
OfThe result of the transformation is a transformation,
Figure BDA0002565100670000085
the invention has the beneficial effects that:
1. by carrying out angular domain resampling on the time domain vibration signal of the rolling bearing, the non-stable time domain signal can be converted into a stable angular domain signal, and the influence of the rotation speed change on the vibration signal analysis can be eliminated; by extracting the fault impact component by using an energy weight method, the characteristic of the fault component can be enhanced, so that the accuracy and effectiveness of fault diagnosis are improved.
2. The vibration data is adopted for diagnosis and analysis, and a vibration signal analysis method is an effective state detection method and is particularly suitable for rotary mechanical equipment such as a rolling bearing and the like.
3. The vibration signal of the rolling bearing in the process of changing the rotating speed is a non-stationary signal, and the traditional analysis method cannot achieve a good effect. Angular domain resampling is an effective method for analyzing non-stationary vibration signals, and converts non-stationary time domain signals into stationary angular domain signals, thereby eliminating the influence of rotation speed change on vibration signal analysis.
4. The Gabor transform is also the optimal short-time fourier transform, and is an excellent method for rotational speed estimation based on time-frequency spectral ridge fitting because the Gabor expansion can well describe the transient characteristics of the severely changing signal.
5. The invention adopts a keyless phase order tracking method, does not need to install a rotation speed sensor, and is suitable for occasions where the rotation speed sensor cannot be installed.
6. CEEMDAN can effectively inhibit interference caused by noise, and improve accuracy and effectiveness of fault diagnosis.
7. CEEMDAN adaptively decomposes the bearing signal into several eigenmode function components, thereby separating fault-related modulation information.
8. The energy weighting method can enhance the fault characteristics of the signals, thereby extracting fault components from strong noise.
Drawings
FIG. 1 is a flow chart of a bearing fault diagnosis method.
Fig. 2 is a graph of simulated vibration signals.
Fig. 3 is an envelope spectrum of a simulated vibration signal.
FIG. 4 is a plot of the rotational speed of the simulated signal.
FIG. 5 is a graph of a simulated vibration signal after resampling.
Fig. 6 is a simulated vibration signal envelope spectrogram after resampling.
Fig. 7 is an order-frequency spectrum of a resampled signal.
Fig. 8 is a binary spectrogram of a resampled signal.
Fig. 9 is an energy weight order sequence chart.
Fig. 10 is an order power spectrum.
Fig. 11 is a diagram of a hoist vibration signal.
Fig. 12 is a diagram of a vibration signal of the hoist after resampling.
Fig. 13 is a resampled envelope spectrum of the vibration signal of the hoist.
Fig. 14 is a time-frequency diagram of a resampled signal.
Fig. 15 is a 2 nd order binary spectrum.
Fig. 16 is a set energy weight order sequence diagram.
Fig. 17 is an order power spectrum.
FIG. 18 is a broken bearing view.
Fig. 19 is a partial enlarged view at I in fig. 18.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Referring to fig. 1, the present invention provides a method for diagnosing a bearing fault under a strong noise and variable rotation speed condition based on an energy weight method, which includes the following steps:
step 1: acquiring a vibration signal of a bearing by using an acceleration sensor;
step 2: carrying out Gabor expansion on the acquired signals to obtain a Gabor time-frequency diagram;
and step 3: selecting an obvious certain-order component in a Gabor time-frequency diagram, placing control points on ridge lines of the component, and connecting the control points by using straight lines to obtain a filtering center frequency line;
and 4, step 4: solving Gabor coefficient C of the order component selected in the step 2 through a covering algorithmmq,nObtaining the time frequency spectrum M of the selected order componentq(t,f)；
And 5: carrying out instantaneous frequency estimation and carrying out quadratic fitting according to a local extremum searching algorithm;
step 6: performing key phase time scale calculation and equal-angle resampling by using the fitted instantaneous frequency function;
and 7: performing Hilbert-Huang transformation order-frequency analysis based on CEEMDAN on the equiangular resampling signals;
and 8: extracting the occurrence position of impact energy in the order-frequency spectrum, and binarizing the order-frequency spectrum;
and step 9: obtaining an energy weight order sequence capable of reflecting impact through multi-scale binary spectrum analysis, summing the energy weight order sequence to obtain an aggregate energy weight time sequence, and performing power spectrum analysis on the aggregate energy weight time sequence to obtain a fault-related impact component;
step 10: the obtained impact components related to the fault are compared with the theoretical fault characteristic order of the bearing, so that the fault diagnosis of the rolling bearing under the conditions of strong noise and variable rotating speed is realized;
the theoretical fault characteristic order of the bearing comprises the following contents:
theoretical fault characteristic order of bearing outer ring:
Figure BDA0002565100670000111
theoretical fault characteristic order of bearing inner race:
Figure BDA0002565100670000112
theoretical fault characteristic order of bearing rolling body:
Figure BDA0002565100670000113
theoretical fault characteristic order of bearing retainer:
Figure BDA0002565100670000114
wherein n is the number of the balls; d is the diameter of the rolling body; d is the pitch diameter of the bearing; beta is the rolling element contact angle.
The Gabor expansion algorithm in the step 2 is as follows:
Figure BDA0002565100670000121
Figure BDA0002565100670000122
wherein L is the period of the signal, M and N are the time domain sampling number and the frequency domain sampling number respectively, Cm,nIs a Gabor coefficient, hL(k) And gammaL(k) Is defined as
hL(k)＝h(k-mΔM)e2πnΔNk/L
γL(k)＝γ(k-mΔM)e2πnΔNk/L
hL(k) And gammaL(k) The two are in a biorthogonal relationship of
Figure BDA0002565100670000123
In the formula, Δ M and Δ N are a time sampling interval and a frequency sampling interval, respectively, and Δ M Δ N is L in the case of critical sampling.
The masking algorithm in the step 4 is as follows: setting one and C according to time-varying filtering neighborhoodm,nBinary mask array having the same dimensions
Figure BDA0002565100670000124
Extraction of
Figure BDA0002565100670000125
Formula (2)
Figure BDA0002565100670000126
Wherein phim,nIs composed of
The filtering neighborhood is calculated by adopting an equal frequency method or an equal order method;
the bandwidth method, i.e. the filter bandwidth, does not change with the change of the center frequency, if the q-th order center frequency is fq(t), if the constant frequency bandwidth delta f is adopted, the filtering neighborhood calculation formula is
Figure BDA0002565100670000127
The bandwidth of the order of magnitude Δ o varies with the center frequency of the filter, which is related to the center frequency fqThe ratio of (t) is constant, and the filtering neighborhood calculation formula of the equal order method is
Figure BDA0002565100670000128
The formula of the local extremum searching algorithm is
IFq(t+1)＝Argmax|Mq(t,f)|,f∈[fq(t)-Δf,fq(t)+Δf]。
The equation of quadratic fit in the step 5 is IFq(t)＝at2+bt+c。
The key phase time scale T of the equal angle resamplingnIs given by the formula
Figure BDA0002565100670000131
Wherein T is0Is the time domain sampling start time.
The equal-angle resampling adopts Lagrange linear interpolation algorithm, and the formula is
Figure BDA0002565100670000132
The specific method of the step 7 comprises the following steps:
and 7-11: dividing the signal into a plurality of intrinsic mode functions IMF by using a CEEMDAN method;
and 7-12: performing Hilbert transform on each IMF component;
Figure BDA0002565100670000133
where y (t) is the Hilbert transform signal, c (t) is the IMF component, and x is the convolution sign.
The CEEMDAN algorithm steps are as follows:
and 7-21: adding white noise N to the original signal X (t)i(t) constructing a signal X to be analyzedi(t)＝X(t)+β0Ni(t)，i＝1，2，…，I；
And 7-22: to Xi(t) EMD, defining operator Ek(. h) the k-th IMF is taken and the first IMF of CEEMDAN is calculated:
Figure BDA0002565100670000134
and 7-23: calculate the first residual:
Figure BDA0002565100670000141
and 7-24: noise is added to the first residue and its first IMF is determined, which is defined as the second IMF of CEEMDAN:
Figure BDA0002565100670000142
and 7-25: calculating the kth residue to obtain the kth +1 IMF of CEEMDAN, where K is 1, 2, …, K:
Figure BDA0002565100670000143
Figure BDA0002565100670000144
and 7-26: repeating the step 5 to obtain K IMFs and residues, wherein the original signal x (t) can be expressed as:
Figure BDA0002565100670000145
the algorithm steps of the EMD are as follows
Step 7-221: let h0(t) finding h0(t) obtaining the upper envelope e described by all the maximum points through cubic spline interpolation for all the maximum points and minimum points+(t) and a lower envelope e characterized by all minima points-(t)；
Step 7-222: calculate the mean of the upper and lower envelopes
Figure BDA0002565100670000146
Step 7-223: average value m of envelope0(t) from h0(t) subtracting to obtain a signal h to be detected1(t)
h1(t)＝h0(t)-m0(t)
Step 7-224: inspection h1(t) whether IMF is satisfied, if not, h1(t) is regarded as h0(t) repeating the above process to obtain h2(t),h3(t),...,hi(t) up to hi(t) satisfies the condition of IMF: the extreme points and the zero-crossing points have equal numbers or differ by one, and the local mean value defined by the upper envelope and the lower envelope is 0, at this time imf1＝hi(t)；
And 7-225: subtracting the first IMF from the original signal to obtain a first residual:
r1(t)＝X(t)-imf1(t)
step 7-226: let h0(t)＝r1(t) repeating the above sieveThe processes are divided until two adjacent signals h to be detected in the IMF screening processi-1(t) and hi(t) standard deviation less than a set value, i.e.:
Figure BDA0002565100670000151
where T is the time span, usually 0.2-0.3,
and 7-227: finally obtaining n IMFs and residual rn(t), then the original signal can be represented as:
Figure BDA0002565100670000152
the conditions of the IMF include: (1) the number of extreme points and the number of zero-crossing points must be equal or differ by one at most; (2) at any time, the average of the upper envelope formed by the local maximum point and the lower envelope formed by the local minimum point is equal to zero, that is, the upper and lower envelopes are locally symmetrical with respect to the time axis.
The multi-scale binary spectrum comprises the following specific steps:
step 9-1, extracting energy time series X under different frequency intervals from the time-frequency spectrum matrix of M × Nn(t)，n∈(1,N)；
Step 9-2: in the energy time series XnSetting a sliding window with the length of 2d +1 points in (t), when | xn(ti)|＝max{|xn(tk) When i-d is not less than k is not more than i + d, let Bn(ti) 1, otherwise Bn(ti) 0. When the midpoint energy value of the window is a local energy extreme value, the weight is set to be 1, otherwise, the weight is 0;
step 9-3: binarizing the energy time sequences on different frequency intervals, and repeating the step (2) for N times to obtain N binary time sequences, namely a binary matrix B (t, f) with the size of M multiplied by N, wherein the matrix is called a binary spectrum;
step 9-4: setting different window lengths 2d +1, and repeating the steps 9-2 and 9-3 for multiple times to obtain the multi-scale binary spectrum.
The formula of d in the sliding window length 2d +1 point is as follows:
Figure BDA0002565100670000161
in the formula fsAs the sampling rate, c as the number of calculations, ffIn order to be the characteristic frequency of the fault,
Figure BDA0002565100670000162
the operator rounding up the even.
The energy weight order sequence formula is
Figure BDA0002565100670000163
Wherein B iscIs a binary spectrum at the c-th scale.
The set energy weight order sequence formula is
Figure BDA0002565100670000164
Wherein C is the number of scales in the multi-scale binary spectrum.
The power spectrum analysis formula is
Figure BDA0002565100670000165
In the formula FW(omega) is
Figure BDA0002565100670000166
The fourier transform of (a) the signal,
Figure BDA0002565100670000167
the specific implementation process of the diagnosis method provided by the invention comprises the following steps:
in order to verify the effectiveness of the proposed method, the following model is constructed to simulate the vibration signal when the bearing fails:
Figure BDA0002565100670000168
in the formula, AiAmplitude of the i-th impact, TiThe frequency of impact generation is 1.75 times of the rotation frequency at the moment of the ith impact generation, BnAmplitude of the nth harmonic, βnIs the initial phase of the nth harmonic, s (t) is the impulse signal, f (t) is the instantaneous frequency conversion, N (t) is the noise, and s (t) and f (t) are the following formulas:
s(t)＝e-500tsin(4000πt)
f(t)＝[250+400cos(0.25πt)]/60
get B1＝0.3，B2＝0.5，B3＝0.4，β1＝π/6，β2＝-π/3，β3The time domain waveform of the pi/2 simulation signal is shown in fig. 2, wherein the abscissa in fig. 2 is time, and the ordinate is the dimensionless amplitude of the time domain waveform of the simulation signal. The envelope spectrum is shown in fig. 3, the abscissa in fig. 3 is frequency, and the ordinate is dimensionless amplitude of the envelope spectrum of the simulation signal. Spectral lines that reflect the characteristic frequencies of the impacts cannot be seen by the envelope spectrum.
The simulation signal is subjected to order tracking analysis, the rotation speed estimation value and the theoretical value obtained through analysis are shown in fig. 4, the abscissa in fig. 4 is time, the ordinate is the rotation speed value, wherein the diagram (a) is the rotation speed estimation value, the diagram (b) is the rotation speed theoretical value, and the rotation speed unit is revolutions per minute. The estimated value of the rotating speed in the figure is basically completely overlapped with the theoretical value, which shows that the estimated result of the rotating speed is ideal, and the difference between the estimated value and the theoretical value is not large.
And calculating instantaneous frequency by using the obtained rotating speed estimated value and performing equal-angle resampling on the time domain vibration signal to obtain an angular domain periodic signal as shown in fig. 5, wherein the abscissa in fig. 5 is the order, and the ordinate is the dimensionless amplitude of the time domain waveform of the simulation signal. The envelope spectrum is directly processed to obtain the spectrum as shown in fig. 6, wherein the abscissa in fig. 6 is the order, and the ordinate is the dimensionless amplitude of the envelope spectrum of the simulation signal. Because the noise signal energy is large and the impact signal is completely submerged, the impact component can not be distinguished in the signal envelope spectrum after the equal-angle sampling.
The resampled signal is analyzed by an energy weighted method, an order-frequency spectrum is obtained as shown in fig. 7, the abscissa in fig. 7 is the number of rotation turns, the ordinate is the order, and the image is the energy of the simulation signal.
Performing binary spectrum analysis on the order-frequency spectrum, and taking the parameter fs＝92，ffIf 1 binary spectral calculation is performed, then d is calculated1The resulting binary spectrum is shown in fig. 8 at 32, where the abscissa is the number of revolutions, the ordinate is the order, and the image is the binary weight. The energy weight order sequence is shown in fig. 9. In fig. 9, the abscissa represents the number of rotations and the ordinate represents the energy weight integrated value. And performing power spectrum analysis on the energy weight order sequence to obtain a final order power spectrum as shown in fig. 10, wherein the abscissa in fig. 10 is the order, and the ordinate is the dimensionless amplitude of the order envelope spectrum. The impact energy component at the 1.75 order can be seen by the order power spectrum, consistent with the impact component in the simulated signal.
Example verification analysis
In order to verify the effectiveness of the method, fault diagnosis is carried out on the winch bearing. The test diagnoses the brake end bearing of the winch, and the parameters are shown in table 1.
TABLE 1 measured bearing parameters
Figure BDA0002565100670000181
The vibration signal of the section of the collected winch in the acceleration process is analyzed, and the time domain waveform of the vibration signal refers to fig. 11. In fig. 11, the abscissa represents time, and the ordinate represents the amplitude of the time-domain waveform of the vibration acceleration signal.
The vibration signal after resampling obtained by performing non-key phase order tracking analysis based on Gabor transform and equal angle resampling on the signal is shown in fig. 12. In fig. 12, the abscissa is the number of turns, and the ordinate is the amplitude of the vibration acceleration signal order domain waveform. The envelope spectrum analysis is performed to obtain the frequency spectrum of fig. 13, and due to the interference of strong noise, the obvious characteristic impact component cannot be seen from the frequency spectrum. In fig. 13, the abscissa is the order and the ordinate is the amplitude of the order envelope spectrum.
The resampled vibration signal is analyzed by energy weighted method, and the time-frequency diagram of fig. 14 is obtained first. In fig. 14, the abscissa is the number of rotations, the ordinate is the order, and the image is the signal energy.
Performing binary spectrum analysis on the time-frequency diagram, and taking f as a parameters＝336，ffWhen 6.427 is satisfied, 4 binary spectrum calculations are performed, d1＝28，d2＝56，d3＝84，d4112. Referring to fig. 15, the obtained 2 nd order binary spectrum has more distinct features, in fig. 15, the abscissa is the number of rotations, the ordinate is the order, and the image is binary weight.
Blank region processing and set energy weight calculation are performed on the obtained 4 binary spectra, and the obtained set energy weight order sequence is shown in fig. 16. In fig. 16, the abscissa represents the number of rotations and the ordinate represents the energy weight integrated value. And performing power spectrum analysis on the set energy weight order sequence to obtain a final order power spectrum as shown in fig. 17. In fig. 17, the abscissa is the order and the ordinate is the dimensionless magnitude of the order envelope spectrum.
The impact energy component at 6.427 th order and the frequency doubling component at 12.85 th order can be seen through the order power spectrum, and the bearing is indicated to have outer ring faults. After a period of use, the maintenance personnel disassemble the apparatus and find the failure at a in fig. 19, and the bearing surface is damaged, which corresponds to the type of failure indicated in the experiment.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims (17)

1. The bearing fault diagnosis method under the condition of strong noise and variable rotating speed based on the energy weight method is characterized by comprising the following steps of:
step 1: acquiring a vibration signal of a bearing by using an acceleration sensor;
step 2: carrying out Gabor expansion on the acquired signals to obtain a Gabor time-frequency diagram;
and step 3: selecting a certain order component in a Gabor time-frequency diagram, placing control points on ridge lines of the order component, and connecting the control points by straight lines to obtain a filtering center frequency line;
and 4, step 4: solving Gabor coefficient of the order component selected in the step 2 through a covering algorithm
Figure FDA0002565100660000011
Obtaining a time-frequency spectrum M of the selected order componentq(t,f)；
And 5: carrying out instantaneous frequency estimation and carrying out quadratic fitting according to a local extremum searching algorithm;
step 6: performing key phase time scale calculation and equal-angle resampling by using the fitted instantaneous frequency function;
and 7: performing Hilbert-Huang transformation order-frequency analysis based on self-adaptive noise complete set empirical mode decomposition on the equiangular resampling signals;
and 8: extracting the occurrence position of impact energy in the order-frequency spectrum, and binarizing the order-frequency spectrum;
and step 9: obtaining an energy weight order sequence capable of reflecting impact through multi-scale binary spectrum analysis, summing the energy weight order sequence to obtain an aggregate energy weight time sequence, and performing power spectrum analysis on the aggregate energy weight time sequence;
step 10: searching theoretical fault characteristic orders of the bearing in the obtained power spectrum, and if corresponding order characteristics exist, indicating that corresponding faults exist, so that fault diagnosis of the rolling bearing under the conditions of strong noise and variable rotating speed is realized;
the theoretical fault characteristic order of the bearing comprises the following contents:
theoretical fault characteristic order of bearing outer ring:
Figure FDA0002565100660000021
theoretical fault characteristic order of bearing inner race:
Figure FDA0002565100660000022
theoretical fault characteristic order of bearing rolling body:
Figure FDA0002565100660000023
theoretical fault characteristic order of bearing retainer:
Figure FDA0002565100660000024
wherein n is the number of the balls; d is the diameter of the rolling body; d is the pitch diameter of the bearing; beta is the rolling element contact angle.
2. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method according to claim 1, wherein the Gabor expansion algorithm in the step 2 is as follows:
Figure FDA0002565100660000025
Figure FDA0002565100660000026
wherein x (k) is a signal, L is a period of the signal, M and N are a time domain sample number and a frequency domain sample number, respectively, Cm,nIs a Gabor coefficient, hL(k) And gammaL(k) Is defined as
hL(k)＝h(k-mΔM)e2πnΔNk/L
γL(k)＝γ(k-mΔM)e2πnΔNk/L
hL(k) And gammaL(k) The two are coincided withThe orthogonal relation is
Figure FDA0002565100660000027
0≤m≤ΔM-1,0≤n≤ΔN-1
In the formula, Δ M and Δ N are a time sampling interval and a frequency sampling interval, respectively, and Δ M Δ N is L in the case of critical sampling.
3. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method as claimed in claim 1, wherein the masking algorithm in the step 4 is as follows: setting one and C according to time-varying filtering neighborhoodm,nBinary mask array phi with same dimensionm,nExtracting
Figure FDA0002565100660000031
Is of the formula
Figure FDA0002565100660000032
4. The bearing fault diagnosis method under the condition of strong noise and variable rotating speed based on the energy weight method is characterized in that the filtering neighborhood is calculated by adopting an equal frequency method or an equal order method;
the bandwidth method, i.e. the filter bandwidth, does not change with the change of the center frequency, if the q-th order center frequency is fq(t), if the constant frequency bandwidth delta f is adopted, the filtering neighborhood calculation formula is
Figure FDA0002565100660000033
The bandwidth of the order of magnitude Δ o varies with the center frequency of the filter, which is related to the center frequency fqThe ratio of (t) is constant, and the filtering neighborhood calculation formula of the equal order method is
Figure FDA0002565100660000034
5. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method as claimed in claim 1, wherein the formula of the local extremum search algorithm is IFq(t+1)＝Arg max|Mq(t,f)|,f∈[fq(t)-Δf,fq(t)+Δf]
In the formula, IFq(t) is a function of instantaneous frequency, fq(t) is the center frequency, t is time, f is frequency, Arg is the average operator, max is the maximum operator.
6. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method according to claim 1, wherein the equation of quadratic fit in the step 5 is IFq(t)＝at2+bt+c
In the formula, IFqAnd (t) is an instantaneous frequency function, t is time, and a, b and c are constants obtained after fitting.
7. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method according to claim 1, wherein the key phase time scale T of the equal-angle resampling isnIs given by the formula
Figure FDA0002565100660000041
(n＝1,2,…N)
In the formula T0A, b and c are constants obtained after the fitting in the step 5, Q, which is the time domain sampling starting momentmaxIs the maximum analysis order.
8. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method as claimed in claim 1, wherein the equal-angle resampling adopts a Lagrange linear interpolation algorithm, and the formula is
Figure FDA0002565100660000042
(ti≤Tn≤ti+1)
Where x (T) is a time domain signal, TnFor the key phase time scale for equiangular resampling, t is time.
9. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method according to claim 1, wherein the concrete method of the step 7 is as follows:
and 7-11: dividing the signal into a plurality of intrinsic mode functions IMF by using a CEEMDAN method;
and 7-12: performing Hilbert transform on each IMF component;
Figure FDA0002565100660000051
where y (t) is the Hilbert transform signal, c (t) is the IMF component, and x is the convolution sign.
10. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method as claimed in claim 1, wherein the CEEMDAN algorithm steps are as follows:
and 7-21: adding white noise N to the original signal X (t)i(t) constructing a signal X to be analyzedi(t)＝X(t)+β0Ni(t)，i＝1，2，…，I；
And 7-22: to Xi(t) EMD, defining operator Ek(. h) the k-th IMF is taken and the first IMF of CEEMDAN is calculated:
Figure FDA0002565100660000052
and 7-23: calculate the first residual:
Figure FDA0002565100660000053
and 7-24: noise is added to the first residue and its first IMF is determined, which is defined as the second IMF of CEEMDAN:
Figure FDA0002565100660000054
and 7-25: calculating the kth residue to obtain the kth +1 IMF of CEEMDAN, where K is 1, 2, …, K:
Figure FDA0002565100660000055
Figure FDA0002565100660000061
and 7-26: repeating the step 5 to obtain K IMFs and residues, wherein the original signal x (t) can be expressed as:
Figure FDA0002565100660000062
11. the method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method according to claim 1, wherein the EMD comprises the following algorithm steps:
step 7-221: let h0(t) finding h0(t) obtaining the upper envelope e described by all the maximum points through cubic spline interpolation for all the maximum points and minimum points+(t) and a lower envelope e characterized by all minima points-(t)；
Step 7-222: calculate the mean of the upper and lower envelopes
Figure FDA0002565100660000063
Step 7-223: average value m of envelope0(t) from h0(t) subtracting to obtain a signal h to be detected1(t)
h1(t)＝h0(t)-m0(t)
Step 7-224: inspection h1(t) whether IMF is satisfied, if not, h1(t) is regarded as h0(t) repeating the above process to obtain h2(t),h3(t),…,hi(t) up to hi(t) satisfies the condition of IMF: the extreme points and the zero-crossing points have equal numbers or differ by one, and the local mean value defined by the upper envelope and the lower envelope is 0, at this time imf1＝hi(t)；
And 7-225: subtracting the first IMF from the original signal to obtain a first residual:
r1(t)＝X(t)-imf1(t)
step 7-226: let h0(t)＝r1(t), repeating the sieving process until two adjacent signals h to be detected are obtained in the IMF sieving processi-1(t) and hi(t) standard deviation less than a set value, i.e.:
Figure FDA0002565100660000071
where T is a time span, is a constant, usually taken from 0.2 to 0.3,
and 7-227: finally obtaining n IMFs and residual rn(t), then the original signal can be represented as:
Figure FDA0002565100660000072
12. the energy weight method-based bearing fault diagnosis method under strong noise and variable speed conditions according to claim 11, wherein the IMF conditions comprise: (1) the number of extreme points and the number of zero-crossing points must be equal or differ by one at most; (2) at any time, the average of the upper envelope formed by the local maximum point and the lower envelope formed by the local minimum point is equal to zero, that is, the upper and lower envelopes are locally symmetrical with respect to the time axis.
13. The method for diagnosing the bearing fault under the strong noise variable-speed condition based on the energy weight method according to claim 1, wherein the multi-scale binary spectrum comprises the following specific steps:
step 9-1, extracting energy time series X under different frequency intervals from the time-frequency spectrum matrix of M × Nn(t)，n∈(1,N)；
Step 9-2: in the energy time series XnSetting a sliding window with the length of 2d +1 points in (t), when | xn(ti)|＝max{|xn(tk) When i-d is not less than k is not more than i + d, let Bn(ti) 1, otherwise Bn(ti) When the midpoint energy value of the window is a local energy extreme value, the weight is set to be 1, otherwise, the midpoint energy value is 0;
step 9-3: binarizing the energy time sequences on different frequency intervals, and repeating the step (2) for N times to obtain N binary time sequences, namely a binary matrix B (t, f) with the size of M multiplied by N, wherein the matrix is called a binary spectrum;
step 9-4: setting different window lengths 2d +1, and repeating the steps 9-2 and 9-3 for multiple times to obtain the multi-scale binary spectrum.
14. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method as claimed in claim 13, wherein the formula of d in the point of 2d +1 of the length of the sliding window is as follows:
Figure FDA0002565100660000081
in the formula fsAs the sampling rate, c as the number of calculations, ffIn order to be the characteristic frequency of the fault,
Figure FDA0002565100660000082
the operator rounding up the even.
15. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method according to claim 1, wherein the energy weight order sequence formula is
Figure FDA0002565100660000083
Wherein B iscIs a binary spectrum at the c-th scale.
16. The method for diagnosing the bearing fault under the condition of strong noise and variable rotating speed based on the energy weight method as claimed in claim 1, wherein the integrated energy weight order sequence formula is
Figure FDA0002565100660000084
Wherein C is the number of scales in the multi-scale binary spectrum.
17. The method for diagnosing the bearing fault under the strong noise and variable rotating speed conditions based on the energy weight method as claimed in claim 1, wherein the power spectrum analysis formula is
Figure FDA0002565100660000091
In the formula FW(omega) is
Figure FDA0002565100660000092
The fourier transform of (a) the signal,
Figure FDA0002565100660000093
CN202010626838.7A 2020-07-01 2020-07-01 Bearing fault diagnosis method under strong noise variable-speed condition based on energy weight method Pending CN111665051A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010626838.7A CN111665051A (en) 2020-07-01 2020-07-01 Bearing fault diagnosis method under strong noise variable-speed condition based on energy weight method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010626838.7A CN111665051A (en) 2020-07-01 2020-07-01 Bearing fault diagnosis method under strong noise variable-speed condition based on energy weight method

Publications (1)

Publication Number Publication Date
CN111665051A true CN111665051A (en) 2020-09-15

Family

ID=72390790

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010626838.7A Pending CN111665051A (en) 2020-07-01 2020-07-01 Bearing fault diagnosis method under strong noise variable-speed condition based on energy weight method

Country Status (1)

Country Link
CN (1) CN111665051A (en)

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112484998A (en) * 2020-11-16 2021-03-12 苏州大学文正学院 Wind turbine generator bearing fault diagnosis method based on synchronous modal spectrum
CN112539827A (en) * 2020-12-04 2021-03-23 五凌电力有限公司 Unit mechanical vibration evaluation method and system based on frequency energy ratio
CN112577746A (en) * 2020-12-07 2021-03-30 东南大学 Method for extracting rolling bearing envelope order spectrum fault features under rotation speed fluctuation
CN112629862A (en) * 2020-11-04 2021-04-09 南京工业大学 Rolling bearing weak fault diagnosis method based on bistable stochastic resonance and CEEMDAN-TEO
CN112629850A (en) * 2020-12-06 2021-04-09 北京工业大学 Fault diagnosis method for variable-speed planetary gearbox
CN112665851A (en) * 2020-10-28 2021-04-16 西安交通大学 Key-free phase change rotating speed gearbox fault diagnosis method
CN113405823A (en) * 2021-05-17 2021-09-17 同济大学 Rotary machine fault diagnosis method based on iterative expansion eigenmode decomposition
CN113532835A (en) * 2021-08-18 2021-10-22 中国国家铁路集团有限公司 Railway contact net hard spot diagnosis method and device
CN113702046A (en) * 2021-09-13 2021-11-26 长沙理工大学 Bearing fault diagnosis method based on mobile equipment under variable rotating speed working condition
CN114486254A (en) * 2022-02-09 2022-05-13 青岛迈金智能科技股份有限公司 Bicycle bearing detection method based on time/frequency double-domain analysis
CN114484286A (en) * 2021-12-27 2022-05-13 西安理工大学 Self-adaptive filtering and noise reduction method for water hammer impact pressure signal
CN114563186A (en) * 2022-01-21 2022-05-31 苏州大学 Rolling bearing unbalance fault diagnosis method and system
CN114659790A (en) * 2022-03-14 2022-06-24 浙江工业大学 Method for identifying bearing fault of variable-speed wind power high-speed shaft
CN115876475A (en) * 2023-02-06 2023-03-31 山东大学 Fault diagnosis method, system, equipment and storage medium
CN116754234A (en) * 2023-08-17 2023-09-15 山东经典印务有限责任公司 Automatic printing production equipment running state detection method
CN117708574A (en) * 2024-02-02 2024-03-15 江苏南高智能装备创新中心有限公司 CNN variable-speed rolling bearing fault diagnosis method embedded with physical information

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105510711A (en) * 2015-12-24 2016-04-20 合肥工业大学 Empirical mode decomposition-based improved harmonic analysis method
CN105784366A (en) * 2016-03-30 2016-07-20 华北电力大学（保定） Wind turbine generator bearing fault diagnosis method under variable speed
CN106289774A (en) * 2016-07-26 2017-01-04 北京工业大学 A kind of rolling bearing fault identification and trend forecasting method
CN110617964A (en) * 2019-07-29 2019-12-27 中国铁道科学研究院集团有限公司城市轨道交通中心 Synchronous compression transformation order ratio analysis method for fault diagnosis of rolling bearing
CN111307452A (en) * 2020-03-05 2020-06-19 江苏天沃重工科技有限公司 Intelligent fault diagnosis method for rotating machinery at time-varying rotating speed

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105510711A (en) * 2015-12-24 2016-04-20 合肥工业大学 Empirical mode decomposition-based improved harmonic analysis method
CN105784366A (en) * 2016-03-30 2016-07-20 华北电力大学（保定） Wind turbine generator bearing fault diagnosis method under variable speed
CN106289774A (en) * 2016-07-26 2017-01-04 北京工业大学 A kind of rolling bearing fault identification and trend forecasting method
CN110617964A (en) * 2019-07-29 2019-12-27 中国铁道科学研究院集团有限公司城市轨道交通中心 Synchronous compression transformation order ratio analysis method for fault diagnosis of rolling bearing
CN111307452A (en) * 2020-03-05 2020-06-19 江苏天沃重工科技有限公司 Intelligent fault diagnosis method for rotating machinery at time-varying rotating speed

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
佟德纯、李华彪 编著: "《振动监测与诊断》", 31 December 1997, 上海科技技术文献出版社 *
周亦人 等: ""基于 CEEMDAN 能量权重法的轴承故障诊断"", 《轴承》 *
周平 主编: "《轨道交通齿轮箱状态监测与故障诊断技术》", 31 December 2012, 西南交通大学出版社 *
张兰 等: ""改进的CEEMDAN故障诊断算法及在加工装备中的应用"", 《机械动力学》 *
杨永锋、吴亚锋 著: "《经验模态分级在振动分析中的应用》", 31 December 2013, 国防工业出版社 *
黄民、肖兴明 编著: "《机械故障诊断技术及应用》", 31 December 2002, 中国矿业大学出版社 *

Cited By (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112665851A (en) * 2020-10-28 2021-04-16 西安交通大学 Key-free phase change rotating speed gearbox fault diagnosis method
CN112629862A (en) * 2020-11-04 2021-04-09 南京工业大学 Rolling bearing weak fault diagnosis method based on bistable stochastic resonance and CEEMDAN-TEO
CN112484998B (en) * 2020-11-16 2022-12-27 苏州大学文正学院 Wind turbine generator bearing fault diagnosis method based on synchronous modal spectrum
CN112484998A (en) * 2020-11-16 2021-03-12 苏州大学文正学院 Wind turbine generator bearing fault diagnosis method based on synchronous modal spectrum
CN112539827A (en) * 2020-12-04 2021-03-23 五凌电力有限公司 Unit mechanical vibration evaluation method and system based on frequency energy ratio
CN112629850B (en) * 2020-12-06 2022-04-22 北京工业大学 Fault diagnosis method for variable-speed planetary gearbox
CN112629850A (en) * 2020-12-06 2021-04-09 北京工业大学 Fault diagnosis method for variable-speed planetary gearbox
CN112577746B (en) * 2020-12-07 2022-07-22 东南大学 Method for extracting rolling bearing envelope order spectrum fault features under rotation speed fluctuation
CN112577746A (en) * 2020-12-07 2021-03-30 东南大学 Method for extracting rolling bearing envelope order spectrum fault features under rotation speed fluctuation
CN113405823A (en) * 2021-05-17 2021-09-17 同济大学 Rotary machine fault diagnosis method based on iterative expansion eigenmode decomposition
CN113532835A (en) * 2021-08-18 2021-10-22 中国国家铁路集团有限公司 Railway contact net hard spot diagnosis method and device
CN113702046A (en) * 2021-09-13 2021-11-26 长沙理工大学 Bearing fault diagnosis method based on mobile equipment under variable rotating speed working condition
CN114484286B (en) * 2021-12-27 2023-06-27 西安理工大学 Adaptive filtering and noise reduction method for water hammer impact pressure signal
CN114484286A (en) * 2021-12-27 2022-05-13 西安理工大学 Self-adaptive filtering and noise reduction method for water hammer impact pressure signal
CN114563186A (en) * 2022-01-21 2022-05-31 苏州大学 Rolling bearing unbalance fault diagnosis method and system
CN114563186B (en) * 2022-01-21 2023-01-06 苏州大学 Rolling bearing unbalance fault diagnosis method and system
CN114486254A (en) * 2022-02-09 2022-05-13 青岛迈金智能科技股份有限公司 Bicycle bearing detection method based on time/frequency double-domain analysis
CN114659790A (en) * 2022-03-14 2022-06-24 浙江工业大学 Method for identifying bearing fault of variable-speed wind power high-speed shaft
CN114659790B (en) * 2022-03-14 2023-12-01 浙江工业大学 Identification method for bearing faults of variable-rotation-speed wind power high-speed shaft
CN115876475A (en) * 2023-02-06 2023-03-31 山东大学 Fault diagnosis method, system, equipment and storage medium
CN115876475B (en) * 2023-02-06 2023-06-09 山东大学 Fault diagnosis method, system, equipment and storage medium
CN116754234A (en) * 2023-08-17 2023-09-15 山东经典印务有限责任公司 Automatic printing production equipment running state detection method
CN116754234B (en) * 2023-08-17 2023-11-21 山东经典印务有限责任公司 Automatic printing production equipment running state detection method
CN117708574A (en) * 2024-02-02 2024-03-15 江苏南高智能装备创新中心有限公司 CNN variable-speed rolling bearing fault diagnosis method embedded with physical information
CN117708574B (en) * 2024-02-02 2024-04-12 江苏南高智能装备创新中心有限公司 CNN variable-speed rolling bearing fault diagnosis method embedded with physical information

Similar Documents

Publication Publication Date Title
CN111665051A (en) Bearing fault diagnosis method under strong noise variable-speed condition based on energy weight method
Liu et al. Vibration analysis for large-scale wind turbine blade bearing fault detection with an empirical wavelet thresholding method
Liu et al. Fault diagnosis of industrial wind turbine blade bearing using acoustic emission analysis
CN109682601B (en) Early fault identification method for rolling bearing under variable rotating speed working condition
Wang et al. Rolling element bearing fault diagnosis via fault characteristic order (FCO) analysis
CN110617964A (en) Synchronous compression transformation order ratio analysis method for fault diagnosis of rolling bearing
Zhao et al. A tacho-less order tracking technique for large speed variations
Yoon et al. On the use of a single piezoelectric strain sensor for wind turbine planetary gearbox fault diagnosis
CN110987438B (en) Method for detecting periodical vibration impact signals of hydraulic generator in variable rotating speed process
CN113125179B (en) Keyless phase order tracking method for rotating speed fluctuation of rotary machine
CN109883706B (en) Method for extracting local damage weak fault features of rolling bearing
CN109520738B (en) Rotating machinery rolling bearing fault diagnosis method based on order spectrum and envelope spectrum
Wang et al. A time–frequency-based maximum correlated kurtosis deconvolution approach for detecting bearing faults under variable speed conditions
Guo et al. Tooth root crack detection of planet and sun gears based on resonance demodulation and vibration separation
Lin et al. A review and strategy for the diagnosis of speed-varying machinery
JPH09113416A (en) Method for diagnosing damage of rolling bearing
CN110108467A (en) Active sounding speed-measuring method based on portable mobile apparatus
CN117836599A (en) Method for detecting bearing defects in a rotating system and monitoring system for implementing said method
Wu et al. A modified tacho-less order tracking method for the surveillance and diagnosis of machine under sharp speed variation
Zhao et al. Rolling element bearing instantaneous rotational frequency estimation based on EMD soft-thresholding denoising and instantaneous fault characteristic frequency
Wang et al. Tacholess order-tracking approach for wind turbine gearbox fault detection
Gryllias et al. A peak energy criterion (pe) for the selection of resonance bands in complex shifted morlet wavelet (csmw) based demodulation of defective rolling element bearings vibration response
CN113984282B (en) Rotor dynamic balance fault feature extraction method under keyless phase-change rotating speed
CN111323233A (en) Local mean decomposition method for low-speed rotating machine fault diagnosis
Song et al. Multispectral Balanced Automatic Fault Diagnosis for Rolling Bearings under Variable Speed Conditions

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20200915