CN112557038A - Bearing early fault diagnosis method based on multiple noise reduction processing - Google Patents

Bearing early fault diagnosis method based on multiple noise reduction processing Download PDF

Info

Publication number
CN112557038A
CN112557038A CN202011623754.4A CN202011623754A CN112557038A CN 112557038 A CN112557038 A CN 112557038A CN 202011623754 A CN202011623754 A CN 202011623754A CN 112557038 A CN112557038 A CN 112557038A
Authority
CN
China
Prior art keywords
bearing
fault
signal
vibration signal
frequency
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
CN202011623754.4A
Other languages
Chinese (zh)
Inventor
王林军
徐洲常
蔡康林
刘洋
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.)
China Three Gorges University CTGU
Original Assignee
China Three Gorges University CTGU
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 China Three Gorges University CTGU filed Critical China Three Gorges University CTGU
Priority to CN202011623754.4A priority Critical patent/CN112557038A/en
Publication of CN112557038A publication Critical patent/CN112557038A/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
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G06F2218/04—Denoising
    • G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching

Abstract

The invention discloses a bearing early fault diagnosis method based on multiple noise reduction processing, which comprises the following steps: collecting a vibration signal of a bearing; carrying out short-time Fourier transform on the bearing vibration signal, and preliminarily judging whether the bearing has a fault or not; decomposing and reconstructing a bearing vibration signal by utilizing wavelet packet transformation, and carrying out preliminary denoising; decomposing, screening and reconstructing wavelet packet reconstruction signals by using a set empirical mode decomposition method; eliminating aliasing interference signals contained in the reconstructed signal, and performing multilayer noise reduction on the bearing vibration signal; demodulating the reconstructed signal after noise reduction, and extracting the fault frequency of the bearing; and comparing with the theoretical calculation fault frequency, and diagnosing to obtain a fault conclusion of the bearing. The fault analysis method combining wavelet packet transformation, ensemble empirical mode decomposition and autocorrelation calculation denoising, which is adopted by the invention, highlights weak fault characteristics, is beneficial to early diagnosis and identification of bearing abnormity at the early fault stage of the bearing, and avoids or reduces loss caused by equipment fault.

Description

Bearing early fault diagnosis method based on multiple noise reduction processing
Technical Field
The invention belongs to the field of fault identification and diagnosis, and particularly relates to a bearing early fault diagnosis method based on multiple noise reduction processing.
Background
The rotary machine is widely applied to industries related to national civilization, such as aerospace, military industry, petroleum industry and the like. Rolling bearings are a very vulnerable or malfunctioning component within rotating machinery. The relevant data indicate that about one third of the failures of rotating machines are caused by the occurrence of failures of rolling bearings; in the case of a motor failure, the failure due to bearing failure is roughly two fifths; in a part of motor cars in China, all used rolling bearings need to be detected, wherein about one third of the rolling bearings need to be exchanged, otherwise, irreversible serious accidents are easily caused. The facts show that in order to ensure stable and efficient operation of equipment, corresponding monitoring means and fault diagnosis means are very necessary, and the method has practical significance for ensuring personnel safety and reducing economic loss.
The rolling bearing plays a role in realizing transmission and connection functions in the rotating machinery, and the operation state of the rolling bearing is directly related to the safe operation of an equipment system. The bearing early fault diagnosis is to judge the damaged part of the bearing at the early fault stage, which is beneficial to taking the equipment maintenance measures immediately and reducing the risk of the unplanned shutdown of the equipment. However, in the early failure stage of the rolling bearing, the rolling bearing often has the characteristics of weak impact components, large noise interference of the surrounding environment and the like, so that on one hand, whether the bearing has early failure or not is difficult to identify, and on the other hand, the difficulty in diagnosing the early failure type of the bearing is increased.
Disclosure of Invention
The invention has the technical problems that the impact component is weak in the early fault stage of a bearing, the environmental noise interference is large, the existing variational modal Decomposition method VMD, Empirical modal Decomposition method EMD or Ensemble Empirical Mode Decomposition (EEMD) can not effectively screen and distinguish the modal component containing weak fault signal components of the bearing, and the methods such as a wavelet packet transformation method, an Ensemble Empirical Mode Decomposition method and the like can not effectively remove a large amount of interference and noise in a vibration signal and can not obtain clear fault characteristic signals.
The invention aims to solve the problems, and provides a bearing early fault diagnosis method for multiple noise reduction processing.
The technical scheme of the invention is a bearing early fault diagnosis method by multiple noise reduction treatment, which comprises the following steps:
step 1: collecting a vibration signal of a bearing;
step 2: carrying out short-time Fourier transform on the bearing vibration signal, and preliminarily judging whether the bearing has a fault or not;
and step 3: decomposing and reconstructing a bearing vibration signal by utilizing wavelet packet transformation, and carrying out preliminary denoising;
and 4, step 4: decomposing, screening and reconstructing wavelet packet reconstruction signals by using a set empirical mode decomposition method;
and 5: eliminating aliasing interference signals contained in the reconstructed signal, and performing multilayer noise reduction on the bearing vibration signal;
step 6: demodulating the reconstructed signal after noise reduction, and extracting the fault frequency of the bearing;
and 7: and comparing with the theoretical calculation fault frequency, and determining the fault type of the current bearing.
Further, in step 2, given a window function γ (t) with a short time width, and sliding the window, the short-time fourier transform of the signal z (t) is calculated as follows
Figure BDA0002874397570000021
Wherein denotes complex conjugation, t and f denote time and frequency, respectively, STFTz(t, f) represents the result of the short-time Fourier transform of the signal z (t).
Further, in step 3, the bearing vibration signal is decomposed and reconstructed, and the calculation formula of the dyadic wavelet packet decomposition is as follows:
Figure BDA0002874397570000022
wherein f (t) represents a time signal;
Figure BDA0002874397570000023
represents the 1 st wavelet packet, i.e., the time signal itself; j ═ 0,1, …, J-2, J-1; 1,2, …,2j-1,2j；
Figure BDA0002874397570000024
j is the number of decomposition layers, i is the number of decompositions in each layer, N is an integer of power of 2, k is 1,2, …, N; wavelet packet coefficient
Figure BDA0002874397570000025
Represents the ith wavelet packet on the jth layer; G. h are all wavelet decomposition filter functions. The calculation formula for reconstructing the bearing vibration signal is as follows:
Figure BDA0002874397570000026
in the formula
Figure BDA0002874397570000027
And h and g are wavelet reconstruction filtering functions.
Further, step 4 specifically includes:
1) decomposing the wavelet packet reconstructed signal x (t) into a sum of several modal and residual components r (t),
Figure BDA0002874397570000028
in the formula cj(t) is the j-th modal component obtained by ensemble empirical mode decomposition of the signal, j being 1,2, …, n; n represents the total number of modal components obtained by decomposition;
2) calculating kurtosis values and cross-correlation coefficients of the modal components:
Figure BDA0002874397570000031
Figure BDA0002874397570000034
wherein K represents the kurtosis value, PS,CjRepresenting the cross-correlation coefficient between the jth modal component and the vibration signal; xmsA root mean square value representing the vibration signal; x (i) is a vibration signal; n is the number of points of the vibration signal; rS,Cj(t) denotes the j-th modal component cj(t) a cross-correlation function with the vibration signal; rS(t) is the autocorrelation function of the vibration signal;
3) comprehensively considering the kurtosis value and the cross correlation coefficient, and selecting effective modal components;
4) and superposing the selected modal components to complete the reconstruction of the vibration signal.
Further, step 5, aliasing interference signals contained in the reconstructed signal are removed by an ensemble empirical mode decomposition method through autocorrelation processing, a signal x (t) is a periodic signal, a noise signal n (t) is superimposed to become x (t), and an autocorrelation function of the signal x (t) is as follows:
Figure BDA0002874397570000032
in the formula Rxx(τ)、Rnn(τ) is the autocorrelation function of signals x (t), n (t), respectively; rxn(τ)、Rnx(τ) is the cross-correlation function between signal x (t) and signal n (t); when τ is not equal to 0 and the value is large, Rxn(τ)、Rnx(τ) and RnnThe values of (tau) all tend to be 0, and R (tau) is approximately equal to Rxx(τ)。
The measured bearing vibration signal is a periodic function mixed with noise and interference, wherein the bearing signal has small correlation with the noise signal and has large correlation with the bearing signal. Therefore, denoising of the reconstructed signal can be achieved by autocorrelation processing.
Further, step 6 specifically includes:
1) performing Hilbert demodulation processing on the denoised reconstruction signal:
Figure BDA0002874397570000033
wherein X (t)' represents the convolution result of the vibration signal and the shock response h (t) of the system; τ is a time parameter in the signal; x (tau) is the index of X (t); t is a time parameter in the impulse response;
2) and obtaining a signal envelope spectrogram after demodulation processing, and extracting the bearing fault frequency according to the envelope spectrogram.
Further, step 7 specifically includes:
1) calculating theoretical fault frequency and frequency conversion of the bearing:
Figure BDA0002874397570000041
Figure BDA0002874397570000042
Figure BDA0002874397570000043
wherein Z represents the number of balls; d represents the ball diameter; d represents a pitch circle diameter; α represents a contact angle; n isiRepresenting the rotating speed of the bearing inner ring; f. ofiRepresenting the theoretical failure frequency of the inner race of the bearing, fjBearing outer race failure frequency, F, representing theoryrRepresenting a theoretical bearing frequency;
2) and comparing the acquired fault information of the bearing with the calculated theoretical fault frequency of the bearing, and determining the fault type of the current bearing.
Compared with the prior art, the invention has the beneficial effects that:
1) according to the invention, a fault analysis mode combining wavelet packet transformation, an ensemble empirical mode decomposition method and autocorrelation calculation denoising is adopted, so that the impact characteristic of the early fault signal of the bearing is effectively increased, the weak fault characteristic is highlighted, early diagnosis and identification of bearing abnormity at the early fault stage of the bearing are facilitated, and the loss caused by equipment fault is avoided or reduced;
2) according to the method, after the bearing vibration signals are subjected to preliminary denoising through wavelet transformation, autocorrelation calculation denoising is performed, the bearing vibration signals are subjected to multiple denoising, noise components are effectively eliminated, fault characteristic signals are separated from original signals, and the accurate diagnosis of the faults of the inner ring and the outer ring of the bearing is realized;
3) according to the invention, whether the collected bearing vibration signal is a fault signal or not is preliminarily judged by time-frequency analysis, and only the judged fault signal is further diagnosed and classified, so that the fault diagnosis efficiency is improved, and time and labor are saved;
4) according to the invention, multiple signal processing methods such as Fourier change, wavelet packet transformation, ensemble empirical mode decomposition, autocorrelation processing and Hilbert transformation are cross-fused, so that a new composite fault diagnosis method is formed, the new method improves the diagnosis speed and precision, and advantage complementation is realized.
5) According to the method, the effective modal components are comprehensively judged and screened by utilizing the kurtosis value and the cross-correlation coefficient, the information contained in the modal components is mined, and the accuracy of bearing fault diagnosis is improved;
6) the method has good anti-noise effect on early fault diagnosis of the bearing, is easy to realize, is easy to popularize and apply to other rotary mechanical equipment for health monitoring of the mechanical equipment, and has good application prospect.
Drawings
The invention is further illustrated by the following figures and examples.
Fig. 1 is a schematic flow chart of a bearing early failure diagnosis method according to an embodiment of the present invention.
FIG. 2a is a time domain waveform diagram of a normal vibration signal of a bearing according to an embodiment of the present invention.
Fig. 2b is a time domain waveform diagram of a bearing inner race fault vibration signal according to an embodiment of the invention.
Fig. 3a is a time-frequency distribution diagram of a normal vibration signal of a bearing according to an embodiment of the present invention.
Fig. 3b is a time-frequency distribution diagram of a fault vibration signal of the bearing inner ring according to the embodiment of the invention.
Fig. 4a is a time domain characteristic diagram of a fault vibration signal of a bearing inner ring according to an embodiment of the invention.
Fig. 4b is a frequency domain characteristic diagram of a fault vibration signal of the bearing inner ring according to the embodiment of the invention.
Fig. 5 is a frequency band energy distribution diagram of a bearing inner ring fault vibration signal obtained by wavelet packet decomposition according to an embodiment of the present invention.
Fig. 6a is a time domain waveform diagram of a bearing inner race fault vibration signal obtained by wavelet packet transformation according to an embodiment of the present invention.
Fig. 6b is a frequency domain waveform diagram of a bearing inner ring fault vibration signal obtained by wavelet packet transformation according to an embodiment of the present invention.
Fig. 7 is a waveform diagram of an inner ring fault vibration signal of modal component reconstruction according to an embodiment of the present invention.
Fig. 8 is a schematic diagram of an autocorrelation calculation result of a reconstructed bearing inner race fault vibration signal according to an embodiment of the present invention.
Fig. 9 is an envelope spectrum obtained by demodulating a reconstructed inner ring fault vibration signal according to an embodiment of the present invention.
Fig. 10 is a schematic diagram of an autocorrelation calculation result of a reconstructed bearing outer ring fault vibration signal according to an embodiment of the present invention.
Fig. 11 is an envelope spectrum obtained by demodulating a reconstructed bearing outer ring fault vibration signal according to an embodiment of the present invention.
Fig. 12a is a waveform diagram of a normal vibration signal of a 6406 type bearing reconstructed according to an embodiment of the present invention.
Fig. 12b is a waveform diagram of the reconstructed vibration signal of the inner race fault of the 6406 type bearing according to the embodiment of the invention.
Fig. 12c is a waveform diagram of the reconstructed vibration signal of the 6406 type bearing outer ring fault according to the embodiment of the invention.
Detailed Description
As shown in fig. 1, the method for diagnosing the early failure of the bearing by multiple noise reduction processes includes the following steps:
step 1: collecting a vibration signal of a bearing;
step 2: carrying out short-time Fourier transform on the bearing vibration signal, and preliminarily judging whether the bearing has a fault or not;
given a window function γ (t) with a short time width, and sliding the window, the short-time Fourier transform of the signal z (t) is calculated as follows
Figure BDA0002874397570000051
Wherein denotes complex conjugation, t and f denote time and frequency, respectively, STFTz(t, f) represents the result of the short-time Fourier transform of the signal z (t).
And step 3: decomposing and reconstructing a bearing vibration signal by utilizing wavelet packet transformation, and carrying out preliminary denoising;
the calculation formula of the dyadic wavelet packet decomposition is as follows:
Figure BDA0002874397570000061
wherein f (t) represents a time signal;
Figure BDA0002874397570000062
represents the 1 st wavelet packet, i.e., the time signal itself; j ═ 0,1, …, J-2, J-1; 1,2, …,2j-1,2j；
Figure BDA0002874397570000063
j is the number of decomposition layers, i is the number of decompositions in each layer, N is an integer of power of 2, k is 1,2, …, N; wavelet packet coefficient
Figure BDA0002874397570000064
Represents the ith wavelet packet on the jth layer; G. h are all wavelet decomposition filter functions. The calculation formula for reconstructing the bearing vibration signal is as follows:
Figure BDA0002874397570000065
in the formula
Figure BDA0002874397570000066
And h and g are wavelet reconstruction filtering functions.
And 4, step 4: decomposing, screening and reconstructing wavelet packet reconstruction signals by using a set empirical mode decomposition method;
step 4.1: decomposing the wavelet packet reconstructed signal x (t) into a sum of several modal and residual components r (t),
Figure BDA0002874397570000067
in the formula cj(t) is the j-th modal component obtained by ensemble empirical mode decomposition of the signal, j being 1,2, …, n; n represents the total number of modal components obtained by decomposition;
step 4.2: calculating kurtosis values and cross-correlation coefficients of the modal components:
Figure BDA0002874397570000068
Figure BDA0002874397570000069
wherein K represents the kurtosis value, PS,CjRepresenting the cross-correlation coefficient between the jth modal component and the vibration signal; xmsA root mean square value representing the vibration signal; x (i) is a vibration signal; n is the number of points of the vibration signal; rS,Cj(t) denotes the j-th modal component cj(t) a cross-correlation function with the vibration signal; rS(t) is the autocorrelation function of the vibration signal;
step 4.3: comprehensively considering the kurtosis value and the cross correlation coefficient, and selecting effective modal components;
step 4.4: and superposing the selected modal components to complete the reconstruction of the vibration signal.
And 5: eliminating aliasing interference signals contained in the reconstructed signals by an ensemble empirical mode decomposition method through autocorrelation processing, and performing multilayer noise reduction on the bearing vibration signals;
the signal x (t) is a periodic signal, and becomes x (t) after the noise signal n (t) is superimposed, and the autocorrelation function of the signal x (t) is as follows:
Figure BDA0002874397570000071
in the formula Rxx(τ)、Rnn(τ) is the autocorrelation function of signals x (t), n (t), respectively; rxn(τ)、Rnx(τ) is the cross-correlation function between signal x (t) and signal n (t); when τ is not equal to 0 and the value is large, Rxn(τ)、Rnx(τ) and RnnThe values of (tau) all tend to be 0, and R (tau) is approximately equal to Rxx(τ)。
The measured bearing vibration signal is a periodic function mixed with noise and interference, wherein the bearing signal has small correlation with the noise signal and has large correlation with the bearing signal. Therefore, denoising of the reconstructed signal can be achieved by autocorrelation processing.
Step 6: demodulating the reconstructed signal after noise reduction, and extracting the fault frequency of the bearing;
step 6.1: performing Hilbert demodulation processing on the denoised reconstruction signal:
Figure BDA0002874397570000072
wherein X (t)' represents the convolution result of the vibration signal and the shock response h (t) of the system; τ is a time parameter in the signal; x (tau) is the index of X (t); t is a time parameter in the impulse response;
step 6.2: and obtaining a signal envelope spectrogram after demodulation processing, and extracting the bearing fault frequency according to the envelope spectrogram.
And 7: comparing with the theoretical calculation fault frequency, and diagnosing to obtain a fault conclusion of the bearing;
step 7.1: calculating theoretical fault frequency and frequency conversion of the bearing:
Figure BDA0002874397570000073
Figure BDA0002874397570000074
Figure BDA0002874397570000075
wherein Z represents the number of balls; d represents the ball diameter; d represents a pitch circle diameter; α represents a contact angle; n isiRepresenting the rotating speed of the bearing inner ring; f. ofiRepresenting the theoretical failure frequency of the bearing inner ring; f. ofjRepresenting theoretical bearing outer ring fault frequency; frRepresenting a theoretical bearing frequency;
step 7.2: and comparing the acquired fault information of the bearing with the calculated theoretical fault frequency of the bearing, and determining the fault type of the current bearing.
In the examples, the rolling bearing data set disclosed in American Case Western Reserve University was used as experimental data, and a 6205-2 RS type deep groove ball bearing was used as an experimental subject. The bearing-related parameters used include: the number of the balls is 9, the diameter of the balls is 7.94mm, the inner diameter of the bearing is 25mm, the outer diameter of the bearing is 52mm, the contact angle is 65 degrees, and the rotating speed of the inner ring is 1750 r/min. The experimental data comprise inner ring faults and outer ring faults, and the faults are single-point damages. The sampling frequency of the bearing data used was 12kHz, and the first 4096 points of a partial data set were taken for case analysis.
As shown in fig. 2a, 2b, 3a, and 3b, the embodiment compares and analyzes the measured normal vibration signal of the bearing and the fault vibration signal of the inner ring of the bearing from the angle of time-frequency distribution. As shown in fig. 3a and 3b, when the bearing works normally, the energy of the vibration signal is mainly concentrated in the low frequency part; when the inner ring of the bearing is in fault, the energy of the low-frequency part is obviously reduced, and the energy is mainly concentrated to the high-frequency part, so that a preliminary diagnosis result can be obtained, namely whether the bearing is in fault or not is judged, and further processing of a fault signal is facilitated. When a large amount of noise is mixed in the actually measured signal, wavelet denoising or wavelet packet denoising can be performed on the signal, and then short-time Fourier transform is performed and analysis is performed.
As shown in fig. 4a and 4b, the embodiment analyzes the measured bearing inner ring fault signal from the angles of the time domain and the frequency domain respectively. As can be seen from fig. 4b, the original vibration signal of the bearing contains a lot of random noise and interference, and it is difficult to obtain the fault information of the bearing by means of the frequency domain analysis result. Therefore, in the embodiment, the fault signal is subjected to wavelet packet conversion processing.
The bearing vibration signal was subjected to wavelet packet decomposition processing with the number of decomposition layers set to 3, and the energy distribution of each band of layer 3 was calculated, with the result shown in fig. 5. Analyzing fig. 5, it is found that the energy of the signal is mainly concentrated in the 5 th band (node), and the signal is reconstructed using this band, and the reconstruction results are shown in fig. 6a and 6 b.
Comparing fig. 4a and 4b with fig. 6a and 6b, it can be seen that most of noise signals are eliminated from the original vibration signals after wavelet packet denoising processing, but the fault characteristic signals are still not obvious, and clear fault signals cannot be obtained only by wavelet packet conversion processing.
Decomposing the signal after wavelet packet conversion by EEMD method, i.e. adding Gaussian white noise into the vibration signal after wavelet packet conversion to obtain mixed noise signal. And carrying out empirical mode decomposition on the mixed noise signal to obtain each modal component. Repeatedly adding different white noises into the signal and repeatedly carrying out empirical mode decomposition. And performing integrated average processing on the modal components obtained by each empirical mode decomposition and taking the modal components as final results. Kurtosis values and correlation coefficient values (absolute values) of the processed modal components (IMF components) are calculated, as shown in table 1.
TABLE 1 indicator table for each modal component of bearing inner ring fault signal
Figure BDA0002874397570000091
And analyzing the index values of all IMF components in the table 1, and finding that the kurtosis values of IMFs 1-9 are all larger than 3, but the cross correlation coefficients of a part of components are small, so that the fault requirements are not met. Only IMF1, IMF2, IMF3 and IMF4 simultaneously meet the requirements that the correlation coefficient is greater than 0.001 and the kurtosis value is greater than 3, IMF1, IMF2, IMF3 and IMF4 are selected for reconstruction, the components are superposed to realize signal reconstruction, and the reconstruction result is shown in figure 7.
The reconstructed signal only shows partial impact characteristics without obvious periodic characteristics, which indicates that the signal denoising is not thorough, and the required fault characteristics and information cannot be obtained by utilizing wavelet packet transformation and EEMD processing. In order to make the fault characteristics of the reconstructed signal more obvious, the reconstructed signal is subjected to autocorrelation calculation denoising again, the result of autocorrelation calculation denoising is shown in fig. 8, a signal time domain waveform is obtained, a signal envelope spectrogram is obtained through hilbert demodulation and is shown in fig. 9, the inner ring fault frequency and the frequency of the bearing can be obtained through analyzing the signal envelope spectrogram shown in fig. 9, and the separated bearing inner ring fault frequency is 156.7 Hz. And the characteristic frequency of the fault of the bearing inner ring obtained by theoretical calculation is 157.76 Hz. The relative error between the failure frequency of the bearing inner ring obtained by the method and the failure characteristic frequency of the bearing inner ring calculated theoretically is 0.67 percent, and the error is in a reasonable range, which shows that the failure information of the bearing inner ring is separated from the original signal.
Comparing the processing results of the inner ring fault signals of fig. 4a, 4b, fig. 6a, 6b, fig. 7 and fig. 8, the following conclusions are obtained by analysis: the bearing vibration signals with more complex components are processed by utilizing Fourier transform, and the bearing fault frequency is covered by excessive frequency domain information and is difficult to distinguish; the fault signal is processed by utilizing a wavelet packet transformation method, only a clear fault signal can be obtained, the fault characteristic is not obvious, and the required fault characteristic information cannot be obtained; the method provided by the invention is used for processing the bearing vibration signal, so that not only can a 'clean' fault signal be obtained, but also the fault characteristic frequency can be obtained.
The method provided by the invention is adopted to process the bearing outer ring fault signal, and the kurtosis value and the correlation coefficient value (absolute value) of each modal component are obtained, as shown in table 2. According to data in table 2, IMF1, IMF2, IMF3 and IMF4 are superimposed and subjected to autocorrelation calculation, denoising and demodulation processing, so as to obtain a signal time domain waveform diagram and a bearing outer ring fault envelope spectrogram, which are respectively shown in fig. 10 and fig. 11.
TABLE 2 indicator table for each modal component of bearing outer ring fault vibration signal
Figure BDA0002874397570000101
The bearing fault frequency extracted by the envelope spectrum is 105.5Hz, and the characteristic frequency of the bearing outer ring fault obtained by theoretical calculation is 104.57 Hz. The relative error between the two signals is only 0.88%, which indicates that the outer ring fault frequency is separated from the original signal.
The time domain waveform of the bearing vibration signal can be obtained by using the method provided by the invention, and whether the bearing has a fault or not is judged according to the waveform. In the embodiment, a 6406 type rolling bearing data set is used for verification, three groups of signals, namely, a normal signal, an inner ring fault signal and an outer ring fault signal, of the 6406 type rolling bearing are processed by the method, and the processing results of the bearing signals under different states are obtained, as shown in fig. 12a, 12b and 12c, it can be seen that the time domain characteristics of the bearings under different working conditions are obviously different, and accordingly, the primary judgment of the bearing fault can be realized.
Through the comparison and analysis of the processing results of the bearing signals of different models, the method can effectively remove noise and interference, and separate the characteristic frequency of the bearing fault from the original signal, thereby diagnosing and determining the fault type of the bearing.
The bearing early fault diagnosis method adopting the multiple noise reduction processing can predict the early fault of the bearing in advance, and identify the damaged part, namely the fault type, in the early stage of the fault, so that the method has higher real-time performance and sensitivity. The method is not only suitable for identifying the abnormal state of the bearing in the embodiment, but also can be applied to other rotating mechanical equipment and key parts thereof, and provides reasonable reference for health evaluation. The invention has good expandability and provides a certain reference function for other technical personnel in the technical field.
It is to be noted that the flow charts in the attached drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. It will also be noted that each block of the flowchart illustrations, and combinations of blocks in the block diagrams, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

Claims (6)

1. The method for diagnosing the early fault of the bearing subjected to the multiple noise reduction processing is characterized by comprising the following steps of:
step 1: collecting a vibration signal of a bearing;
step 2: carrying out short-time Fourier transform on the bearing vibration signal, and preliminarily judging whether the bearing has a fault or not;
and step 3: decomposing and reconstructing a bearing vibration signal by utilizing wavelet packet transformation, and carrying out preliminary denoising;
and 4, step 4: decomposing, screening and reconstructing wavelet packet reconstruction signals by using a set empirical mode decomposition method;
and 5: eliminating aliasing interference signals contained in the reconstructed signal, and performing multilayer noise reduction on the bearing vibration signal;
step 6: demodulating the reconstructed signal after noise reduction, and extracting the fault frequency of the bearing;
and 7: and comparing with the theoretical calculation fault frequency, and diagnosing to obtain a fault conclusion of the bearing.
2. The method for diagnosing the early failure of the bearing with multiple noise reduction processes according to claim 1, wherein in step 2, given a window function γ (t) with a short time width, the window is slid, and then the short-time fourier transform of the signal z (t) is calculated as follows:
Figure FDA0002874397560000011
wherein denotes complex conjugate, f denotes frequency, t denotes time, STFTz(t, f) represents the result of the short-time Fourier transform of the signal z (t).
3. The method for diagnosing the early failure of the bearing with multiple noise reduction processes according to claim 1, wherein in step 3, the bearing vibration signal is decomposed and reconstructed, and the calculation formula of the binary wavelet packet decomposition is as follows:
Figure FDA0002874397560000012
wherein f (t) represents a time signal;
Figure FDA0002874397560000013
represents the 1 st wavelet packet, i.e., the time signal itself; j ═ 0,1, …, J-2, J-1; 1,2, …,2j-1,2j；
Figure FDA0002874397560000014
j is the number of decomposition layers, i is the number of decompositions in each layer, N is an integer of power of 2, k is 1,2, …, N; wavelet packet coefficient
Figure FDA0002874397560000015
Represents the ith wavelet packet on the jth layer; G. h is all wavelet decomposition filter functions; the calculation formula for reconstructing the bearing vibration signal is as follows:
Figure FDA0002874397560000016
in the formula
Figure FDA0002874397560000021
Representing the ith wavelet on the jth layerAnd the packets h and g are wavelet reconstruction filtering functions.
4. The method for diagnosing the early failure of the bearing with the multiple noise reduction treatments according to claim 3, wherein the step 4 specifically comprises:
1) decomposing the wavelet packet reconstructed signal x (t) into a sum of several modal and residual components r (t),
Figure FDA0002874397560000022
in the formula cj(t) is the j-th modal component obtained by ensemble empirical mode decomposition of the signal, j being 1,2, …, n; n represents the total number of modal components obtained by decomposition;
2) calculating kurtosis values and cross-correlation coefficients of the modal components:
Figure FDA0002874397560000023
Figure FDA0002874397560000025
wherein K represents the kurtosis value, PS,CjRepresenting the jth modal component cj(t) cross-correlation coefficient with the vibration signal; xmsA root mean square value representing the vibration signal; x (i) is a vibration signal; n is the number of points of the vibration signal; rS,Cj(t) denotes the j-th modal component cj(t) a cross-correlation function with the vibration signal; rS(t) is the autocorrelation function of the vibration signal;
3) comprehensively considering the kurtosis value and the cross correlation coefficient, and selecting effective modal components;
4) and superposing the selected modal components to complete the reconstruction of the vibration signal.
5. The method for diagnosing the early failure of the bearing with the multiple noise reduction treatments according to claim 4, wherein the step 6 specifically comprises:
1) performing Hilbert demodulation processing on the denoised reconstruction signal:
Figure FDA0002874397560000024
wherein X (t)' represents the convolution result of the vibration signal and the shock response h (t) of the system, tau is a time parameter in the signal, X (tau) is the element change of X (t), and t is the time parameter in the shock response;
2) and obtaining a signal envelope spectrogram after demodulation processing, and extracting the bearing fault frequency according to the envelope spectrogram.
6. The method for diagnosing the early failure of the bearing with the multiple noise reduction treatments according to claim 5, wherein the step 7 specifically comprises:
1) calculating theoretical fault frequency and frequency conversion of the bearing:
Figure FDA0002874397560000031
Figure FDA0002874397560000032
Figure FDA0002874397560000033
wherein Z represents the number of balls; d represents the ball diameter; d represents a pitch circle diameter; α represents a contact angle; n isiRepresenting the rotating speed of the bearing inner ring; f. ofiRepresenting the theoretical failure frequency of the inner race of the bearing, fjBearing outer race failure frequency, F, representing theoryrRepresenting a theoretical bearing frequency;
2) and comparing the acquired fault information of the bearing with the calculated theoretical fault frequency of the bearing, and determining the fault type of the current bearing.
CN202011623754.4A 2020-12-30 2020-12-30 Bearing early fault diagnosis method based on multiple noise reduction processing Pending CN112557038A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011623754.4A CN112557038A (en) 2020-12-30 2020-12-30 Bearing early fault diagnosis method based on multiple noise reduction processing

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011623754.4A CN112557038A (en) 2020-12-30 2020-12-30 Bearing early fault diagnosis method based on multiple noise reduction processing

Publications (1)

Publication Number Publication Date
CN112557038A true CN112557038A (en) 2021-03-26

Family

ID=75034990

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011623754.4A Pending CN112557038A (en) 2020-12-30 2020-12-30 Bearing early fault diagnosis method based on multiple noise reduction processing

Country Status (1)

Country Link
CN (1) CN112557038A (en)

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113281042A (en) * 2021-06-28 2021-08-20 江苏大学 Early fault diagnosis system and method for walking gearbox of combine harvester
CN113375939A (en) * 2021-05-28 2021-09-10 三峡大学 Mechanical part fault diagnosis method based on SVD and VMD
CN113375940A (en) * 2021-05-28 2021-09-10 三峡大学 Fault bearing diagnosis method based on SVD and CEEMDAN
CN113486750A (en) * 2021-06-29 2021-10-08 国家电网有限公司 Oil-immersed transformer partial discharge signal denoising method based on improved VMD algorithm and wavelet packet denoising algorithm
CN113567129A (en) * 2021-07-28 2021-10-29 大连海天兴业科技有限公司 CEEMD-based noise reduction method for train bearing vibration signal
CN113865871A (en) * 2021-11-03 2021-12-31 沈阳航空航天大学 Rolling bearing fault diagnosis method based on fault characteristic frequency multiplication energy judgment
CN113916536A (en) * 2021-09-03 2022-01-11 山东科技大学 Sinusoidal roller screen coupling fault diagnosis method and system
CN114235396A (en) * 2021-11-26 2022-03-25 电子科技大学 Gear reducer fault feature extraction method
CN114486263A (en) * 2022-02-15 2022-05-13 浙江大学 Noise reduction and demodulation method for vibration signal of rolling bearing of rotary machine
CN114781427A (en) * 2022-03-17 2022-07-22 华电河南新能源发电有限公司 Wind generating set rolling bearing fault diagnosis system based on information fusion technology
CN117250004A (en) * 2023-11-13 2023-12-19 山东省宇捷轴承制造有限公司 Bearing fault diagnosis method and system based on digital signal processing
CN114781427B (en) * 2022-03-17 2024-05-14 华电河南新能源发电有限公司 Wind generating set antifriction bearing fault diagnosis system based on information fusion technology

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101038231A (en) * 2006-03-17 2007-09-19 中国石油天然气股份有限公司 Vibration monitoring and failure diagnosis system of gas turbine
CN108168811A (en) * 2017-12-12 2018-06-15 电子科技大学 The Portable acquiring analytical equipment and method of a kind of vibration signal
CN110514441A (en) * 2019-08-28 2019-11-29 湘潭大学 A kind of Fault Diagnosis of Roller Bearings based on vibration signal denoising and Envelope Analysis

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101038231A (en) * 2006-03-17 2007-09-19 中国石油天然气股份有限公司 Vibration monitoring and failure diagnosis system of gas turbine
CN108168811A (en) * 2017-12-12 2018-06-15 电子科技大学 The Portable acquiring analytical equipment and method of a kind of vibration signal
CN110514441A (en) * 2019-08-28 2019-11-29 湘潭大学 A kind of Fault Diagnosis of Roller Bearings based on vibration signal denoising and Envelope Analysis

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
张超: "《基于振动分析的旋转机械故障诊断原理及应用》", 30 April 2014 *
李刚: "基于EEMD和小波包的机车轴箱轴承故障诊断方法", 《兰州交通大学学报》 *
王中明: "《信号与系统[M]》", 31 December 2019 *

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113375939B (en) * 2021-05-28 2022-04-08 三峡大学 Mechanical part fault diagnosis method based on SVD and VMD
CN113375939A (en) * 2021-05-28 2021-09-10 三峡大学 Mechanical part fault diagnosis method based on SVD and VMD
CN113375940A (en) * 2021-05-28 2021-09-10 三峡大学 Fault bearing diagnosis method based on SVD and CEEMDAN
CN113281042B (en) * 2021-06-28 2022-09-16 江苏大学 Early fault diagnosis system and method for walking gearbox of combine harvester
CN113281042A (en) * 2021-06-28 2021-08-20 江苏大学 Early fault diagnosis system and method for walking gearbox of combine harvester
CN113486750A (en) * 2021-06-29 2021-10-08 国家电网有限公司 Oil-immersed transformer partial discharge signal denoising method based on improved VMD algorithm and wavelet packet denoising algorithm
CN113567129A (en) * 2021-07-28 2021-10-29 大连海天兴业科技有限公司 CEEMD-based noise reduction method for train bearing vibration signal
CN113916536A (en) * 2021-09-03 2022-01-11 山东科技大学 Sinusoidal roller screen coupling fault diagnosis method and system
CN113865871A (en) * 2021-11-03 2021-12-31 沈阳航空航天大学 Rolling bearing fault diagnosis method based on fault characteristic frequency multiplication energy judgment
CN114235396A (en) * 2021-11-26 2022-03-25 电子科技大学 Gear reducer fault feature extraction method
CN114486263A (en) * 2022-02-15 2022-05-13 浙江大学 Noise reduction and demodulation method for vibration signal of rolling bearing of rotary machine
CN114781427A (en) * 2022-03-17 2022-07-22 华电河南新能源发电有限公司 Wind generating set rolling bearing fault diagnosis system based on information fusion technology
CN114781427B (en) * 2022-03-17 2024-05-14 华电河南新能源发电有限公司 Wind generating set antifriction bearing fault diagnosis system based on information fusion technology
CN117250004A (en) * 2023-11-13 2023-12-19 山东省宇捷轴承制造有限公司 Bearing fault diagnosis method and system based on digital signal processing

Similar Documents

Publication Publication Date Title
CN112557038A (en) Bearing early fault diagnosis method based on multiple noise reduction processing
He et al. Tunable Q-factor wavelet transform denoising with neighboring coefficients and its application to rotating machinery fault diagnosis
Zheng et al. Incipient fault detection of rolling bearing using maximum autocorrelation impulse harmonic to noise deconvolution and parameter optimized fast EEMD
Vishwakarma et al. Vibration analysis & condition monitoring for rotating machines: a review
Yin et al. Weak fault feature extraction of rolling bearings based on improved ensemble noise-reconstructed EMD and adaptive threshold denoising
Singh et al. Rolling element bearing fault diagnosis based on over-complete rational dilation wavelet transform and auto-correlation of analytic energy operator
Yang et al. Vibration feature extraction techniques for fault diagnosis of rotating machinery: a literature survey
Guo et al. A compound fault diagnosis method of rolling bearing based on wavelet scattering transform and improved soft threshold denoising algorithm
Guo et al. Faulty bearing signal recovery from large noise using a hybrid method based on spectral kurtosis and ensemble empirical mode decomposition
Zhang et al. Multi-fault diagnosis of gearbox based on resonance-based signal sparse decomposition and comb filter
Assaad et al. Vibration based condition monitoring of a multistage epicyclic gearbox in lifting cranes
CN111521400B (en) Bearing early fault diagnosis method based on EDM and spectral kurtosis
CN102607845B (en) Bearing fault characteristic extracting method for redundantly lifting wavelet transform based on self-adaptive fitting
Wang et al. Kurtogram manifold learning and its application to rolling bearing weak signal detection
CN113375939B (en) Mechanical part fault diagnosis method based on SVD and VMD
Luo et al. Real-time condition monitoring by significant and natural frequencies analysis of vibration signal with wavelet filter and autocorrelation enhancement
Pang et al. Enhanced singular spectrum decomposition and its application to rolling bearing fault diagnosis
Fan et al. Machine fault feature extraction based on intrinsic mode functions
Zhou et al. Using frequency domain analysis techniques for diagnosis of planetary bearing defect in a CH-46E helicopter aft gearbox
CN109883706A (en) A kind of rolling bearing local damage Weak fault feature extracting method
Jena et al. Radial ball bearing inner race defect width measurement using analytical wavelet transform of acoustic and vibration signal
Kong et al. Fault feature extraction of planet gear in wind turbine gearbox based on spectral kurtosis and time wavelet energy spectrum
Bao et al. Anti-aliasing lifting scheme for mechanical vibration fault feature extraction
Singh et al. A review of vibration analysis techniques for rotating machines
Xu et al. Rolling bearing fault feature extraction via improved SSD and a singular-value energy autocorrelation coefficient spectrum

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