CN112098093A - Bearing fault feature identification method and system - Google Patents

Bearing fault feature identification method and system Download PDF

Info

Publication number
CN112098093A
CN112098093A CN202010968520.7A CN202010968520A CN112098093A CN 112098093 A CN112098093 A CN 112098093A CN 202010968520 A CN202010968520 A CN 202010968520A CN 112098093 A CN112098093 A CN 112098093A
Authority
CN
China
Prior art keywords
signal
fault
momeda
swd
bearing
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
CN202010968520.7A
Other languages
Chinese (zh)
Inventor
朱亚军
胡建钦
李武
林青云
易灿灿
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.)
Lishui Special Equipment Testing Institute
Original Assignee
Lishui Special Equipment Testing Institute
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 Lishui Special Equipment Testing Institute filed Critical Lishui Special Equipment Testing Institute
Priority to CN202010968520.7A priority Critical patent/CN112098093A/en
Publication of CN112098093A publication Critical patent/CN112098093A/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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching

Landscapes

  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • General Physics & Mathematics (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)

Abstract

The invention belongs to the technical field of bearing fault feature identification, and discloses a bearing fault feature identification method and system, wherein a group decomposition (SWD) is utilized to carry out mode decomposition on a vibration signal to obtain an oscillation modal component; performing noise reduction and frequency domain vibration signal feature extraction on the selected oscillation mode component by using a multipoint optimal minimum deconvolution MOMEDA (maximum deconvolution), so as to realize the enhancement of weak signal features; and finally, identifying the fault characteristics of the vibration signals. Firstly, SWD decomposition is carried out, and then MOMEDA is utilized to carry out feature extraction on the decomposed signals, so that the method aims to obtain useful signal components through SWD decomposition, improve the signal-to-noise ratio of the signals and remove the interference of irrelevant components; meanwhile, filtering of signals and accurate extraction of impact characteristics are achieved by using MOMEDA. The effectiveness of the method in identifying the fault characteristics of the rolling bearing is verified by carrying out numerical simulation analysis and experiment table vibration signal analysis.

Description

Bearing fault feature identification method and system
Technical Field
The invention belongs to the technical field of bearing fault feature identification, and particularly relates to a bearing fault feature identification method and system.
Background
At present, a rolling bearing is one of important parts in rotary machinery, and is widely applied to the fields of petrochemical industry, energy, electric power, material metallurgy and the like. Meanwhile, the rolling bearing is one of the most vulnerable parts, and 30% of rotating machinery faults are caused by structural damage of the rolling bearing, and the normal operation of the whole equipment is directly influenced by the operating state of the rolling bearing. Therefore, the method has important theoretical and engineering significance for realizing accurate diagnosis of the rolling bearing fault.
The method is one of the current key research directions for enhancing and extracting the weak fault characteristics of the rolling bearing. Among them, an Empirical Mode Decomposition (EMD) method is first proposed and applied to bearing fault diagnosis, but EMD has a large bandwidth for decomposing high-frequency components and a low frequency resolution, and tends to cause modal aliasing when decomposing a complex fault signal having a relatively close frequency component. Therefore, researchers have proposed improved algorithms based on EMD, such as Ensemble Empirical Mode Decomposition (EEMD), Local Mean Decomposition (LMD), and the like. From the actual analysis effect, multi-component modulation and strong interference are main characteristics of bearing composite faults, fault characteristics in signals are quite weak and difficult to extract, and the traditional decomposition method is limited by problems of mode aliasing, predefined mode number and the like, so that the weak fault signals are difficult to accurately extract in the composite faults. Apostolisis and the like propose a Swarm Decomposition (SWD) algorithm, and by setting Swarm filter parameters, the main modal frequency of the oscillation component can be controlled, so that two harmonic signals with similar frequencies can be separated when the SWD performs modal Decomposition, and compared with other EMD improvement methods, the method has higher frequency discrimination capability.
The Minimum Entropy Deconvolution (MED) is a time-domain blind convolution technique, which was first proposed and applied to fault diagnosis of rolling bearings. But the filter solved by the MED is not necessarily a globally optimal filter and often only a few pulse components can be extracted. With respect to these limitations of the MED, on the basis of the correlation Kurtosis, some researchers have proposed a Maximum correlation Kurtosis Deconvolution Method (MCKD), which can extract more pulse components than the MED, but can extract only a limited number of pulses locally, and needs a priori knowledge to set the fault period and the filter parameters. Therefore, McDonald et al propose a Multipoint Optimal Minimum Entropy Deconvolution (MOMEDA) method, and define the weight and position of the pulse sequence obtained by Deconvolution by using a target vector, and obtain an Optimal filter without an iterative algorithm. However, this method can only extract a unique periodic impulse during each noise reduction process, and is prone to misdiagnosis in the case of strong noise.
Through the above analysis, the problems and defects of the prior art are as follows:
(1) the existing EMD has large bandwidth of high-frequency components and low frequency resolution, and can easily cause mode aliasing when complex multi-component fault signals with relatively close frequency components are decomposed.
(2) The traditional mode decomposition method is often limited by the problems of mode aliasing, predefined mode number and the like, and is difficult to accurately extract weak fault signals in multi-component fault signals.
(3) The filter solved by the minimum entropy deconvolution method MED is not necessarily a globally optimal filter and often only a few pulse components can be extracted.
(4) The maximum correlation kurtosis deconvolution method MCKD can extract more pulse components than MED, but can only extract a limited number of pulses locally, and requires a priori knowledge to set the fault period and filter parameters.
(5) The optimal minimum entropy deconvolution method MOMEDA can only extract unique periodic impact in each noise reduction process, and is easy to have a misdiagnosis phenomenon under the condition of strong noise.
The difficulty in solving the above problems and defects is: 1) the actually measured bearing fault signal is often a typical multi-component signal, which contains useful signal components such as fault characteristic components and frequency multiplication thereof, and also contains irrelevant components such as strong noise interference, so that the key for processing the fault signal is to decompose the multi-component signal and find oscillation mode components related to fault characteristics; 2) the key of fault diagnosis is to effectively identify fault features, so after a desired oscillation mode component is obtained, a suitable method needs to be found for signal enhancement and feature amplification, and the fault feature analysis capability is further improved.
The significance of solving the problems and the defects is as follows: on one hand, theoretical innovation can be realized, a novel method for decomposing SWD and deconvoluting MOMEDA through the optimal minimum entropy is provided, and a powerful technical means can be provided for decomposition and feature enhancement of bearing fault signals; on the other hand, the bearing can be applied to the fields of metallurgy, chemical engineering and the like widely existing in the rotary machinery, the service life of the bearing in the rotary machinery is prolonged, the predicted maintenance is realized, and the occurrence of major safety accidents is avoided.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a bearing fault feature identification method based on SWD and MOMEDA.
The invention is realized in such a way that a bearing fault feature recognition method based on SWD and MOMEDA comprises the following steps:
step one, performing mode decomposition on the vibration signal by using SWD to obtain an oscillation mode component.
And step two, performing noise reduction and frequency domain vibration signal feature extraction on the selected oscillation mode component by using MOMEDA (motion estimation and motion estimation), and realizing the enhancement of weak signal features.
And step three, identifying fault characteristics and analyzing the numerical simulation signals.
Further, in the first step, the multi-component signal is decomposed into a sum of a plurality of oscillation components by the iterative group filter, and the SWD process is as follows:
(1) determination of initial center frequency from signal power spectrum peak
Figure BDA0002683224030000031
Figure BDA0002683224030000032
Wherein S isy(w) denotes the Welch power spectrum of the signal, q denotes the qth frequency w as the center frequency during SWD, PhIs a threshold value. By
Figure BDA0002683224030000033
SwF parameters M and:
Figure BDA0002683224030000034
wherein odd (·) represents a rounding operation.
(2) SwF filtering is carried out on the signal y (t) to obtain an output signal u (t), and the variance of the input and output signals is calculated as follows:
Figure BDA0002683224030000041
if the variance D is larger than the variance threshold DhRepeating SwF filtering with u (t) as input until D < DhWhen the output signal is recorded as yi(t)。
(3) Updating the input signal:
Figure BDA0002683224030000042
wherein the content of the first and second substances,
Figure BDA0002683224030000043
for the cross-correlation function, τ denotes the time delay.
(4) Repeating the above steps (1) - (3) until S using the updated input signaly(w)≤PhThe input signal at this time is the residual r (t).
(5) Calculating a plurality of oscillation mode components according to the following formula:
Figure BDA0002683224030000044
wherein omegad＝{w:w＝wdAnd k is the number of oscillation mode components.
Further, in step two, the purpose of the MOMEDA method is to find an optimal FIR filter f (l) coefficient, so that the objective function of the output signal after MOMEDA enhancement is maximized. For input vibration signal vector
Figure BDA0002683224030000049
The objective function is described as:
Figure BDA0002683224030000045
wherein the content of the first and second substances,
Figure BDA0002683224030000046
is the filter vector, and T is the fault period.
Figure BDA0002683224030000047
A target vector for determining the pulse weight and position.
To find
Figure BDA0002683224030000048
Derivative of (a):
Figure BDA0002683224030000051
where N is the total number of samples and L is the filter length. The above formula is converted into a matrix form and further simplified to obtain:
Figure BDA0002683224030000052
wherein, X0In the form of a matrix of pulse signals. Due to the fact that
Figure BDA0002683224030000053
Suppose that
Figure BDA0002683224030000054
Substituting equation (7) with existence, we can get:
Figure BDA0002683224030000055
can be obtained from the formula (9)
Figure BDA0002683224030000056
I.e. the optimum filter coefficients sought.
Further, in step three, the method for analyzing a numerical simulation signal includes:
rolling bearings are used primarily for supporting rotating parts in mechanical equipment, whose vibration signals always contain a lot of information, including fault signatures and noise components. The key step of fault diagnosis is to extract the frequency domain vibration signal characteristics. There are many types of simulated signal models for bearing faults, most typically the model proposed by Randall. For the sake of no loss of generality, the outer ring fault numerical analog signal is represented as follows:
Figure BDA0002683224030000057
Figure BDA0002683224030000058
x(t)＝x1(t)+x2(t)+n(t) (18)
wherein the numerical simulation signal x (t) is a bearing fault simulation signal x1(t) and a modulation signal x2(t), and noise signal composition. At bearing fault simulation signal x1In (t), A0The value is 1.2 for the resonance intensity; f. ofmIs the modulation frequency (outer ring failure f)m＝0)；φA，φwAnd CAAre all arbitrary constants; b is the attenuation coefficient; the period T of the fault impact is 0.0071 s; tau isiA slight fluctuation of the ith impact with respect to the period T; f. ofnIs the resonant frequency of the system; x is the number of2(t) is a modulation signal, wherein f1＝15Hz，f250 Hz; n (t) is white Gaussian noise with a variance of 0.5. Frequency conversion frIs 20Hz, sampling frequency fs10000Hz, 10000N as sampling point, and f as outer ring fault frequency0Is 140 Hz.
And finally, carrying out spectrum analysis on the MOMEDA processed IMF1 vibration signal component.
Another object of the present invention is to provide a bearing fault signature recognition system comprising:
the oscillation mode component acquisition module is used for carrying out mode decomposition on the vibration signal by using SWD to acquire an oscillation mode component;
the weak signal characteristic enhancement module is used for carrying out noise reduction and frequency domain vibration signal characteristic extraction on the selected oscillation mode component by using MOMEDA (motion estimation and motion estimation), so that the weak signal characteristic is enhanced;
and the signal analysis module is used for identifying fault characteristics and analyzing the numerical simulation signal.
It is a further object of the invention to provide a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of:
carrying out mode decomposition on the vibration signal by using SWD to obtain an oscillation mode component;
performing noise reduction and frequency domain vibration signal feature extraction on the selected oscillation mode component by using MOMEDA (motion estimation and motion estimation), so as to realize the enhancement of weak signal features;
and identifying fault characteristics and analyzing the numerical simulation signals.
It is another object of the present invention to provide a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
carrying out mode decomposition on the vibration signal by using SWD to obtain an oscillation mode component;
performing noise reduction and frequency domain vibration signal feature extraction on the selected oscillation mode component by using MOMEDA (motion estimation and motion estimation), so as to realize the enhancement of weak signal features;
and identifying fault characteristics and analyzing the numerical simulation signals.
By combining all the technical schemes, the invention has the advantages and positive effects that:
according to the bearing fault feature identification method based on SWD and MOMEDA, provided by the invention, firstly, the SWD is used for carrying out mode decomposition on a vibration signal to obtain a useful component, then, the MOMEDA is used for realizing the enhancement of weak signal features on the selected component, and finally, the fault feature is identified. The effectiveness of the method in identifying the fault characteristics of the rolling bearing is verified by carrying out numerical simulation analysis and experiment table vibration signal analysis on the method.
The bearing fault feature identification method based on SWD and MOMEDA provided by the invention can effectively reduce interference items under the condition of strong noise and realize accurate identification of the bearing fault features.
Aiming at multi-component complex vibration signals, SWD decomposition is carried out on the multi-component complex vibration signals, and then characteristic extraction is carried out on the decomposed signals by using MOMEDA. The method aims to obtain useful signal components through SWD decomposition, improve the signal-to-noise ratio of signals and remove the interference of irrelevant components; meanwhile, filtering of signals and accurate extraction of impact characteristics are achieved by using MOMEDA. The method is applied to the analysis of the inner ring fault experimental data in the multi-component numerical simulation signal and mechanical fault public data set, and the result shows that the method provided by the invention has a good effect on the extraction of the fault characteristics of the rolling bearing.
The technical effect or experimental effect of comparison comprises the following steps:
compared with the bearing fault feature identification method based on EEMD, the bearing fault feature identification method based on SWD and MOMEDA provided by the invention has the advantages that the intrinsic mode function components obtained respectively cannot be prepared to identify the inner ring fault feature frequency from the bearing fault feature identification method diagram based on EEMD, and the one-frequency multiplication to the four-frequency multiplication of the inner ring fault frequency can be clearly identified from the bearing fault feature identification method diagram based on SWD and MOMEDA provided by the invention, so that the fault type can be determined to be the inner ring fault, which is consistent with the actual situation. The bearing fault feature identification method based on SWD and MOMEDA provided by the invention has more outstanding advantages and more reliable applicability.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments of the present application will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained from the drawings without creative efforts.
Fig. 1 is a flowchart of a bearing fault feature identification method based on SWD and MOMEDA according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of a bearing fault feature identification method based on SWD and MOMEDA according to an embodiment of the present invention.
FIG. 3 is a time domain diagram of the components of an emulated signal provided by an embodiment of the present invention;
in the figure: graph (a) is a time domain plot of the fault signal; graph (b) is a time domain graph of the modulated signal.
FIG. 4 is a diagram of the time domain and the frequency domain of a simulation signal containing a strong noise value according to an embodiment of the present invention;
in the figure: graph (a) is a time domain graph of a noisy numerical simulation signal; graph (b) is a spectrum of a noise-containing numerical simulation signal.
FIG. 5 is a graphical representation of the results after SWD decomposition provided by embodiments of the present invention.
Fig. 6 is a spectrum diagram of a mode component after SWD decomposition according to an embodiment of the present invention.
Fig. 7 is a schematic diagram of a result of noise reduction of MOMEDA according to an embodiment of the present invention.
Fig. 8 is a schematic diagram of a result of extracting a fault feature frequency according to an embodiment of the present invention.
Fig. 9 is a schematic view of a bearing failure experiment table provided by an embodiment of the invention.
FIG. 10 is a time domain and a frequency spectrum of a vibration signal collected according to an embodiment of the present invention;
in the figure: graph (a) is a time domain plot of the vibration signal; graph (b) is the vibration signal spectrum.
FIG. 11 is a diagram illustrating the results of envelope spectrum analysis and wavelet de-noising provided by an embodiment of the present invention;
in the figure: graph (a) is an envelope spectrum analysis; graph (b) is the result of wavelet de-noising.
FIG. 12 is a graphical representation of the results of EEMD decomposition provided by an embodiment of the present invention.
FIG. 13 is a graphical representation of the results of SWD decomposition provided by embodiments of the present invention.
Fig. 14 is a schematic diagram of a result of noise reduction of MOMEDA according to an embodiment of the present invention.
Fig. 15 is a schematic diagram of a spectrum analysis result of the 1 st order component subjected to MOMEDA denoising according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Aiming at the problems in the prior art, the invention provides a bearing fault feature identification method based on SWD and MOMEDA, and the invention is described in detail below with reference to the accompanying drawings.
As shown in fig. 1, the method for identifying a bearing fault feature based on SWD and MOMEDA provided in an embodiment of the present invention includes the following steps:
and S101, performing mode decomposition on the vibration signal by using SWD to obtain a useful component.
S102, denoising and frequency domain vibration signal feature extraction are carried out on the selected oscillation mode component by using MOMEDA, and weak signal feature enhancement is achieved.
And S103, identifying fault characteristics and analyzing the numerical simulation signals.
The present invention will be further described with reference to the following examples.
1. Summary of the invention
The rolling bearing is widely applied to rotary mechanical equipment, and has important significance in health monitoring and fault diagnosis of the rotary mechanical equipment. The invention provides a fault diagnosis method based on group decomposition (SWD) and multipoint optimal minimum deconvolution (MOMEDA). Firstly, the SWD is used for carrying out mode decomposition on the vibration signal to obtain a useful component, then MOMEDA is used for realizing the enhancement of weak signal characteristics on the selected component, and finally fault characteristics are identified. The effectiveness of the method in identifying the fault characteristics of the rolling bearing is verified by carrying out numerical simulation analysis and experiment table vibration signal analysis on the method.
2. Description of the theory
2.1 SWD Algorithm
SWD, an intelligent mode decomposition algorithm, decomposes a multi-component signal into a sum of a plurality of Oscillation Components (OC) by using an iterative group filter (SwF), and the SWD process is as follows:
(1) determination of initial center frequency from signal power spectrum peak
Figure BDA0002683224030000091
Figure BDA0002683224030000092
Wherein S isy(w) represents the Welch Power Spectrum of the signal, q represents the qth frequency w as the center frequency during SWD, PhIs a threshold value. By
Figure BDA0002683224030000101
SwF parameters M and:
Figure BDA0002683224030000102
wherein odd (·) represents a rounding operation.
(2) SwF filtering is carried out on the signal y (t) to obtain an output signal u (t), and the variance of the input and output signals is calculated as follows:
Figure BDA0002683224030000103
if the variance D is larger than the variance threshold DhRepeating SwF filtering with u (t) as input until D < DhWhen the output signal is recorded as yi(t)。
(3) Updating input signals
Figure BDA0002683224030000104
Wherein the content of the first and second substances,
Figure BDA0002683224030000105
for the cross-correlation function, τ denotes the time delay.
(4) Repeating the above steps (1) - (3) until S using the updated input signaly(w)≤PhThe input signal at this time is the residual r (t).
(5) Calculating a plurality of oscillation mode components according to the following formula:
Figure BDA0002683224030000106
in the above formula, omegad＝{w:w＝wdAnd k is the number of oscillation mode components.
2.2 MOMEDA Algorithm
The purpose of the MOMEDA method is to find an optimal FIR filter f (l) coefficient to maximize the objective function of the output signal after MOMEDA enhancement. For input vibration signal vector
Figure BDA0002683224030000107
The objective function is described as:
Figure BDA0002683224030000108
in the above formula, the first and second carbon atoms are,
Figure BDA0002683224030000111
is the filter vector, and T is the fault period.
Figure BDA0002683224030000112
A target vector for determining the pulse weight and position.
To find
Figure BDA0002683224030000113
Derivative of (a):
Figure BDA0002683224030000114
where N is the total number of samples and L is the filter length. The above formula is converted into a matrix form and further simplified to obtain:
Figure BDA0002683224030000115
in the above formula, X0In the form of a matrix of pulse signals. Due to the fact that
Figure BDA0002683224030000116
Suppose that
Figure BDA0002683224030000117
Substituting equation (6) with existence, we can get:
Figure BDA0002683224030000118
can be obtained from the above formula
Figure BDA0002683224030000119
I.e. the optimum filter coefficients sought.
In the invention, firstly, the vibration signal is subjected to SWD decomposition, and then MOMEDA noise reduction and feature extraction are carried out on the decomposed oscillation mode component, wherein the algorithm flow is shown in figure 2.
3. The invention is further described below in connection with specific simulations.
3.1 numerical simulation Signal analysis
Rolling bearings are used primarily for supporting rotating parts in mechanical devices, the vibration signals of which always contain a lot of information, such as fault characteristics and noise components. The key step of fault diagnosis is to extract the frequency domain vibration signal characteristics. There are many types of simulated signal models for bearing faults, most typically the model proposed by Randall. For the sake of no loss of generality, the outer ring fault numerical analog signal is represented as follows:
Figure BDA00026832240300001110
Figure BDA0002683224030000121
x(t)＝x1(t)+x2(t)+n(t) (18)
wherein the numerical simulation signal x (t) is a bearing fault simulation signal x1(t) and a modulation signal x2(t), and noise signal composition. At bearing fault simulation signal x1In (t), A0The value is 1.2 for the resonance intensity; f. ofmIs the modulation frequency (outer ring failure f)m＝0)；φA，φwAnd CAAre all arbitrary constants; b is the attenuation coefficient; the period T of the fault impact is 0.0071 s; tau isiA slight fluctuation of the ith impact with respect to the period T; f. ofnIs the resonant frequency of the system; x is the number of2(t) is a modulation signal, wherein f1＝15Hz，f250 Hz; n (t) is white Gaussian noiseThe difference was 0.5. Frequency conversion frIs 20Hz, sampling frequency fs10000Hz, 10000N as sampling point, and f as outer ring fault frequency0Is 140 Hz.
x1(t) and a modulation signal x2(t) in time domain, as shown in fig. 3(a) and 3(b), respectively, where the fault simulation signal has very distinct impact characteristics. In order to verify the effect of SWD on multi-component complex signal decomposition, strong background noise was added, and graphs of time domain and frequency domain of noisy numerical simulation signals are shown in fig. 4. Due to interference of multi-component signals and strong noise, outer-ring fault characteristics and modulation characteristics cannot be correctly identified through the spectrum analysis result shown in fig. 4 (b).
Next, modal decomposition of the multi-component simulation using SWD was performed, and a total of 6 modal components were obtained, the results of which are shown in fig. 5. Then, an FFT operation is performed for each of the decomposed mode components, and the result is shown in fig. 6. In fig. 6, the spectrum of IMF3 corresponds to the modulated signal, from which we can find the center frequency f2And modulation phenomenon (f)2+f1And f2-f1). Through calculation, the correlation coefficient of the first component IMF1 and the noise-containing numerical simulation signal x is found to be the largest, and the signal quality enhancement and noise reduction are carried out on the first component IMF1 by utilizing the MOMEDA method, as shown in FIG. 7, the purpose of the method is to extract the impact signal characteristic and identify the outer ring fault characteristic frequency fo。
Finally, the MOMEDA processed IMF1 vibration signal component is subjected to frequency spectrum analysis, the result is shown in FIG. 8, and the characteristic frequency f of the outer ring fault can be clearly identified from the graph0And frequency multiplication 2f thereof0、3f0. Therefore, the method provided by the invention has obvious effects on noise reduction of multi-component signals and outer ring fault feature extraction.
3.2 analysis of Experimental data
The experimental data are from a public data set of rolling bearing faults of the university of transport in west safety. The whole fault bearing experiment table consists of an alternating current induction motor, a motor speed regulator (frequency converter), two-end supporting bearings and a hydraulic load. The fault rolling bearing is of the type LDK UER204, and the number of the rolling bodies is8, the diameter of the outer ring is 39.8mm, the diameter of the inner ring is 29.3mm, and the contact angle is 0 degree. In order to obtain vibration data of the tested bearing, two PCB acceleration sensors are placed on a bearing seat at an angle of 90 degrees with each other, and the model of the sensor is 352C 33. The sampling frequency f of the vibration signal in this experiments25.6kHz, and a frequency of fr35 HZ. The type of failure of the bearing is an inner race failure. The structural parameters and the rotation frequency of the bearing are calculated according to the theory, and the fault frequency f of the inner ring can be obtainedi168 Hz. The bearing failure test bench is shown in fig. 9.
First, time domain analysis and FFT calculation are performed on the measured vibration signal, and the result is shown in fig. 10. In fig. 10(a) it can be seen that the measured signal contains significant impact characteristics and noise signals. The result of the spectrum analysis is shown in FIG. 10(b), and the frequency conversion f can be identifiedrAnd a frequency doubling of 2frIn addition, two distinct peak frequencies can be found, which correspond to 232Hz and 442Hz, respectively, but which are independent of the fault signature frequency. Envelope spectrum analysis and wavelet noise reduction are common signal analysis means for bearing fault diagnosis. The result of the envelope spectrum analysis is shown in fig. 11(a), from which the failure characteristic frequency cannot be identified. The result of the wavelet analysis is shown in fig. 11(b), from which the frequency conversion and frequency doubling, and the peak frequency 232Hz independent of the fault characteristics can be found. Obviously, these methods have certain limitations for feature extraction of the data of the present invention.
Next, modal decomposition is performed on the bearing vibration signal by using a common EEMD method, 12 IMF components are obtained in total, the results of the frequency spectrum analysis of the first 8 components are shown in fig. 12, and the characteristic frequency of the inner ring fault cannot be identified in preparation from the graph. Finally, the vibration signal of the actually measured faulty bearing is subjected to pattern decomposition by SWD, and as a result, as shown in fig. 13, a total of 6 components, i.e., IMF1 to IMF6, are obtained. Since the first component IMF1 has the greatest correlation with the original signal, the first component is selected and subjected to signal quality enhancement and impact feature extraction using MOMEDA, the result of which is shown in fig. 14. From fig. 14, the impact characteristics are found to be obvious, and then FFT calculation is performed thereon, and the result is shown in fig. 15. From which the inner circle can be clearly identifiedFrom one to four times the barrier frequency, i.e. fi～4fiThus, it can be determined that the failure type is an inner ring failure, which is in accordance with the actual situation. The effectiveness of the method is further illustrated by experimental data analysis.
4. Aiming at the defects of the weak fault feature identification method for the rotating machinery, the invention provides a rolling bearing weak fault feature identification method based on SWD and MOMEDA. Aiming at multi-component complex vibration signals, SWD decomposition is carried out on the multi-component complex vibration signals, and then characteristic extraction is carried out on the decomposed signals by using MOMEDA. The method aims to obtain useful signal components through SWD decomposition, improve the signal-to-noise ratio of signals and remove the interference of irrelevant components; meanwhile, filtering of signals and accurate extraction of impact characteristics are achieved by using MOMEDA. The method is applied to the analysis of the inner ring fault experimental data in the multi-component numerical simulation signal and mechanical fault public data set, and the result shows that the method provided by the invention has a good effect on the extraction of the fault characteristics of the rolling bearing.
The above description is only for the purpose of illustrating the present invention and the appended claims are not to be construed as limiting the scope of the invention, which is intended to cover all modifications, equivalents and improvements that are within the spirit and scope of the invention as defined by the appended claims.

Claims (7)

1. A bearing fault feature identification method is characterized by comprising the following steps:
carrying out mode decomposition on the vibration signal by using SWD to obtain an oscillation mode component;
performing noise reduction and frequency domain vibration signal feature extraction on the selected oscillation mode component by using MOMEDA (motion estimation and motion estimation), so as to realize the enhancement of weak signal features;
and analyzing the numerical simulation signal and the actual measurement fault signal to realize the identification of the fault characteristics.
2. The bearing fault signature identification method of claim 1, wherein the decomposing of the multi-component signal into a sum of the plurality of oscillating components by the iterative group filter is characterized by an SWD process of:
(1) determination of initial center frequency from signal power spectrum peak
Figure FDA0002683224020000011
Figure FDA0002683224020000012
Wherein S isy(w) denotes the Welch power spectrum of the signal, q denotes the qth frequency w as the center frequency during SWD, PhIs a threshold value; by
Figure FDA0002683224020000013
SwF parameters M and:
Figure FDA0002683224020000014
wherein odd (·) represents a rounding operation;
(2) SwF filtering is carried out on the signal y (t) to obtain an output signal u (t), and the variance of the input and output signals is calculated as follows:
Figure FDA0002683224020000015
if the variance D is larger than the variance threshold DhRepeating SwF filtering with u (t) as input until D < DhWhen the output signal is recorded as yi(t)；
(3) Updating the input signal:
Figure FDA0002683224020000016
wherein the content of the first and second substances,
Figure FDA0002683224020000017
for the cross-correlation function, τ represents the time delay;
(4) repeating the above steps (1) - (3) until S using the updated input signaly(w)≤PhThe input signal at this time is the residual r (t);
(5) calculating a plurality of oscillation mode components according to the following formula:
Figure FDA0002683224020000021
wherein omegad＝{w:w＝wdAnd k is the number of oscillation mode components.
3. The method for identifying a fault in a bearing according to claim 1, wherein the MOMEDA method is aimed to find an optimal FIR filter f (l) coefficient to maximize an objective function of an output signal after MOMEDA enhancement; for input vibration signal vector
Figure FDA00026832240200000211
The objective function is described as:
Figure FDA0002683224020000022
wherein the content of the first and second substances,
Figure FDA0002683224020000023
is the filter vector, T is the fault period;
Figure FDA0002683224020000024
a target vector for determining the pulse weight and position;
to find
Figure FDA0002683224020000025
Derivative of (a):
Figure FDA0002683224020000026
wherein, N is the total number of sampling points, and L is the length of the filter; converting the formula (7) into a matrix form, and further simplifying the formula:
Figure FDA0002683224020000027
wherein, X0In the form of a matrix of pulse signals; due to the fact that
Figure FDA0002683224020000028
Suppose that
Figure FDA0002683224020000029
If exists, substituting into formula (7), then:
Figure FDA00026832240200000210
is obtained from the formula (9)
Figure FDA00026832240200000212
I.e. the optimum filter coefficients sought.
4. The bearing fault signature identification method of claim 1, wherein the method of numerical simulation signal analysis comprises:
the rolling bearing is mainly used for supporting a rotating part in mechanical equipment, and a vibration signal of the rolling bearing always contains much information including fault characteristics and noise components; the key step of fault diagnosis is to extract the frequency domain vibration signal characteristics; there are many types of simulated signal models for bearing faults, most typically the model proposed by Randall; the outer ring fault numerical analog signal is represented as follows:
Figure FDA0002683224020000031
Figure FDA0002683224020000032
x(t)＝x1(t)+x2(t)+n(t) (12)
the numerical simulation signal x (t) is composed of a bearing fault simulation signal x1(t) and a modulation signal x2(t), and a noise signal composition; at bearing fault simulation signal x1In (t), A0The value is 1.2 for the resonance intensity; f. ofmIs modulation frequency, outer ring fault fm＝0；φA，φwAnd CAAre all arbitrary constants; b is the attenuation coefficient; the period T of the fault impact is 0.0071 s; tau isiA slight fluctuation of the ith impact with respect to the period T; f. ofnIs the resonant frequency of the system; x is the number of2(t) is a modulation signal, wherein f1＝15Hz，f250 Hz; n (t) is white Gaussian noise with a variance of 0.5; frequency conversion frIs 20Hz, sampling frequency fs10000Hz, 10000N as sampling point, and f as outer ring fault frequency0Is 140 Hz;
and finally, carrying out spectrum analysis on the MOMEDA processed IMF1 vibration signal component.
5. A bearing fault feature recognition system based on SWD and MOMEDA methods according to any one of claims 1 to 4, characterized in that the bearing fault feature recognition system comprises:
the oscillation mode component acquisition module is used for carrying out mode decomposition on the vibration signal by using SWD to acquire an oscillation mode component;
the weak signal characteristic enhancement module is used for carrying out noise reduction and frequency domain vibration signal characteristic extraction on the selected oscillation mode component by using MOMEDA (motion estimation and motion estimation), so that the weak signal characteristic is enhanced;
and the signal analysis module is used for identifying fault characteristics and analyzing the numerical simulation signal.
6. A computer device, characterized in that the computer device comprises a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to carry out the steps of:
carrying out mode decomposition on the vibration signal by using SWD to obtain an oscillation mode component;
performing noise reduction and frequency domain vibration signal feature extraction on the selected oscillation mode component by using MOMEDA (motion estimation and motion estimation), so as to realize the enhancement of weak signal features;
and identifying fault characteristics and analyzing the numerical simulation signals.
7. A computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
carrying out mode decomposition on the vibration signal by using SWD to obtain an oscillation mode component;
performing noise reduction and frequency domain vibration signal feature extraction on the selected oscillation mode component by using MOMEDA (motion estimation and motion estimation), so as to realize the enhancement of weak signal features;
and analyzing the numerical simulation and the actual measurement signal to complete the identification of the fault characteristics.
CN202010968520.7A 2020-09-15 2020-09-15 Bearing fault feature identification method and system Pending CN112098093A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010968520.7A CN112098093A (en) 2020-09-15 2020-09-15 Bearing fault feature identification method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010968520.7A CN112098093A (en) 2020-09-15 2020-09-15 Bearing fault feature identification method and system

Publications (1)

Publication Number Publication Date
CN112098093A true CN112098093A (en) 2020-12-18

Family

ID=73760280

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010968520.7A Pending CN112098093A (en) 2020-09-15 2020-09-15 Bearing fault feature identification method and system

Country Status (1)

Country Link
CN (1) CN112098093A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112836583A (en) * 2021-01-05 2021-05-25 沈阳工业大学 Wind turbine fault diagnosis method
CN113029566A (en) * 2021-02-02 2021-06-25 王晓东 Rolling bearing fault acoustic emission feature extraction method based on improved EEMD and MED
CN113933035A (en) * 2021-09-30 2022-01-14 中国船舶重工集团公司第七一九研究所 Rotary mechanical equipment fault diagnosis method and system based on correlation analysis

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108168886A (en) * 2017-12-22 2018-06-15 合肥工业大学 Rolling Bearing Fault Character extracts and method for monitoring operation states
CN108507783A (en) * 2018-03-14 2018-09-07 湖南大学 A kind of combined failure of rotating machinery diagnostic method decomposed based on group
CN111413095A (en) * 2020-04-13 2020-07-14 西安电子科技大学 Instantaneous angular velocity-based distributed fault diagnosis and analysis method for planetary bearing
CN111623982A (en) * 2020-06-15 2020-09-04 大连理工大学 Planetary gearbox early fault diagnosis method based on APEWT and IMOMEDA

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108168886A (en) * 2017-12-22 2018-06-15 合肥工业大学 Rolling Bearing Fault Character extracts and method for monitoring operation states
CN108507783A (en) * 2018-03-14 2018-09-07 湖南大学 A kind of combined failure of rotating machinery diagnostic method decomposed based on group
CN111413095A (en) * 2020-04-13 2020-07-14 西安电子科技大学 Instantaneous angular velocity-based distributed fault diagnosis and analysis method for planetary bearing
CN111623982A (en) * 2020-06-15 2020-09-04 大连理工大学 Planetary gearbox early fault diagnosis method based on APEWT and IMOMEDA

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
张志强 等: "基于自适应VMD和MOMEDA的变转速轴承早期故障诊断", 《军事交通学院学报》 *
李娟 等: "基于SWD-AVDIF的齿轮箱复合故障诊断方法", 《噪声与振动控制》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112836583A (en) * 2021-01-05 2021-05-25 沈阳工业大学 Wind turbine fault diagnosis method
CN112836583B (en) * 2021-01-05 2024-04-12 沈阳工业大学 Wind turbine fault diagnosis method
CN113029566A (en) * 2021-02-02 2021-06-25 王晓东 Rolling bearing fault acoustic emission feature extraction method based on improved EEMD and MED
CN113933035A (en) * 2021-09-30 2022-01-14 中国船舶重工集团公司第七一九研究所 Rotary mechanical equipment fault diagnosis method and system based on correlation analysis
CN113933035B (en) * 2021-09-30 2023-08-29 中国船舶重工集团公司第七一九研究所 Rotary machinery equipment fault diagnosis method and system based on correlation analysis

Similar Documents

Publication Publication Date Title
Hemmati et al. Roller bearing acoustic signature extraction by wavelet packet transform, applications in fault detection and size estimation
Golafshan et al. SVD and Hankel matrix based de-noising approach for ball bearing fault detection and its assessment using artificial faults
Su et al. Rolling element bearing faults diagnosis based on optimal Morlet wavelet filter and autocorrelation enhancement
CN112098093A (en) Bearing fault feature identification method and system
Cong et al. Short-time matrix series based singular value decomposition for rolling bearing fault diagnosis
Zhang et al. Time-varying singular value decomposition for periodic transient identification in bearing fault diagnosis
Yang Interpretation of mechanical signals using an improved Hilbert–Huang transform
Yan et al. Harmonic wavelet-based data filtering for enhanced machine defect identification
Osman et al. An enhanced Hilbert–Huang transform technique for bearing condition monitoring
Hao et al. Morphological undecimated wavelet decomposition for fault diagnostics of rolling element bearings
He et al. A joint adaptive wavelet filter and morphological signal processing method for weak mechanical impulse extraction
JP6133112B2 (en) Rolling bearing diagnostic device and rolling bearing diagnostic method
Jiang et al. A novel method for adaptive multiresonance bands detection based on VMD and using MTEO to enhance rolling element bearing fault diagnosis
Klausen et al. Cross-correlation of whitened vibration signals for low-speed bearing diagnostics
He et al. Feature extraction of gear and bearing compound faults based on vibration signal sparse decomposition
Williams et al. Helicopter transmission fault detection via time-frequency, scale and spectral methods
Zheng et al. Faults diagnosis of rolling bearings based on shift invariant K-singular value decomposition with sensitive atom nonlocal means enhancement
CN112945546A (en) Accurate diagnosis method for complex fault of gear box
Chi et al. Bearing fault diagnosis through stochastic resonance by full-wave signal construction with half-cycle delay
Fan et al. A joint wavelet lifting and independent component analysis approach to fault detection of rolling element bearings
CN106980722B (en) Method for detecting and removing harmonic component in impulse response
Babu et al. Fault diagnosis in bevel gearbox using coiflet wavelet and fault classification based on ANN including DNN
Gong et al. Fault detection for rolling element bearing based on repeated single-scale morphology and simplified sensitive factor algorithm
Zhang et al. A joint kurtosis-based adaptive bandstop filtering and iterative autocorrelation approach to bearing fault detection
Singh et al. Condition monitoring of wind turbine gearbox using electrical signatures

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20201218

RJ01 Rejection of invention patent application after publication