CN111024398A - Deconvolution method for maximum correlation kurtosis without period - Google Patents

Deconvolution method for maximum correlation kurtosis without period Download PDF

Info

Publication number
CN111024398A
CN111024398A CN201911352882.7A CN201911352882A CN111024398A CN 111024398 A CN111024398 A CN 111024398A CN 201911352882 A CN201911352882 A CN 201911352882A CN 111024398 A CN111024398 A CN 111024398A
Authority
CN
China
Prior art keywords
signal
vibration signal
deconvolution
iteration
period
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.)
Granted
Application number
CN201911352882.7A
Other languages
Chinese (zh)
Other versions
CN111024398B (en
Inventor
赵明
苗永浩
李博闻
孔子豪
苟超
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.)
Xian Jiaotong University
Original Assignee
Xian Jiaotong University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xian Jiaotong University filed Critical Xian Jiaotong University
Priority to CN201911352882.7A priority Critical patent/CN111024398B/en
Publication of CN111024398A publication Critical patent/CN111024398A/en
Application granted granted Critical
Publication of CN111024398B publication Critical patent/CN111024398B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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

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

Firstly, carrying out truncation and mean value removing processing on the acquired vibration signal without a periodic maximum correlation kurtosis deconvolution method; then, performing Hilbert transform on the vibration signal to obtain an analytic signal of the vibration signal, performing deconvolution operation by selecting a maximum value point of a part after a first zero-crossing point in an autocorrelation spectrum of the vibration signal as an iteration period, and then continuously calculating the maximum value point of the part after the first zero-crossing point in the autocorrelation spectrum of the vibration signal by using the signal after each iteration filtering as the iteration period to update a filter so as to obtain an optimal filter coefficient; finally, envelope analysis is carried out on the filtered signals, and fault characteristic frequency can be extracted from an envelope spectrum.

Description

Deconvolution method for maximum correlation kurtosis without period
Technical Field
The invention relates to the technical field of fault diagnosis of mechanical equipment, in particular to a maximum correlation kurtosis deconvolution method without a period.
Background
Vibration analysis is one of the most effective ways to diagnose the fault of the mechanical equipment at present stage, and the state degradation of the mechanical equipment is often represented as the change or the abnormity of vibration information. At present, signal processing methods based on vibration information, such as a time domain method, a frequency domain method and a time-frequency domain method, are successfully applied to bearing fault diagnosis and have good effects. However, many challenges still face in the field of fault diagnosis of rolling bearings, and extraction of bearing faults still has many difficulties. 1. The lengthy and complicated transfer path between the test sensor and the fault source can seriously affect the transfer function, thereby reducing the amplitude and prolonging the time of the impact signal, so that the pulse caused by the fault is easily covered by the noise. 2. Random fluctuations of the rollers in the bearing can cause further blurring of the spectral envelope spectrum of the otherwise quasi-periodic fault impact. 3. The effects from non-periodic noise and periodic disturbances in the mechanical system add more challenges to extracting bearing fault impacts.
At present, the deconvolution method is considered to be one of the most effective methods for eliminating the influence of the transmission path, and the characteristic that the deconvolution method can adaptively design a filter is critical for accurately and completely filtering fault information. In the field of mechanical fault diagnosis, two types of deconvolution methods are most commonly used, namely a minimum entropy deconvolution method and a maximum correlation kurtosis deconvolution method, but the minimum entropy deconvolution method is easily affected by random impact interference in signals, while the maximum correlation kurtosis deconvolution method can overcome the defect of random impact interference, but the method also needs an accurate period as priori knowledge. In the failure diagnosis of the rolling bearing, it is difficult to accurately determine the failed bearing in advance. Firstly, the rotation of mechanical equipment is difficult to keep constant completely, so that the calculation of a fault period caused by speed fluctuation is inaccurate and difficult to avoid; secondly, the number of parts in the equipment is large, the number of rolling bearings is usually large, and it is difficult to determine the failure source in advance, so it is not practical to calculate the period of the failed bearing in advance.
Disclosure of Invention
In order to overcome the above-mentioned existing disadvantages, the present invention aims to provide a deconvolution method without periodic maximum correlation kurtosis, and without prior knowledge, to realize accurate fault diagnosis.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a method of deconvolution of maximum correlation kurtosis without periodicity, comprising the steps of:
the method comprises the following steps: adsorbing a vibration acceleration sensor on a bearing seat of a rolling bearing to be tested, carrying out high-frequency sampling, truncation and mean value removing processing on a vibration signal, and recording the processed vibration signal as x (n);
step two: subjecting the vibration signal x (n) to Hilbert transform to obtain its analytic signal
Figure BDA0002335093420000021
Step three: the number of iterations k of deconvolution, the filter length L, and the initialization filter coefficient f are set in advance [0,1,0,0](ii) a Then calculating to obtain analytic signal
Figure BDA0002335093420000022
The maximum point of all the components after the first zero-crossing point in the autocorrelation spectrum is selected as the number of sampling points T in the iteration period by using the modes of the equations (1) to (2)sAnd substituting this information into the following iteration steps for selection of optimal filter coefficients;
Figure BDA0002335093420000031
Figure BDA0002335093420000032
wherein N is an analytic signal
Figure BDA0002335093420000033
Is equal to (1), τ is the offset,
Figure BDA0002335093420000034
is the zero position of the autocorrelation function;
step four: performing maximal correlation kurtosis deconvolution iterative filtering on the vibration signal x (n), wherein the filter coefficients are implemented according to equation (3),
Figure BDA0002335093420000035
wherein M is a shift order of the correlation kurtosis, the shift order is set to 1, y is a filtered signal after deconvolution filtering of the maximum correlation kurtosis of the vibration signal x (n), T is a transposed symbol, | | | is an Euclidean norm operation,
Figure BDA0002335093420000036
Figure BDA0002335093420000037
step five: obtaining a first filtered signal y1And performing Hilbert transform to obtain its analytic signal
Figure BDA0002335093420000038
Selecting the maximum value point of all the components after the first zero-crossing point in the autocorrelation spectrum as the number of sampling points in a new iteration period according to the modes of the equations (4) to (5)
Figure BDA0002335093420000039
Figure BDA0002335093420000041
Figure BDA0002335093420000042
Step six: sampling point number T according to formula (3) by using new iteration periods 1Updating the filter coefficient to carry out the next filtering until the iteration number reaches the specified iteration number k times and then ending, and selecting the final filtering signal ykIs the best filtered signal;
step seven: to filterWave signal ykAnd carrying out envelope analysis to obtain an envelope spectrum, analyzing the envelope spectrum, further extracting fault characteristic frequency, and finally identifying the fault of the rolling bearing.
Compared with the prior art, the invention has the following beneficial effects:
a) compared with the traditional maximum correlation kurtosis deconvolution method, the method does not need any priori knowledge or system fault characteristic frequency prediction, and has robustness.
b) Compared with the traditional maximum correlation kurtosis deconvolution method, the method is simple to operate, does not need resampling operation, and has few input parameters.
c) The present invention overcomes many of the drawbacks of conventional deconvolution methods, such as sensitivity to random shock components.
Drawings
FIG. 1 is a schematic view of a bearing test bench according to an embodiment of the present invention.
FIG. 2 is a flow chart of the method of the present invention.
FIG. 3 shows an exemplary vibration signal x (n).
FIG. 4 is an analytic signal of the vibration signal of the embodiment
Figure BDA0002335093420000043
FIG. 5 shows the signal after the maximum correlation kurtosis deconvolution process.
Fig. 6 is an envelope spectrum of a signal processed by the maximum correlation kurtosis deconvolution method.
Fig. 7 is a signal processed by a maximum correlation kurtosis deconvolution method without periodicity.
Fig. 8 is an envelope spectrum of a signal processed by a maximum correlation kurtosis deconvolution method without periodicity.
Detailed Description
The invention is described in detail below with reference to the figures and examples.
In the embodiment, a locomotive bearing test bed is adopted, as shown in fig. 1, the test bed is composed of a hydraulic motor, a driving wheel, a bearing and a locomotive wheel in an equivalent mode, the hydraulic motor drives the driving wheel to move so as to drive an outer ring of the bearing to move, an inner ring of the bearing is fixed on an axle of the locomotive wheel pair, an acceleration sensor is fixed at one end of the bearing, and vibration signals of the bearing are measured.
The wheel bearing in the test bed is diagnosed by a maximal correlation kurtosis deconvolution method without a period, experimental data are analyzed, and the experimental data are compared with a traditional maximal correlation kurtosis deconvolution method.
As shown in fig. 2, a method for deconvolving a maximum correlation kurtosis without a period includes the steps of:
the method comprises the following steps: adsorbing a vibration acceleration sensor on a bearing seat of a rolling bearing to be tested, and carrying out high-frequency sampling, truncation and mean value removing processing on a vibration signal, wherein the sampling frequency is 76.8k Hz, the duration is 1s, and the processed vibration signal x (n) is shown in FIG. 3;
step two: subjecting the vibration signal x (n) to Hilbert transform to obtain its analytic signal
Figure BDA0002335093420000051
As shown in fig. 4;
step three: the number of iterations k of deconvolution is set in advance to 30, the filter length L is set to 100, and the initialization filter coefficient f is set to [0,1,0,0](ii) a Then calculating to obtain analytic signal
Figure BDA0002335093420000052
The maximum point of all the components after the first zero-crossing point in the autocorrelation spectrum is selected as the number of sampling points T in the iteration period by using the modes of the equations (1) to (2)sAnd substituting this information into the following iteration steps for selection of optimal filter coefficients;
Figure BDA0002335093420000061
Figure BDA0002335093420000062
wherein N is an analytic signal
Figure BDA0002335093420000063
Is equal to (1), τ is the offset,
Figure BDA0002335093420000064
is the zero position of the autocorrelation function;
step four: performing maximal correlation kurtosis deconvolution iterative filtering on the vibration signal x (n), wherein the filter coefficients are implemented according to equation (3),
Figure BDA0002335093420000065
wherein M is a shift order of the correlation kurtosis, the shift order is set to 1, y is a filtered signal after deconvolution filtering of the maximum correlation kurtosis of the vibration signal x (n), T is a transposed symbol, | | | is an Euclidean norm operation,
Figure BDA0002335093420000066
Figure BDA0002335093420000067
step five: obtaining a first filtered signal y1And performing Hilbert transform to obtain its analytic signal
Figure BDA0002335093420000071
Selecting the maximum value point of all the components after the first zero-crossing point in the autocorrelation spectrum as the number of sampling points in a new iteration period according to the modes of the equations (4) to (5)
Figure BDA0002335093420000072
Figure BDA0002335093420000073
Figure BDA0002335093420000074
Step six: sampling point number T according to formula (3) by using new iteration periods 1Updating the filter coefficient to perform the next filtering until the number of iterations reaches the specified number of iterations 30, and selecting the final filtering signal y30Is the best filtered signal;
step seven: for the filtered signal y30And carrying out envelope analysis to obtain an envelope spectrum, analyzing the envelope spectrum, further extracting fault characteristic frequency, and finally identifying the fault of the rolling bearing.
As shown in fig. 5, fig. 5 is a signal filtered by the maximum correlation kurtosis deconvolution method, and it can be seen that the filtered signal is basically a large interference impact signal, and no obvious periodic bearing fault impact signal is found; as shown in fig. 6, fig. 6 is an envelope spectrum signal corresponding to fig. 5, from which it is difficult to find the characteristic frequency of the bearing failure. FIG. 7 is a filtered signal, as shown in FIG. 7, after application of the method of the present invention to the same signal, from which a significant periodic impulse component is seen; as shown in fig. 8, fig. 8 is the envelope spectrum signal corresponding to fig. 7, in which the inner ring fault characteristic frequencies and their harmonic frequency components are quite obvious. Therefore, the method has obvious advantages compared with the traditional maximum correlation kurtosis deconvolution method.

Claims (1)

1. A method of deconvolving a maximum correlation kurtosis without a period, comprising the steps of:
the method comprises the following steps: adsorbing a vibration acceleration sensor on a bearing seat of a rolling bearing to be tested, carrying out high-frequency sampling, truncation and mean value removing processing on a vibration signal, and recording the processed vibration signal as x (n);
step two: subjecting the vibration signal x (n) to Hilbert transform to obtain its analytic signal
Figure FDA0002335093410000011
Step three: the number of deconvolution iterations k, the filter length L and the initial filter coefficient f are preset[0,1,0,0,...,0](ii) a Then calculating to obtain analytic signal
Figure FDA0002335093410000012
The maximum point of all the components after the first zero-crossing point in the autocorrelation spectrum is selected as the number of sampling points T in the iteration period by using the modes of the equations (1) to (2)sAnd substituting this information into the following iteration steps for selection of optimal filter coefficients;
Figure FDA0002335093410000013
Figure FDA0002335093410000014
wherein N is an analytic signal
Figure FDA0002335093410000015
Is equal to (1), τ is the offset,
Figure FDA0002335093410000016
is the zero position of the autocorrelation function;
step four: performing maximal correlation kurtosis deconvolution iterative filtering on the vibration signal x (n), wherein the filter coefficients are implemented according to equation (3),
Figure FDA0002335093410000017
wherein M is a shift order of the correlation kurtosis, the shift order is set to 1, y is a filtered signal after deconvolution filtering of the maximum correlation kurtosis of the vibration signal x (n), T is a transposed symbol, | | | is an Euclidean norm operation,
Figure FDA0002335093410000021
Figure FDA0002335093410000022
step five: obtaining a first filtered signal y1And performing Hilbert transform to obtain its analytic signal
Figure FDA0002335093410000023
Selecting the maximum value point of all the components after the first zero-crossing point in the autocorrelation spectrum as the number of sampling points in a new iteration period according to the modes of the equations (4) to (5)
Figure FDA0002335093410000024
Figure FDA0002335093410000025
Figure FDA0002335093410000026
Step six: sampling point number T according to formula (3) by using new iteration periods 1Updating the filter coefficient to carry out the next filtering until the iteration number reaches the specified iteration number k times and then ending, and selecting the final filtering signal ykIs the best filtered signal;
step seven: for the filtered signal ykAnd carrying out envelope analysis to obtain an envelope spectrum, analyzing the envelope spectrum, further extracting fault characteristic frequency, and finally identifying the fault of the rolling bearing.
CN201911352882.7A 2019-12-25 2019-12-25 Deconvolution method for maximum correlation kurtosis without period Active CN111024398B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911352882.7A CN111024398B (en) 2019-12-25 2019-12-25 Deconvolution method for maximum correlation kurtosis without period

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911352882.7A CN111024398B (en) 2019-12-25 2019-12-25 Deconvolution method for maximum correlation kurtosis without period

Publications (2)

Publication Number Publication Date
CN111024398A true CN111024398A (en) 2020-04-17
CN111024398B CN111024398B (en) 2021-01-15

Family

ID=70213489

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911352882.7A Active CN111024398B (en) 2019-12-25 2019-12-25 Deconvolution method for maximum correlation kurtosis without period

Country Status (1)

Country Link
CN (1) CN111024398B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113484019A (en) * 2021-07-26 2021-10-08 西南交通大学 Bearing detection method and device and computer readable storage medium
CN113702030A (en) * 2021-07-14 2021-11-26 西南交通大学 Mechanical fault diagnosis method and system based on maximum average window kurtosis blind deconvolution
CN114462451A (en) * 2022-01-24 2022-05-10 北京航空航天大学 Characteristic modal decomposition method for mechanical fault diagnosis
CN117932302A (en) * 2024-01-22 2024-04-26 浙江大学 Fault signal extraction method, device, equipment and medium for mechanical device

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6161076A (en) * 1997-11-14 2000-12-12 Baker Hughes Incorporated Seismic data acquisition and processing using non-linear distortion in a vibratory output signal
CN105510032A (en) * 2015-12-11 2016-04-20 西安交通大学 Deconvolution method based on harmonic to noise ratio guidance
CN106772646A (en) * 2016-12-19 2017-05-31 吉林大学 A kind of ground nuclear magnetic resonance method for extracting signal
CN107560718A (en) * 2017-07-25 2018-01-09 苏州微著设备诊断技术有限公司 One kind is based on the maximized deconvolution method of auto-correlation kurtosis
US10102444B2 (en) * 2016-11-22 2018-10-16 Lunit Inc. Object recognition method and apparatus based on weakly supervised learning

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6161076A (en) * 1997-11-14 2000-12-12 Baker Hughes Incorporated Seismic data acquisition and processing using non-linear distortion in a vibratory output signal
CN105510032A (en) * 2015-12-11 2016-04-20 西安交通大学 Deconvolution method based on harmonic to noise ratio guidance
US10102444B2 (en) * 2016-11-22 2018-10-16 Lunit Inc. Object recognition method and apparatus based on weakly supervised learning
CN106772646A (en) * 2016-12-19 2017-05-31 吉林大学 A kind of ground nuclear magnetic resonance method for extracting signal
CN107560718A (en) * 2017-07-25 2018-01-09 苏州微著设备诊断技术有限公司 One kind is based on the maximized deconvolution method of auto-correlation kurtosis

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
唐道龙 等: "基于参数优化MCKD的行星齿轮箱微弱故障诊断研究", 《机电工程》 *
王建国 等: "基于局部均值分解和最大相关峭度解卷积的滚动轴承早期故障提取", 《机械传动》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113702030A (en) * 2021-07-14 2021-11-26 西南交通大学 Mechanical fault diagnosis method and system based on maximum average window kurtosis blind deconvolution
CN113484019A (en) * 2021-07-26 2021-10-08 西南交通大学 Bearing detection method and device and computer readable storage medium
CN114462451A (en) * 2022-01-24 2022-05-10 北京航空航天大学 Characteristic modal decomposition method for mechanical fault diagnosis
CN114462451B (en) * 2022-01-24 2024-06-07 北京航空航天大学 Characteristic modal decomposition method for mechanical fault diagnosis
CN117932302A (en) * 2024-01-22 2024-04-26 浙江大学 Fault signal extraction method, device, equipment and medium for mechanical device
CN117932302B (en) * 2024-01-22 2024-07-12 浙江大学 Fault signal extraction method, device, equipment and medium for mechanical device

Also Published As

Publication number Publication date
CN111024398B (en) 2021-01-15

Similar Documents

Publication Publication Date Title
CN111024398B (en) Deconvolution method for maximum correlation kurtosis without period
CN109682601B (en) Early fault identification method for rolling bearing under variable rotating speed working condition
CN107505135B (en) Rolling bearing composite fault extraction method and system
Du et al. Application of the EMD method in the vibration analysis of ball bearings
CN108388692B (en) Rolling bearing fault feature extraction method based on layered sparse coding
Jiang et al. A new l0-norm embedded MED method for roller element bearing fault diagnosis at early stage of damage
CN109855874B (en) Random resonance filter for enhancing detection of weak signals in vibration assisted by sound
CN112101245B (en) Short-time Fourier transform mechanical impact feature extraction method based on frequency domain window function
CN110987434A (en) Rolling bearing early fault diagnosis method based on denoising technology
CN102840907B (en) Rolling bearing vibration signal characteristic extracting and analyzing method under early fault state
CN103499445A (en) Time-frequency slice analysis-based rolling bearing fault diagnosis method
CN113375939B (en) Mechanical part fault diagnosis method based on SVD and VMD
Miao et al. Period-oriented multi-hierarchy deconvolution and its application for bearing fault diagnosis
CN110779724B (en) Bearing fault diagnosis method based on frequency domain group sparse noise reduction
CN109063672A (en) A kind of early stage bearing outer ring method for diagnosing faults based on adaptive M CKD
Elasha et al. A comparative study of the effectiveness of adaptive filter algorithms, spectral kurtosis and linear prediction in detection of a naturally degraded bearing in a gearbox
Xu et al. An enhanced multipoint optimal minimum entropy deconvolution approach for bearing fault detection of spur gearbox
Zheng et al. Faults diagnosis of rolling bearings based on shift invariant K-singular value decomposition with sensitive atom nonlocal means enhancement
CN105699082A (en) Sparse maximum signal-to-noise ratio deconvolution method
Xin et al. Extraction of second-order cyclostationary sources by matching instantaneous power spectrum with stochastic model–application to wind turbine gearbox
CN108398260B (en) Method for quickly evaluating instantaneous angular speed of gearbox based on mixed probability method
CN110543820A (en) Fault rolling bearing modal parameter extraction method based on edited cepstrum
CN107941511B (en) A kind of implementation method of the frequency based on signal Time-frequency Decomposition-kurtosis figure
Tse et al. Enhanced eigenvector algorithm for recovering multiple sources of vibration signals in machine fault diagnosis
Hua et al. Matching linear Chirplet strategy-based synchroextracting transform and its application to rotating machinery fault diagnosis

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant