CN115712822A

CN115712822A - Bearing fault diagnosis method, system and medium based on CEEMDAN-OMEDA

Info

Publication number: CN115712822A
Application number: CN202211652941.4A
Authority: CN
Inventors: 徐徐; 钱进; 杨世飞; 孙磊; 邹小勇; 刘宗斌
Original assignee: Nanjing Chaos Data Technology Co ltd
Current assignee: Nanjing Chaos Data Technology Co ltd
Priority date: 2022-12-22
Filing date: 2022-12-22
Publication date: 2023-02-24

Abstract

The invention discloses a CEEMDAN-OMEDA-based bearing fault diagnosis method, a system and a medium, belonging to the technical field of industrial diagnosis and data processing, wherein the method comprises the steps of performing CEEMDAN decomposition on an original bearing acceleration signal; and enhancing impact components in the IMF components by an OMEDA method, extracting the components with the most impact components in the IMF components by using the index of the impact energy average value, and finally performing envelope demodulation by using the component signals to extract the fault characteristics of the bearing. And checking the amplitude of the characteristic frequency of the bearing in the envelope spectrum, and if the amplitude of a certain characteristic frequency exceeds an envelope threshold value, the bearing component corresponding to the characteristic frequency has a fault. The method can effectively reduce the possibility of missing fault impact signals, thereby ensuring the integral denoising effect and effectively identifying the fault part of the bearing.

Description

Bearing fault diagnosis method, system and medium based on CEEMDAN-OMEDA

Technical Field

The invention belongs to the technical field of industrial diagnosis and data processing, and particularly relates to a CEEMDAN-OMEDA-based bearing fault diagnosis method, system and medium aiming at weak impact generated by early bearing fault.

Background

With the rapid development of emerging technologies such as big data, artificial intelligence, industrial internet and the like, industrial digitization is an inevitable direction for future development. The continuous generation of industrial data has made traditional processing data processing methods challenging with low efficiency and insufficient accuracy.

The bearing is used as an indispensable part of a transmission system and widely applied in the fields of industry, transportation industry and military, and the condition of the running state of the bearing is directly related to the working condition of the whole equipment. When the bearing has early failure, weak impact signals can be generated, the resonance frequency band usually appears in a high-frequency band range, and the common time-frequency analysis method has limitation on the weak impact signals.

In a conventional CEEMDAN (Complete Empirical Mode Decomposition) algorithm, white Noise is adaptively added to the whole Decomposition process, and IMF components are obtained by calculating a unique residual signal, so that the problems of Mode aliasing and the like are solved to a great extent.

However, when many scholars use the CEEMDAN method, the correlation between each IMF and the original signal is calculated, the IMF component with high correlation is selected, the IMF component with low correlation is discarded, that is, the selected IMF is the component with information as the main component, and then the IMF component with more noise is discarded, which may miss the impact signal, resulting in that the effective information in the IMF cannot be fully utilized, thereby affecting the overall denoising effect.

Therefore, how to reduce the possibility of missing the impulse signal on the premise of ensuring that the impulse signal is extracted accurately enough is an urgent problem to be solved.

Disclosure of Invention

Aiming at the defects or improvement requirements of the prior art, the invention provides a CEEMDAN-OMEDA-based bearing fault diagnosis method, system and medium, which can effectively solve the relevant problems in extraction and processing of weak impact signals generated by early faults of a bearing, and reduce the possibility of missing the impact signals on the premise of ensuring that enough and accurate fault impact signals are extracted, thereby ensuring the integral denoising effect and effectively identifying the fault impact signals corresponding to the fault parts of the bearing.

In order to solve the above technical problem, according to an aspect of the present invention, there is provided a bearing fault diagnosis method based on CEEMDAN-omeeda, comprising the steps of:

s1, acquiring an original acceleration signal of a bearing;

s2, performing CEEMDAN decomposition on the original acceleration signal;

s3, drawing a fast kurtosis spectrum of each IMF component signal in the basic mode, and acquiring a resonance frequency band of each IMF component signal;

s4, obtaining a filtering step length according to the resonance frequency band, and then performing enhancement processing on the impact component of each IMF component signal by using an OMEDA method;

s5, performing frequency domain filtering on each processed IMF component signal;

s6, calculating an impact energy average value of each filtered IMF component signal;

s7, carrying out envelope demodulation on the IMF component signal with the largest average impact energy value to obtain an envelope spectrum;

s8, checking the amplitude of the characteristic frequency of the bearing in the envelope spectrum, and if the amplitude of a certain characteristic frequency exceeds an envelope threshold value, enabling the bearing component corresponding to the characteristic frequency to have a fault.

In the above technical solution, the step length of step S4 is to subtract the initial frequency from the cutoff frequency of the resonance frequency band.

In the above technical solution, the specific process of applying the OMEDA method in step S4 is to set a filter with a finite length, where the finite length is a filtering step length, and extract an impact component of the bearing fault signal according to the length.

According to the technical scheme, the bearing characteristic frequency is obtained according to the type and the rotating speed of the bearing, and the bearing characteristic frequency respectively corresponds to an inner ring, an outer ring, a retainer and a rolling body of the bearing.

In the above technical solution, step S5 includes performing bandpass parameter setting on the wired sensor filtering and the wireless sensor filtering, respectively.

In some optional embodiments, step S5 sets the band pass frequency to 20 to 25khz for the wired sensor; for a wireless sensor, the band pass frequency is set to be 3K to 5KHz.

Specifically, the impact energy average value of each IMF is calculated after CEEMDAN decomposition, the IMF component with the largest impact energy average value is selected as the impact signal of the bearing to be analyzed, the impact component in the signal is extracted, and the integrity of the signal extraction source is ensured. Meanwhile, the method takes the maximum kurtosis as a criterion, extracts the periodic pulse characteristics of the bearing fault and simultaneously minimizes the noise component. When the fault periodicity signal times are extracted and the noise is reduced, the fault signal in the reconstructed generated signal is enhanced, i.e. the impact component of the bearing fault signal is enhanced.

According to another aspect of the present invention, the present invention also provides a CEEMDAN-omeeda-based bearing fault diagnosis system, comprising:

the signal acquisition device is used for acquiring an original acceleration signal of the bearing;

the signal processing device is used for carrying out CEEMDAN decomposition on the original acceleration signal, drawing a rapid kurtosis spectrum of each basic mode IMF component signal and acquiring a resonance frequency band of each IMF component signal; obtaining a filtering step length according to the resonance frequency band, wherein the step length is obtained by subtracting the initial frequency from the cut-off frequency of the resonance frequency band, and then reinforcing the impact component of each IMF component signal by using an OMEDA method; performing frequency domain filtering on each processed IMF component signal; calculating the impact energy average value of each filtered IMF component signal; carrying out envelope demodulation on the IMF component signal with the largest impact energy average value to obtain an envelope spectrum;

and the signal comparison device is used for checking the amplitude of the characteristic frequency of the bearing in the envelope spectrum and judging whether the amplitude of a certain characteristic frequency exceeds an envelope threshold value so as to determine whether the bearing component corresponding to the characteristic frequency has a fault.

In the technical scheme, when the signal processing device carries out frequency domain filtering on each processed IMF component signal, the band-pass frequency band of a wired sensor is set to be 20 to 25KHz; for a wireless sensor, the band pass frequency is set to be 3K to 5KHz.

In the technical scheme, when the signal processing device adopts an OMEDA method, a filter with a limited length is arranged, the limited length is a filtering step length, and an impact component of a bearing fault signal is extracted according to the length.

According to another aspect of the invention, there is also provided a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of any of the methods described above.

In general, compared with the prior art, the above technical solution contemplated by the present invention can achieve the following beneficial effects:

according to the invention, the impact component in the IMF component is strengthened by an OMEDA method, the component with the most impact component in the IMF component is extracted by using the index of the impact energy average value, and finally the component signal is used for envelope demodulation to extract the fault characteristic of the bearing.

Compared with the prior art, the OMEDA method can highlight the impact component in the signal and easily extract the impact signal of the bearing fault.

And a signal closely related to the bearing fault is extracted through the impact energy average value, and an interference signal is suppressed.

All impact signals of faults are reserved while the bearing fault signals are denoised, the integrity and the comprehensiveness of the signals are guaranteed, the possibility of missing the impact signals is avoided, the integral denoising effect is guaranteed, and the effectiveness and the accuracy of a diagnosis method are guaranteed.

Drawings

Fig. 1 is a schematic flow chart of a CEEMDAN-OMEDA-based bearing fault diagnosis method provided by an embodiment of the invention.

FIGS. 2-7 are fast kurtosis spectra of the first six orders of IMF (IMF 1-IMF6, respectively) according to an embodiment of the present invention.

Figure 8 is a component signal envelope spectrum of an 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 described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

Example 1:

fig. 1 is a flow chart of an analysis method of the present invention, which shows a bearing fault diagnosis method based on CEEMDAN-omeeda according to the present invention, specifically including the following steps:

s1, acquiring an original acceleration signal of a bearing;

s2, performing CEEMDAN decomposition on the original acceleration signal;

s3, drawing a fast kurtosis spectrum of each fundamental mode component IMF, and acquiring a resonance frequency band of each component signal IMF;

s4, obtaining a filtering step length according to the resonance frequency band, wherein the step length is obtained by subtracting the initial frequency from the cut-off frequency of the resonance frequency band, and then reinforcing the impact component of each IMF component signal by using an OMEDA method; the OMEDA method is to design a filter with limited length, the length is the filtering step length, the impact component of the bearing fault signal is extracted, and the noise is reduced, so the signal noise after being processed by the OMEDA is reduced, and the impact component is enhanced;

s5, performing frequency domain filtering on each processed IMF component;

s6, calculating an impact energy average value of each filtered IMF component;

The Optimal Minimum Entropy Deconvolution (OMEDA) method is to design a filter with a limited length, wherein the length is the filter step length, impact components of bearing fault signals are extracted, and noise is reduced, so that the signal noise after the OMEDA processing is reduced, and the impact components are enhanced.

And drawing a fast kurtosis spectrum of each IMF according to each component signal according to the steps S1-S3, and displaying the fast kurtosis spectrum of the first six orders, as shown in the figures 2-7.

Step size of filtering IMF1 is set to 2560, step size of IMF2 is 853, step size of IMF3 is 640, step size of IMF4 is 853, step size of IMF5 is 426, step size of IMF6 is 320, step size of IMF7 is 1706, step size of IMF8 is 1706, step size of IMF9 is 640, step size of IMF10 is 320, step size of IMF11 is 213, step size of IMF12 is 80 according to step S4, and an OMEDA method is adopted to highlight the impulse component in each IMF component signal.

Each IMF component signal is frequency domain filtered according to S5. For a wired sensor, the band pass band is set to 20 to 25KHz. For a wireless sensor, the band pass frequency is set to be 3K to 5KHz.

The mean impact energy value for each IMF component after filtering is calculated according to S6, as shown in table 1.

As can be seen from table 1, the component signal having the largest average value of impact energy is IMF8.

The bearing model is 6314 deep groove ball bearing. Bearing parameters: number of rolling elements

Diameter of rolling element

Pitch diameter of bearing

Contact angle of

. The bearing failure frequency at 1800r/min is shown in Table 2.

TABLE 1 IMF average impact energy

TABLE 2 bearing characteristic frequency (Hz)

Structure of the device	Inner ring	Outer ring	Holding rack	Rolling body
					Characteristic frequency	147.7	92.3	12.0	123.0

According to the step S7, performing envelope demodulation on the IMF8, as shown in the graph 8, obtaining an envelope spectrum, according to the step S8, bringing the characteristic frequency of the bearing into the envelope spectrum to search for the amplitude of the corresponding frequency, and finding that the amplitude of 147.6Hz is more prominent, thereby preliminarily judging that the inner ring of the bearing has faults.

The verification result shows that the method can effectively identify the fault part of the bearing, so that the fault of the bearing can be effectively diagnosed.

Example 2:

according to another aspect of the present invention, the present invention also provides an embodiment of a CEEMDAN-omeeda based bearing fault diagnosis system, comprising:

the signal processing device is used for carrying out CEEMDAN decomposition on the original acceleration signal, drawing a rapid kurtosis spectrum of each IMF component signal in the basic mode and acquiring a resonance frequency band of each IMF component signal; obtaining a filtering step length according to the resonance frequency band, wherein the step length is obtained by subtracting the initial frequency from the cut-off frequency of the resonance frequency band, and then reinforcing the impact component of each IMF component signal by using an OMEDA method; performing frequency domain filtering on each processed IMF component signal; calculating the impact energy average value of each filtered IMF component signal; carrying out envelope demodulation on the IMF component signal with the largest impact energy average value to obtain an envelope spectrum;

In the technical scheme, when the signal processing device carries out frequency domain filtering on each processed IMF component signal, the band-pass frequency band of a wired sensor is set to be 20 to 25KHz; and for the wireless sensor, setting the band-pass frequency band to be 3K to 5KHz.

Example 3:

according to another aspect of the invention, the invention also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, carries out the steps of any of the methods described above.

It should be noted that, according to implementation requirements, each step/component described in the present application can be divided into more steps/components, and two or more steps/components or partial operations of the steps/components can also be combined into a new step/component to achieve the purpose of the present invention.

It will be understood by those skilled in the art that the foregoing is only an exemplary embodiment of the present invention, and is not intended to limit the invention to the particular forms disclosed, since various modifications, substitutions and improvements within the spirit and scope of the invention are possible and within the scope of the appended claims.

Claims

1. A bearing fault diagnosis method based on CEEMDAN-OMEDA is characterized by comprising the following steps:

s1, acquiring an original acceleration signal of a bearing;

s2, performing CEEMDAN decomposition on the original acceleration signal;

s7, carrying out envelope demodulation on the IMF component signal with the largest average impact energy value to obtain an envelope spectrum, and extracting the fault characteristics of the bearing;

2. The CEEMDAN-OMEDA-based bearing failure diagnosis method as set forth in claim 1, wherein the method of obtaining the filter step size from the resonance frequency band in step S4 is to subtract an initial frequency from a cutoff frequency of the resonance frequency band.

3. The CEEMDAN-OMEDA-based bearing fault diagnosis method as claimed in claim 1, wherein the specific process of applying the OMEDA method in step S4 is to set a filter of finite length, the finite length being a filter step size, and extract an impulse component of the bearing fault signal according to the length.

4. The CEEMDAN-OMEDA-based bearing fault diagnosis method as claimed in claim 1, wherein the step S5 is divided into wired sensor filtering and wireless sensor filtering for band-pass parameter setting, respectively.

5. The CEEMDAN-OMEDA-based bearing failure diagnosis method as claimed in claim 1, wherein the band pass band is set to 20 to 25KHz for the wired sensor at step S5; and for the wireless sensor, setting the band-pass frequency band to be 3K to 5KHz.

6. The CEEMDAN-OMEDA-based bearing fault diagnosis method as claimed in claim 1, wherein the step S8 obtains the bearing characteristic frequencies according to the bearing model and the rotation speed, and the bearing characteristic frequencies respectively correspond to an inner ring, an outer ring, a cage, and a rolling body of the bearing.

7. A CEEMDAN-ome da-based bearing fault diagnosis system, characterized by comprising:

the signal processing device is used for carrying out CEEMDAN decomposition on the original acceleration signal, drawing a rapid kurtosis spectrum of each basic mode IMF component signal and acquiring a resonance frequency band of each IMF component signal; obtaining a filtering step length according to the resonance frequency band, and then strengthening the impact component of each IMF component signal by using an OMEDA method; performing frequency domain filtering on each processed IMF component signal; calculating the impact energy average value of each filtered IMF component signal; carrying out envelope demodulation on the IMF component signal with the largest impact energy average value to obtain an envelope spectrum;

8. The CEEMDAN-OMEDA-based bearing fault diagnosis system as claimed in claim 7, wherein the signal processing means performs band pass parameter setting for wired sensor filtering and wireless sensor filtering, respectively.

9. A CEEMDAN-OMEDA-based bearing fault diagnosis system as claimed in claim 7, wherein when the signal processing means employs the OMEDA method, a filter of finite length is provided, the finite length being a filter step size, and the impulse component of the bearing fault signal is extracted from the length.

10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method of any one of the preceding claims 1 to 6.