Unlock instant, AI-driven research and patent intelligence for your innovation.

Feature extraction method of vibration signal under strong noise background

A vibration signal and feature extraction technology, applied in vibration measurement, vibration measurement in solids, measurement devices, etc., can solve the problems of noise sensitivity of characteristic signal components, easy generation of false components, and inability to effectively obtain signal characteristic components, etc. The effect of high analysis efficiency

Inactive Publication Date: 2018-03-30
NO 719 RES INST CHINA SHIPBUILDING IND
View PDF2 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] A large number of simulations and engineering examples have shown that wavelet analysis has some shortcomings in the practical application of vibration signal processing: (1) Wavelet transform is very sensitive to the singular point of the signal, and can accurately locate the sudden point of the singular signal, But it is difficult to judge those mutation points are the points of interest; (2) Selecting the appropriate time domain and frequency domain resolution has not been well resolved in engineering; (3) The wavelet analysis lacks effective and fast algorithms, so it is difficult Meet the real-time requirements; (4) The selection of wavelet function in wavelet transform is the same as the selection of window function in spectral analysis. It needs to be considered according to the characteristics of different application problems. It is very difficult to choose the appropriate wavelet basis function for specific engineering problems.
Although the clear instantaneous frequency in HHT can express the instantaneous change of the signal, it has been shown in simulation and engineering applications that when the signal-to-noise ratio is small, the HHT method cannot effectively obtain the characteristic components in the signal
The HHT method has the shortcomings of the empirical mode decomposition stop criterion is not perfect, it is easy to produce false components in the decomposition process, and the uncertainty of the decomposition results, etc.
[0012] To sum up, a common problem existing in the existing vibration signal analysis methods is that they are sensitive to the aliased noise in the characteristic signal components. Signal analysis is very limited

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Feature extraction method of vibration signal under strong noise background
  • Feature extraction method of vibration signal under strong noise background
  • Feature extraction method of vibration signal under strong noise background

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0092] Object: s 1 (t)=[1+sin(5πt)]·cos[20πt+0.2sin(10πt)]+sin(80πt) (15)

[0093] The time sequence diagram described by the above formula (15) is as follows figure 2 shown.

[0094] For the time series given by equation (15), using the method provided by the present invention, the marginal spectrum obtained is as image 3 shown.

[0095] The Daubechies wavelet in the orthogonal wavelet is selected as the wavelet basis function, and the db10 in the MATLAB wavelet toolbox is used to decompose the time series given by equation (15) with 5 layers of wavelet, and then the obtained detail signal is reconstructed, Then do marginal spectrum analysis on the reconstructed signal, the marginal spectrum obtained in this way is as follows Figure 4 shown.

[0096] Perform 1.5-dimensional spectral analysis on the time series given by equation (15), and the obtained spectrum is as follows Figure 5 shown.

[0097] For the HHT analysis of the time series given by equation (15), first,...

Embodiment 2

[0100] Object: s 2 (t)=[1+sin(5πt)] cos[20πt+0.2sin(10πt)]+sin(80πt)+n(t) (16)

[0101] The above formula is obtained by adding the AM-FM signal constructed by equation (15) to the noise signal n(t) with zero mean value and variance of 15, and the time sequence diagram described by it is as follows Figure 7 shown.

[0102] For the time series given by equation (16), using the method provided by the present invention, the marginal spectrum obtained is as Figure 8 shown.

[0103] Apply db10 in the MATLAB wavelet toolbox to decompose the time series given by equation (16) with 5 layers of wavelet, and then reconstruct the obtained detail signal, and then perform marginal spectrum analysis on the reconstructed signal, the obtained The marginal spectrum is as Figure 9 shown.

[0104] The 1.5-dimensional spectrum analysis is performed on the AM-FM sequence containing noise, and the obtained spectrum is shown in Figure 10.

[0105] Performing HHT analysis on the time series ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to a feature extraction method of a vibration signal under a strong noise background. The method comprises a step of sampling to obtain a piece of mechanical vibration signal A,a step of performing an even number of periodic extension on the signal A to obtain a new time series B after extension, wherein the even number is larger than 4, a step of performing more than threemulti-correlation processing on the time sequence B to obtain a sum of a characteristic signal correlation sequence C and a constant term, a step of performing EMD decomposition on the sequence C to obtain each IMF component ci and margin rn, wherein ci is denoted as D, a step of carrying out cross-correlation calculation on each IMF component D and the time sequence B, comparing calculation results and a set threshold value lambda, screening a calculation result which is larger than lambda and denoting the calculation result as E, and a step of carrying out marginal spectrum analysis on E toform a marginal spectrum curve, wherein the a prominent one in the marginal spectrum curve is a feature of the vibration signal. The method has the advantages of high analysis efficiency, the online detection on the vibration signal can be carried out, and hidden characteristic signal and state characterization signal can be extracted under a condition of a low signal to noise ratio.

Description

technical field [0001] The invention relates to mechanical system vibration signal processing, in particular to a method for feature extraction of vibration signals under strong noise background. Background technique [0002] With the continuous emergence of high-precision testing equipment and advanced testing methods, higher requirements are put forward for the analysis methods of vibration signals of mechanical systems. However, some characteristic signals and potential state characterization signals of the mechanical system are often submerged in strong background noise, so that some early features of the mechanical system, such as fault information, equipment status change information, and system comprehensive performance information, cannot be detected. Early detection will bring safety hazards and economic losses to the operation of the entire industrial system, and even cause casualties. [0003] Existing methods for processing vibration signals of mechanical system...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G01H1/00
CPCG01H1/00
Inventor 廖庆斌谢小华蔡如桦陈刚
Owner NO 719 RES INST CHINA SHIPBUILDING IND