CEEMD and wavelet packet-based ultrasonic signal denoising method

Abstract

The invention discloses a CEEMD and wavelet packet-based ultrasonic signal denoising method. The method comprises the steps of firstly performing mode decomposition on a signal by utilizing a CEEMD algorithm to obtain a series of intrinsic mode functions and a trend term; secondly performing soft threshold denoising on noise dominant modes in the intrinsic mode functions, and performing adaptive rule denoising of an unbiased risk estimate principle for signal dominant modes; and finally further denoising the signal by utilizing a fine decomposition capability of wavelet packet analysis. According to the CEEMD algorithm, two opposite white noises are added to an original signal, EMD is performed and results are averaged; the end effect and the mode aliasing problem can be effectively eliminated; and the signal is further denoised under the condition of not adding auxiliary noises through wavelet and wavelet packet decomposition. The denoising method has better performance in comparisonwith that of a conventional denoising method. The denoising method can be widely applied to the material defect signal processing.

Description

technical field [0001] The invention relates to the technical field of defect echo signal processing, in particular to an ultrasonic signal noise reduction method based on CEEMD and wavelet packets. Background technique [0002] In ultrasonic nondestructive testing, noise is often mixed with useful signals, so the key to obtaining defect information is how to reduce the noise of echo signals. Due to the nonlinearity and non-stationarity of the echo signal, it is difficult for the traditional Fourier transform to accurately identify useful signals and effectively denoise them. As an improvement of Fourier transform, wavelet transform has good time-frequency local analysis ability. The disadvantage of this method is that the denoising effect is greatly affected by the selection of wavelet basis function, decomposition layer and threshold, and the selection principle is often based on experience. to make sure. Wavelet packet transform is a new development of wavelet transform...

AI Technical Summary

Problems solved by technology

Due to the nonlinearity and non-stationarity of the echo signal, it is difficult for the traditional Fourier transform to accurately identify the useful signal and effectively denoise it.

As an improvement of Fourier transform, wavelet transform has good time-frequency local analysis ability. The disadvantage of this method is that the denoising effect is greatly affected by the selection of wavelet basis function, decomposition layer and threshold, and the selection principle is often based on experience. to make sure

Wavelet packet transform is a new development of wavelet transform, which can provide a more refined analysis method for the signal, and further decompose the high-frequency part that is not subdivided by wavelet transform, thereby improving the processing ability of the signal, but there are still differences between wavelet transform and wavelet transform same disadvantage

Empirical mode decomposition (EMD) is a new method of time-frequency analysis, which can better deal with nonlinear and non-stationary signals, and has strong adaptability. It decomposes the signal into a series of natural modes arranged from high frequency to low frequency. state function (IMF) and a residual, which overcomes the shortcomings of the wavelet threshold denoising method, but it also has its shortcomings, such as modal aliasing, endpoint effects, etc.

Images