基于EMD和小波熵阈值算法的超声回波信号降噪  

Noise reduction in ultrasonic echo signal based on EMD and wavelet entropy threshold algorithm

作　　者：杜必强[1] 孙立江[1] 

机构地区：[1]华北电力大学(保定)机械工程系,河北保定071003

出　　处：《中国测试》2017年第1期101-105,共5页China Measurement & Test

基　　金：中央高校基本科研业务费项目(2014MS118)

摘　　要：超声检测信号中通常包含大量噪声,而其中材料晶界散射的噪声是一种相关噪声。鉴于传统的方法难以将这种噪声和缺陷回波信号区分,提出一种EMD和小波熵阈值联合降噪的算法。该算法首先对目标信号进行EMD分解,提取具有噪声特性的IMF分量进行小波分解,利用含噪系统熵增的特性,在分解各尺度层的细节部分选用小波熵自适应阈值降噪,然后将剩余分量和降噪处理后的信号进行重构。仿真信号结果表明:该降噪方法(EMD-WET)输出信号的信噪比(SNR)为7.9 d B、均方根误差(RMSE)为18.1、相似系数(NCC)为0.92,优于传统的小波软、硬阈值方法。对实测信号进行处理,该方法降低信号中的大部分噪声,更好地还原回波信号的波形。Ultrasonic testing signal often contains a lot of noise, the noise scattered by grain boundary is a correlated noise. Considering that it is difficult for traditional approach to distinguish between this kind of noise and defect echo signal, the paper presents a de-noising method that combines EMD and wavelet entropy threshold. With this algorithm, the target signal is decomposed by EMD firstly and then the IMF component with noise characteristic is extracted for wavelet decomposition. After that, according to the entropy increase characteristics of noisy system, wavelet entropy adaptive threshold is used for de-noising in the decomposition of some details of each scale and then the remaining components and de-noised signals are reconstituted. Simulation results show that the output signal of the de-noising method （EMD-WET） in this paper has a signal-to-noise ratio（SNR） of 7.9dB, root-mean-square error（RMSE） of 18.1 and normalized correlation coefficient （NCC） of 0.92. Thus it is superior to traditional wavelet soft threshold and hard threshold method. During the process of disposition on tested signal, this method efficiently reduces most noise of ultrasonic echo signal and better restores the waveform of echo signal.

关 键 词：超声检测 降噪 小波熵 经验模态分解 

分 类 号：TP311.52[自动化与计算机技术—计算机软件与理论]