基于EEMD-FastICA的单通道超声回波信号去噪研究

Noise Filtering of Single-channel Ultrasonic Echo Signal Based on EEMD-FastICA

针对单通道超声检测回波信号易受到噪声信号的影响导致缺陷诊断精度低的问题,提出基于集合经验模态分解(EEMD)和快速独立成分分析(Fast ICA)的单通道超声回波信号盲源分离方法(EEMD-Fast ICA)。首先应用EEMD算法对回波信号进行自适应分解,得到多个不同尺度的固有模态函数(IMF)分量,利用主成分分析(PCA)预估源信号的数目,通过相关系数法筛选出相应的IMF分量进行重构,并将重构信号与原始采集信号组合构成新的多维观测信号,解决了原始采集信号盲源分离中存在的欠定问题,然后进行快速独立成分分析(Fast ICA)运算,实现了单通道超声检测信号的噪声分离。实验结果表明:EEMD-Fast ICA方法能对采集超声回波信号进行有效的降噪处理,并且能保护真实信号的频域特征信息。

Aiming at the situation that the echo signal of single-channel ultrasonic testing is easily affected by noise signals, which can reduce the accuracy of defect detection, a blind source separation of single-channel ultrasonic echo signals is proposed based on ensemble empirical mode decomposition(EEMD) and fast independent component analysis(FastICA). The EEMD algorithm is used to adaptively decompose the echo signal, and multiple intrinsic mode function(IMF) components are obtained. The principal component analysis(PCA) algorithm is used to evaluate the quantity of source signals, and correlation coefficients are used to filter out IMF components for reconstruction signal. Combining the reconstructed signal and the original signal into a new observation signal solves the under-determined problem of blind source separation in the original signal. Finally, fast independent component analysis(FastICA) is performed on the new observation signal, so as to realize the noise separation of the single-channel ultrasonic detection signal. The experimental results show that the EEMD-FastICA method can effectively reduce the noise of the collected ultrasonic echo signals, and can protect the frequency domain characteristic information of the real signal.