基于经验模态分解的管道超声回波信号识别

Recognition of pipeline ultrasonic echo signal based on empirical mode decomposition

在管道超声无损检测中,超声回波信号常受到噪声干扰或者因管道壁缺陷原因导致回波波形重叠,故检测信号很难判断缺陷大小。现采用一种基于经验模态分解(Empirical Model Decomposition,EMD)的方法,对超声回波信号的到达时间进行准确识别。该方法首先把原始采集的超声回波信号进行经验模态分解,预处理回波信号中的噪声干扰,处理后的信号通过正交检波处理得到该信号的峰值包络,则该包络的峰值时刻对应了回波信号的到达时刻。通过对实测缺陷信号的试验,验证了该方法的可行性和准确性。

Pipeline ultrasonic NDT echo signals are difficult to determine the sizes of defects because of low SNR or pipeline wall flaw with causing echo overlapped. A technique based on empirical model decomposition （EMD） was proposed to improve the arrival time identification of ultrasonic echo. A signal is decomposed by EMD for pretreating the noise in the echo, and then is changed into the envelope by quadrature detection transform which the arrival times of echoes corresponds to the peaks of the envelope of reconstructed signal. Meanwhile, the effectiveness of this technique was testified by the experimental results with the real ultrasonic data.