超声信号的改进EMD阈值方法降噪研究

在进行工业超声无损检测的过程中，由于环境、仪器等原因产生噪声，对后续缺陷分析等工作形成一定干扰，不利于完成后期数据分析。因此为了使信号的信噪比有所提高，提出了一种改进的经验模态分解阈值降噪算法。同时对比小波软阈值降噪法与经验模态分解清除重复间隔阈值的方法，在MATLAB仿真软件中建立超声回波数学模型并进行算法验证。实验结果表明，经验模态分解降噪方法优于小波阈值降噪，改进后的经验模态分解阈值降噪方法在信噪比、均方误差、光滑度3个方面均有所提高，达到了较好的降噪效果。

Study on noise reduction ofultrasonic signals by improved EMD threshold method

In the process of ultrasonic nondestructive testing in the industry, due to environmental reasons, from instrument noise, form a certain interference to the subsequent defect analysis, is not conducive to the completion of data analysis. In order to improve the signal to noise ratio, the signal waveform closer to no noise, proposes an improved empirical mode decomposition the threshold denoising algorithm. Compared with wavelet soft threshold denoising method and EMD method to eliminate the repetition interval of the threshold, the establishment of mathematical model of ultrasonic echo in MATLAB simulation software and the algorithm is verified. The experimental results show that the EMD denoising method is better than wavelet threshold denoising, the improved empirical mode decomposition threshold denoising method based on SNR, mean square error, smoothness are the three aspects to improve, achieve better noise reduction effect.