基于EEMD和分层阈值的磁记忆信号降噪方法研究

针对现有缺陷管道的磁记忆信号降噪效果不佳及信号完整性鲜有考虑等问题，提出了基于总体平均经验模态分解（ensemble empirical mode decomposition，EEMD）和分层阈值的磁记忆信号降噪方法。首先，设计了以STM32F407为控制核心的金属磁记忆检测系统，用于采集缺陷管道的磁信号；然后，对磁信号进行EEMD预处理，得到其本征模函数（intrinsic mode function， IMF）分量，并根据频谱分析和相似度计算选择最佳分解层数；最后，利用分层阈值降噪算法重构在最佳分解层数下的IMF分量，得到降噪后的信号。通过仿真分析和实验测试，对EEMD分层阈值降噪方法进行定量评价。结果表明：该方法适用于信噪比较小的含噪信号；与小波阈值降噪方法相比，其降噪后信号的信噪比和平滑度较高，均方根误差较小，缺陷特征信号完整，可更直观地显示缺陷位置。研究结果为金属管道磁信号降噪提供了一种切实可行的方法，为管道缺陷的在线检测奠定了基础。

Research on noise reduction method of magnetic memory signal based on EEMD and layered threshold

For the poor noise reduction effect of existing defective pipe and the rare consideration of defect signal integrity, a method of magnetic memory signal noise reduction based on ensemble empirical mode decomposition (EEMD) and layered threshold was proposed. Firstly, a metal magnetic memory detecting system controlled by STM32F407 was designed to collect the magnetic signals of the defective pipe. Then, through EEMD pretreatment of magnetic signal, the intrinsic mode function (IMF) components were obtained and the optimal decomposition layer number was selected according to spectrum analysis and similarity calculation. Finally, the IMF components with the optimal decomposition layer number were reconstructed through the layered threshold noise reduction algorithm, and the de-noised signal was obtained. By means of simulation analysis and experimental test, the EEMD layered threshold noise reduction method was quantitatively evaluated.The results showed that this method was suitable for the noisy signal with small signal-to-noise ratio (SNR) and the SNR and smoothness of the de-noised signal were higher and the root mean square error was smaller than that of wavelet threshold noise reduction. The defect feature signal was complete and the defect location was more intuitively displayed.The research result provides a practical method for noise reduction of magnetic signal of metal pipe and a reference for online detection of pipe defects.