利用经验模态分解的高频水声信号滤波方法*

研究了一种高频水声信号的滤波问题,提出了一种改进的经验模态分解加小波阈值滤波方法。首先对信号进行带通滤波处理,再进行经验模态分解,将分解得到的各个模态转换为频域信号,采用小波软阈值方法对这些频域信号进行滤波,最后对信号进行重构,并转换为时域信号。经数值仿真与试验数据验证表明此方法是可行有效的,与原基于经验模态分解的小波阈值滤波方法相比,本方法滤波效果较好:对不同输入信噪比的仿真信号进行滤波后,本方法的输出信噪比最大提高17.41 d B,滤波后所得信号与加噪前纯信号的相关系数最大提高0.90;对实验数据进行滤波后,不同时间段信号的相关系数最大提高0.62。

A filtering method for high frequency underwater acoustic signal using a improved empirical mode decomposition*

The purpose of this work is to study a method for filtering high frequency underwater acoustic signal based on the ensemble empirical mode decomposition (EEMD) and the wavelet soft threshold (WST) methods. Firstly, the band-pass filter is used to denoise the signal with noise. Secondly, the EEMD method is used to process the signal, then the intrinsic mode functions (IMFs) are transformed to signals in frequency domain, respectively. Thirdly, analyzing the characteristic of the IMFs, and finding the main component IMFs, then the IMFs are filtered by using the WST method. Finally, the IMFs are added to reconstruct the signal in frequency domain, and then the signal in time domain is obtained. The given method is proved feasibly and effectively by numerical simulation and experiment data, comparing with FIR digital filter method, the following acquaintances can be observed: (1) There is above 30% output signal noise ratio (SNR) improved for simulation signal under different input SNR, respectively. (2) The correlative coefficient between the signal filtered and the simulation signal without noise can be improved 22% at most. (3) The correlative coefficient between different periods of timeSfor experiment data can also be improved to 1.6 times at most.