摘要
针对微地震信号所具有的随机非平稳特点,基于小波包分解重构原理提出一种微地震信号降噪方法。首先根据拾震器采样频率对微地震信号进行N层小波包分解,得到2~N个子频带,计算各子频带与原信号之间的互相关系数,互相关系数较大的子频带不做处理,中等相关的子频带采用小波软阈值降噪,剩余子频带直接排除,最后重构处理后的信号分量,得到最终的降噪信号。引入信噪比、均方根误差以及降噪后信号占原信号的能量百分比作为降噪效果评估标准。仿真结果表明,在对微地震信号降噪时,该方法可以有效地压制噪声,信噪比提升了10 d B以上,均方根误差降低至0.035以下,能量百分比在90%以上。工程上的实际微地震信号通过该方法处理后,也取得了较好的效果。
Because of the random non-stationarity of micro-seismic signal,a denoising method is proposed based on wavelet packet decomposition and reconstruction. Firstly,the noisy signal is decomposed into 5 layers by wavelet packet to obtain 2 Nsub-bands. The cross-correlation coefficient between each sub-bands signal and the original signal is calculated. The sub-bands with large correlation number are not processed,and the sub-bands of medium correlation is denoised by soft threshold,and the remaining sub-bands are eliminated directly. These sub-bands are reconstructed,and a new micro-seismic signal is produced. The signal-to-noise ratio( SNR),the root means square error( RMSE) and the energy percentage of the signal are introduced after denoised and the original signal( Esn)as criteria for evaluating the effect of noise reduction. The simulation results show that this method can suppress the noise signal efficiently. The SNR can be improved more than 10 d B. The RMSE can be reduced to less than 0. 035 and Esn remained above 90%.The method can achieve good performance even when applied on the field signals.
出处
《电子测量与仪器学报》
CSCD
北大核心
2018年第4期134-143,共10页
Journal of Electronic Measurement and Instrumentation
基金
山东省重点研发计划(2017GSF20115)
中国博士后科学基金(2015M582117)资助项目
关键词
微震信号降噪
小波包分解
信号重构
互相关
micro-seismic signal denoising
wavelet packet decomposition
signal reconstruction
cross correlation
