基于变分模态分解和独立成分分析的矿山微震信号降噪

微震信号降噪对P、S波初至拾取、震源定位和震源机制反演等具有重要意义，为此本文提出一种基于变分模态分解和独立成分分析的微震信号降噪方法。首先采用变分模态分解将微震信号分解为特定数目的模态分量，再根据模态分量与原微震信号的相关系数剔除噪音特别大的模态分量。针对噪音与有用信号混合的模态分量，采用独立成分分析提取有用信号，再与剩余的低频模态分量相加得到VMD_ICA降噪信号。此外，采用正弦函数拟合的方法去除VMD_ICA降噪信号存在的工频噪音。信号测试表明：VMD_ICA法和VMD_ICA_Sine法降噪均能有效保留微震信号的局部特征，且其信噪比大于直接去除部分模态分量。矿山微震信号应用进一步表明：VMD_ICA法和VMD_ICA_Sine法均能提高微震信号的信噪比，有效地提高了PAI-K法P波初至拾取效果，且VMD_ICA_Sine法优于VMD_ICA法降噪效果。综上所述，VMD_ICA_Sine法能为矿山微震信号降噪提供一种较好的分析方法。
Abstract：Microseismic signal denoising plays an important role in P and S phase arrival picking,seismic event location,focal mechanism inversion,and so on.To handle this problem,a variational mode decomposition (VMD) and independent component analysis (ICA) based method was proposed.Firstly,VMD was applied to decompose microseismic into certain number mode functions,then correlation coefficient between each mode function and original microseismic signal was used to remove mode functions which have a large noise.For noise and useful signal mixed mode functions,the ICA method was adopted to extract the useful signal,then the extracted useful signal was combined with the rest low frequency mode functions,which was called the VMD_ICA denoised signal.In addition,a Sine function was used to remove the power frequency noise which remained in the VMD_ICA denoised signal.A signal test shows that both the VMD_ICA method and the VMD_ICA_Sine method can retain microseismic signal local features effectively,and their signal to noise ratios (SNRs) are higher than that based on removing some mode functions directly.The mine microseismic signal application further indicates that the VMD_ICA method and the VMD_ICA_Sine method can improve microseismic signal's SNR and P phase arrival picking quality of the PAI-K method,and the VMD_ICA_Sine method has a better denoising performance than the VMD_ICA method.In conclusion,the VMD_ICA_Sine method provides a good way for mine microseismic signal denoising.