低信噪比微震P波震相初至自动拾取方法  

Method of automatic detection on micro-seismic P-arrival time under low signal-to-noise ratio

作　　者：贾瑞生[1,2] 谭云亮[1,3] 孙红梅[2] 洪永发[2] 

机构地区：[1]山东科技大学矿山灾害预防控制省部共建国家重点实验室培育基地,山东青岛266590 [2]山东科技大学信息科学与工程学院,山东青岛266590 [3]山东科技大学矿业与安全工程学院,山东青岛266590

出　　处：《煤炭学报》2015年第8期1845-1852,共8页Journal of China Coal Society

基　　金：山东省自然科学基金资助项目(ZR2013EEM019);国家"十二五"科技支撑计划资助项目(2012BAK04B06)

摘　　要：针对微震信号固有的低信噪比、非平稳性、随机性等特征,发展了一种基于希尔伯特-黄变换(HHT)和AIC准则相结合的震相初至自动拾取方法,该方法首先对含噪信号进行经验模态分解(EMD)及内蕴模态函数(IMF)重构,在保留微震信号固有特征基础上实现降噪;其次,基于Hilbert变换计算出归一化包络信号,通过设置包络阈值搜索震相初至的大致位置,并以该位置为基础为AIC函数选择合适的计算时窗以降低计算复杂度;最后,在选择的时窗内应用AIC准则计算出P波震相初至。应用该方法对随机选取的高、低信噪比两类共80个微震信号进行震相初至拾取,若以人工拾取结果为基准、时差在10 ms以内视为准确拾取,结果显示,高信噪比信号拾取准确率为100%,低信噪比信号拾取准确率为92%。According to the low signal-to-noise ratio,non-stationary,and inherent randomness of micro-seismic signals,a new method based on Hilbert-Huang Transform( HHT) and AIC criterion was proposed. Firstly,to implement denoising and filtering,the signals with noise were decomposed with empirical mode decomposition( EMD) and then reconstructed with intrinsic mode function( IMF); Secondly,based on the Hilbert transform,the normalized envelope signal was calculated,the coarse position of P-arrival was searched by setting the envelope threshold,and on the basis of the position,the time window of AIC function was determined; Finally,the P-arrival time was calculated by using AIC criterion in the selected time window. This method was applied to 80 random micro-seismic signals with both high SNR and low SNR. Considering the manual picking results as a benchmark and the time difference within 10 ms as an accurate result,the result shows that for the signals with high SNR,the picking accuracy is 100%. For the signals with low SNR,the accuracy is 92%.

关 键 词：微震P波 震相初至拾取 经验模态分解 希尔伯特-黄变换 AIC准则 

分 类 号：TD326[矿业工程—矿井建设]