基于ICEEMDAN-盲源分离联合的微震信号降噪方法研究  

Research on Microseismic Signal Denoising Based on ICEEMDAN and Blind Source Separation

作　　者：黄港 郑禄林 王英乐 左宇军[1] 郑禄璟 刘晓蓉 HUANG Gang;ZHENG Lulin;WANG Yingle;ZUO Yujun;ZHENG Lujing;LIU Xiaorong(College of Mining,Guizhou University,Guiyang 550025,Guizhou,China;Guizhou Jinfeng Mining Co Ltd,Zhenfeng 562200,Guizhou,China)

机构地区：[1]贵州大学矿业学院,贵州贵阳550025 [2]贵州锦丰矿业有限公司,贵州贞丰562200

出　　处：《矿冶工程》2023年第3期24-29,共6页Mining and Metallurgical Engineering

基　　金：贵州省科技支撑计划项目(黔科合支撑[2021]一般516、黔科合支撑[2022]一般247);贵州省矿山动力灾害预警与控制技术科技创新人才团队(黔科合平台人才[2019]5619号)。

摘　　要：针对黔西南锦丰金矿巷道施工采集的微震信号非平稳特征和背景噪声干扰问题,引入一种基于完善的自适应噪声完备集成经验模态分解(ICEEMDAN)与盲源分离联合的降噪方法。该方法通过ICEEMDAN算法对微震信号进行初步分解,再利用MATLAB平台计算出信号的相关系数和边际频谱,筛选出含噪模态分量和信号的主频率分量,最后通过FastICA算法进行盲源分离,实现降噪。实际应用结果表明,与经验模态分解(EMD)和小波包阈值传统方法相比,该方法信噪比更大(24.142 5 dB)、标准误差更小(0.012 18)、降噪效果更好。Microseismic signals collected during roadway construction in Jinfeng Gold Mine in the southwest Guizhou Province are non⁃stationary and also contaminated with background noise.For solving this problem,a noise reduction method based on the improved complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN)combined with blind source separation(BSS)was proposed.With this method,a microseismic signal can be preliminarily decomposed by ICEEMDAN algorithm,and then correlation coefficient and marginal spectrum of signal can be calculated with MATLAB.Components of noise⁃containing mode and main frequency of signal can be filtered,and FastICA algorithm is finally used for BSS to achieve noise reduction.The practical application in mine shows that compared to empirical mode decomposition(EMD)and traditional wavelet packet threshold method,the proposed method,with higher signal⁃to⁃noise ratio(24.1425 dB),can bring better noise reduction effect with smaller standard error(0.01218).

关 键 词：ICEEMDAN 盲源分离 FASTICA算法 微震信号 降噪 微震监测 

分 类 号：TD853[矿业工程—金属矿开采]