隧道爆破振动信号畸变校正与混沌多重分形特征研究  

Distortion correction and fractal characteristics of vibration signals of a tunnel blasting

作　　者：付晓强 俞缙[2] 刘纪峰[1] 杨仁树 戴良玉 FU Xiaoqiang;YU Jin;LIU Jifeng;YANG Renshu;DAI Liangyu(School of Civil Engineering,Sanming University,Sanming 365004,China;Fujian Research Center for Tunneling and Urban Underground Space Engineering,Huaqiao University,Xiamen 361021,China;Sanming Coffer Fine Chemical Industrial Co.,Ltd.,Sanming 365500,China;School of Civil and Resources Engineering,University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]三明学院建筑工程学院,福建三明365004 [2]华侨大学福建省隧道与城市地下空间工程技术研究中心,福建厦门361021 [3]三明科飞产气新材料股份有限公司,福建三明365500 [4]北京科技大学土木与资源工程学院,北京100083

出　　处：《振动与冲击》2022年第6期76-85,共10页Journal of Vibration and Shock

基　　金：三明学院国家基金培育计划(PYT2008);福建省自然科学基金(2020J01390);国家自然科学基金面上项目(51874144)。

摘　　要：受测试环境影响,隧道爆破监测信号中普遍包含噪声和趋势项干扰。针对爆破信号干扰项消除难题,选取典型地铁隧道工程监测到的畸变爆破信号为分析对象,采用稀疏化基线估计与去噪(baseline estimation and denoising with sparsity,BEADS)算法实现了噪声和趋势项成分的提取,得到反映真实爆破信息的校正信号。利用多重分形去趋势波动分析(multi-fractal detrended fluctuation analyses,MF-DFA)捕捉到三个分量信号的混沌分形特征,并根据小波相关性凝聚谱对三个分量信号与原始信号的时频域相关性进行了精确表征。结果表明:隧道爆破信号高频噪声、低频趋势项和校正信号三者的混沌分形特征具有显著差异。校正信号吸引子轨迹形态为反复周期性有序波动且具有持续性和反持续性分形谱特征,其递归图具有周期模式;低频趋势项吸引子形态表现为近似直线且具有持续性分形谱特征,其递归图具有对角线分布突变模式;高频噪声吸引子形态为杂乱无章的随机波动且具有反持续性分形谱特征,其递归图具有漂移模式。在置信度为95%的小波影响锥范围内,校正信号、趋势项和噪声分量与原始信号分别具有持续正相关、局部负相关和无相关性特征。三类信号的有效分离和混沌分形特征提取为爆破信号成分的准确辨识和归类提供了客观表征和量化指标。Affected by the test environment,the monitoring vibration signals of a tunnel blasting generally contain noise and trend interference components.To eliminate the interference items,the distorted blasting signals detected in a typical tunnel were selected as the analysis objects.The approach of baseline estimation and denoising with sparsity(BEADS)was used to extract the noise and trend item and to obtain the calibrated signal that could reflect the true information.The chaotic fractal characteristics of the three components were captured by the multi-fractal detrended fluctuation analyses(MF-DFA),and the time-frequency domain correlations between them and the original signal were accurately characterized according to the wavelet correlation aggregation spectrum.The results show that the chaotic fractal characteristics of the high frequency noise,low frequency trend item and calibrated signal of the tunnel blasting are significantly different.The trajectory of the calibrated signal attractor is characterized by a fractal spectrum with persistent and anti-persistent periodic fluctuation and its recursion graph has periodic pattern.The attractors of the low-frequency trend item have the shape of approximate straight line and the persistent fractal spectrum characteristics,and its recursion graph has the mutation pattern of diagonal distribution.The attractor of the high frequency noise is characterized by random fluctuation and anti-persistent fractal spectrum,and its recursion graph has drift pattern.In the range of wavelet influence cone with confidence of 95%,the calibrated signal,trend item and noise have the characteristics of persistent positive correlation,local negative correlation,and no correlation with the original signal respectively.The effective separation of the three components and the extraction of chaotic fractal features provide objective characterization and quantified indexes for the identification and classification of blasting signal components.

关 键 词：爆破信号 稀疏化基线估计与去噪(BEADS) 多重分形去趋势波动分析(MF-DFA) 混沌特征 时频相关性 

分 类 号：TD235.1[矿业工程—矿井建设]