岩石破坏声发射时频分析算法与瞬时频率前兆研究  

RESEARCH ON TIME-FREQUENCY ANALYSIS ALGORITHM AND INSTANTANEOUS FREQUENCY PRECURSORS FOR ACOUSTIC EMISSION DATA FROM ROCK FAILURE EXPERIMENT

作　　者：宫宇新[1] 何满潮[2,3] 汪政红[1] 尹雨婷[1] 

机构地区：[1]中南民族大学数学与统计学学院,湖北武汉430074 [2]中国矿业大学深部岩土力学与地下工程国家重点实验室,北京100083 [3]中国矿业大学力学与建筑工程学院,北京100083

出　　处：《岩石力学与工程学报》2013年第4期787-799,共13页Chinese Journal of Rock Mechanics and Engineering

基　　金：国家重点基础研究发展计划(973)项目(2006CB202200);国家级大学生创新创业训练计划项目(GCX12031)

摘　　要：为得到以时间、频率、幅度和动态过程为特征量的多维度瞬时频率前兆信息,为岩爆等地质灾害的预测提供原理上的支持,开展岩石破坏声发射波形分析的研究。以花岗岩单轴压缩为实例,对声发射大数据的分割、时频分析理论与算法,以及时频分析窗函数的滤波性能的优化进行了理论分析与计算机实验。建立分割短数据的时间/频率分辨率的计算公式与大数据优化分割的基本原则;阐明时频分析窗宽度与频率域指标的优选方法;提出时频分析的综合优化算法,即:先用对声发射大数据进行最优分割、再根据短数据的特点对分析窗进行优选、用优化的短时傅里叶变换进行主频带定位、再用维格纳变换进行主频带中心频率的识别、最后对变换后的时频图进行精细化处理,以充分展现其中蕴含的动态信息。时频分析综合优化算法的应用表明:分割的短数据较好地反映了在不同应力水平下裂纹扩展的动态过程;分析窗的优化提高了时频变换的频率局部化性能并减少了频谱泄漏效应;解决了短时傅里叶变换不能精确描述主频分布的问题,避免了维格纳变换中"交叉项"的识别问题。应用多维度瞬时频率前兆可以更好地描述岩石破坏非线性过程的前兆信息,为岩石破坏的预测与机制认识提供新的、有力的手段。In order to obtain the instantaneous frequency precursors containing such information as time,frequency,amplitude and dynamical information,which are the basis for rockburst prediction,wave-form based investigation on acoustic emission（AE） was carried out.Taking the uniaxially loaded granite specimen as a practical case,theoretical and computational analyses were conducted on the segmentation of the large AE data,theory and algorithm of the time-frequency transform and optimization of filtering performance for the analysis window functions.Such analytical or computational based outcomes were obtained as the formulas for assessment of temporal and spectral resolutions of the segmented data,principle for optimized segmentation of the large AE data,optimal design of window width and spectral parameters for the analysis window function,and establishment of the comprehensive optimal time-frequency analysis algorithm involving：（1） the optimal segmentation of the large AE data;（2） optimal design of the analysis window function;（3） locating the dominant frequency band using short-time Fourier transform（STFT）;（4） determining exactly the central frequencies for the dominant frequency band using Wigner-Ville distribution（WVD）,and enhancement of the time-frequency contour graphs for a detailed representation of the dynamical information.Results from application of the proposed algorithm in analysis of the case demonstrate that the segmented data sets match well with the dynamical process at which fractures propagate at specific stress level;the increase in frequency localization performance and reduction of spectral leaks;good ability in accurately determining central frequency of the dominant frequency band obtained by STFT and avoiding identification of the＂cross terms＂in WVD.The instantaneous frequency precursors containing the dynamical information can better describe the precursory information for nonlinear process of the rock failure and provide a new,robust means for rock failure pred

关 键 词：岩石力学 时频分析 前兆信息 声发射 短时傅里叶变换 维格纳变换 

分 类 号：TU45[建筑科学—岩土工程]