基于经验小波变换的矿山微震信号识别研究

针对微震信号与爆破震动信号自动识别难的问题，提出了基于经验小波变换(EWT)的矿山微震信号识别方法。运用仿真信号对EWT和经验模态分解(EMD)进行对比检验，表明EWT分解效果要优于EMD，而且可以减少模态混叠问题；对矿山实测的400组爆破震动和微震信号进行EWT分解，得到紧支集频谱的内禀模态分量，借助互信息量筛选得到f1~f7共7个分量，进而分别利用分量f1~f7构建Hankel矩阵，计算每个Hankel矩阵的奇异值平均值、方均根值、标准差，并作为特征量；利用支持向量机(SVM)对微震和爆破震动信号进行分类。结果表明：爆破震动信号分量f1~f7的奇异值方均根值和标准差都要大于微震信号，分量f1~f5的奇异值平均值要大于微震信号；EWT_Hankel_SVD特征提取法识别效果要优于应用较为广泛的EWT_SVD，且基于EWT_Hankel_SVD分类准确率达到92.5%。

Research on mine microseismic signal recognition based on empirical wavelet transform

In view of the problem that it is difficult to identify microseismic signal and blasting vibration signal automatically, a method based on empirical wavelet transform (EWT) for mine microseismic signal identification is proposed. Firstly, using the simulated signal to compare EWT and empirical mode decomposition (EMD), the results showed that the EWT decomposition effect was better than the EMD, and the modal aliasing problem can be reduced; after that, the EWT decomposition of the measured 400 sets of blasting vibration and microseismic signals was carried out, and the intrinsic modal components of the compact supported spectrum were obtained, the seven principal components from f1 to f7 were obtained by mutual information filtering, and then the 7 components were respectively utilized to construct Hankel matrix, and the singular value mean, root mean square value, and standard deviation of each Hankel matrix were calculated and used as feature quantities; finally, microseismic and blasting vibration signals were classified by support vector machine (SVM). The results show that the singular value square root and standard deviation of the blasting vibration signal components from f1 to f7 are larger than the microseismic signal, and the singular value of the component from fl to f5 is larger than the microseismic signal; in terms of the recognition effect, EWT_Hankel_SVD feature extraction method is better than the widely used EWT_SVD, and the classification accuracy rate based on EWT_Hankel_SVD reaches 92.5%.