基于谱矩心和线性度的微震事件分类

Classification of Micro earthquakes based on Spectral Centroid and Rectilinearity

岩体破裂产生的应力波触发微震事件,传统的互相关匹配难以识别微震事件深浅类型。本文利用谱矩心和线性度提取地震波的特征信息,并通过多层感知器网络进行分类识别,将微震事件分为浅源和深源。实验结果表明该方法的识别准确率为86.14%,谱矩心对微震事件分类精度高于线性度,且精度均高于传统的互相关方法。该方法不仅可以识别地震波形,也可为岩爆、滑坡等动力灾害监测提供预警信息。

Micro earthquakes,triggered by stress wave from rock mass fracture,are difficult to be identified using traditional cross-correlation method.In this paper,we propose a method to discriminate the depth of micro earthquake,in which one can use spectral centroid and rectilinearity to extract the features of seismicity and implement the multilayer perceptron to classify the micro earthquake.The results indicate that the accuracy of combination of the two features is 86.14%.For the solely classification,the result by spectral centroid is more accurate than that by rectilinearity.In addition,comparing with the cross-correlation,the feature-based classification in this paper is more effective.This method not only offers guidance for seismic events identification,but also provides early warning information for rock burst,landslide and other disaster monitoring.