基于波形聚类分析的微地震事件成像研究

Research on microseismic event imaging based on waveform clustering analysis

微地震数据的质量极大地影响震源定位结果的准确性,尤其在地面微地震数据中P波、S波初至不明显或难以识别的低信噪比信号,基于初至拾取的定位方法往往具有较大误差。由于诱发地震的震源机制不同,不同检波器接收到的信号特征也不尽相同,这些因素都会对微地震事件准确定位造成一定的影响。本文提出一种基于聚类分析的微地震数据优选方法,以提高微地震事件振幅叠加定位的成像质量。首先,通过聚类分析计算各数据道之间的欧式距离,通过聚类图去除距离较大和较小的数据道,即去除信号特征不明显或噪声干扰严重数据道;其次,对各道信号进行互相关计算,提取信号特征相关性强数据道,去除P波特征不明显的数据道;最后,对筛选数据道进行振幅叠加定位。与原始数据道的定位结果对比表明,微地震聚类分析定位效果更加聚焦,最大震源点位置更加清晰,定位结果得到明显改善。

The quality of microseismic data greatly affects the accuracy of the source location results,especially in surface microseismic data where the first arrivals of P waves and S waves are not obvious or the low signal-to-noise ratio signals are difficult to identify.Secondly,due to the different focal mechanisms that induce microseismic,the signal characteristics received by different geophones are not the same.All of these factors have a certain impact on the accurate location of microseismic events.This paper proposes a method for selecting microseismic data based on cluster analysis to improve the imaging quality of microseismic event amplitude superposition locating.First,the Euclidean distance between the trace channels is calculated by cluster analysis,and the trace channels with larger and smaller distances are removed through the clustering graph,that is,trace channels with insignificant signal characteristics or severe noise interference are removed.Second,the cross-correlation calculation is performed on the signals of each channel,the trace channels with strong signal characteristic correlation being extracted,and those with unobvious P wave characteristics removed.Finally,the screened trace channels are su-perimposed and located with amplitude superposition.The comparison of the locating results with the o-riginal trace channels shows that the locating effect of the microseismic cluster analysis is more focused,the location of the focal point is clearer,and the locating results are significantly improved.