论文详情
在水力压裂过程中, 常规微地震事件检测方法中基于能量的长短时窗能量比值(STA/LTA)方法检测精度低, 而基于波形的模板匹配方法检测速度慢。为此, 提出了以指纹和相似性阈值(FAST)方法为主, 联合同态反褶积去噪、STA/LTA、带噪声的基于密度的聚类(DBSCAN)方法的一种检测精度高并且检测速度快的微地震事件检测方法。首先使用同态反褶积方法对微地震数据去噪; 然后利用STA/LTA方法获取高信噪比微地震事件并作为模板, 进一步利用FAST方法将模板和连续波形制作为指纹, 再通过比较指纹的杰卡德相似度检测低信噪比微地震事件, 得到各个台站的P波初至时间; 最后利用DBSCAN方法将多个台站同一震相进行关联以去除错误检测。利用人工合成的171个具有不同信噪比的微地震事件, 运用该方法能够检测到所有微地震事件, 验证了方法的实用性。对四川盆地威远页岩气开发水平井2014年11月10号第19级压裂段的井下微地震数据集进行处理, 并与模板匹配方法、STA/LTA方法进行了对比, 结果显示该方法能够检测到STA/LTA检测不到的低信噪比微地震事件, 检测结果与模板匹配方法相近, 其计算效率比模板匹配方法高。
Two methods commonly used in microseismic event detection during hydraulic fracturing:short term averaging / long term averaging (STA/LTA) algorithm based on energy and template matching based on waveform similarity, have the problems of low accuracy and low speed of detection, respectively.Thus, we establish a technique which integrates fingerprinting and similarity thresholding (FAST) with homomorphic deconvolution for noise reduction, STA/LTA, and density-based spatial clustering of applications with noise (DBSCAN) for fast microseismicity detection with high precision.After microseismic data denoising through homomorphic deconvolution and STA/LTA detection of microseismic events with high signal-to-noise ratios which will be taken as the templates, we use FAST for template and continuous waveform fingerprinting and then examine the Jaccard similarity of fingerprints to check out those events with low signal-to-noise ratios and obtain first arrivals of P-waves at each station.DBSCAN is finally utilized for the correlation of the same microseismic facies at different stations to eliminate false events.According to the model study, we successfully detect all the 171 synthetic microseismic events with different signal-to-noise ratios using our method.We also compare our method with template matching and STA/LTA in handling microseismic data acquired from the 19th stage of a horizontal well fractured on November 102014 for shale gas production in Weiyuan, the Sichuan Basin.Our method is superior to STA/LTA in the detection of microseismic events with low signal-to-noise ratios and template matching in computational efficiency.