基于卷积神经网络的微地震事件识别方法研究

Research on Recognition Methods of Micro-seismic Events Based on Convolutional Neural Network

近些年来,科学技术的发展为社会带来了可观的收益。利用深度学习进行微地震事件识别也成为了一个研究热点。非常规油气勘探开发成为当前油气资源的主要途径,非常规勘探开发又需要微地震事件识别,针对微地震事件识别,主要解决的是快速、准确地检测微地震事件,这对石油勘探工作有着重大意义。为解决提取特征引入不确定性等缺点,论文利用雷克子波正演生成微地震信号数据再添加高斯嗓声进行模型研究。通过对构建数据集、搭建网络模型、评价模型输出结果等步骤,实现识别方法。经过反复试验与仿真实验,用卷积神经网络的方法可以对微地震有效信号快速准确地检测以及去掉冗余信息,提高微地震有效数据传输。

In recent years,the development of science and technology has brought considerable benefits to society.Mi⁃cro-seismic event identification using deep learning has also become a research hotspot.Unconventional oil and gas exploration and development has become the main way of current oil and gas resources.Unconventional exploration and development also requires micro-seismic event identification.For micro-seismic event identification,the main solution is to detect micro-seismic events quickly and accurately,which is of great significance to oil exploration work.In order to solve the shortcomings of introducing uncer⁃tainty in extracting features,in this paper,the micro-seismic signal data is generated by the forward modeling of the rake pre-wave,and the Gaussian voice is added for the model research,and the convolutional neural network is used to identify the mi⁃cro-seismic events.The identification method is realized through the steps of constructing a data set,building a network model,and evaluating the output results of the model.After repeated experiments and simulation experiments,the method of convolutional neu⁃ral network can quickly and accurately detect the effective signal of micro-seismic and remove redundant information,and improve the effective data transmission of microseismic.