基于深度学习的地面微地震P波极性分类方法研究

微地震监测技术是监测水力压裂过程及评价压裂效果的重要手段。对于地面监测，P波极性可以直接、快速地反演震源机制，同时极性校正能够提高绕射叠加定位方法的精度。因此，准确而迅速地确定P波极性对地面微地震实时监测具有重要意义。卷积神经网络是一种深度学习算法，具有强大的特征学习与分类能力，通过学习大量已标定极性的训练样本，可准确高效地进行极性分类。由于地面监测多采用规则观测系统，本项目拟使用多个相邻检波器记录作为训练样本，构建P波极性分类网络，使其能够同时学习P波极性特征和相邻道的极性分布规律。首先研究基于分类模型的多道记录标定方法，并构建一个混合算法将微地震事件检测与极性分类相融合，以实现直接对连续波形进行极性分类；然后将两套实际数据训练获得的模型应用于另外一个区域的数据上，以测试模型的泛化能力。本项目预期将获得准确高效且适用性较强的P波极性分类模型，可为地面微地震实时监测提供技术支撑。

Microseismic monitoring technology is an important method for hydraulic fracturing process monitoring and fracturing effect evaluation. For surface monitoring, the P-wave polarity can directly and quickly invert the focal mechanism, while the polarity correction can also improve the imaging accuracy of the diffraction-based location method. Therefore, accurate and rapid determination of P-wave polarity is of great significance for real-time surface monitoring. Convolutional neural network is a deep learning algorithm with powerful feature learning and classification capabilities. By learning a large number of polarized training samples, it can accurately and efficiently implement polarity classification. Because surface microseismic monitoring mostly uses star, grid, or other regular acquisitions, this project intends to use several neighboring seismograms as a training sample to build a P-wave polarity classification network so that it can simultaneously learn P-wave polarity characteristics and the polarity information of neighbor receivers of the receiver array. First, a multi-trace labeled method based on a classification model will be studied, and a hybrid algorithm will be constructed to combine microseismic event detection and polarity classification to directly perform polarity classification on continuous waveforms; then the model obtained by training on two field data will be applied to the other field data to test the generalization of the model. This project is expected to obtain an accurate, efficient and applicable P-wave polarity classification model, which can provide technical support for real-time monitoring of surface microseismic.