MSNet: A Seismic Phase Picking Network Applicable to Microseismic Monitoring

Jun 12,2025


IFEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 20, 2023 Chengyu Feng , Yang Yang , Xin Hu, Guixi Liu , and Kezhu Song

This paper introduces MSNet, a fully convolutional neural network designed for seismic phase picking in microseismic monitoring. MSNet improves upon existing models by using a dual-decoder structure and advanced training strategies. It is trained on the Stanford Earthquake Dataset (STEAD) and tested on microseismic data from a coal mine in China. MSNet achieves higher recall and precision rates compared to traditional methods like STA/LTA and PhaseNet, demonstrating its effectiveness in real-time microseismic monitoring.

Conclusion

The study demonstrates that MSNet, with its improved architecture and training strategy, effectively enhances the generalization performance for microseismic phase picking. The model shows superior recall and precision compared to existing methods like PhaseNet and STA/LTA. The results indicate that MSNet can be a valuable tool for real-time microseismic monitoring, providing accurate phase picking with lower computational cost and labor requirements. Future work could focus on further improving the model's performance in low SNR conditions and exploring multi-trace data processing techniques.
 

CopyRight:Hefei Guowei Electronics Co., Ltd.   Powered by 300.cn   SEO  

Business license