Identification method for microseismic, acoustic emission and electromagnetic radiation interference signals of rock burst based on deep neural networks
Yangyang Di,Enyuan Wang,Tao Huang
DOI: https://doi.org/10.1016/j.ijrmms.2023.105541
2023-07-22
International Journal of Rock Mechanics and Minings Sciences
Abstract:Microseismic, acoustic emission, and electromagnetic radiation (MS, AE, and EMR) are promising methods for the monitoring and early warning of rock bursts. In the underground mining process, personnel activities and electromechanical equipment produce MS, AE, and EMR interference signals that affect the accuracy of MS, AE, and EMR monitoring. This study is aimed at overcoming the difficulties in identifying MS, AE, and EMR interference signals by means of on-site monitoring and deep learning methods. First, MS waveform signals and AE and EMR time series signals are collected by the SOS monitoring system and GDD12 monitor. According to the field records, the MS effective waveform, MS interference waveform, AE and EMR interference signals were labeled, and training sets and verification sets were made. An identification model for MS waveforms based on the ResNet-50 convolutional neural network and an identification model for AE and EMR interference signals based on the recurrent neural network were constructed using the original MS, AE, and EMR data. In addition, the identification accuracy and generalization ability of the models were analyzed. The results show that the proposed method can respond positively to MS, AE, and EMR interferences and accurately eliminate MS, AE, and EMR interference signals. It can significantly improve the reliability of MS, AE, and EMR monitoring data and effectively monitor rock burst disasters.