Real-Time Identification of Rock Failure Stages Using Deep Learning: A Case Study of Acoustic Emission Analysis in Rock Engineering

Author:

Pu Yuanyuan1,Chen Jie1,Apel Derek B.2,Shang Xueyi1

Affiliation:

1. Chongqing University

2. University of Alberta

Abstract

Abstract

The accurate and timely determination of rock failure processes is crucial for various rock engineering applications, especially for preventing dynamic disasters such as rock bursts and roof failures. The primary aim of this study was to determine the current rock failure stage using a single acoustic emission (AE) event signal recorded during the failure process. To achieve this, we proposed a deep learning model that employs advanced convolutional modules and a soft-threshold technique to extract the full waveform features of AE events from four different stages of rock failure in a laboratory uniaxial compressive strength (UCS) test. Once fully trained, our model can instantaneously determine the current rock failure stage from a raw waveform of a single recorded AE event. Subsequently, the trained model was applied to on-site microseismic data analysis at a coal mine working face. Compared to traditional methods of microseismic data analysis that consider large-energy events, our model can identify the rock failure stage at the time of a specific microseismic event. Furthermore, by analyzing microseismic events triggered by post-peak rock fracturing, we identified potential hazard areas for rock bursts in the working face, and the results closely matched the site's burst prevention logs. This study successfully developed a real-time method for determining rock failure stages using deep learning, which can be effectively applied to microseismic data analysis in engineering sites to provide more precise early warnings of rock dynamic disasters.

Publisher

Springer Science and Business Media LLC

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3