Scholarlink

<p>Microseismic monitoring technology is an important method for evaluating the stability of rock masses underground in coal mines, and microseismic localization is an important part of this technology. To achieve both high accuracy and rapid speed in microseismic localization within coal mines, we combine supervised machine learning(SML) algorithm with 3D(three-dimensional) ray tracing method to construct a microseismic localization method that combines SML and 3D ray tracing(called SML-Tracing algorithms). Based on the layered velocity model and 3D ray tracing methods, a novel objective function is proposed for determining optimal ray travel times, and training and testing data for travel times required by the SML-Tracing algorithms are synthesized using the objective function. A Normalization and event adjustment method for minimum non-zero value is developed to adapt travel time training data to arrival time detection data. Based on this, the SML-Tracing algorithms is trained and tested, revealing that the Random Forest (RF)-Tracing algorithm within the SML-Tracing algorithms exhibits the best localization performance. It requires approximately 0.0136&#xa0;s to locate a single event and demonstrates strong robustness to velocity errors, travel time errors, and sensor quantities. Finally, coal mine blasting data is used as testing data to verify the localization accuracy of the RF-Tracing algorithm. The results indicate that the maximum localization error of the RF-Tracing algorithm for microseismic localization of blasting events in coal mines is 1.47&#xa0;m. This study provides valuable experience for fast and high-precision microseismic localization based on machine learning.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on microseismic source localization method combining supervised machine learning and 3-dimensional ray tracing method

  • 期刊论文

  • 发表日期：

    2026年6月24日

查看全文
  • Qi Ma,
  • Peng Chen,
  • Yunpeng Zhang,
  • Tiancheng Shan,
  • Kailong Qian,
  • Jiahui Du,
  • Xinke Chang,
  • Xue Li,
  • Shanxi Wu

摘要

Microseismic monitoring technology is an important method for evaluating the stability of rock masses underground in coal mines, and microseismic localization is an important part of this technology. To achieve both high accuracy and rapid speed in microseismic localization within coal mines, we combine supervised machine learning(SML) algorithm with 3D(three-dimensional) ray tracing method to construct a microseismic localization method that combines SML and 3D ray tracing(called SML-Tracing algorithms). Based on the layered velocity model and 3D ray tracing methods, a novel objective function is proposed for determining optimal ray travel times, and training and testing data for travel times required by the SML-Tracing algorithms are synthesized using the objective function. A Normalization and event adjustment method for minimum non-zero value is developed to adapt travel time training data to arrival time detection data. Based on this, the SML-Tracing algorithms is trained and tested, revealing that the Random Forest (RF)-Tracing algorithm within the SML-Tracing algorithms exhibits the best localization performance. It requires approximately 0.0136 s to locate a single event and demonstrates strong robustness to velocity errors, travel time errors, and sensor quantities. Finally, coal mine blasting data is used as testing data to verify the localization accuracy of the RF-Tracing algorithm. The results indicate that the maximum localization error of the RF-Tracing algorithm for microseismic localization of blasting events in coal mines is 1.47 m. This study provides valuable experience for fast and high-precision microseismic localization based on machine learning.