Scholarlink

<p>Mining-induced seismicity is a big concern as geological conditions and rock mechanics play a crucial role in underground spaces. This paper focuses on the assessment of moment magnitude (<i>M</i><sub><i>w</i></sub>) from recorded seismic events at Laohutai coal mine in China using historical data to better understand seismic impact. We developed a self-organizing neural network (SONN) for a <i>M</i><sub><i>w</i></sub> assessment. To improve the accuracy of our results, we developed an optimized version APSO-SONN model by incorporating accelerated particle swarm–based metaheuristic optimization algorithm (APSO). Conventional AI models including multilayer perceptron (MLP), stochastic gradient boosting (SGB), support vector machine (SVM) were also applied and tuned to serve as benchmark for comparison with proposed APSO-SONN model. Results show that APSO-SONN model provides a very accurate <i>M</i><sub><i>w</i></sub> assessment with relative error from − 0.107 to 0.079%, MAE = 0.055, RMSE = 0.075, <i>R</i><sup>2</sup> = 0.965. This paper provides valuable insights into the application of AI techniques for the <i>M</i><sub><i>w</i></sub> assessment and contributes to hazard control and early warning system in mining area.</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

Application of an Optimized Self-Organizing Neural Network for Accurate Seismic Moment Magnitude Assessment in Mining: A Case Study of Laohutai Coal Mine, China

  • 期刊论文

  • 发表日期：

    2025年3月10日

查看全文
  • Tenglong Huang,
  • Yong Luo,
  • Hui Chen,
  • Yong Shi

摘要

Mining-induced seismicity is a big concern as geological conditions and rock mechanics play a crucial role in underground spaces. This paper focuses on the assessment of moment magnitude (Mw) from recorded seismic events at Laohutai coal mine in China using historical data to better understand seismic impact. We developed a self-organizing neural network (SONN) for a Mw assessment. To improve the accuracy of our results, we developed an optimized version APSO-SONN model by incorporating accelerated particle swarm–based metaheuristic optimization algorithm (APSO). Conventional AI models including multilayer perceptron (MLP), stochastic gradient boosting (SGB), support vector machine (SVM) were also applied and tuned to serve as benchmark for comparison with proposed APSO-SONN model. Results show that APSO-SONN model provides a very accurate Mw assessment with relative error from − 0.107 to 0.079%, MAE = 0.055, RMSE = 0.075, R2 = 0.965. This paper provides valuable insights into the application of AI techniques for the Mw assessment and contributes to hazard control and early warning system in mining area.