Scholarlink

<p>The geotechnical characteristics of strata are crucial indicators for geotechnical investigations and design. To overcome the existing challenges in this field, such as low automation levels, underutilization of borehole data, and difficulties in identifying lithology during the drilling process, this study developed a monitoring-while-drilling system, and proposed an intelligent data processing scheme. Onsite tests were conducted in Chongqing and Guangdong, China, collecting substantial effective borehole data. The particle swarm optimization (PSO) algorithm was used to optimize the parameters of the kernel function of the support vector machine (SVM), ensuring the best learning and prediction performance. The particle swarm optimization support vector machine (PSO–SVM) algorithm successfully identified three rock types (sandstone, sandy mudstone, and limestone) with an accuracy of 87.5% and accurately predicted uniaxial compressive strength (UCS) and rock indentation hardness. Furthermore, the pull-wire displacement sensor demonstrated superior stability and cost-effectiveness as compared to the laser displacement sensor, making it a more favorable option for engineering applications. Abrupt changes in drilling rate and oil pressure were identified as critical indicators of changes in rock type or strength, offering a practical approach for real-time stratigraphic identification. The proposed monitoring-while-drilling system and PSO–SVM algorithm provide significant benefits for geological drilling by enhancing automation and intelligence. Future research will focus on extending the application of this system and algorithm to other rock types and geological settings to further validate their effectiveness and robustness.</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

Development and Application of a Monitoring-While-Drilling System with an Optimized Machine Learning Algorithm for Lithology Identification and Rock Strength Prediction

  • 期刊论文

  • 发表日期：

    2025年6月16日

查看全文
  • Minglong You,
  • Fei Tan,
  • Yu Zhang,
  • Danjie Sheng,
  • Changqun Zuo,
  • Yuyong Jiao

摘要

The geotechnical characteristics of strata are crucial indicators for geotechnical investigations and design. To overcome the existing challenges in this field, such as low automation levels, underutilization of borehole data, and difficulties in identifying lithology during the drilling process, this study developed a monitoring-while-drilling system, and proposed an intelligent data processing scheme. Onsite tests were conducted in Chongqing and Guangdong, China, collecting substantial effective borehole data. The particle swarm optimization (PSO) algorithm was used to optimize the parameters of the kernel function of the support vector machine (SVM), ensuring the best learning and prediction performance. The particle swarm optimization support vector machine (PSO–SVM) algorithm successfully identified three rock types (sandstone, sandy mudstone, and limestone) with an accuracy of 87.5% and accurately predicted uniaxial compressive strength (UCS) and rock indentation hardness. Furthermore, the pull-wire displacement sensor demonstrated superior stability and cost-effectiveness as compared to the laser displacement sensor, making it a more favorable option for engineering applications. Abrupt changes in drilling rate and oil pressure were identified as critical indicators of changes in rock type or strength, offering a practical approach for real-time stratigraphic identification. The proposed monitoring-while-drilling system and PSO–SVM algorithm provide significant benefits for geological drilling by enhancing automation and intelligence. Future research will focus on extending the application of this system and algorithm to other rock types and geological settings to further validate their effectiveness and robustness.