Scholarlink

<p>The morphology characteristics of structural plane have a significant impact on the mechanical properties of fractured rock mass and the fluid flow characteristics within rock mass. To ensure the safety and stability of large-scale rock engineering projects, it is necessary to conduct in-depth research on the morphology characteristics of structural plane and their quantitative methods. In this study, systematic experimental research and theoretical analysis were conducted to address the shortcomings in existing studies, such as the lack of rigor in the <i>JRC</i> values of profiles and the inadequacy of existing quantitative methods in fully reflecting the surface morphology of structural plane. The eight most representative parameters were identified from hundreds of morphology parameters by examining their physical significance, frequency in existing literature and correlation with the roughness coefficient <i>JRC</i>. The limitations of existing profile-<i>JRC</i> datasets were discussed in detail by analyzing the profiles and their corresponding <i>JRC</i> values, the source, lithology and dimension of each specimen, the extraction method and length of each profile and the quantitative methods of profile <i>JRC</i> values. The entropy weight-TOPSIS method was adopted to rank the extensive profiles extracted from 85 sandstone specimens according to roughness level. Hundred profiles were selected at equal intervals. Based on the 100 profiles, the CNC engraving technology was applied to manufacturing structural plane specimens. Subsequently, direct shear test was conducted. The <i>JRC</i> value of each profile was inversely calculated via the <i>JRC</i>-<i>JCS</i> model, according to which a profile-morphology parameter-<i>JRC</i> dataset was compiled. On this basis, the relationship between representative morphology parameters and <i>JRC</i> values was established. Factor analysis was performed on the eight representative morphology parameters of the 100 profiles to eliminate overlapping information. A PSO-RBF neural network-based model was constructed to evaluate the morphology characteristics of structural plane, with the common factors of morphology parameters as input and profile <i>JRC</i> values as output.</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

Assessment Method of Surface Morphology Based on 100 Shear Tests and PSO-RBF Neural Network

  • 期刊论文

  • 发表日期：

    2025年4月28日

查看全文
  • Yongchao Tian,
  • Yujie Wang,
  • Quansheng Liu,
  • Zhicheng Tang,
  • Bin Liu,
  • Yucong Pan,
  • Penghai Deng,
  • Xing Huang,
  • Shuang Gong,
  • Shuxue Ding

摘要

The morphology characteristics of structural plane have a significant impact on the mechanical properties of fractured rock mass and the fluid flow characteristics within rock mass. To ensure the safety and stability of large-scale rock engineering projects, it is necessary to conduct in-depth research on the morphology characteristics of structural plane and their quantitative methods. In this study, systematic experimental research and theoretical analysis were conducted to address the shortcomings in existing studies, such as the lack of rigor in the JRC values of profiles and the inadequacy of existing quantitative methods in fully reflecting the surface morphology of structural plane. The eight most representative parameters were identified from hundreds of morphology parameters by examining their physical significance, frequency in existing literature and correlation with the roughness coefficient JRC. The limitations of existing profile-JRC datasets were discussed in detail by analyzing the profiles and their corresponding JRC values, the source, lithology and dimension of each specimen, the extraction method and length of each profile and the quantitative methods of profile JRC values. The entropy weight-TOPSIS method was adopted to rank the extensive profiles extracted from 85 sandstone specimens according to roughness level. Hundred profiles were selected at equal intervals. Based on the 100 profiles, the CNC engraving technology was applied to manufacturing structural plane specimens. Subsequently, direct shear test was conducted. The JRC value of each profile was inversely calculated via the JRC-JCS model, according to which a profile-morphology parameter-JRC dataset was compiled. On this basis, the relationship between representative morphology parameters and JRC values was established. Factor analysis was performed on the eight representative morphology parameters of the 100 profiles to eliminate overlapping information. A PSO-RBF neural network-based model was constructed to evaluate the morphology characteristics of structural plane, with the common factors of morphology parameters as input and profile JRC values as output.