Scholarlink

<p>Identification and dominant partitioning of rock mass discontinuities are the basis for rock slope stability analysis. In this paper, a modified self-organizing mapping (SOM) neural network is proposed to automatically cluster the orientations of rock mass discontinuity sets. The new method uses the competitive mechanism network model and takes the winning neuron as the cluster center, which can obtain the global optimization. The negative sine-squared value of the acute angle(SSA) between the normal vectors of discontinuous is used instead of Euclidean distance as the similarity measurement for cluster analysis. The Silhouette validity index is introduced to determine the optimal clustering number. The new method is verified on artificial data sets and publish data sets, and the Precision, Recall and F1 value are innovatively introduced to analyze the accuracy of the new method. Finally, the method is applied to the discontinuity grouping of a rocky slope on Nujiang River in Southwest China. Meanwhile, the new method is compared with the classical KPSO clustering algorithm, FCM algorithm and spectral clustering algorithm. The results show that the new method has high accuracy and good clustering results with stronger 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

A new method of rock discontinuity sets using modified self-organizing mapping neural network

  • 期刊论文

  • 发表日期：

    2025年1月17日

查看全文
  • Mingzhe Zhou,
  • Haiying Fu,
  • Yanyan Zhao,
  • Yangli Zhou,
  • Tao Yang,
  • Wangming Huang,
  • Xiongwei Hu

摘要

Identification and dominant partitioning of rock mass discontinuities are the basis for rock slope stability analysis. In this paper, a modified self-organizing mapping (SOM) neural network is proposed to automatically cluster the orientations of rock mass discontinuity sets. The new method uses the competitive mechanism network model and takes the winning neuron as the cluster center, which can obtain the global optimization. The negative sine-squared value of the acute angle(SSA) between the normal vectors of discontinuous is used instead of Euclidean distance as the similarity measurement for cluster analysis. The Silhouette validity index is introduced to determine the optimal clustering number. The new method is verified on artificial data sets and publish data sets, and the Precision, Recall and F1 value are innovatively introduced to analyze the accuracy of the new method. Finally, the method is applied to the discontinuity grouping of a rocky slope on Nujiang River in Southwest China. Meanwhile, the new method is compared with the classical KPSO clustering algorithm, FCM algorithm and spectral clustering algorithm. The results show that the new method has high accuracy and good clustering results with stronger robustness.