Scholarlink

<p>Earthquake classification is important for effective seismic signal analysis and enhances earthquake detection. This paper discusses the analysis and classification of seismic signals based on an entropy-based Wavelet Packet (WP) tree structure and a Self-Organizing Feature Map (SOFM). The significant attributes of the seismic signals are extracted from the WP tree structures, that were constructed using an entropy-based node selection (EBNS) algorithm. Four different types of best wavelet packet tree structures, namely, Shannon entropy-based best wavelet packet (SEBWP), Threshold entropy-based best wavelet packet (TEBWP), Log energy entropy-based best wavelet packet (LEBWP), and Sure entropy-based best wavelet packet (SUEBWP) tree structures, are proposed for the extraction of important features of seismic signals. The proposed classification method categories earthquake based on the given values. By using the Self-Organizing Feature Map (SOFM), dimensionality of the seismic signal features can be reduced allowing for the visualization of complex data in two or three dimensions and thereby facilitating analysis. Different earthquake datasets, namely the Chiba earthquake record having 20 events, the Fukushima earthquake record having 30 events, and the Hyuga-nada earthquake record having 9 events, were collected from the Building Research Institute (BRI) network website for the analysis purpose. Totally 64 earthquake events were collected from three channels with variable waveform lengths. The experimental results were evaluated based on their quality measurement metrics, such as Quantization Error (QE), Topographic Error (TE), Silhouette Score (SS), and Davis Bouldin Index (DBI). The experimental results highlight the importance of the proposed methodology over other traditional methods, such as Principal Component analysis (PCA), K-means clustering, Hierarchical clustering, and Fuzzy c-means clustering.</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

Analysis and classification of seismic signals based on wavelet packet tree structures and self-organizing feature map

  • 期刊论文

  • 发表日期：

    2025年1月3日

查看全文
  • T. Jayasree,
  • N. Malini

摘要

Earthquake classification is important for effective seismic signal analysis and enhances earthquake detection. This paper discusses the analysis and classification of seismic signals based on an entropy-based Wavelet Packet (WP) tree structure and a Self-Organizing Feature Map (SOFM). The significant attributes of the seismic signals are extracted from the WP tree structures, that were constructed using an entropy-based node selection (EBNS) algorithm. Four different types of best wavelet packet tree structures, namely, Shannon entropy-based best wavelet packet (SEBWP), Threshold entropy-based best wavelet packet (TEBWP), Log energy entropy-based best wavelet packet (LEBWP), and Sure entropy-based best wavelet packet (SUEBWP) tree structures, are proposed for the extraction of important features of seismic signals. The proposed classification method categories earthquake based on the given values. By using the Self-Organizing Feature Map (SOFM), dimensionality of the seismic signal features can be reduced allowing for the visualization of complex data in two or three dimensions and thereby facilitating analysis. Different earthquake datasets, namely the Chiba earthquake record having 20 events, the Fukushima earthquake record having 30 events, and the Hyuga-nada earthquake record having 9 events, were collected from the Building Research Institute (BRI) network website for the analysis purpose. Totally 64 earthquake events were collected from three channels with variable waveform lengths. The experimental results were evaluated based on their quality measurement metrics, such as Quantization Error (QE), Topographic Error (TE), Silhouette Score (SS), and Davis Bouldin Index (DBI). The experimental results highlight the importance of the proposed methodology over other traditional methods, such as Principal Component analysis (PCA), K-means clustering, Hierarchical clustering, and Fuzzy c-means clustering.