Pipeline monitoring using acoustic principal component analysis recognition with the Mel scale

Chunfeng Wan, Akira Mita

研究成果: Article査読

12 被引用数 (Scopus)


In modern cities, many important pipelines are laid underground. In order to prevent these lifeline infrastructures from accidental damage, monitoring systems are becoming indispensable. Third party activities were shown by recent reports to be a major cause of pipeline damage. Potential damage threat to the pipeline can be identified by detecting dangerous construction equipment nearby by studying the surrounding noise. Sound recognition technologies are used to identify them by their sounds, which can easily be captured by small sensors deployed along the pipelines. Pattern classification methods based on principal component analysis (PCA) were used to recognize the sounds from road cutters. In this paper, a Mel residual, i.e.the PCA residual in the Mel scale, is proposed to be the recognition feature. Determining if a captured sound belongs to a road cutter only requires checking how large its Mel residual is. Experiments were conducted and results showed that the proposed Mel-residual-based PCA recognition worked very well. The proposed Mel PCA residual recognition method will be very useful for pipeline monitoring systems to prevent accidental breakage and to ensure the safety of underground lifeline infrastructures.

ジャーナルSmart Materials and Structures
出版ステータスPublished - 2009

ASJC Scopus subject areas

  • 信号処理
  • 土木構造工学
  • 原子分子物理学および光学
  • 材料科学一般
  • 凝縮系物理学
  • 材料力学
  • 電子工学および電気工学


「Pipeline monitoring using acoustic principal component analysis recognition with the Mel scale」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。