音を用いた開閉器の状態監視手法の開発

カテゴリ: 全国大会

論文No: 6-044

グループ名: 【全国大会】令和6年電気学会全国大会論文集

発行日: 2024/03/01

タイトル(英語): Development of Condition Monitoring Method for Contactor Using Sound

著者名: 藤原海渡（東海旅客鉄道）,井上雄仁（東海旅客鉄道）,榎本淳（東海旅客鉄道）

著者名(英語): Kaiot Fujiwara (Central Railway),Katsuhito Inoue (Central Railway),Jun Enomoto (Central Railway)

キーワード: 機械学習|状態監視|動作音解析|Machine learning|Condition monitoring|Acoustic analysis of operating sound

要約(日本語): Along th route of the Chuo Shinkansen, a large number of contactors are installed, which requires a significant amount of effort for maintenance. In order to reduce maintenance costs, a condition monitoring method has been developed to diagnose faults from the operation sound. This paper compared the accuracy of MFCC and Mel spectrogram, which are commonly used as acoustic features. It was confirmed that using Mel spectrogram enables more accurate anomaly detection.

本誌掲載ページ: 56-58 p

原稿種別: 日本語

PDFファイルサイズ: 593 Kバイト