講演抄録/キーワード |
講演名 |
2018-10-31 10:25
[ポスター講演]A Comparison of Machine Learning Algorithms for Motor Sound Fault Detection ○Arpith Paida(AIT)・Prerapong・Aimaschana Niruntasukrat・Koonlachat Meesublak・Panita(NECTEC) |
抄録 |
(和) |
Automation plays important role in order to make human activities easier. In industries, machines /motors are used for manufacturing and maintenance of its condition is very important. Its condition can be checked by its operating sound. It is difficult for humans to monitor the machines/motors every time. Hence detecting the fault in motor sounds by using machine learning algorithms will reduce human resource and indicate the faults more accurately and spontaneously. In this paper, the sounds are collected in form of wav files. Features are extracted using Signal processing techniques, which are later used for training machine learning algorithms. In this paper, the algorithms are evaluated in terms of accuracy. Support Vector Machine and K-Nearest Neighbors are the optimal algorithms resulted after evaluation. |
(英) |
Automation plays important role in order to make human activities easier. In industries, machines /motors are used for manufacturing and maintenance of its condition is very important. Its condition can be checked by its operating sound. It is difficult for humans to monitor the machines/motors every time. Hence detecting the fault in motor sounds by using machine learning algorithms will reduce human resource and indicate the faults more accurately and spontaneously. In this paper, the sounds are collected in form of wav files. Features are extracted using Signal processing techniques, which are later used for training machine learning algorithms. In this paper, the algorithms are evaluated in terms of accuracy. Support Vector Machine and K-Nearest Neighbors are the optimal algorithms resulted after evaluation. |
キーワード |
(和) |
Audio detection / Spectral features / Feature extraction / Supervised learning / Unsupervised learning / Accuracy / / |
(英) |
Audio detection / Spectral features / Feature extraction / Supervised learning / Unsupervised learning / Accuracy / / |
文献情報 |
信学技報 |
資料番号 |
|
発行日 |
|
ISSN |
|
査読に ついて |
本技術報告は査読を経ていない技術報告であり,推敲を加えられていずれかの場に発表されることがあります. |
PDFダウンロード |
|