Electromagnetic Noise Classification and Novelty Detection Using Machine Learning for Frequency Sharing among Various Communications in the Manufacturing Field
-
- MIYAMOTO Michio
- Advanced Telecommunications Research Institute International
-
- OHNISHI Ayano
- Advanced Telecommunications Research Institute International
-
- TAKEUCHI Yoshio
- Advanced Telecommunications Research Institute International
-
- MAEYAMA Toshiyuki
- Advanced Telecommunications Research Institute International,Takushoku University
-
- HASEGAWA Akio
- Advanced Telecommunications Research Institute International
-
- YOKOYAMA Hiroyuki
- Advanced Telecommunications Research Institute International
Bibliographic Information
- Other Title
-
- 製造現場における多用途周波数共用のための電磁ノイズの機械学習による分類と新規性検出
Abstract
The time waveform and spectral pattern of electromagnetic noise generated from industrial equipment in the manufacturing site differ depending on the equipment that is the source. Therefore, there is a high possibility that unknown patterns will occur when classifying by machine learning. Novelty detection can be used to detect unknown patterns. However, limited computational resources on the shop floor limit the parallel execution of electromagnetic noise classification and novelty detection. In this paper, we propose a method to detect unknown noise from posterior probability values obtained during classification, and demonstrate it using actual measurement data.
Journal
-
- 電子情報通信学会論文誌B 通信
-
電子情報通信学会論文誌B 通信 J105-B (9), 710-718, 2022-09-01
電子情報通信学会
- Tweet
Details 詳細情報について
-
- CRID
- 1390293202184348288
-
- ISSN
- 18810209
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
-
- Abstract License Flag
- Disallowed