Online learning for active odor sensing system based on a sensor array

カテゴリ: 研究会(論文単位)

論文No: CHS22005,BMS22010

グループ名: 【E】センサ・マイクロマシン部門 ケミカルセンサ/【E】センサ・マイクロマシン部門 バイオ・マイクロシステム合同研究会

発行日: 2022/02/28

タイトル(英語): Online learning for active odor sensing system based on a sensor array

著者名: Aleixandre Manuel(東京工業大学),中本 高道(東京工業大学)

著者名(英語): Manuel Aleixandre(Tokyo Institute of Technology),Takamichi Nakamoto(Tokyo Institute of Technology)

キーワード: Active Sensing|Gas sensors|Quartz Crystal Microbalances|Adaptive Control|Odor blender|Artificial odor perception

要約(日本語): In active sensing the parameters of the measurement are changed to optimize the measurement or extract some additional information. In this work the composition of a gas mixture, produced by an odor blender, is adjusted to match that of target odor, thus odor reproduction is achieved. Drift and environmental conditions can modify the sensor characteristics with the parameters required for active sensing algorithms. In this paper we examine an online training algorithm that modifies the calibration of the sensor characteristics during the active sensing. This allows a higher stability against ambient conditions and eliminates frequent recalibration.

要約(英語): In active sensing the parameters of the measurement are changed to optimize the measurement or extract some additional information. In this work the composition of a gas mixture, produced by an odor blender, is adjusted to match that of target odor, thus odor reproduction is achieved. Drift and environmental conditions can modify the sensor characteristics with the parameters required for active sensing algorithms. In this paper we examine an online training algorithm that modifies the calibration of the sensor characteristics during the active sensing. This allows a higher stability against ambient conditions and eliminates frequent recalibration.

本誌: 2022年3月3日ケミカルセンサ/バイオ・マイクロシステム合同研究会

本誌掲載ページ: 17-22 p

原稿種別: 英語

PDFファイルサイズ: 672 Kバイト