Paper Abstract and Keywords |
Presentation |
Estimation of CO2 Concentration Using MOX Sensors and Neural Network Harunobu Taguchi, Takuya Sano (NITTC), Shinichi Kondo, Koji Takahashi, Yutaka Tamura, Toshimitsu Kitamura (Toshiba Info. Sys. Co.), Shinichiro Mito (NITTC) |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
A visualization of indoor environments by measuring CO2 concentration has been attracting attention, because CO2 causes drowsiness and fatigue. CO2 concentration is a key indicator for the ventilation, which is the effectual countermeasure for COVID-19. NDIR sensors are usually used for measuring CO2 concentration, but they are not very popular at present due to their high price. Therefore, we focused on inexpensive MOX sensors that can measure the concentration of organic compounds including CO2. In this study, we estimated CO2 value from the output of the MOX sensor using supervised machine learning. The CO2 concentration of the classrooms were estimated by MLP and RNN. The root mean squared error of the MLP and the RNN estimation were 299 ppm and 334 ppm respectively. This result would expand use of CO2 measurement that make indoor environment more comfortable and safe. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Machine Learning / CO2 Sensor / MLP / RNN / Indoor Air Quality / IoT / COVID-19 / |
Reference Info. |
IEICE Tech. Rep. |
Paper # |
|
Date of Issue |
|
ISSN |
|
Download PDF |
|