Data Analysis-Based Gas Identification with a Single Metal Oxide Sensor Operating in Dynamic Temperature Regime

Nicolas Morati (1) , Thierry Contaret (1) , Jean-Luc Seguin (1) , Marc Bendahan (1) , Oussama Djedidi (2) , Mohand Djeziri (2)

1 IM2NP - Institut des Matériaux, de Microélectronique et des Nanosciences de Provence
2 LIS - Laboratoire d'Informatique et des Systèmes (LIS) (Marseille, Toulon)

This paper deals with a preliminary evaluation of a new data-driven approach aiming to discriminate several gases in a gas mixture, with a single Metal Oxide (MOX) sensor, under real conditions of use, i.e. in the permanent presence of a gas concentration. This method relies on a database obtained using temperature modulation of the sensor, for fixed gas concentrations. The database is increased by extracting the temporal attributes of the signal, and then reduced using principal component analysis to extract only useful information. Gas discrimination is achieved by the online projection of measurements in the principal space consisting of the two main components. Experimental results obtained on a test bench show the effectiveness of the proposed approach: carbon monoxide, ozone and nitrogen dioxide are clearly discriminated using one single MOX gas micro sensor.

Mots clés: MOX gas sensor; temperature modulation; temporal attributes; multivariate analysis; gas discrimination

Domaines: Analyse de données, Statistiques et Probabilités [physics.data-an]; Instrumentations et Détecteurs [physics.ins-det]

Nicolas Morati, Thierry Contaret, Jean-Luc Seguin, Marc Bendahan, Oussama Djedidi, et al.. Data Analysis-Based Gas Identification with a Single Metal Oxide Sensor Operating in Dynamic Temperature Regime. ALLSENSORS 2020, The Fifth International Conference on Advances in Sensors, Actuators, Metering and Sensing, Nov 2019, Valencia, Spain. pp.20-23. ⟨hal-02575436⟩