Identification of CO and NO2 using a thermally resistive microsensor and support vector machine

Abstract

Levels of carbon monoxide and nitrogen dioxide in air are currently monitored using two different thick-film resistive gas sensors. The resultant high power consumption of thick-film-based gas sensors is problematic for portable multi-gas monitors. The use of a single low-power thermally-modulated resistive gas sensor to monitor simultaneously both gases is reported. The silicon micromachined Substrate not only reduces the DC power consumption to 100mW Lit 300degreesC but also permits AC temperature modulation through a small thermal mass. Uniquely, a Support vector machine is employed to classify the wavelet coefficients of the AC resistive signal. This simple method permits the rapid classification of CO/NO2, gas mixtures with a high level of confidence (94% or better) using just one low-power gas microsensor. Thus demonstrating the potential application of a single low-power thermally-modulated resistive gas sensor in portable multi-gas monitors.