Gas concentration estimation in ternary mixtures with room temperature operating sensor array using tapped delay architectures

In this work, we present a hybrid multisensor system equipped with sensor fusion architectures for continuous gas concentration estimation, in a gas mixture scenario. Complex gas mixtures pose significant issues to quantification capability of room temperature operating sensors due to the often poor stability and selectivity of this class of devices. In this work, we show how these problems can take advantage by the use of pattern recognition algorithms. Here, the use of ad-hoc sensor fusion algorithms based on neural networks and support vector machines, together with an array of heterogeneous, room temperature operating sensors, is investigated to enhance array performances. Results obtained by different architectures for individual analyte quantification in NO2–NH3–humid air mixtures are presented and discussed.