Assessment of VOCs in air using sensor array under various exposure conditions

Sensor array exposure conditions were examined in this work regarding their influence on the assessment of volatile organic compounds (VOCs) in air. Measurements were performed using sensor array composed of fifteen TGS sensors. Eight VOCs were considered together with air featured by different humidity levels. It was shown that misclassification rates of VOCs patterns could be reduced to zero by selecting best conditions of exposure and by considering responses of selected sensors in these conditions as the basis for classification. Combinations of best sensors and best exposure conditions allowed to achieve mean relative error of VOCs concentration prediction at the level of several percent. The considerable improvement was associated with using a nonlinear model of relationship between VOC concentrations and sensor responses as compared to a linear one.