Discrimination of VOCs along with concentration change detection applying a combination of DWT and Machine Learning tools

The issue of selectivity has been a major concern for metal-oxide based chemiresistive gas sensors. Also, the accurate and quantified detection of change in concentration is another major problem faced by conventional gas sensors. Here we report, a systematic approach for the selective identification of volatile organic compounds (VOCs) where the corresponding concentrations of each tested VOC were varied in different combinations (low to high). The gas sensing study was performed by a single sensor device employing tin-oxide hollow-spheres as the sensing material. The discrete wavelet transform (DWT) technique was applied over the transient response curves to extract features from the signal. Finally, machine learning algorithms were engaged with the extracted features to obtain perfect discrimination among the VOCs along with recognition of concentration change.