Feature Selection to Identify Durability of Perfume Scent using Electronic Nose Randi Septiawan, Adhytia Tungga, Riser Fahdiran, Bambang Heru Iswanto
Complex Systems Research Group, Department of Physics, Universitas Negeri Jakarta, Indonesia
Abstract
This paper discusses feature extraction from e-nose responses to identify the durability of perfume scents over time. Ten volunteer panelists took part in sensory analysis in relation to the strength of the perfume scent. The sample used was a branded perfume and five artificial perfumes were analyzed. Aroma durability is measured every three hours using an e-nose consisting of eight types of gas sensors, namely TGS 813, TGS 822, TGS 2600, TGS 826, TGS 2611, TGS 2620, TGS 2612, and TGS 2602. Three methods, namely relative amplitude (RA), survey (S), and wavelet decomposition scale 3 (WD) are used for the extraction of sensor response features. Feature analysis is performed to compare the rate of e-nose decay response using the exponential curve fitting method. The experimental results showed that the S and RA features correlated with the level of aroma that was responded by the panelist team. The results of numerical analysis show the decay of the original perfume scent four times faster than the fake perfume.
