Identify Smells Using Time Series Data from Metal Oxide Gas Sensors

抄録

Nowadays, various metal oxide gas sensors (MOGSs) are widely combined and used as an electronic nose (EN). Most of the ENs always use only single feature (e.g. peak signal, average signal of saturation stage) from each MOGS without considering the signals before reaching the saturation points. In this letter, we increase the ability of an EN to identify smells that yield nearly the same saturation points without increasing the number of MOGS by using the information from the time series signals of MOGSs during absorbing the tested odors. The results from multivariate analysis show perfectly classification in all tested odors.