A new gases identification method based on noise spectroscopy using metal-oxide gas sensors

Electronic nose is a system, which can be used to identify the nature of gases using a sensors array. In this paper, we propose to present an electronic nose (E-nose) based on noise spectroscopy of metal-oxide (MOX) gas sensor response. After measuring the power density spectrum (PDS) of the gas sensor noise generated when the sensor is exposed to the tested gas, the obtained signal (PDS) is plotted and computed in order to extract a new feature that will be considered as a gas signature and used to identify the gas. The data array composed of eight measures (four different concentrations for nitrogen dioxide and ozone) and three variables (three sensors) has been inserted into principal component analysis (PCA) process. Results indicate that successful classifications have been gotten in the discrimination of two sort of gas using support vector machine (SVM) which shows 100% of success rate. The result of data analysis demonstrates that the E-nose technology combined with noise spectroscopy could be perfectly applied to identify pollutant gases.