A Study on Binary Gas Mixture

In this study, quantitative classification of tricholoroethylene and carbontetrachloride was tried using steady state responses of sensors. For this purpose, Artificial Neural Networks (ANN) were used. ANN was used for gas concentration estimation and quantitative classification of the gas mixture. For gas concentration estimation, the gas sensor transient state responses were taken and for quantitative classification of the gas mixture, the gas sensor steady state responses were taken. A feed-forward multi-layer neural network with hidden layers trained by a back- propagation and Levenberg-Marquardt learning algorithms has been implemented. Acceptable performance is obtained for this system and the appropriateness of ANN for the quantitative classification of volatile organic compounds is observed.