Portable Electronic Nose System for Identification of Synthesized Gasoline Using Metal Oxide Gas Sensor and Pattern Recognition

摘要

This paper describes a portable electronic nose (e-nose) system for use in the identification of synthesized gasoline, comprised of a single semiconductor type of gas sensor and pattern recognition neural networks. The designed e-nose system consists of a one-chip microcontroller, a pre-concentrator, and a gas sensor. Two different neural networks, a multi-layer perceptron (MLP) neural network and a fuzzy ARTMAP neural network were applied to discriminate synthesized gasoline from normal gasoline. The results of the classification showed 100percent and 85percent recognition rates for the training data set and testing data set, respectively.