Study On Gas Electronic Nose System Based On Mixing Array And Its Application In Gas Quantitative Determination

The electronic nose, also named as Artificial Olfaction, is an intelligent electronic device which is the simulation of human olfactory. Electronic nose has been widely applied in some field, such as the environmental information monitoring, control of food quality and safety. The research of electronic nose focuses on the design of sensor and sensor array, the optimization of pattern recognition algorithm.In this paper, we combined the needs of project and the research of lab, discussed the gas sensing principle of metal oxide semiconductor sensor and carbon nanotube sensor and PANI nanofiber sensor, made multi wall carbon nanotubes sensor and PANI nanofiber sensor which consisted mixing sensors array with metal oxide semiconductor sensor. We designed a complete electronic nose and studied the optimal designing of sensor array in electronic nose and its application in mixed gas quantitative determination. The pattern recognition algorithm achieved the qualitative and quantitative determination of mixture, which included data preprocessing, feature extraction, BP neural network and Fuzzy ART neural network.The LDA algorithm and BP neural network achieved the qualitative identification, and the accuracy rate could be 100%. The feature selection method combining with the Fuzzy ART neural network and Fuzzy theory achieved the quantitative determination. The results showed:as to the trained data, the estimation accuracy reached 100%; as to the untrained data but in the scope of training, the estimation accuracy of ammonia gas reached above 96.7%, the estimation accuracy of ethanol and the mixed both reached above 95.8%.The electronic nose with mixing sensors array based the sensitivity ratio difference of the different gas sensors as to different gases, which could take the advantage of high accuracy in mixed gas quantitative determination.