ELECTRONIC NOSE FOR VOLATILE AND NONVOLATILE ODORS IDENTIFICATION

Abstract

The developed nose can be used in normal environment with no extra odor delivery interface needed. The system comprises a sensor array of four MOS sensor, ATMEL microcontroller and a developed feature extraction algorithm based on a newly proposed feature extraction algorithm. Classification is performed using KNN, where K is optimized for the best accuracy. The system is supported by an extendable interface that allows the nose to learn more odors and use other classification techniques if required. The proposed E-nose has been tested and proved its efficiency in detecting non-volatile odors such as pure fruit juices (Orange, Apple, Pineapple and Grenade juices) and rotten egg and volatile odors such as Butane gas and Grenade perfume with accuracy of 98.6%using KNN(K=1).