A novel spectrum analysis technique for odor sensing in optical electronic nose

Abstract

With the advantages of light, such as high speed, high sensitivity, and wide response, the optical electronic nose draws more and more attention from the academia and industry. In an optical electronic nose, the optical gas sensing system is one of the crucial parts. In this study, optical gas sensing was realized by detecting the absorption spectrum of odors, where polychromatic light was employed to provide the sensing array according to its cross-sensitive and widely responsive to odors. To facilitate the odor identification of the electronic nose, the sensing responsive spectra of the optical gas sensing system were analyzed in this paper. Specifically, a feature-based analysis technique for the spectra in the form of two-dimensional matrix was firstly proposed. With the proposed spectrum analysis technique, the synthesized features of odors which are reflected by the sensing responsive spectra can be extracted. Afterwards, the theoretical validity of both the feature extraction and fault tolerance using the proposed feature-based spectrum analysis technique has been proved mathematically. Finally, nine hazardous gases were chosen in the experiments and the discrimination as well as fault-tolerant capability of the proposed feature-based spectrum analysis technique was experimentally demonstrated. Experiment results show that the proposed technique, compared with the counterpart techniques, has superior results on discriminating odors of similar response under noisy condition, which also indicates that the theoretical results and experimental results are consistent.