Electronic Nose

Author

Amit Jain

BITS-PILANI

Abstract

There is currently a great interest in developing new techniques for identifying an unknown gas. By employing gas sensors specifically made for the purpose, such identification can be made, by using a suitable algorithm. Such a system is often termed as an Electronic Nose. For this purpose, a LabVIEW Virtual Instrument (VI) was developed for acquiring data from the gas sensors through an 8-channel Data acquisition system, providing a graphical user interface to the user and for providing additional options like training the system with sample gases, saving results of experiment, loading past experiments etc. The system uses a Probabilistic Neural Network (PNN) to identify the unknown gas by comparing the data gathered from the unknown gas with the previously stored training data, and delivers the results in real-time.

The project was successful, and can correctly identify the unknown gas with great accuracy. Tests were carried out with 8 gases, which were identified correctly by the system in all cases after training with just one sample data for each gas. The system also gives a probability of match for each gas, which the user can use to determine the confidence interval of the results.