Groupe d’études et de recherche en analyse des décisions

Statistical measurement validation with application to electronic nose technology

Mina Mirshahi Polytechnique Montréal, Canada

A crucial element in assessment of indoor and outdoor air quality is auditing the odorants. Electronic nose (e-nose) is an artificial olfactory system which consists of an array of gas sensors. The e-nose is designed for recognizing complex odors in its surrounding environment. The gas sensor array receives chemical information about gaseous mixtures as input and converts it to measurable signals. Gas sensor’s performance is affected by different elements which make it unstable and less sensitive to odors. The sensor’s output is used to quantify odor concentration. Pattern recognition methods are employed in order to predict the odor concentration for each set of sensor values. To assess the accuracy of predictions, the validity of sensor values must be ensured. Sensors in the e-nose structure may report incorrect values or some stop functioning for a short period of time. In this talk, we discuss about a computationally efficient algorithm for diagnosing these anomalies in real time.

Entrée gratuite.
Bienvenue à tous!