Conception et fabrication d'un prototype de nez électronique basé sur un système d'apprentissage et de reconnaissance évolutif des composants organiques volatiles

Abstract

Several actions are taken by the human being according to the perception of smell such as taking out the garbage, changing the layers of baby, taking measures of safety in case of gas leak, etc. But the sense of smell of the man is limited because there are gases that are very toxic and the human being cannot detect them by the nose like carbon monoxide. Thus, the sense of smell is used in many industrial applications in production (perfume industries) or in safety (oil and gas industry), medical applications (bacteria detection) and national security applications (detection of cannabis). For several decades, the sensor community has been trying to artificially reproduce the capacity of smell. The first appearance of electronic noses or artificial nose was in the 1980s. This device is a set of gas sensors with learning and recognition techniques used to distinguish many odors. Several works have been published on the use of the electronic nose in specific applications. However, there is not a lot of work on artificial noses that can be used in many applications. This project aims at designing and manufacturing an electronic nose that can be used in many different applications. User needs. A design and fabrication of the hardware part was made from scratch. In which there is the sampling system which facilitates the reaction of the gases with the sensors and the electronic card which translates these reactions into values comprehensible by the software part. A globally optimal design for all applications was made at the end of this part. For the software part, a learning and recognition process has been proposed using evolutionary fuzzy rules-based learning systems (FRB). The evolution of the software part ensures the flexibility of the whole (hardware part and software part) to the needs of users. In order to decrease the dependence of the system on the user, an active supervision method has been used with the learning and recognition system.

Date	15 Nov 2019
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Abdelouahed Gherbi (Supervisor)