Decision Trees Based Odor Detection Method with Fusion Data Model Using Semiconductor Gas Sensor Array

Abstract

Odor, one of our senses, is an important sense of life. In some cases, smell can be vitally important. A smell can also help in the correct recognition of a substance. Odor is composed of gas molecules. In this study, 8 simple semiconductor gas sensors were used to detect the odor of various substances. A gas sensor fusion setup was created with gas sensors and a candidate data set was created by collecting sensor data with the help of Arduino Mega embedded system. With the help of this data set, the odor of 7 different cleaning agents and similar substances was detected with the help of Decision Trees (DT). The results obtained from the decision tree (DT) classifier using the data set obtained from the fusion setup (95.44%) are close to the state-of-the-art results. As a result of the study, the feasibility of an embedded odor detection device has been demonstrated.