Detection Optimization Using a Transient Feature from a Metal Oxide Gas Sensor Array

Metal oxide sensors are largely studied in electronic nose field for gas detection or identification. They are robust, easy to use and have good sensitivity to many gases in the ppm range. However, the drift and low response-time of these chemical sensors constitute an important barrier in their utilization especially when low concentrations of gaseous atmospheres should be detected. To resolve these limitations many possibilities have been investigated to enhance either the sensor properties or the measurement conditions and the data analysis procedures. This paper presents the performance comparison of several features extracted from the sensor response signals. Particularly, a new transient feature corresponding to the inflexion point of the sensor response is processed. This feature is obtained during the first minutes of the gas exposure and shows a good performance, for example to detect a slight concentration variation of a gaseous atmosphere.