Data driven approaches to characterizing cross-sensitive sensors and to improve calibration transer
(English)Manuscript (preprint) (Other academic)
Many different data analysis procedures can be developed for applications where chemical sensors are used, and it is sometimes even necessary to do so in order to get any reasonable results. Many of these procedures put certain demands on the used sensors, e.g. that they remain stable during operation. For this reason, it is useful to first learn the characteristics of a device before initiating developments of advanced procedures.
ln this paper, data driven modeling techniques are used to approximate sensor responses and cross sensitivity patterns. By keeping record of parameters in an adapted model structure the characteristics of a sensor can be extracted. The extracted information is valuable when developing analysis procedures.
The paper shows how the complexity of the calibration transfer problem increases when cross-sensitive sensors are used, and also shows a simple example of how the problem can be solved using application know-how for measurements using gas sensitive Metal-Insulator-Silicon-Carbide-Transistors. These procedures are simple and can be performed by sensor manufacturers to develop calibration transfer algorithms that reduce or eliminate the need of calibration measurements for the user.
calibration transfer, cross sensitivity, chemical sensors, pattern recognition
Engineering and Technology
IdentifiersURN: urn:nbn:se:liu:diva-100824OAI: oai:DiVA.org:liu-100824DiVA: diva2:663771