Rapid Identification of CW Agents by Artificial Neural Networks Pruning of Temperature Programmed Microsensor Databases

Published

December 1, 2003

Author(s)

Zvi Boger, Richard E. Cavicchi, Douglas C. Meier, Stephen Semancik

Abstract

The need for reliable, quick-acting low-cost detectors for chemical warfare (CW) gases is evident, and many sensors systems have been proposed 1,2. Mobile and hand-held battlefield CW sensor systems are already deployed, but their high cost prevents their widespread use in homeland security applications. CW monitoring requires rapid and certain identification of agents and agent class compounds with high sensitivity and selectivity in order to avoid non-detection or delayed response. Differentiation from similar commercial chemicals is also essential, as false alarms will degrade the confidence in these sensors. In this communication, we seek to illustrate aspects of the methodology that could be employed for developing fast CW warnings based on solid-state micro-sensors, and signal processing methods that could be integrated into custom-made electronic devices that efficiently use only the most relevant inputs necessary for positive agent identification.

Citation

Sensor Letters

Volume

1

Issue

1

Pub Type

Journals

Download Paper

DOI Link

Keywords

artificaI neural networks, chemical warfare agents, microhotplate sensors, temperature programmed sequence, tin oxide, titanium oxide

Citation

Boger, Z. , Cavicchi, R. , Meier, D. and Semancik, S. (2003), Rapid Identification of CW Agents by Artificial Neural Networks Pruning of Temperature Programmed Microsensor Databases, Sensor Letters, [online], https://doi.org/10.1166/Sl.2003.003 (Accessed October 7, 2025)