Share Email Print
cover

Proceedings Paper

Calibration of a portable cost-effective chemical residue detection system with adaptive neural net control
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

The Sensory Research Institute at the Florida State University has quantitatively characterized a chemical residue detection system with adaptive neural net data processing. Two separate configurations, "Stormy" and "Gaea", were trained by exposure to decreasing amounts of n-amyl acetate from chemical emitters randomly distributed among a collection of non-emitters. The concentration of chemical in the sampled air stream was controlled precisely. The detection threshold for "Stormy" was 1.14 ppt; that for "Gaea" was 1.9 ppt. Cycle time for sampling and chemical analysis of each sample port was on the order of seconds. Possible effects on the sensors of environmental factors such as ambient humidity, temperature, and air velocity were not considered. Besides processing individual air sample data, the neural nets can sense concentration gradients and track to chemical source. The adaptive neural nets are accessed by a voice recognition system and are capable of point testing or free-ranging search. The service life of the detectors, the neural net processors, and auxiliary packaging is approximately 8 years under normal field use. Maintenance requires a good quality kibble and an occasional romp in the park.

Paper Details

Date Published: 16 July 2003
PDF: 8 pages
Proc. SPIE 5048, Nondestructive Detection and Measurement for Homeland Security, (16 July 2003); doi: 10.1117/12.484789
Show Author Affiliations
Alan C. Tripp, Univ. of Utah (United States)
James C. Walker, Florida State Univ. (United States)


Published in SPIE Proceedings Vol. 5048:
Nondestructive Detection and Measurement for Homeland Security
Steven R. Doctor; Yoseph Bar-Cohen; A. Emin Aktan, Editor(s)

© SPIE. Terms of Use
Back to Top