Related Experiment Video
Updated: May 12, 2026
07:57
Quantitative Detection of Trace Explosive Vapors by Programmed Temperature Desorption Gas Chromatography-Electron Capture Detector
Published on: July 25, 2014
Recognition of explosives fingerprints on objects for courier services using machine learning methods and
J Moros1, J Serrano, F J Gallego
1Department of Analytical Chemistry, University of Malaga, E-29071 Malaga, Spain.
Talanta
|April 27, 2013
Summary
Laser-induced breakdown spectroscopy (LIBS) can detect explosives, but struggles with selectivity. A new multi-stage algorithm using neural networks effectively sorts explosive fingerprints with low error rates.
Area of Science:
- Analytical Chemistry
- Spectroscopy
- Machine Learning
Background:
- Laser-induced breakdown spectroscopy (LIBS) shows promise for trace explosive detection.
- Limited selectivity in LIBS hinders accurate identification of explosive fingerprints due to spectral similarities with interferents.
Purpose of the Study:
- To develop and validate a multi-stage algorithm for classifying explosive fingerprints using LIBS.
- To improve the selectivity and accuracy of LIBS in identifying hazardous materials amidst common interferents.
Main Methods:
- Collected LIBS spectra from six explosives (chloratite, ammonal, DNT, TNT, RDX, PETN) and various harmless interferents.
- Employed a multi-stage algorithm combining three learning classifiers, including neural networks trained with the Levenberg-Marquardt rule.
- Utilized 3D scatter plots projected onto feature subspaces for spectral data analysis.
Main Results:
- The algorithm successfully sorted explosive fingerprints from interferents, even with spectrally similar materials.
- Achieved low rates of false negatives (<10%) and false positives (<10%) in fingerprint classification.
- Demonstrated improved technology readiness for LIBS in defense and security applications.
Conclusions:
- The developed multi-stage algorithm significantly enhances the selectivity of LIBS for explosive fingerprint detection.
- This advancement represents a crucial step towards reliable, real-world application of LIBS in homeland security and defense.
- The method offers a robust solution for distinguishing hazardous materials based on spectral analysis.
