Energetic materials identification by laser-induced breakdown spectroscopy combined with artificial neural network.

Abstract

In this study, for the first time to the best of our knowledge, a combination of the laser-induced breakdown spectroscopy (LIBS) technique and artificial neural network (ANN) analysis has been implemented for the identification of energetic materials, including TNT, RDX, black powder, and propellant. Also, aluminum, copper, inconel, and graphite have been used for more accurate investigation and comparison. After the LIBS test and spectrum acquisition on all samples in both air and argon ambient, optimized neural networks were designed by LIBS data. Based on input data, three ANN algorithms are proposed: the first is fed with the whole LIBS spectra in air (ANN1) and the second with the principle component analysis (PCA) scores of each spectrum in air (ANN2) and the other with the PCA scores of the spectrum in Ar (ANN3). According to the results, error of the network is very low in ANN2 and 3 and the best identification and discrimination was obtained by ANN3. After these, in order to validate and for more investigation of this combined method, we also used Al/RDX standard samples for analysis.