TY JOUR TI NEURAL NETWORK SOLUTION OF INVERSE PROBLEM OF IDENTIFICATION OF LIGHT PETROLEUM PRODUCTS BY RAMAN SPECTROSCOPY FOR FIRE INVESTIGATION KW light petroleum products KW Raman spectrometry KW neural network architecture KW fire investigation JO NATURAL AND MAN-MADE RISKS (PHYSICO-MATHEMATICAL AND APPLIED ASPECTS) AU Otgon, F.V. AU Dementiev, F.A. AU Ivanov, A.V. PY 2018 IS 2018 PB Saint-Petersburg University of State Fire Service of Emercom of Russia