Problemas diretos e inversos em cinética química e ressonância magnética nuclear

Abstract

Chemical and Physics direct problems are represented by mathematical models in which properties are determined without ill conditioned matrix inversion. Conversely, inverse problems are characterized also by systems properties determination, but along the resolution process matrix inversion is required. Studies on direct and inverse problems are presented in this work. At first, a general analysis was performed, as well as the methodologies employed to address them, especially artificial neural networks: MLP and Hopfield. In the second phase, the theoretical and numerical background were applied in the chemical kinetic area in two works: 1) decomposition of solid polymer study using experimental thermal analysis data and 2) combustion of methane study to recover rate constant and model certification. In the nuclear magnetic resonance area, the theoretical basis was also applied in two works: 3) simulations of NMR experiments by solving the Bloch equations and 4) dosimetric systems studies using echo spin experiment