Eventos Anais de eventos
ENCIT 2020
18th Brazilian Congress of Thermal Sciences and Engineering
AN INVERSE PROBLEM OF TWO-DIMENSIONAL AND TIME DEPENDENT HEAT FLUX ESTIMATION VIA HAMILTONIAN MONTE CARLO METHOD
Submission Author:
Gabriel Neves , RJ
Co-Authors:
Gabriel Neves, Luiz A. S. Abreu, Diego Knupp, Antônio Silva Neto
Presenter: Gabriel Neves
doi://10.26678/ABCM.ENCIT2020.CIT20-0442
Abstract
This work addresses the problem of heat flux estimation with two-dimensional spatial and temporal variations. The inverse problem is conducted via a Hamiltonian Monte Carlo (a.k.a. Hybrid Monte Carlo) method, applied to a set of simulated experimental data. CPU time and regularization processes are taken into account in order to compare with the standard, Markov Chain Monte Carlo Method. As results, the HMC method proved able to reconstruct the heat flux much quicker than MCMC method. Furthermore, it dismiss the need of regularization processes, while the MCMC implementation uses a Total Variation prior for regularization.
Keywords
Heat Flux Estimate, Bayesian inference, Inverse problem, hamiltonian monte carlo, markov chain monte carlo
DOWNLOAD PDF VIEW PRESENTATION
