LOGIN / Acesse o sistema

Esqueceu sua senha? Redefina aqui.

Ainda não possui uma conta? Cadastre-se aqui!

REDEFINIR SENHA

Insira o endereço de email associado à sua conta que enviaremos um link de redefinição de senha para você.

Ainda não possui uma conta? Cadastre-se aqui!

Este conteúdo é exclusivo para membros ABCM

Inscreva-se e faça parte da comunidade

CADASTRE-SE

Tem uma conta?

Torne-se um membros ABCM

Veja algumas vantagens em se manter como nosso Associado:

Acesso regular ao JBSMSE
Boletim de notícias ABCM
Acesso livre aos Anais de Eventos
Possibilidade de concorrer às Bolsas de Iniciação Científica da ABCM.
Descontos nos eventos promovidos pela ABCM e pelas entidades com as quais mmantém acordo de cooperação.
Estudantes de gradução serão isentos no primeiro ano de afiliação.
10% de desconto para o Associado que pagar anuidade anntes de completar os 12 meses da última anuidade paga.
Desconto na compra dos livros da ABCM, entre eles: "Engenharia de Dutos" e "Escoamento Multifásico".
CADASTRE-SE SEGUIR PARA O VIDEO >

Tem uma conta?

Eventos Anais de eventos

Anais de eventos

COBEM 2023

27th International Congress of Mechanical Engineering

OPTIMAL PREPROCESSING FOR PROPER ORTHOGONAL DECOMPOSITION OF THE OSCILLATORY METHANE JET DIFFUSION FLAMES DATASET

Submission Author: Fernanda Spilotros Costa Cordeiro , RJ
Co-Authors: Fernanda Spilotros Costa Cordeiro, Leonardo Santos de Brito Alves, Davi Saadi de Almeida Lettieri
Presenter: Davi Saadi de Almeida Lettieri

doi://10.26678/ABCM.COBEM2023.COB2023-1426

 

Abstract

This study aims to establish an experimental image processing to study oscillatory diffusion flames of the single methane jet, and its impact on the Proper Orthogonal Decomposition (POD). Modal decomposition techniques are widely used to characterize the behavior of unsteady dynamical systems. They provide a systematic way to generate lowdimensional approximations of a large set of high-dimensional dynamical systems. In particular, the POD has numerous applications, among which are experimental image analysis, order reduction in dynamical systems, turbulence modeling and neural network training. Noting the ultimate goal of reducing the order of the dynamical system that models the oscillatory diffusion flames, it is necessary to topologically analyze the trajectories of the POD coefficients through their phase portraits, in order to identify information about the existence of attractors, repulsors and limit cycles. However, since the input data are experimental images, the signal-to-noise ratio can be quite high. Hence, filtering techniques employed on experimental images allow the extraction and identification of essential elements of the images, and improve the visual quality of certain structural aspects, facilitating their computational interpretation. In this paper, we present the image processing technique known as Block-Matching and 3D Filtering (BM3D), which aims to remove as much noise as possible from the original data. The removal of noisy signals through such processing is fundamental for the accurate generation of higher-order POD coefficients. Based on this statement, several filtering parameters of the BM3D processing were tested on experimental images, generating the phase portraits obtained from the POD decomposition. The obtained results were compared with each other to evaluate which parameter provides the highest number of accurate POD coefficients.

Keywords

image processing, Proper orthogonal decomposition (POD), Order reduction, Block-Matching and 3D Filtering (BM3D), Phase Portraits

 

DOWNLOAD PDF

 

‹ voltar para anais de eventos ABCM