Eventos Anais de eventos
ENCIT 2022
19th Brazilian Congress of Thermal Sciences and Engineering
MODELING SINGLE JET DIFFUSION FLAME OSCILLATIONS BASED ON UNDER-SAMPLED DATA
Submission Author:
Davi Saadi de Almeida Lettieri , RJ , Brazil
Co-Authors:
Davi Saadi de Almeida Lettieri, Leonardo Santos de Brito Alves, Juan Carlos Assis da Silva
Presenter: Juan Carlos Assis da Silva
doi://10.26678/ABCM.ENCIT2022.CIT22-0035
Abstract
Since first described, the Sparse Identification of Nonlinear Dynamics (SINDy) received significant attention from the scientific community as it represents one of the latest tools of symbolic regression in the machine learning field. Even though the original paper is relatively recent, there is already a large literature concerning its implementation and performance. However, most of it is focused on the rediscovery of a previously known dynamical system. Time resolved images from an acoustic forcing experiment for the single jet diffusion flame oscillations are decomposed via Proper Orthogonal Decomposition (POD). After that, the most energetic modes with coherent structures are selected. However, the sampling rate of images were not enough to satisfy the Shannon Nyquist Theorem, and a spectral analysis alone is not sufficient to model the temporal behavior of each mode. On the other hand, the results indicate that looking at the 2D phase portraits can overturn the under-sampling problem.
Keywords
machine learning, Single Jet Diffusion Flame, SINDy, 2D Phase Portrait
