Original title: Artificial intelligence for simulation of soot deposits- from 2D to 3D estimates
Authors: Khýr, Matyáš ; Isoz, Martin ; Plachá, M.
Document type: Papers
Conference/Event: Conference on Topical Problems of Fluid Mechanics 2025, Praha (CZ), 20250212
Year: 2025
Language: eng
Abstract: The accumulation of soot within the porous microstructure of catalytic filters (CFs) impacts the overall performance of the device, making the prediction of soot distribution a major challenge in the design of CFs. We previously developed a neural network that leverages convolutional autoencoders and fully-connected layers and demonstrated that this network can effectively estimate the distribution of deposited soot within two-dimensional model microstructures. In this contribution, we expand on the previously devised model, by exploring its possible extension to processing three-dimensional data from real-world catalytic filter microstructures. Specifically, we examine and compare three principally different approaches to generating three-dimensional estimates, spanning from a posteriori reconstructions to the application of fully three-dimensional convolutional networks.
Keywords: ANN; CFD; CNN
Project no.: EH23_020/0008501, StrategieAV21/20
Funding provider: GA MŠk, AV ČR
Host item entry: Topical Problems of Fluid Mechanics 2025, ISBN 978-80-87012-05-5, ISSN 2336-5781
Note: Související webová stránka: https://tpfm.it.cas.cz/im/im/proceeding/2025/18

Institution: Institute of Thermomechanics AS ČR (web)
Document availability information: Fulltext is available in the digital repository of the Academy of Sciences.
Original record: https://hdl.handle.net/11104/0372365

Permalink: http://www.nusl.cz/ntk/nusl-692739


The record appears in these collections:
Research > Institutes ASCR > Institute of Thermomechanics
Conference materials > Papers