On Fostering Predictions in Data-Driven Reduced Order Model for Eulerian–Lagrangian Simulations: Decision of Sufficient Training Data

The development of a data-driven surrogate model (SM) is extensively studied in Eulerian–Lagrangian simulations for its advantage of high computational speed. However, in the application of granular systems with violent fluid-solid flows, how to select sufficient training data to ensure consistency...

Full description

Saved in:
Bibliographic Details
Main Authors: Kai-en Yang, Shuo Li, Guangtao Duan, Mikio Sakai
Format: Article
Language:English
Published: Taylor & Francis Group 2024-12-01
Series:Journal of Chemical Engineering of Japan
Subjects:
Online Access:https://www.tandfonline.com/doi/10.1080/00219592.2024.2316155
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850123474716065792
author Kai-en Yang
Shuo Li
Guangtao Duan
Mikio Sakai
author_facet Kai-en Yang
Shuo Li
Guangtao Duan
Mikio Sakai
author_sort Kai-en Yang
collection DOAJ
description The development of a data-driven surrogate model (SM) is extensively studied in Eulerian–Lagrangian simulations for its advantage of high computational speed. However, in the application of granular systems with violent fluid-solid flows, how to select sufficient training data to ensure consistency between the high-fidelity model and SM remains unknown and highly challenging. The accuracy of SM can be easily deteriorated due to insufficient training data. This necessitates a trial-and-error process and hinders its industrial applications. To address this issue, this study newly reveals a finding that data density is a key to sufficient training, and we propose a novel technique for deciding the sufficient training data of SM. Specifically, a feasibility index is proposed based on posterior error analysis. It is demonstrated that when the training data is determined under the proposed feasibility index [Formula: see text] 2, the consistency of granular dynamics between SM and the high-fidelity model can be guaranteed. Employed in a representative SM, a reduced order model (ROM), this technique enables the successful decision of sufficient training data, resulting in the remarkable predictability in violent fluid-solid flows without trial-and-error. This technique holds great potential in solving the predicament of deciding training data for data-driven models.
format Article
id doaj-art-c5da0f4de59a41479cc881a7d95093fc
institution OA Journals
issn 0021-9592
1881-1299
language English
publishDate 2024-12-01
publisher Taylor & Francis Group
record_format Article
series Journal of Chemical Engineering of Japan
spelling doaj-art-c5da0f4de59a41479cc881a7d95093fc2025-08-20T02:34:35ZengTaylor & Francis GroupJournal of Chemical Engineering of Japan0021-95921881-12992024-12-0157110.1080/00219592.2024.2316155On Fostering Predictions in Data-Driven Reduced Order Model for Eulerian–Lagrangian Simulations: Decision of Sufficient Training DataKai-en Yang0Shuo Li1Guangtao Duan2Mikio Sakai3Department of Nuclear Engineering and Management, The University of Tokyo, Tokyo, JapanDepartment of Nuclear Engineering and Management, The University of Tokyo, Tokyo, JapanDepartment of Nuclear Engineering and Management, The University of Tokyo, Tokyo, JapanDepartment of Nuclear Engineering and Management, The University of Tokyo, Tokyo, JapanThe development of a data-driven surrogate model (SM) is extensively studied in Eulerian–Lagrangian simulations for its advantage of high computational speed. However, in the application of granular systems with violent fluid-solid flows, how to select sufficient training data to ensure consistency between the high-fidelity model and SM remains unknown and highly challenging. The accuracy of SM can be easily deteriorated due to insufficient training data. This necessitates a trial-and-error process and hinders its industrial applications. To address this issue, this study newly reveals a finding that data density is a key to sufficient training, and we propose a novel technique for deciding the sufficient training data of SM. Specifically, a feasibility index is proposed based on posterior error analysis. It is demonstrated that when the training data is determined under the proposed feasibility index [Formula: see text] 2, the consistency of granular dynamics between SM and the high-fidelity model can be guaranteed. Employed in a representative SM, a reduced order model (ROM), this technique enables the successful decision of sufficient training data, resulting in the remarkable predictability in violent fluid-solid flows without trial-and-error. This technique holds great potential in solving the predicament of deciding training data for data-driven models.https://www.tandfonline.com/doi/10.1080/00219592.2024.2316155Data-driven ROMFluid-solid flowsSufficient training dataPredictions
spellingShingle Kai-en Yang
Shuo Li
Guangtao Duan
Mikio Sakai
On Fostering Predictions in Data-Driven Reduced Order Model for Eulerian–Lagrangian Simulations: Decision of Sufficient Training Data
Journal of Chemical Engineering of Japan
Data-driven ROM
Fluid-solid flows
Sufficient training data
Predictions
title On Fostering Predictions in Data-Driven Reduced Order Model for Eulerian–Lagrangian Simulations: Decision of Sufficient Training Data
title_full On Fostering Predictions in Data-Driven Reduced Order Model for Eulerian–Lagrangian Simulations: Decision of Sufficient Training Data
title_fullStr On Fostering Predictions in Data-Driven Reduced Order Model for Eulerian–Lagrangian Simulations: Decision of Sufficient Training Data
title_full_unstemmed On Fostering Predictions in Data-Driven Reduced Order Model for Eulerian–Lagrangian Simulations: Decision of Sufficient Training Data
title_short On Fostering Predictions in Data-Driven Reduced Order Model for Eulerian–Lagrangian Simulations: Decision of Sufficient Training Data
title_sort on fostering predictions in data driven reduced order model for eulerian lagrangian simulations decision of sufficient training data
topic Data-driven ROM
Fluid-solid flows
Sufficient training data
Predictions
url https://www.tandfonline.com/doi/10.1080/00219592.2024.2316155
work_keys_str_mv AT kaienyang onfosteringpredictionsindatadrivenreducedordermodelforeulerianlagrangiansimulationsdecisionofsufficienttrainingdata
AT shuoli onfosteringpredictionsindatadrivenreducedordermodelforeulerianlagrangiansimulationsdecisionofsufficienttrainingdata
AT guangtaoduan onfosteringpredictionsindatadrivenreducedordermodelforeulerianlagrangiansimulationsdecisionofsufficienttrainingdata
AT mikiosakai onfosteringpredictionsindatadrivenreducedordermodelforeulerianlagrangiansimulationsdecisionofsufficienttrainingdata