Physics-Guided Machine Learning for Uncertainty Quantification in Turbulence Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chu, Minghan, Qian, Weicheng
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912695688626176
author Chu, Minghan
Qian, Weicheng
author_facet Chu, Minghan
Qian, Weicheng
contents Predicting the evolution of turbulent flows is central across science and engineering. Most studies rely on simulations with turbulence models, whose empirical simplifications introduce epistemic uncertainty. The Eigenspace Perturbation Method (EPM) is a widely used physics-based approach to quantify model-form uncertainty, but being purely physics-based it can overpredict uncertainty bounds. We propose a convolutional neural network (CNN)-based modulation of EPM perturbation magnitudes to improve calibration while preserving physical consistency. Across canonical cases, the hybrid ML-EPM framework yields substantially tighter, better-calibrated uncertainty estimates than baseline EPM alone.
format Preprint
id arxiv_https___arxiv_org_abs_2511_05633
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Physics-Guided Machine Learning for Uncertainty Quantification in Turbulence Models
Chu, Minghan
Qian, Weicheng
Machine Learning
Fluid Dynamics
Predicting the evolution of turbulent flows is central across science and engineering. Most studies rely on simulations with turbulence models, whose empirical simplifications introduce epistemic uncertainty. The Eigenspace Perturbation Method (EPM) is a widely used physics-based approach to quantify model-form uncertainty, but being purely physics-based it can overpredict uncertainty bounds. We propose a convolutional neural network (CNN)-based modulation of EPM perturbation magnitudes to improve calibration while preserving physical consistency. Across canonical cases, the hybrid ML-EPM framework yields substantially tighter, better-calibrated uncertainty estimates than baseline EPM alone.
title Physics-Guided Machine Learning for Uncertainty Quantification in Turbulence Models
topic Machine Learning
Fluid Dynamics
url https://arxiv.org/abs/2511.05633