Multiphysics-informed machine learning platform for interface study
Bansal, Parth
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/124524
Description
Title
Multiphysics-informed machine learning platform for interface study
Author(s)
Bansal, Parth
Issue Date
2024-04-19
Director of Research (if dissertation) or Advisor (if thesis)
Li, Yumeng
Doctoral Committee Chair(s)
Li, Yumeng
Committee Member(s)
Wang, Pingfeng
Shao, Chenhui
Allison, James
Department of Study
Industrial&Enterprise Sys Eng
Discipline
Industrial Engineering
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Physics-Informed Machine Learning
Corrosion
Li-Ion Battery
Adaptive Sampling, Interface Design
Abstract
With the increasing focus on sustainable technologies both in terms of newer developments and increasing the life of existing ones, there is a need to efficiently and accurately assess these technological systems. This can be achieved through using less expensive and really accurate finite element computational models. However, these Monte-Carlo simulations are still too computationally expensive and require a lot of resources. Hence, this thesis develops finite element models that work together with machine learning techniques to provide a robust framework to perform various studies such as uncertainty quantification, state of health prognostics and design of different physical and electrical systems. The main contribution of this thesis is to demonstrate frameworks that can be used to evaluate the system performance (e.g. corrosion related material loss, capacity loss in batteries) and help in designing better systems by understanding and quantifying the sources of uncertainty in them by the use of physics-informed machine learning. The first step in this process of physics-informed machine learning is to develop the finite element models, whose results are used to inform or train the machine learning algorithms. This thesis focuses on two main systems: galvanic corrosion in dissimilar material joints and the capacity fade in silicon anode based lithium-ion batteries. The finite element models for both these processes include a variety of failure modes that can accurately and reliably predict the system life cycle. Experimental work is also used to partially verify the finite element models.
The results from these finite element models are then used with machine learning models such as Gaussian Process Regression models to reduce the overall cost burden. Processes such as probablistic-confidence based adaptive sampling techniques can further reduce the computational costs by thoroughly exploring the design space in an efficient manner. The trained machine learning models can then be used for a variety of applications such as state of health analysis, uncertainty quantification and better system design.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
