Project # 52
Multiscale Modeling of Two-dimensional Lattice Metamaterials Using Machine Learning

Team Lead:	Emmanuel Odukogbe
Other Team Members:	Deeva Khanuja
Faculty Advisor(s):	Dr. Xin Liu
Department:	MAE
Project Type:	REU
Title:	Multiscale Modeling of Two-dimensional Lattice Metamaterials Using Machine Learning
Abstract:	Two-dimensional (2D) lattice metamaterials are emerging lightweight materials showing improved mechanical, thermal and acoustic properties. The microstructures of these materials can be designed and fabricated to achieve specific properties to fulfill the needs of different applications. Due to the extended design space, it is computationally expensive to predict the material properties given their microstructures. Although some machine learning (ML) models have been developed to reduce computational costs, different 2D lattices often need training in different ML models, requiring significant efforts and computational resources. In this project, we will develop a new parameterization method using heat kernel images to generate a unified input to represent different 2D lattices. An ML model will be developed to take the heat kernel images and predict the corresponding effective material properties. We will benchmark the accuracy of the proposed method against the ML model with direct image input. This project will train undergraduate students in the design, modeling, and manufacturing of advanced lattice materials.
Poster:	View Poster #52 (PDF)
Judges Scoring:	Submit Evaluation (judges only)

Project # 52Multiscale Modeling of Two-dimensional Lattice Metamaterials Using Machine Learning