Forward and inverse design of kirigami via supervised autoencoder

Paul Z Hanakata,David K Campbell,Ekin D Cubuk,Harold S Park

doi:10.1103/physrevresearch.2.042006

Paul Z Hanakata, David K Campbell + Show 2 more

Open Access

https://doi.org/10.1103/physrevresearch.2.042006

Copy DOI

Abstract

Machine learning (ML) methods have recently been used as forward solvers to predict the mechanical properties of composite materials. Here, we use a supervised-autoencoder (sAE) to perform inverse design of graphene kirigami, where predicting the ultimate stress or strain under tensile loading is known to be difficult due to nonlinear effects arising from the out-of-plane buckling. Unlike the standard autoencoder, our sAE is able not only to reconstruct cut configurations but also to predict mechanical properties of graphene kirigami and classify the kirigami witheither parallel or orthogonal cuts. By interpolating in the latent space of kirigami structures, the sAE is able to generate novel designs that mix parallel and orthogonal cuts, despite being trained independently on parallel or orthogonal cuts. Our method allows us to both identify novel designs and predict, with reasonable accuracy, their mechanical properties, which is crucial for expanding the search space for materials design.

Highlights

There has been growing interest in investigating the nonlinear mechanics of perforated thin sheets across length scales ranging from the macroscale [1,2,3,4] down to nanoscale systems [5,6,7,8]
The molecular dynamics (MD) simulation procedure is similar to our previous work [8] and the simulation details can found in the Supplemental Material (SM)
As we allow either only 0–15 orthogonal cuts or 0–15 parallel cuts, we obtain a total of 62 558 configurations, of which 29 791 are nondetached configurations with orthogonal cuts while the remaining are the configurations with parallel cuts which have nondetached configurations

Summary

Introduction

There has been growing interest in investigating the nonlinear mechanics of perforated thin sheets across length scales ranging from the macroscale [1,2,3,4] down to nanoscale systems [5,6,7,8]. The cuts in a thin sheet—known as kirigami cuts—induce buckling and other motions (e.g., rotations). These mechanisms result in new properties, such as enhanced ductility [7] and auxeticity [9] that are different from the pristine (cut-free) counterpart. This simple strategy has led to programable kirigami actuators which are the building blocks of soft robots [1,10]. An analytical model that can describe how the mechanical properties of kirigami sheets depend on the interaction of different types of cuts still needs to be developed

Objectives

Results

Conclusion