TY - JOUR

T1 - Stochastic interpretable machine learning based multiscale modeling in thermal conductivity of Polymeric graphene-enhanced composites

AU - Liu, Bokai

AU - Lu, Weizhuo

AU - Olofsson, Thomas

AU - Zhuang, Xiaoying

AU - Rabczuk, Timon

N1 - Funding Information: We gratefully acknowledge the support of the EU project H2020-AURORAL Grant agreement ID: 101016854 (Architecture for Unified Regional and Open digital ecosystems for Smart Communities and Rural Areas Large scale application) and the Kempe Foundation Sweden (Kempestiftelserna - Stiftelserna J.C. Kempes och Seth M. Kempes minne). This work is also funded and supported by J. Gust. Richert stiftelse, SWECO, Sweden (Grant agreement ID: 2023–00884). The computations handling were enabled by resources provided by the Swedish National Infrastructure for Computing (SNIC) and Academic Infrastructure for Supercomputing in Sweden (NAISS) at High-Performance Computing Center North (HPC2N) partially funded by the Swedish Research Council through grant agreement no. 2018-05973 and no. 2022–06725.

PY - 2024/1/1

Y1 - 2024/1/1

N2 - We introduce an interpretable stochastic integrated machine learning based multiscale approach for the prediction of the macroscopic thermal conductivity in Polymeric graphene-enhanced composites (PGECs). This method encompasses the propagation of uncertain input parameters from the meso to macro scale, implemented through a foundational bottom-up multi-scale framework. In this context, Representative Volume Elements in Finite Element Modeling (RVE-FEM) are employed to derive the homogenized thermal conductivity. Besides, we employ two sets of techniques: Regression-tree-based methods (Random Forest and Gradient Boosting Machine) and Neural networks-based approaches (Artificial Neural Networks and Deep Neural Networks). To ascertain the relative influence of factors on output estimations, the SHapley Additive exPlanations (SHAP) algorithm is integrated. This interpretable machine learning methodology demonstrates strong alignment with published experimental data. It holds promise as an efficient and versatile tool for designing new composite materials tailored to applications involving thermal management.

AB - We introduce an interpretable stochastic integrated machine learning based multiscale approach for the prediction of the macroscopic thermal conductivity in Polymeric graphene-enhanced composites (PGECs). This method encompasses the propagation of uncertain input parameters from the meso to macro scale, implemented through a foundational bottom-up multi-scale framework. In this context, Representative Volume Elements in Finite Element Modeling (RVE-FEM) are employed to derive the homogenized thermal conductivity. Besides, we employ two sets of techniques: Regression-tree-based methods (Random Forest and Gradient Boosting Machine) and Neural networks-based approaches (Artificial Neural Networks and Deep Neural Networks). To ascertain the relative influence of factors on output estimations, the SHapley Additive exPlanations (SHAP) algorithm is integrated. This interpretable machine learning methodology demonstrates strong alignment with published experimental data. It holds promise as an efficient and versatile tool for designing new composite materials tailored to applications involving thermal management.

KW - Data-driven technique

KW - Interpretable Integrated Learning

KW - Polymeric graphene-enhanced composites (PGECs)

KW - Stochastic multi-scale modeling

KW - Thermal properties

UR - http://www.scopus.com/inward/record.url?scp=85175088621&partnerID=8YFLogxK

U2 - 10.1016/j.compstruct.2023.117601

DO - 10.1016/j.compstruct.2023.117601

M3 - Article

AN - SCOPUS:85175088621

VL - 327

JO - Composite structures

JF - Composite structures

SN - 0263-8223

M1 - 117601

ER -