Predicting the microstructure of composite plates using deep learning based on thermal expansion

• Published in: AIP advances Vol. 14; no. 10; pp. 105036 - 105036-5
• Main Authors: Zhang, Faling, Wang, Ziping
• Format: Journal Article
• Language: English
• Published: Melville American Institute of Physics 01-10-2024; AIP Publishing LLC
• Subjects: Composite structures; Datasets; Deep learning; Design optimization; Finite element method; Microstructure; Multilayer perceptrons; Predictions; Software; Thermal expansion
• ISSN: 2158-3226; 2158-3226
• DOI: 10.1063/5.0231089

In this paper, we introduce a novel method to predict composite plate microstructures based on thermal expansion using deep learning. Traditional analysis methods are cumbersome and resource intensive. We created a detailed dataset from finite element simulations, capturing edge displacements during thermal expansion. This dataset feeds into our multilayer perceptron model, generating precise microstructure matrices. Our approach achieves high accuracy and efficiency, markedly reducing computational overhead. By combining deep learning with finite element analysis, we streamline predictions and enhance precision. This integrated approach serves as a potent tool for engineers and materials scientists, facilitating composite structure design and optimization.