Physics-Constrained Neural Network for design and feature-based optimization of weave architectures

Abstract

Woven fabrics play an essential role in everyday textiles for clothing/sportswear, water filtration, retaining walls, and reinforcements in stiff composites for lightweight structures in aerospace, sporting, automotive, and marine industries. Several possible weave architectures (combinations of weave patterns and material choices) present a challenging question about how they could influence the physical and mechanical properties of woven fabrics and reinforced structures. This paper presents a novel Physics-Constrained Neural Network (PCNN) to predict the mechanical properties (like modulus) of weave architectures and the inverse problem of predicting pattern/material sequence for a design/target modulus value. The inverse problem is particularly challenging as it usually requires many iterations to find the appropriate architecture using traditional optimization approaches. We show that the proposed PCNN can more accurately predict weave architecture for the desired modulus than several baseline models considered. We present a feature-based optimization strategy to improve predictions using features in the Gray Level Co-occurrence Matrix space. We combine PCNN with feature-based optimization to discover near-optimal weave architectures and facilitate the initial design of weave architecture. The proposed frameworks will primarily enable the woven composite analysis and optimization process and be a starting point to introduce knowledge-guided neural networks into the complex structural analysis.