Extracting fiber paths from the optimized lamination parameters of variable-stiffness laminated shells based on physic-informed neural network

Xinming Li, Lujie Ma, Bowen Ji, Kuan Fan, Zhengdong Huang

Abstract: Highlights•The physics-informed neural network is utilized to extract fiber paths from the lamination parameters of laminated shells.•The streamlines in parameter space are translated into the corresponding fiber paths in physical space.•The Riemannian geometry is used to derive the formula for geodesic curvature and the intrinsic distance between adjacent fiber paths.•A fiber path extraction algorithm is employed to minimize the loss function at randomly sampled points through gradient descent.