Solution of Contact Problems by Deep Learning
摘要
Abstract
A method for solving contact problems that involve deformable bodies is proposed on the basis of physics-informed neural networks (PINNs) and machine learning. A special type of loss function is formulated for the neural network, taking account of the inequality in the mathematical formulation of the contact conditions. An algorithm for solving contact problems involving beams and absolutely solid bodies is described, on the basis of physics-informed deep learning.