A Study on Behavior of Functionally Graded Microbeam Under Static Deflection Using Physics-Informed Neural Network

This article presents a physics-informed neural network (PINN) approach to model functionally graded (FG) microbeams. The governing equation to predict the behavior of an FG cantilever microbeam under static deflection is obtained using reformulated strain gradient theory. A deep neural network (DNN) is utilized to approximate the deflection in the beam. A backpropagation algorithm dealing with the gradient loss is used to train the network, and then a Quasi-Newton optimizer is used to achieve the minimization. The trained network can obtain solutions faster than the classical approach, like the finite element method. Finally, we compared our results for the FG microbeam with available FEM solutions.