Predicting the Loading Parameters of a Square Panel Upon Linear Deflection

This paper presents a machine learning methodology for predicting the static deflection of an isotropic squared panel in the linear-elastic regime. Our method uses a series of simulations via FEM (Finite Element Method) for obtaining the contour plot images of flexural deflections due to point loads at different locations on the panel. From here, we obtain, using sliding windows of different sizes, segments of images that are transformed into features using a state-of-the-art pre-trained deep neural network. We trained three different machine learning models for multi-output regression. Our results show that it is possible to predict with a high score the coordinates of the point of load and the amount of deflection with partial visual information.