Impact Characterization on Reinforced Aerospace Structures via Machine Learning

In the present manuscript Machine Learning algorithms are compared in order to implement a model able to predict the coordinates of impacts on panels used in aerospace field. Experimental activities are conducted to build a proper impacts’ dataset. Polynomial regression and artificial neural network are applied and optimised to panels without stringer to test their capability to identify the location of the impacts. Subsequently, the algorithms are applied to reinforced panels, with much more complexity of dynamic features of the system to test: the goal is not only on the impact position’s detection but also on the event’s severity. The work demonstrates the validity and the computing efficiency of the Machine Learning application to the impact localisation on realistic structures, regardless of specimen complexity.