# Artificial Neural Networks for Impact Strength Prediction of Composite Barriers

## TL;DR

This paper uses artificial neural networks to predict the impact strength of composite barriers, improving simulation efficiency and accuracy.

## Contribution

A novel approach using ANNs trained on FEM results to address computational challenges in impact simulations.

## Key findings

- ANNs effectively predict impact strength of composite barriers with reduced computational demand.
- The method overcomes numerical instabilities in high-stress contact zones during simulations.
- Results show the approach is applicable to complex impact problems with high strain rates.

## Abstract

This study considers the impact and penetration of composite targets by steel projectiles. Firstly, experiments on the impact of homogeneous polymethyl methacrylate (PMMA) targets were simulated using the finite element method (FEM) and the incubation time fracture criterion (ITFC). Next, targets were assumed to be composed of cells with weakened mechanical properties, forming a composite barrier. The composite impact problems were then used to demonstrate an approach, which can be applied to overcome the typical difficulties for impact simulations—high demands on computing resources, long computation times, and potential numerical instabilities arising from high stresses in the contact zone and high strain rates. The approach is based on the use of artificial neural networks (ANNs) trained on arrays of numerical results obtained via finite element method.