Tensile Strength Prediction of Fiberglass Polymer Composites Using Artificial Neural Network Model

Abstract

Highlighting the properties of polymer composites is a complex process given their great diversity and the wide range in which their characteristics could vary. An Artificial Neural Network model for predicting tensile strength was designed using LabVIEW software. The proposed model was developed for randomly reinforced polymeric composite materials with 30%, 40% and 50% fiber-glass. Volume fraction of glass fibre has represented the independent variable for this study. The dependence of the tensile strength on the volume fraction was investigated and highlighted by modelling using neural networks. The designed Artificial Neural Network behaves as a computational system that process data input into a desired output using a network of functions composed of layers. The training process was developed with different Artificial Neural Network architectures with two hidden layers to produce the best prediction results. For each hidden layer the number of neurons was varied be-tween 3 to 50.