“Mechanical Characterization and Predictive Modeling of Abaca-Bagasse Polyester Sandwich Composites Using an Artificial Neural Network”

This study explores the mechanical properties of abaca-bagasse polyester sandwich composites. Abaca fiber, known for its exceptional tensile and flexural strength, is combined with bagasse fiber powder, a renewable byproduct of sugarcane, and polyester resin to create hybrid composites. The composite’s performance was evaluated through tensile, flexural, impact, and interlaminar shear strength tests, with microstructural analysis, performed using scanning electron microscopy (SEM). Furthermore, a Multi-Layer Perceptron (MLP) regression model, a type of artificial neural network (ANN), was employed to predict the ultimate tensile strength (UTS) and flexural strength of the composites. The MLP model, consisting of interconnected neurons across multiple layers, was trained on experimental data to achieve accurate predictions. The results showed that the composite with 30 wt.% abaca fiber, 15 wt.% bagasse powder, and 55 wt.% polyester resin (BA-1) exhibited outstanding mechanical performance. Specifically, the BA-1 sample achieved a tensile modulus increase of 36.94% and tensile strength improvement of 30.12%, along with superior flexural modulus (53.11%) and flexural strength (50.80%). It also demonstrated higher energy absorption (44.28%) and impact resistance (20.85%) compared to other composite formulations. The ANN model successfully predicted these strengths, achieving R² values of 0.96 for tensile strength and 0.90 for flexural strength.