Harnessing Machine Learning for Wear Prediction in Sustainable Fibre-Reinforced Hybrid Composites

The escalating demand for sustainable materials in engineering applications has led to the emergence of natural fibre-reinforced polymer composites (NFRPCs).This study introduces a hybrid composite integrating glass, raffia, and acacia fibres with an epoxy matrix, fabricated using the hand lay-up technique.The composite's design emphasises fibre arrangement, with glass fibre as outer layers, to enhance surface finish and structural integrity.Thermal performance evaluated through the thermogravimetric analysis (TGA) confirmed stability across diverse temperature ranges.Dry sliding wear tests assessed the composite's tribological properties, highlighting the influence of load, sliding velocity, and distance.A novel machine-learning framework, utilizing deep learning neural networks (DNNs), was employed to predict wear rates based on these parameters.Developed in MATLAB R2023a, the DNN model showed a high correlation between the predicted and the experimental data, substantiating its accuracy and reliability.This study underscores the hybrid composite's potential in automotive applications, such as brake systems, by demonstrating superior thermal stability, wear resistance, and predictive modelling capabilities, paving the way for optimised material design.