An Artificial Neural Network and Taguchi Prediction on Wear Characteristics of Kenaf–Kevlar Fabric Reinforced Hybrid Polyester Composites

Abstract

Fiber-reinforced composites have found their prominent place in various applications, including aerospace, automobile and marine manufacturing industries, because of outstanding properties obtained during composite preparation. One such aspect of improving the composite property is hybridization, where natural fibers (or both natural and synthetic fibers) are combined to obtain different composite structures for diverse applications. This research aims to hybridize the composite considering Kenaf fabric and Kevlar fabric reinforced in an unsaturated polyester matrix with different proportions. Three different laminate sequences (L1, L2, and L3) were developed by considering the fabric's stacking sequence, weave pattern, and orientation. The composite laminates prepared were tested where Taguchi's method (L9 orthogonal array) and artificial neural network were used to study influencing parameters for tribological behavior of the composite. From the practical information, a prediction model from the artificial neural network is applied to forecast the wear rate of the laminates at a broader range of operating factors beyond and within the test phase. The microstructures of the worn surfaces were investigated from a scanning electron microscope to confirm the wear principle of the laminates under different cases.