Optimization of Tensile Strength in Nano Silica and Calotropis Gigantea Reinforced Hybrid Polymer Nanocomposites using ANN-BBD Modelling

Abstract

Machine learning, artificial neural networks (ANNs) and artificial intelligence (AI) have all found applications in the engineering field. In this study, a reinforced hybrid polymer nanocomposite (NC) was created utilizing nano silica (nSiO2) and natural fibers (NF) derived from Calotropis gigantea fiber (CGF). The nanocomposite is designed for advanced composite applications. To promote interaction and adhesion between the CGF/polymer matrix, the fibers were modified by applying a chemical solution comprising KMnO4 in C3H6O. This altered the surface of the fibers. A convolutional neural network (CNN) model was used to optimize and predict the tensile strength (TS) of CGF/nSiO2 hybrid nanocomposite (CGF/nSiO2nc). With a hidden layer of five neurons and a one-layer perceptron structure of 3-5-1, this model, which was built using the Box-Behnken Design (BBD), showcased very different optimization. Evidence of the effect of the KMnO4 alteration on the hybridized nanocomposite's TS may be seen in scanning electron micrographs. As demonstrated by ANOVA with R2 = 0.9985, the mechanical test output demonstrate that these variables had an effect on the CGF/nSiO2 TS. The experimental results were quite close to matching the predicted outcomes. The ideal TS, according to the model, is 53.2025 MPa. The validity of the empirical experimental study was confirmed by doing TS analysis at the optimal parameters that were predicted. The average strength was found to be 40.28 MPa in the TS data. The model captures around 98.68% of the predicted TS. Finding the right values for mechanical properties fast, while cutting production costs and conserving resources, is possible with the help of the ANN-BBD modelling approach.