Artificial Intelligence prediction and optimization of the mechanical strength of modified Natural Fibre/MWCNT polymer nanocomposite

Abstract

Artificial Intelligence (AI), techniques like artificial neural networks (ANN), and machine learning have been used to solve a variety of engineering problems. In this study, Multiwall Carbon Nanotube (MWCNT) and Natural Fibres (NF) from plantain (Musa Paradisiaca) fiber (PF), were utilized to prepare a reinforced hybrid polymer nanocomposite for advanced composite applications. A chemical solution containing potassium permanganate (KMnO₄) in acetone (C₃H₆O) was applied to modify the fibers to alter their surface and improve adhesion and interaction between the PF/polymer matrix. To predict and optimize the tensile strength (TS) of the prepared PF/MWCNT hybrid nanocomposite (PFMNC), the ANN model with hyper-parameter optimization in a single-layer-perceptron architecture of 3-5-1 was used with 5 neurons in the hidden layer, and Box-Behnken Design (BBD) was utilized. Scanning electron microscope (SEM) micrographs demonstrate that KMnO₄ modification has impacted the TS of the hybridized nanocomposite. Mechanical Test results show that these variables impacted the TS of the PFMNC as shown by analysis of variance (ANOVA) with R² = 0.9986. The expected findings were nearly identical to the experimental results. The model predicted an optimal tensile strength of 46.1563 Mpa. To substantiate the reliability of the empirical experimental investigation, TS analysis was performed at predicted optimal settings. TS results showed an average strength of 45.4401 Mpa. About 98.45 % of the projected tensile strength is accounted for by the model. The present study has demonstrated the effectiveness of the ANN-BBD modeling technique in achieving the appropriate mechanical property values quickly, reducing production costs, and preserving resources.