Experimental and ANN approach to predict the mechanical performance of hybrid natural fiber composites for sustainable lightweight engineering applications

In this research, banana fiber (BaF), Khas Khas grass (Kkg) and SiC filler in epoxy (E1–E4) and polyester (P1–P4) are developed as eco-friendly hybrid composites. The mechanical, thermal, and water absorption properties were measured. E2 composite had the greatest tensile (87 Mpa) and flexural strength (116 Mpa) and impact resistance and hardness were improved by 15–20 wt.% SiC. TGA, FTIR, and SEM ensured that there was an increase in thermal stability, interfacial bonding, and fracture behavior. ANOVA revealed that SiC is the main factor affecting tensile strength (46.5%), Kkg fiber (28.4%), and the type of matrix (16.0%). ANN model which is a preliminary exploratory tool was found to be very accurate in tensile prediction ( R 2  = 0.950 MAE = 1.832% Mpa) and partially accurate in other properties. SHAP analysis showed that matrix type had an influence of about 90%. These findings demonstrate the strength of incorporating experimental, statistical and data-based models toward sustainable composite design.