Machine learning based prediction on mechanical and wear characterization of sisal fiber reinforced dolomite dust-epoxy hybrid composites

Natural fibers and industrial waste fillers are increasingly utilized in manufacturing, construction, and biomedical fields due to their sustainability and performance benefits. This study focuses on developing a composite material reinforced with dolomite dust (0, 5, 10, and 15 wt.%) and sisal fiber (2 wt.%) using the hand lay-up process. The fabricated composites were analyzed for their physical, mechanical, and sliding wear properties. Results indicated that higher dolomite content led to an increase in density, hardness, tensile strength, and flexural strength. Sliding wear behavior, evaluated using Taguchi’s L 16 orthogonal design, revealed that filler content and sliding speed were the key factors influencing the specific wear rate (SWR). To predict the composite properties, five machine learning models—Linear Regression, Decision Tree, Random Forest, Support Vector Regression (SVR), and Artificial Neural Network (ANN) - were employed. The ANN model exhibited the highest accuracy, achieving R 2 values near 1.0000 and very low mean absolute errors, demonstrating its reliability in predicting material properties. Furthermore, microstructural studies using FE-SEM and XRD confirmed a uniform dispersion of dolomite particles in the epoxy-sisal matrix, which contributed to enhanced mechanical strength and improved wear resistance.