Comparative machine learning analysis of wear characteristics in industrial waste-filled epoxy-jute composites.

This study focuses on the fabrication and analysis of hybrid epoxy based composites using jute fiber (JF) and Linz-Donawitz (LD) sludge as reinforcement materials. The composites were fabricated through a hand-lay-up technique, with LD sludge concentrations varying at 0, 5, 10, 15, 20, and 25 weight percentages and a fixed concentration of JF (20 wt%). The density and microhardness of the composites were evaluated in accordance with ASTM standards. An experimental design approach and analysis of variance (ANOVA) were employed to examine the effects of control parameters on the sliding wear rate of the composites. Further, this study evaluated the performance of machine learning models for predicting composite properties that are enhanced by the addition of reinforcements. The findings showed that the S4 specimen (20 wt% of LD sludge and 20 wt% JF) exhibited enhanced microhardness and improved wear resistance relative to the other specimens. The density of the S4 specimen improved by approximately 18.77% compared to neat epoxy. The specific wear rate decreased, as it dropped to 0.526 mm3/N-m at 20 wt% LD sludge, which is a 60% improvement in wear resistance. The machine learning models were highly predictive, and Gradient Boosting and XGBoost yielded R2 values of 0.9999 and low error rates. The study concluded that the practical outcomes were in close alignment with the optimal predicted results.