Hybrid experimental and machine learning approach for optimizing abrasive wear of microcrystalline cellulose modified hemp/bamboo fiber composites

In this work, hemp/bamboo hybrid fabric-epoxy composites reinforced with 0-9 wt% microcrystalline cellulose (µCC) is examined for their abrasive wear behavior. Compression molding was used to create composites with 0, 3, 6, and 9 wt% µCC. In accordance with ASTM G65 guidelines, wear tests were conducted under controlled dry sand abrasion. Using a Taguchi L₁₆ design, the effects of applied load (5-20 N), abrading distance (250-1000 m), and µCC content on wear loss were assessed. To predict abrasive wear and examine the role of µCC filler, several machine learning models were used, including Linear Regression, K-Nearest Neighbors, Artificial Neural Networks, Random Forest, Gradient Boosting, and eXtreme Gradient Boosting. By increasing the hardness and load-bearing capacity of the composite, µCC mechanistically increases wear resistance and lessens material removal during abrasion. According to ANOVA results, wear loss was most affected by abrading distance (44.08%), load (34.21%), and µCC content (18.01%). The Random Forest model had the lowest error (RMSE = 0.045) and the highest predictive accuracy (R² = 0.942). Abrading distance is the main factor influencing wear resistance, followed by load and µCC content, according to feature importance analysis. Accurately forecasting abrasive wear and creating high-performance, sustainable hybrid composites can be accomplished by combining machine learning and experimental data.