Machine learning guided process optimization and sustainable valorization of coconut biochar filled PLA biocomposites

Abstract

Abstract This study focuses on the valorization of coconut shell biochar (CCB) as a sustainable reinforcement in polylactic acid (PLA) biocomposites, targeting enhanced mechanical performance. PLA/CCB composites were fabricated by varying injection molding parameters at three levels: composition (Pure, 5 wt%, 10 wt%), injection temperature (135 °C, 145 °C, 155 °C), injection speed (50 mm/s, 60 mm/s, 70 mm/s), and injection pressure (30 bar, 40 bar, 50 bar). A Taguchi L27 orthogonal array was employed to systematically assess the effects of these parameters on tensile strength, Young’s modulus, and hardness. ANOVA results indicated that composition and injection temperature were the most influential factors, contributing 50.42% and 42.67% to tensile strength, and 38.58% and 20.14% to Young’s modulus, respectively. For hardness, composition dominated with a 78.3% contribution. To predict the mechanical responses, five machine learning models, including Linear Regression, Support Vector Regression (SVR), Random Forest Regression (RFR), Gradient Boosting, and Extreme Gradient Boosting (XGBoost), were implemented. Gradient Boosting and XGBoost exhibited superior predictive accuracy, with R2 values of 98.77% for tensile strength, 96.28% for Young’s modulus, and 96.45% for hardness. The integration of Taguchi design, ANOVA-based analysis, and advanced machine learning techniques offers a robust framework for optimizing process parameters and valorizing CCB as a high-performance, eco-friendly reinforcement in biodegradable biocomposites.