Work

Strain-rate-dependent optimization of alkali treatment in sugarcane bagasse fibers using ensemble learning and genetic algorithms

Year: 2025

Type: article

Abstract: Natural fibers, particularly sugarcane bagasse, have gained increasing attention as sustainable reinforcements in polymer composites due to their environmental and economic benefits. However, the mechanical properties of bagasse fiber-reinforced composites is highly sensitive to alkali treatment parameters (NaOH concentration and soaking time) and tensile testing conditions (strain rate). This study investigates the combined effects of these parameters on the bagasse fiber tensile strength using a machine learning-based optimization framework. A full factorial experimental design involving 48 experimental runs was conducted, varying NaOH concentrations (1%, 3%, 5%, 10%), soaking durations (2, 4, 6, and 10 hours), and strain rates (3.3 × 10⁻⁴ s⁻¹, 1.0 × 10⁻³ s⁻¹, and 1.67 × 10⁻³ s⁻¹). FTIR spectroscopy confirmed the effective removal of hemicellulose and lignin, enhancing cellulose intensity. A blending ensemble model combining Random Forest, XGBoost, and regression models achieved high predictive accuracy (R² = 86.3%, RMSE = 25.7). Feature importance analysis revealed that strain rate (62.37%) had the greatest influence, followed by NaOH concentration (24.72%) and soaking time (12.90%). Genetic algorithm based optimization identified optimal treatment conditions at each strain rate. The proposed data-driven framework enables the efficient optimization of alkali treatment parameters of bagasse fibers, contributing to the development of high-performance, eco-friendly fiber-reinforced composites.

Source: Results in Engineering

DOI: 10.1016/j.rineng.2025.107394