Data-driven optimization and multiscale characterization of sustainable epoxy-benzoxazine hybrid composites for high-performance structural applications

The polymer composites with enhanced mechanical, thermal, and tribological properties are increasingly important for advanced structural applications. In this study, hybrid composites based on an epoxy-benzoxazine matrix reinforced with basalt fibers (BF), waste cotton textile fibers (WCTF), and graphene nanoplatelets (GNPs) were developed. Optimization of BF content, WCTF content, GNP loading, and curing temperature for tensile strength, flexural strength, impact strength, and Shore D was performed using a Taguchi L9 experimental design. The optimized composite (30% BF, 10% WCTF, 1% GNPs, and a curing temperature of 160 °C) achieved a tensile strength of 138 MPa, flexural strength of 208 MPa, impact strength of 42 kJ m⁻², and Shore D hardness of 87. In addition, the optimized composite exhibited a storage modulus of 5.2 GPa and a glass transition temperature (Tg) of 132 °C. FTIR, AFM, FESEM with EDS, DMA, DSC, TGA, and wear analyses supported enhanced matrix-reinforcement interactions, improved interfacial adhesion, and favorable thermomechanical and tribological performance. Random Forest and XGBoost machine learning models were created for predicting the composite properties, and XGBoost yielded the highest prediction accuracy (R² = 0.9870–0.993). The proposed experimental, statistical, and machine learning strategy is effective for the design of high-performance epoxy–benzoxazine hybrid composites for applications in structural engineering.