Machine learning-based estimation and optimization of phoenix Dactylifera Seed Powder reinforced vinyl ester bio-composites

This study aims to develop a robust machine learning (ML) framework for accurately predicting the mechanical and thermal properties of sustainable vinyl ester composites reinforced with Phoenix Dactylifera Seed Powder (PDSP). To achieve this, multiple ML algorithms such as Linear Regression, Support Vector Machine (SVM), Random Forest and Decision Tree were evaluated to predict the mechanical and thermal using both experimental data generated in-house and data collected from past literature review, Among which the SVM model demonstrated superior predictive accuracy, achieving high R² values for tensile strength (0.91), flexural strength (0.83), hardness (0.86), and heat deflection temperature (0.85). The analysis confirms that filler weight percentage is the most influential parameter and that ML can reliably forecast long-term property retention under environmental exposure. Consequently, this data-driven approach significantly reduces reliance on extensive experimental testing and enables optimized design of PDSP/VE composites for practical applications in automotive components and civil engineering structures, thereby facilitating the integration of sustainable materials into performance-based standards.