Abstract
Abstract Fiber-reinforced polymer composites are extensively used in aerospace, automotive, and structural applications due to their excellent strength-to-weight ratio and design versatility. However, experimental tensile strength testing is both time-consuming and resource-intensive, hindering rapid material development. A significant challenge in advancing fiber-reinforced composites is the scarcity of experimental tensile strength data for newly formulated designs. To address this gap, this study introduces a comprehensive machine learning (ML) framework to predict the tensile strength based on compositional and processing parameters. The framework utilizes publicly accessible datasets containing parameters such as fiber types (Glass, Carbon, Aramid, Basalt), resin systems (Epoxy, Polyester, Vinyl Ester, Phenolic), density, layer count, curing temperature, fiber volume fraction, and void content. Five regression algorithms, including Linear Regression, Support Vector Regression (SVR), Random Forest, XGBoost, LightGBM, Artificial Neural Networks (ANN), and stacking ensemble models, were trained and evaluated using five-fold cross-validation and standard performance metrics such as R², RMSE, and MAE. The ANN and stacking ensemble achieved the highest predictive performance, with an R² of 0.9975 and an RMSE of approximately 31 MPa, demonstrating excellent cross-validation stability. Overall, the proposed framework offers a robust, reproducible, and data-driven method for accurate strength prediction, reducing the experimental workload while providing actionable insights for optimizing composite design.