DelamPredict-X: An Ensemble Learning-Based Approach for Evaluating Delamination in Jute Fiber Reinforced Materials

Jute fiber reinforced polymer composites experience delamination which represents their most important failure mode because it deteriorates structural capabilities and service capabilities. The study presents DelamPredict-X as an ensemble learning-based predictive framework which provides delamination severity evaluation and classification across different material conditions and loading scenarios. The meta-learner consists of both a Gradient Boosting Regressor and soft voting classifier which integrates Random Forest, XGBoost and LightGBM as base learners. The model receives experimental data from Double Cantilever Beam (DCB) and End-Notched Flexure (ENF) testing which uses energy release rate and resin type and fiber volume fraction parameters. A combination of SMOTE and Gaussian noise injection functions as the data augmentation strategy to increase generalization capabilities. Both regression R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score evaluation at 96.92% and classification accuracy at 96.23% from the model show stronger performance than conventional models separately in these tasks. Information provided by SHAP interpretability identifies which features have the greatest influence on delamination development. The delamination prediction system DelamPredict-X maintains outstanding predictive accuracy together with superior robustness while delivering explainable results which leads to practical delamination assessment for sustainable composites. This predictive model shows high suitability for industrial purposes that include manufacturing operations along with structural health assessments and composite material optimization.