Correlating Interfacial Shear Strength and Material Properties in Fiber-Reinforced Composites

Abstract

This study explores the precise forecasting of interfacial shear strength in fiber-reinforced composites by employing the CatBoost machine learning model. The research utilizes partial dependence plots and accumulated local effect techniques to improve interpretability, emphasizing the significance of the impact of each variable on the output. It demonstrates how interpretable methods can offer valuable understanding into the decision-making procedure of machine learning models, identifying influential variables for specific instances and contributing to a comprehensive understanding of overall predictions. The findings of study contribute to the field by enhancing the prediction of interfacial properties in fiber-reinforced composites, underscoring the value of interpretable machine learning methods in unraveling insights from complex predictive models. Overall, this research underscores the importance of transparency and interpretability in model-driven predictions, particularly in critical engineering applications.