Data-Driven Inverse Heterogeneity Mapping for Multi-Scale Fatigue Life Prediction in Layered Composites

**Abstract:** This paper introduces a novel data-driven framework for inverse heterogeneity mapping and fatigue life prediction in layered composite materials. Traditional methods often rely on computationally expensive finite element (FE) models with pre-defined material properties, limiting their applicability to complex, real-world structures. Our approach leverages a hybrid methodology combining Gaussian Process Regression (GPR) for inverse parameter estimation and a Reduced Order Model (ROM) based on Proper Orthogonal Decomposition (POD) for accelerated fatigue life prediction, enabling rapid characterization of microstructural heterogeneity and accurate life estimations for various loading conditions. This research accelerates the material design process and facilitates improved durability assessments for critical engineering applications.