Digital Twin Deep Link: Physics Based Model with Scientific Machine Learning for Simultaneous Crack and Wave Propagation

Abstract

The integration of Digital Twin technology with Physics based models and Scientific Machine Learning (SciML) offers a transformative approach for Nondestructive Evolution (NDE) and Structural Health Monitoring (SHM). Detection and prediction of crack evolution under local stress through wave propagation demands physics-based understanding of crack-wave interaction. Traditional computational methods struggle to simultaneously capture crack growth dynamics and guided wave interactions due to the complexities of remeshing and high computational costs. The study demonstrates the simultaneous simulation of crack propagation and guided wave interactions without requiring mesh updates. Further, by leveraging physics-informed neural networks (PINNs) under SciML approaches, a Digital Twin framework can bridge this gap, providing real-time, data-driven insights into Structural Health Monitoring (SHM) and Nondestructive Evaluation (NDE). This study presents a Digital Twin Deep Link framework that coupled physics-based models with SciML to accurately predict crack initiation, growth patterns, and guided wave interactions in stiffened structures. Case studies illustrate how physics-informed AI enables the identification of crack signatures in sensor data, providing a robust mechanism for defect detection and material state assessment. The results highlight the potential of SciML-powered Digital Twins in SHM and NDE, paving the way for AI-driven diagnostics and autonomous monitoring systems.