Physics-Informed Machine Learning for Structural Health Monitoring of Aerospace Composite Structures

Abstract

This paper presents advanced structural health monitoring (SHM) methods for aerospace composite structures, which pose unique challenges due to sensor placement, cost, and environmental exposure. We introduce novel, physics-disciplined, data-driven approaches developed through two European Union projects. The first technique embeds glass-coated copper microwires in carbon-fibre composites, exploiting their Giant Magnetoimpedance (GMI) response to detect, classify, and quantify damage under stress. The second approach supports damage monitoring in composite liquid hydrogen tanks using Fibre Bragg Grating (FBG) sensors, addressing the challenges of conformal geometry and strict leakage tolerance where conventional diagnostics are inadequate. Physics-informed machine learning algorithms are developed for both systems. Finite element simulations inform neural network architecture and feature selection, while simulated signals guide strain and damage modelling. Experimental and simulation-based validation confirms high accuracy in damage detection and characterisation. This work was funded by the European Union under the Horizon Europe grant 101056884 and by the EU Clean Hydrogen Partnership under Grant Agreement 101101404. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or Clean Hydrogen Joint Undertaking. Neither the European Union nor the granting authority can be held responsible for them.