Detection of hidden damages in Fibre Metal Laminates (FML) is a challenge. Damage detection, classification, and localisation is part of the lower levels of Structural Health Monitoring (SHM). SHM is an extremely useful tool for ensuring integrity and safety, detecting the evolution of damage, and estimating performance deterioration of civil infrastructures. Early damage detection can avoid situations which can be catastrophic. SHM can allow efficient maintenance works and can avoid unnecessary inspections, furthermore, saving time and money.

The paper focuses on designing a suitable damage feature marking algorithm for the early detection and identification of internal defects occurring in FML using advanced Machine Learning methods and state-based models for data series prediction. The aim of this study is to develop a SHM system for composite structures through the use of ML algorithms and optimization theory. The paper presents an early method of accessing the sections of the FML for identifying internal damages using X ray imaging by combining automatic adaptive feature extraction and machine learning methods. A comprehensive image-based data set is collected by means of X ray and CT images containing micro-scale damage mechanisms (fibre breakage, metal cracks etc.) The original scan image data is then transformed into a feature image using a feature extraction method based on an anomaly detector. The anomaly detector is created from unsupervised ML using an Autoencoder architecture with state-based memory neuronal networks. Spatially resolved features then marks material anomalies that can be any kind of damage or material change.

The generated knowledge and the image data collected would further accelerate the development in the field of autonomous SHM of the composite structures which would further reduce the safety risks and total time associated with structural integrity assessment.