Eddy Current Imaging Using Sensor Arrays for the Detection and Characterization of Random Defects

Abstract

This thesis addresses the advancement of Eddy Current Testing (ECT) as a reliable Non-Destructive Testing (NDT) technique for defect detection and characterization in conductive materials. Despite its advantages—such as high sensitivity, portability, and non-contact operation—ECT is limited by shallow penetration depth, geometric complexity, and measurement noise. To overcome these challenges, a comprehensive framework integrating analytical modeling, finite element simulations, experimental validation, and machine learning is developed. Multiplexed sensor arrays are employed to enhance defect imaging, while Radial Basis Function neural networks provide efficient prediction of defect dimensions. The models are validated experimentally on aluminum, stainless steel, and titanium specimens, confirming accuracy and applicability. This work contributes to improving the precision, reliability, and industrial relevance of ECT in structural health monitoring and quality assurance.