Supported by the Croatian Science Foundation under the grant
IP-2024-05-4331

Routine and planned non-destructive testing and evaluation of the integrity, characteristics, and quality of materials, components, systems, or structures without damaging or diminishing their usefulness is essential in ensuring safety, reliability, and quality across industries. Eddy current testing is among the most important methods of non-destructive testing. The challenges in shifting the paradigm from routine and planned testing to continuous eddy current testing of critical structures lie in the energy consumption of the sensor system and the need for human involvement in data interpretation.

Both limitations can be addressed by implementing low -power embedded sensor systems, reducing the amount of data transmitted, and focusing on machine processing of data near the sensor. This project contributes to the research of an autonomous network of intelligent sensors for the continuous monitoring of safety-critical industrial components or structures using the eddy current method. The overall objective of the project is to explore methods for the realistic simulation of eddy current sensors and techniques for integrating a sensor system simulator with machine learning algorithms in a low-power embedded system. The main outcome of the project is a demonstrator of an embedded sensor system for autonomous eddy current testing.

During the project, the research team will develop an expanded analytical model and implement a simulator of the eddy current sensor and sensor interface, design a measurement method based on the interaction of the eddy current sensor and the object during the sensor’s approach to the object, develop machine learning algorithms for the analysis of the eddy current sensor data, and investigate the implementation of these algorithms in a low -power embedded sensor system.