A smart magnetic sensing system for non-destructive material property characterization

Non-destructive testing (NDT) and evaluation methods offer fast and online alternatives to other traditional techniques. This thesis focuses on magnetic techniques such as Magnetic Barkhausen Noise (MBN) and Eddy Current (EC), which have been analyzed and applied for the construction of a novel magnetic sensor. The first part focuses on thickness estimation with a physics based analytical approach: The EC problem is solved analytically to create voltage vs. thickness plots. Namely, setup and material properties are provided to an optimized algorithm for the estimation of the pick-up voltage in the presence of a conductive sheet with varying thicknesses. The voltage signal of the pick-up winding is then measured and processed. The signal processing step involves the extraction of the dominant frequency by sine and cosine projection and reconstructing it by multiplying the amplitude with a fixed-phase sine wave resulting in the complete elimination of background noise and possible phase shifts. The signal is then averaged over 10 measurements and the amplitude is mapped to the analytical graphs enabling the estimation of thickness. The second part details the assessment of material properties by MBN analysis. Specifically, the Barkhausen effect is employed to detect superficial atmospheric rust in steel sheets. The MBN signals each composed of 4 bursts per window are sampled with an analog oscilloscope, and inserted to the signal conditioning program. A novel signal processing algorithm and CNN network with a low computational cost is proposed for accurate and fast rust classification. Experiments performed on DC04 cold-rolled uncoated steel sheets validate the results of the magnetic sensor. The tests demonstrate that the proposed sensor is a suitable alternative where continuous material property monitoring is needed. In addition to their various applications in industrial sectors, these types of sensors hold significant potential for future use in detecting material loss, assessing surface degradation, and evaluating stress in biomedical implants.