Quantitative reconstruction of plate defects from eddy-current scans via position-dependent trajectory sensitivity of a coil array

Reconstruction of defect distributions from eddy-current (EC) measurements is essential for structural health assessment of planar structures. The state-of-the-art approach uses a coil array to scan defect areas, providing increased resolution. However, the images obtained are based on data acquired when the sensor is directly above the defect region, resulting in a largely qualitative reconstruction. The information gathered during the scanning process, arising from the position-dependent interaction between the electromagnetic fields and the defects, is not fully exploited. Drawing on classical sensitivity theory, this work explicitly derives a position-dependent sensitivity for a moving coil array, facilitating quantitative defect conductivity reconstruction. The sensitivities are assembled into a unified linear model spanning all scan positions, coil pairs, and excitation frequencies. The inverse problem is solved accordingly using representative methods of tomographic image reconstruction. Numerical simulations and experiments are conducted to evaluate the method. In numerical simulation, the influence of the number of scan points, coil pairs, and excitation frequencies on reconstruction accuracy is quantified. In experiments, the surface and back-side cracks at various depths on stainless-steel and aluminium plates are scanned and reconstructed. The reconstructed conductivity variation agrees with the measured crack geometry. The results demonstrate that position-dependent trajectory sensitivities enable accurate and quantitative defect reconstruction on plates.