Qualification of the inverse problem of defect reconstruction using optimized mesh database

Purpose - To propose a novel method for defect reconstruction in electromagnetic non-destructive testing (NDT). Design/methodology/approach - The inversion method is based on an optimized database that contains the measured signals for some predefined defect prototypes. The database is supported by an anisotropic: simplex mesh, which has been generated adaptively in the abstract n-dimensional space, spanned by the model parameters of the defect type. The actual reconstruction reduces to a mesh search and interpolation. The described theory is demonstrated in the paper by a solved NDT test problem. Findings - We have realized that in addition to sole defect reconstruction, the database provides meta-information about the quality of the inversion, the suitability of the chosen defect model parameters, as well as the capabilities of the testing experiment. Research limitations/implications - Defect models having several parameters require a sophisticated mesh generation algorithm, which works in higher dimensions. Originality/value - In the authors' opinion the mesh database approach offers a totally new point of view of a given inverse problem, and may help in the better understanding of its nature.