Updating the finite element model for electrical impedance tomography using self-organizing map

Electrical impedance tomography (EIT) is recently demonstrated to be viable for damage localization over a spatial area. The algorithm reconstructs the spatial conductivity distribution within a defined boundary via boundary voltage measurements. To solve this inverse problem, a finite element model (FEM) conforming to the interrogated geometry is required. Previous studies on identifying a center crack’s propagation suggests that an FEM-updating strategy may help identify both the existence of a crack and the plastic zones formed around the crack’s tips. In this paper a data-driven algorithm is applied to automatically update the FEM. The self-organizing map algorithm is adopted to categorize the reconstructed conductivity data, tracing the boundary of the crack to be updated as material-absence. The EIT results from the updated FEM model are able to identify damage location and damage severity with desired accuracy