The non�]destructive inspection is a key�]problem in many industrial processes since the detection of defects or cracks in the products is mandatory. Many tomographic approaches that have been proposed are based on the use of interrogating microwaves and their effectiveness in inspecting dielectric or conductive materials has been demonstrated. In general, a NDT/NDE problem solved through an inverse scattering technique is still ill�]conditioned and non�]linear. However, unlike standard microwave imaging problems where a complete image of the unknown investigation domain is required, a lot of a�]priori information on the scenario under test (the industrial product) is available. In this framework, Caorsi et al. proposed an optimization technique (FGA) [1], where a reduction of the number of problem unknowns was achieved by exploiting such an a�]priori information about the inspected geometry. As a matter of fact, a non�]destructive inspection is aimed at detecting an unknown defect inside a known unperturbed host medium lying on a known background. Starting from this assumption, the a�]priori information on the unperturbed structure was exploited by means of a suitable Genetic Algorithm (GA). As a consequence, the inverse scattering problem in hand was significantly simplified since the unknowns was reduced to an array of geometric parameters of the defect: the position of the defect, its size and orientation, and its electromagnetic properties.
展开▼
